CENTERING STUDENTS’ PERSPECTIVES IN COMPUTATION-INTEGRATED PHYSICS
CURRICULA
By
Patti Cotter Hamerski
A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Physics – Doctor of Philosophy
2021
ABSTRACT
CENTERING STUDENTS’ PERSPECTIVES IN COMPUTATION-INTEGRATED PHYSICS
CURRICULA
By
Patti Cotter Hamerski
Physics education researchers and curriculum developers have recognized the experiential expertise
of students, using students’ perspectives to make improvements to curriculum and pedagogy.
Recently, they have given students more control in this process, sometimes even a direct voice in
curricular decision-making. This dissertation intends to introduce and apply these student-centered
research methods to a new type of curriculum: computation-integrated physics. Even though
computational modeling is being integrated into physics curricula as a learning tool, there is no
consensus on assessment, curriculum, or learning goals. This gap provides an opportunity to build
an understanding of what matters to students in this new context from students’ perspectives and
make recommendations for curriculum and pedagogy.
Using a qualitative case study methodology, I present an in-depth view of how students perceive
their experiences in these computation-integrated classrooms. In total, the dissertation spans four
studies in two research contexts. The first study illustrates how case study can be used to center
a student’s perspective on her experience as an undergraduate learning assistant in a computationintegrated
physics course. Building on the first study, the second study is a more in-depth case
study on the cohort of learning assistants in the course, in effect demonstrating how students’
perspectives can be translated into pedagogical expertise when examined with an attention to
context and a grounding in a theoretical perspective. For the last two studies, I shifted the research
context to a computation-integrated high school physics class. The third study is an exploration of
students’ accounts of the challenges they face when doing computational activities in their physics
class, including those related to computation, the integration of computation with physics, and
the contextual factors in the classroom. Using the students’ perspectives once again, the fourth
study uses a theoretical framework to characterize students’ tendencies to engage productively with
computation. This final study demonstrates that examining students’ perspectives with a theoretical
basis and contextual attentiveness can provide a platform to step into student-centered curricular
change in computation-integrated physics.
Overall, these research studies come together in this dissertation to show that paying attention to
students’ perspectives and affect in computation-integrated physics courses is key to understanding
how to support students when teaching a computation-integrated curriculum. The findings also
bring researchers and curriculum developers a few steps closer to infusing students’ perspectives
directly into curricular and pedagogical decisions in computation-integrated educational settings.
I dedicate this dissertation to Otto, Circe, Beck, Blaine, Joyce, and Ed.
iv
ACKNOWLEDGMENTS
I want to thank my dissertation committee members for guiding me through this project and
believing in me. I am especially grateful for my advisors—Paul and Daryl. They are the people
I want to be around in the worst of times and the greatest of celebrations. They gave me comfort
when I became demoralized, and they celebrated my wins like their own. I am privileged to do
PER with them.
I also want to thank my close friends and colleagues from my first year at MSU: Paul, Daryl,
Abhilash, and Alanna. Some of my best memories are from hanging out in East Lansing having
a drink or working late with the record player spinning in PERL. These memories gave me life
during some of the more toilsome periods of graduate school.
I want to thank the larger PERL research group for its support, especially my fellow graduate
students who understood the grind best of all and whom I have sorely missed running into lately:
Brean, Laura, Nick, Alyssa, Bryan, Cami, and Carissa. I also thank Kim and Jenn from the physics
department and Jeno and Dustin from the Bailey Scholars Program. It’s nice to know there will
always be places on campus where people care for me and have my back.
I am grateful for the ICSAM community, past and present. I thank Dan, Theo, Jacqueline, Julie,
Sunghwan, Nickolaus, and the brave teachers who showed up and engaged earnestly with us in our
workshops. I especially thank the ICSAM teachers’ students, who shared their experiences with me
and genuinely reflected on the multitude of questions I asked them—without them this dissertation
would not exist.
I want to thank my parents for loving me and rooting for me more than I could have imagined—
thanks Mom for all the care packages. I want to thank my friends in Lansing, without whom I could
not have survived mentally or emotionally over the last year: Tom-Erik, Adam, Adam, Brynn, and
Brean. Lastly, I thank Mariana—you calmed my frequent panicking as I wrote this, and you have
been my rock and ever-present source of love over the last several months. It has meant the world
to me.
v
TABLE OF CONTENTS
LIST OF TABLES • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • ix
LIST OF FIGURES • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • x
CHAPTER 1 INTRODUCTION • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 1
CHAPTER 2 LITERATURE REVIEW • • • • • • • • • • • • • • • • • • • • • • • • • • 5
2.1 Incorporating students’ perspectives into curriculum • • • • • • • • • • • • • • • • 5
2.2 A context where students’ perspectives are needed: computation-integrated physics 7
2.3 Laying the foundation for incorporating students’ perspectives in computationintegrated
physics • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 10
2.4 Building on my previous work to highlight and address a curricular need in
computation-integrated physics • • • • • • • • • • • • • • • • • • • • • • • • • • • 13
2.5 Summary of literature review • • • • • • • • • • • • • • • • • • • • • • • • • • • • 15
CHAPTER 3 METHODOLOGY • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 16
3.1 Definition and purposes of case study • • • • • • • • • • • • • • • • • • • • • • • • 17
3.2 Traditions of case study • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 19
3.2.1 Interpretivist case study • • • • • • • • • • • • • • • • • • • • • • • • • • • 19
3.2.2 Realist case study • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 21
3.2.3 Comparative case study • • • • • • • • • • • • • • • • • • • • • • • • • • • 23
3.3 Use of case study • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 24
CHAPTER 4 LEARNINGASSISTANTSASCONSTRUCTORSOFFEEDBACK:HOW
ARE THEY IMPACTED? • • • • • • • • • • • • • • • • • • • • • • • • • • 26
4.1 Introduction • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 26
4.2 Methods • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 27
4.3 Results and Discussion • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 29
4.3.1 How constructing feedback affects decisions Carly makes in other contexts • 30
4.3.2 How receiving feedback as a student affects Carly’s approach to
constructing it • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 32
4.3.3 How the impacts of constructing and receiving feedback are connected
for Carly • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 33
4.4 Conclusions and Future Work • • • • • • • • • • • • • • • • • • • • • • • • • • • 35
CHAPTER 5 LEARNING ASSISTANTS AS STUDENT-PARTNERS IN
INTRODUCTORY PHYSICS • • • • • • • • • • • • • • • • • • • • • • • • 37
5.1 Introduction and Background • • • • • • • • • • • • • • • • • • • • • • • • • • • • 37
5.1.1 P-Cubed • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 43
5.2 Theoretical Framework • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 47
5.2.1 Communities of Practice • • • • • • • • • • • • • • • • • • • • • • • • • • 47
5.2.2 The design behind the development of a community of practice in P-Cubed 51
vi
5.3 Methodology • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 53
5.4 Methods • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 55
5.5 Analysis and Findings • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 57
5.5.1 The Learning Assistant community of practice • • • • • • • • • • • • • • • 57
5.5.2 Development of feedback practice • • • • • • • • • • • • • • • • • • • • • 66
5.5.3 Learning Assistant influence through student-partnership • • • • • • • • • • 73
5.6 Discussion and Conclusion • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 81
5.7 Acknowledgments • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 87
CHAPTER 6 STUDENTS’PERSPECTIVESONCOMPUTATIONALCHALLENGES
IN PHYSICS CLASS • • • • • • • • • • • • • • • • • • • • • • • • • • • • 88
6.1 Introduction and Background • • • • • • • • • • • • • • • • • • • • • • • • • • • • 88
6.2 Methodology • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 97
6.3 Context • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 99
6.4 Methods • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 103
6.4.1 Data Generation and Transcription • • • • • • • • • • • • • • • • • • • • • 106
6.4.2 Data Analysis • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 107
6.5 Student-Perceived Challenges • • • • • • • • • • • • • • • • • • • • • • • • • • • • 108
6.5.1 Stress/Frustration • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 108
6.5.2 Feeling Worse at Physics • • • • • • • • • • • • • • • • • • • • • • • • • • 111
6.5.3 Unbelonging and Stereotypes • • • • • • • • • • • • • • • • • • • • • • • • 113
6.5.4 Repeated Confusion • • • • • • • • • • • • • • • • • • • • • • • • • • • • 115
6.5.5 Interpreting Code • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 118
6.5.6 Interpretations of Implementation • • • • • • • • • • • • • • • • • • • • • 121
6.5.6.1 Assessment and Motivation • • • • • • • • • • • • • • • • • • • 121
6.5.6.2 Solutions and “Right” Answers • • • • • • • • • • • • • • • • • • 123
6.6 Connection Between Challenges and Theory • • • • • • • • • • • • • • • • • • • • 124
6.6.1 Self-efficacy • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 125
6.6.2 Mindset • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 128
6.6.3 Self-concept • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 131
6.6.4 Intersection of Self-efficacy, Mindset, and Self-concept • • • • • • • • • • • 134
6.7 Positive Student Experiences • • • • • • • • • • • • • • • • • • • • • • • • • • • • 135
6.8 Discussion • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 138
6.9 Conclusions and Future Work • • • • • • • • • • • • • • • • • • • • • • • • • • • 143
6.10 Acknowledgments • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 145
CHAPTER 7 DISPOSITIONS AND MINDSET IN COMPUTATION-INTEGRATED
PHYSICS • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 146
7.1 Introduction • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 146
7.2 Theoretical Framework • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 150
7.3 Methodology • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 158
7.4 Context • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 160
7.5 Methods • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 163
7.5.1 Data Generation and Transcription • • • • • • • • • • • • • • • • • • • • • 165
7.5.2 Data Analysis • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 166
vii
7.6 Results • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 172
7.6.1 Dispositions Results • • • • • • • • • • • • • • • • • • • • • • • • • • • • 172
7.6.1.1 Otto (Dispositions) • • • • • • • • • • • • • • • • • • • • • • • • 172
7.6.1.1.1 Tolerance for Ambiguity • • • • • • • • • • • • • • • • • • 173
7.6.1.1.2 Persistence • • • • • • • • • • • • • • • • • • • • • • • • • 177
7.6.1.1.3 Collaboration • • • • • • • • • • • • • • • • • • • • • • • 181
7.6.1.2 Blaine (Dispositions) • • • • • • • • • • • • • • • • • • • • • • • 185
7.6.1.2.1 Ambiguity • • • • • • • • • • • • • • • • • • • • • • • • • 185
7.6.1.2.2 Persistence • • • • • • • • • • • • • • • • • • • • • • • • • 190
7.6.1.2.3 Collaboration • • • • • • • • • • • • • • • • • • • • • • • 193
7.6.1.3 Ed (Dispositions) • • • • • • • • • • • • • • • • • • • • • • • • • 197
7.6.1.3.1 Ambiguity • • • • • • • • • • • • • • • • • • • • • • • • • 197
7.6.1.3.2 Persistence • • • • • • • • • • • • • • • • • • • • • • • • • 201
7.6.1.3.3 Collaboration • • • • • • • • • • • • • • • • • • • • • • • 205
7.6.1.4 Summary of Dispositions Results • • • • • • • • • • • • • • • • 209
7.6.2 Mindset Results • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 211
7.6.2.1 Otto (Mindset) • • • • • • • • • • • • • • • • • • • • • • • • • • 212
7.6.2.2 Blaine (Mindset) • • • • • • • • • • • • • • • • • • • • • • • • • 214
7.6.2.3 Ed (Mindset) • • • • • • • • • • • • • • • • • • • • • • • • • • • 217
7.6.2.4 Summary of Mindset Results • • • • • • • • • • • • • • • • • • • 220
7.7 Discussion and Conclusion • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 222
7.7.1 Application of Dispositions Framework • • • • • • • • • • • • • • • • • • • 223
7.7.2 Application of Mindset Coding Scheme in Connection with Dispositions • • 229
7.7.3 Critique of Dispositions Framework • • • • • • • • • • • • • • • • • • • • 232
7.7.4 Concluding Remarks • • • • • • • • • • • • • • • • • • • • • • • • • • • • 236
CHAPTER 8 DISCUSSION • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 239
APPENDICES • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 246
APPENDIX A MULBERRY HIGH SCHOOL INTERVIEW PROTOCOL • • • • • 247
APPENDIX B BECK RESULTS • • • • • • • • • • • • • • • • • • • • • • • • • • 249
REFERENCES • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 262
viii
LIST OF TABLES
Table 5.1: Five types of data sources: interviews, feedback, course documents, discussion
notes, and emails. • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 55
Table 6.1: Four types of data sources: student interviews, a teacher interview, field notes,
and classroom recordings. • • • • • • • • • • • • • • • • • • • • • • • • • • • • 107
Table 7.1: Definitions, inclinations, sensitivities, and abilities for each CT disposition.
Features of the table are copied from different parts of Pérez [1]. The inclinations,
sensitivities, and abilities as described here align with “high” dispositions.
• • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 153
Table 7.2: Indicators of growth and fixed mindset, to be used as a coding scheme for our
data, and developed from Dweck [2, 3], Blackwell et al. [4], and Yeager and
Dweck [5]. • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 156
Table 7.3: Data sources used in this study (more were generated, detailed in Chapter 6). • • 165
Table 7.4: Transcription Key. Some transcription symbols borrowed from Jefferson [6].
Symbols are used only for in-class transcripts unless otherwise specified. • • • • 167
Table 7.5: Coded instances of alignment between CT dispositions and Otto’s data, separated
by data source. • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 173
Table 7.6: Coded instances of CT dispositions in Blaine’s data, separated by data source. • 186
Table 7.7: Coded instances of CT dispositions in Ed’s data, separated by data source. • • • 198
Table 7.8: Codes for inclinations, sensitivities, and abilities, separated out among the
dispositions for comparison. The codes in this table are binned by data source.
The left four columns are separated between high and developing dispositions,
and the right two columns simplify the information by providing counts that
combine high and developing codes for each disposition. • • • • • • • • • • • • 211
Table 7.9: Coded instances of mindset in each student’s data, separated by data source.
The bottom row provides a total count of fixed and growth mindset codes for
each student’s data. • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 212
Table 7.10: The way mindset overlapped with dispositions for excerpts in which we coded
for both constructs. • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 222
Table B.1: Coded instances of CT dispositions in Beck’s data, separated by data source. • • 249
ix
LIST OF FIGURES
Figure 3.1: Diagram from Erickson [7] about the organization of the links between data
and assertions in a qualitative case study. • • • • • • • • • • • • • • • • • • • • 18
Figure 4.1: This shows direct and indirect influences of Carly’s experiences with feedback
in P-Cubed, as referenced in the discussion. The relationship 1 ) 2 is the
focus of Section 4.3.1. Section 4.3.2 focuses on the relationship 0 ) 1.
Section 4.3.3 demonstrates the relationship between the two influences, and
argues that the first influence strengthens the second. The influences and
connections themselves are detailed in the discussion. • • • • • • • • • • • • • 33
Figure 5.1: LA participation ladder: a visualization of the different levels of influence that
LAs can have on curriculum design and pedagogy. Borrowed from Bovill and
Bulley [8] to conceptualize student-faculty relationships instead of studenttutor
relationships. The arrow represents how student participation changes
with the nature of the partnership. To be clear, this is not a recommendation
for classrooms to strive for the highest rung, rather a model for characterizing
partnerships based on the roles of students. • • • • • • • • • • • • • • • • • • • 40
Figure 5.2: Text exchange between Erica and a graduate TA, exemplifying strong relationships
within the student teaching staff of P-Cubed. The blue bubbles
on the right show Erica’s suggestions for the TA’s feedback, while the grey
bubble on the left shows the TA’s response to Erica’s suggestions. • • • • • • • 60
Figure 5.3: The feedback structure described in the course materials is two similarly
structured paragraphs, one focused on in-classwork by the group, one focused
on in-class work by the individual. • • • • • • • • • • • • • • • • • • • • • • • 68
Figure 7.1: Mr. Buford’s classroom setup. This snapshot shows students gathered around
tables in groups of around three or four. • • • • • • • • • • • • • • • • • • • • 162
Figure 7.2: Partially filled-in bars representing the overlap of mindset codes into howeach
disposition was coded through all the data. Each bar represents the number of
excerpts coded for a particular disposition, and the diagonal shading represents
the portion of the excerpts also coded for mindset. The percentage of overlap,
rounded to the nearest percent, was 27% for tolerance for ambiguity, 30% for
persistence, and 14% for collaboration. • • • • • • • • • • • • • • • • • • • • • 221
Figure 7.3: Waffle diagram showing the number of mindset-coded excerpts in total, the
portion of those excerpts thatwere also coded for dispositions, and the number
of those overlap-coded excerpts that showed alignment (either between growth
mindset and high dispositions or fixed mindset and developing dispositions). • 221
x
CHAPTER 1
INTRODUCTION
Students have a central role to play in how physics curricula continue to develop. Physics education
researchers, curriculum designers, and teachers have used students’ perspectives as insightful
resources [9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23]; institutional science standards
have aligned with student-centered approaches to schooling [24, 25, 26, 27], and students themselves
have been handed more power to make physics classroom decisions [28, 29, 30, 31, 32]. Ultimately,
this focus has led to calls to gain a deeper understanding of the experiences of physics students [33].
My aim in this dissertation is to show how physics students’ perspectives can provide insight into
possibilities for curricular change and further research into their experiences, especially in the
context of computation-integrated physics.
There have been many efforts in a variety of contexts to incorporate the perspectives of students
into curriculum design and pedagogy [29, 32, 34, 35]. These instances of students gaining control
over institutional processes are often spurred on by research into what students and/or institutions
stand to gain from redistributing power [28, 36, 8, 37, 38]. This dissertation aligns with such
research efforts because each study provides a window into what students teach us about a learning
environment and provides a pathway for how those findings could motivate student-centered
curricular change.
One realm where curricula are changing significantly is physics classrooms through the integration
of computation [39, 40]. Rather than being taught as a separate coding class, computational
modeling is being integrated into physics curricula as a tool for learning science and representing
physics concepts in new ways. In the last two decades, there has been serious consideration of
curriculum redesign around computation [41, 42, 43, 26] with more widespread calls to incorporate
computation in the last ten years [24, 27, 44], yet still no widespread agreement on assessment,
curriculum, or learning goals. Due to the relative newness of computation in physics curriculum,
little work has been done on what makes a computational-curricular integration effective for learn-
1
ing, motivation, and attitudes [33]. Because these constructs revolve around student affect, there
is a need for incorporating students’ perspectives into research on these new computation-based
physics curricula.
The role I intend for this dissertation to play is to incorporate students’ perspectives into
computation-integrated physics curricula by studying and communicating what students have to
say about their experiences in physics classes with integrated computation. By listening to students,
we can begin to change the structure of teaching and learning in their favor. Historically,
physics curriculum designers and researchers have attempted to incorporate student perspectives
via understanding how students organize their content knowledge [9, 10, 13, 19, 18], view their
learning [11, 14, 15, 17, 23], and view pedagogical strategies [12, 21, 20]. However, these efforts
have not been extended into the context of a computation-integrated physics curriculum. Due to the
changes a physics curriculum undergoes to incorporate computational modeling, we need research
on how students experience this new learning tool and how its curricular use can be improved.
The research I present here is designed to provide an authentic, in-depth, rigorous view of
how students perceive their experiences in these computation-integrated classrooms. In striving
for authenticity, I base my work in the words of students themselves, mainly via semi-structured
interviews [45] and recordings of students working in their natural classroom context. To provide
depth, my work is qualitative, meaning I focus on social phenomena by studying its participants
and taking detailed accounts of their interpretations. To ensure my work is rigorous, I use a widely
accepted research methodology: qualitative case study [46, 47, 48, 49, 7, 50, 51, 52, 53, 54, 55, 56].
In summary, I am motivated by a research-based need to center students and incorporate their
voices into physics curricula. I explore this need first in a university context where student voices
already have significant power in curriculum design (Chapters 4 and 5) and second in a high school
context where the curriculum is new, changing, and ripe for student voices to shape it (Chapters 6
and 7). I use case study throughout the chapters to illustrate the richness that can be drawn from
studying student perspectives in an in-depth, qualitative fashion. Below, I provide a roadmap with
more details.
2
This dissertation is a presentation of four research studies in two contexts. Chapters 4
and 5 investigate the experiences of undergraduate learning assistants (LAs) in an introductory,
computation-integrated physics course and their dynamic relationships with the curriculum that
they teach. Chapters 6 and 7 are studies on students in a high school physics class into which
computation has been newly integrated. I tie these chapters together by providing a background
and literature review in Chapter 2, where I explore the background of research on physics LAs,
computation, and the use of student perspectives in curriculum development. I also introduce some
of the theories and models that help provide inroads to studying student perspectives in the relevant
contexts.
In Chapter 3, I flesh out the methodology used in all my research studies: qualitative case
study. Case study is ideal for researching a phenomenon in its natural setting via multiple data
sources [47, 46]. Myuse of case study develops in complexitymoving from Chapter 4 to Chapter 7. I
begin with a generic case study, then transition to using more structured case study methodological
choices. Because case study is used sporadically and with much variance in physics education
research [57, 17, 23, 21, 22, 58, 20, 59, 13, 16, 19, 10, 18], I orient Chapter 3 to introducing
qualitative case study and its traditions to an unfamiliar reader.
Chapter 4 represents a published paper [60] on the case study of an LA for an introductory
physics course. I use the study to illustrate an example of how case study and student perspectives
join together constructively in a context where a student has a dynamic relationship with the course
she teaches. In this case, the use of case study is the driving force behind centering the student’s
perspective, and the findings serve as a jumping off point for the next chapter.
Chapter 5 represents an in-press paper [31] (accepted in Physical Review Physics Education
Research). that expands on the research design from Chapter 4 to include the perspectives of
other LAs and a faculty member. The focus of this chapter lies in how the expertise of LAs has
been leveraged via the design of the course and via their relationships with faculty members. This
chapter uses case study to center the interpretations of LAs and make an argument for how the
impact that LAs have on the curriculum could be expanded and formalized. For the purposes of
3
this dissertation, Chapter 5 is also a demonstration of the efficacy of using case study to investigate
the relationships between a curriculum and the some of the students who interact with it.
Chapter 6 is a case study on high school student perceptions of the challenges they face in a
computation-integrated physics class. This chapter serves as a demonstration of case study in a
high school physics classroom setting. Unlike the LAs from Chapters 4 and 5, the participants in
this study do not have much influence over the curriculum. Accordingly, much of my investigation
focuses on describing the perspectives of these high school students and framing the findings as an
opportunity to design computation-integrated curriculum with student perspectives in mind.
Chapter 7 is another case study on the students and context from Chapter 6. This study applies
a framework originally theorized for analyzing a person’s orientation towards learning computationally
in a mathematics context called the Computational Thinking Dispositions framework [1].
I apply the framework to the perspectives of high school students in an effort to demonstrate the
framework’s efficacy in the context of computation-integrated classrooms and build a connection
to the more established construct of mindset. This chapter demonstrates that a case study on
student perspectives can provide motivation for student-centered curricular change in computationintegrated
physics classrooms.
Chapter 8 is a discussion of the ideas I build and test throughout the dissertation. I review the
uses and benefits of qualitative case study, the importance of researching students’ perspectives,
the limitations of this work, and the potential for such research to influence curriculum in ways that
benefit students.
4
CHAPTER 2
LITERATURE REVIEW
This chapter is a literature review that provides context and motivation for the chapters that follow.
I begin by reviewing how students’ perspectives can be incorporated into curriculum design
choices, and I describe the setup of Chapters 4 and 5 to show why it matters to listen to students.
Then, I introduce a curricular context where students’ perspectives need to be leveraged more:
computation-integrated physics. In order to address this need, I lay a foundation of literature that
calls attention to students’ perspectives in computation-integrated physics, which aligns with the
setup for Chapter 6. Lastly, I build on that foundation to set up the background of a study (Chapter 7)
that uses students’ perspectives to explore and apply a framework of student dispositions with the
potential to offer improvements to curriculum in computation-integrated physics.
2.1 Incorporating students’ perspectives into curriculum
My interest in using students’ perspectives is grounded in the history of doing research on what students
have to say about their physics learning and making corresponding suggestions for curricular
improvements [9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 61]. The focuses of these
studies are mostly qualitative and span from understanding how students organize their content
knowledge [9, 10, 13, 19, 18] or view their learning [11, 14, 15, 17, 23] to how they view different
pedagogical strategies or classroom supports and how students think those strategies and supports
could be improved [12, 21, 20, 61]. In general, physics education researchers consult students’
perspectives to catalog experiences from the perspectives of students and use those findings to
think about potential improvements to curriculum and pedagogy. With this dissertation, I aim to
contribute to this collective effort.
To understand how student perspectives can be leveraged more directly and more expansively
in curriculum, I investigated a recent model for gathering and implementing student input called
5
Students as Partners (SaP) [28]. This model allows students to engage with their instructors on
course design. In SaP, students are viewed as experts on their own learning and use that expertise
to help make decisions on curriculum and pedagogy for a course. There is a focus on partnership,
which means students can be positioned as co-developers of curriculum through their collaborative
relationships with faculty or some other curriculum developer. While it is possible that students
can be enrolled in the course for which they are consulted, they do not need to be—in some cases
the student-partners are employed as learning assistants (LAs) (e.g., Jardine [30] and Hamerski et
al., my own work represented in Chapter 5 [31]). Due to the relative recency of higher education’s
embrace [32], SaP has just begun to be used to describe instances where students have significant
voice in curricular decisions in physics contexts [29, 31].
My use of SaP to describe a student-partnership in a physics context [31] was an expansion of an
earlier study on the perspective of a student who helped teach introductory physics [60]. Chapters 4
and 5 represent the work from these two studies. I found in the first study that students had rich
reflections on their experiences and deeply insightful thoughts about how a physics course could
function [60]. This fed my interest in the value of listening to students about their experiences
and leveraging what they say to make meaningful curricular changes. Eventually, this led to the
second study, which expanded my focus to a handful of students who were employed as learning
assistants (LAs) and had opportunities to infuse their perspective into the curriculum of the course
they helped teach [31].
The main takeaway from doing this work was that it is not enough to just listen to what
students have to say. There also needs to be a theoretical framework in place and a consideration
of contextual features to help to interpret students’ perspectives. Listening to students without
the structure afforded by theory and context can lead to reactionary incorporation of students’
voices into the curriculum instead of more thoughtful changes [14, 29]. Context and theory are
important for informing howstudent perspectives can be listened to and used as feedback to facilitate
productive curricular change. For example, in Chapter 5, I used the SaP model [28] to describe the
power undergraduate LAs held in curricular decisions. I also used the theory of Communities of
6
Practice (CoP) [62] to help me describe how LAs can develop their expertise over time and make
decisions that align with the goals of the community of LAs. I chose to use CoP to describe this
phenomenon because the course itself was designed around that theory [63]. It also provided a
mechanism for describing how LAs can become more central voices [62] to the curriculum with
experience. The consideration of context and theory helped show how LAs can be leveraged as
pedagogic consultants in courses where a community of practice might be present.
The precedent of listening to students in qualitative physics education research helps show that
meaningful ideas for curricular improvements can come from students. The experience of using
SaP and CoP in Chapters 4 and 5 helps highlight the importance of providing theoretical structure
and attending to context when consulting students about their experiences. Next, I describe a type
of curriculum that is newly developing and spreading [39, 33, 64, 65, 66, 67, 68, 69, 44], indicating
a significant opportunity to learn from student perspectives at the nascent stages of this curriculum’s
development .
2.2 A context where students’ perspectives are needed: computation-integrated physics
Curricula in physics classrooms across the United States have been changing over the last fifteen to
twenty years to include the integration of computational modeling into curricula as a tool to learn
physics [41, 42, 43]. The goal of this integration is for “students [to] use computing as naturally as
they [now] use traditional mathematics” [39]. Just as mathematics is taken for granted as an essential
tool for learning physics, the hope is for computation to be integrated into physics curricula just as
deeply. Because computational modeling is becoming a critical part of STEM careers [70, 71], it
is also becoming increasingly important in curriculum development [33, 64, 65, 66, 67, 68, 69, 44]
and in national learning standards [24, 25, 26, 27]. Because of the broad scope of this goal and
widespread integration, there are many different types of computational integration [33, 64, 65, 66,
67, 68, 69, 44]. The variance among implementations means there are many different opportunities
to incorporate student perspectives into the new and changing curricula.
This variance is motivated by and reflected in the wide-ranging, often ambiguous national calls
7
and standards that give structure to the integration of computation into school STEM[24, 25, 26, 27].
They describe broad learning goals, like using “computational tools...to analyze, represent, and
model data” [24], or “choose among computational algorithms and computational tools to produce a
solution” [27]. There is not much specificity provided about howto achieve these learning outcomes,
so the implementations have varied greatly, even in the last few years [33, 64, 65, 66, 67, 68, 69, 44].
A recent call has emphasized the lack of direction in how to integrate computation into STEM
courses and provided some guidelines for integration and recommendations for research that can
strengthen and deepen the research-based support for computational integration [33]. One of
the major recommendations from the call was to “examine the development of students’ identity,
agency, positioning and motivation in relation to their engagement in computational tasks” [33].
This need for understanding how students relate to computation on an affective level is a gap that I
intend to address with this dissertation.
Some research has attempted to coordinate the perspectives of faculty and professionals into
computation-integrated physics curricula or learning goals related to it [72, 71, 70, 73], but none
have consulted students themselves about their experiences or for direct input into what matters
when learning computation-integrated physics. Pawlak et al. [74] produced a research study
adjacent to this endeavor by seeking to understand computation-integrated physics learning from
the perspectives of LAs. In the context of physics without computation, student perspectives
have been consulted for rethinking curriculum [29, 75], but this type of work is still needed in
computation-integrated physics. One way to address this need is by observing students working
on computational activities in their physics class, and cataloging learning goals based on their
experiences as done by Weller et al. [76]. However, this method still requires researchers to
interpret students’ experiences rather than hearing students’ interpretations for themselves. A more
direct, affect-based approach would be to focus on interviews as a data source, where students can
say directly what they struggle with and how they feel about it. This is the study presented in
Chapter 6.
Outside of computation or computation-integrated contexts, researchers have used student
8
perspectives and affect-based studies to better understand STEM courses and to motivate change in
STEM pedagogy [77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 14, 88, 89, 90, 91, 92, 93, 94, 95]. For
some examples, Hannula [80] demonstrated connections between affect and success in a middle
school math context, suggesting that teaching and learning can be improved by attending to student
affect in pedagogy. Galloway et al. [84] asked students in an undergraduate chemistry lab course
about their experiences, finding that students had complex, multifaceted affective responses. This
led the authors to develop pedagogical suggestions for cultivating positive affect and making labbased
chemistry more meaningful for students. Alsop and Watts [87] examined how students felt
about and perceived radiation and radioactivity in their physics class, finding that it was possible to
keep students engaged but not off track by striking a balance between staying informed and following
passions and interests. These examples align Section 2.1, where I showed that student perspectives
can enrich curriculum when their input is incorporated in structured ways. This further emphasizes
the importance of addressing the need to research student perspectives in computation-integrated
physics.
While Chapters 4 and 5 focused on the LAs in an introductory-level university physics class,
Chapter 6 specifically emphasizes the student perspective. In Chapters 4 and 5, the LAs operated
in an environment that had been around for a few years [63], and physics LAs have precedent for
participating in physics education research [96, 97, 98, 99, 34, 100, 101, 102, 103], meaning our
studies were built on a longstanding foundation of listening to LA perspectives. In contrast, there is
a lack of foundation for research on student perspectives in computation-integrated physics. Though
faculty perspectives have been consulted in computation-integrated physics education research and
student perspectives have been consulted in other disciplines, as I described above, we need to fill
the specific gap of students’ perspectives in the context of computation-integrated physics. Due
to computation-based integration being a novel research area, the research design in Chapter 6 is
more exploratory than the previous chapters. Since so little is known about computation-integrated
environments, the intention of Chapter 6 is to build a foundation upon which more focused research
can be built. This exploratory research will facilitate future steps towards the act of incorporating
9
student perspectives into curriculum. For Chapter 6, I primarily used students’ interview comments
and the context of the classroom to explore the landscape and meaning of students’ experiences.
To build a structure, so to speak, I sought to understand how affect-based theories interacted in the
context, which I detail in the next section of the literature review and in Chapter 6.
2.3 Laying the foundation for incorporating students’ perspectives in computation-integrated
physics
This section is about providing a background for designing research that listens to students in a
computation-integrated physics environment. Chapter 6 uses a broad study design, and the breadth
of the array of student experiences led to a decision to catalog their perspectives using context to
gain insight into what the students meant. The focus was mainly on how computation brought about
new experiences for the students. However, once I analyzed the students’ perspectives, it became
clear that some of them related to established theoretical constructs from parallel, student-centered
studies that used those theories [104, 105, 92, 93, 94, 95, 106, 107, 108, 109, 110, 111, 14]. I
plan to briefly review the relevant theories below to provide context for Chapter 6. The reason for
considering multiple conceptual framings in the same study is to provide multiple perspectives on an
understudied context. This serves to provide researchers and practitioners with multiple pathways
into examining computation-integrated physics, pathways which can lead to deeper findings as
they are explored. To instigate significant change, I would need to focus more deeply on applying
a theoretical framework to the students’ perspectives and processing their experiences through a
lens that can help understand what these perspectives mean for potential curricular change (like in
Chapter 7). I address that line of reasoning in the next section, but here I review the theories that
featured in Chapter 6, which have been used to explore computation-integrated STEM contexts,
albeit minimally.
The first theoretical lens is self-efficacy. As defined by Bandura, self-efficacy is “concerned
with judgments of how well one can execute courses of action required to deal with prospective
situations” [112]. In relation to students’ motivation and confidence for a given academic subject,
10
he said, “the higher the students’ beliefs in their efficacy to regulate their motivation and learning
activities, the more assured they are in their efficacy to master academic subjects” [113]. This
construct has been used widely in studies focused on understanding how students’ affect relates to
their view of their own abilities [104, 105, 92, 93, 94, 95]. While not quite the same as our context
of computation-integrated physics, self-efficacy has been used in technology-based interventions
aimed at improving self-efficacy for physics [114, 115], physics-based interventions aimed at
improving computational thinking [116], and one experimental study aimed at characterizing and
supporting teachers’ self-efficacies for teaching computation-integrated engineering [117]. These
studies took a focus towards the outcomes of their interventions, and their contexts were physics or
computation but not both. The perspectives gathered in Chapter 6 included students discussing their
abilities in relation to specific computational or physics activities, so it was necessary to provide
a preamble describing previous applications of self-efficacy in computation and physics to ground
our findings in previous research.
The second theoretical lens is self-concept. Marsh and Craven [118] provided a definition
for self-concept, based on the work of Shavelson et al. [119], as “one’s self-perceptions that
are formed through experience with and interpretations of one’s environment. They are influenced
especially by evaluations of significant others, reinforcement, and attributions for one’s
own behavior.” A person has a different self-concept depending on the context (e.g., physics
class) and focus (e.g., computational activities) [119]. This construct has been used widely in
studies focused on understanding students’ affect as related to how they view themselves in an
academic setting [109, 110, 120, 111, 14]. Despite the wide use and focus of self-concept on
students’ perceptions, I could not find any application of self-concept at all in published research
on computation-integrated physics. The closest approximation is research on self-concept related
to students’ perceptions of relevance in their physics curricula [14]. The perspectives I gathered in
carrying out the research of Chapter 6 were relatable to self-concept, which provided an opportunity
to explore the data’s relationship to theory given the lack of studies on self-concept in physics or
computation, let alone computation-integrated physics.
11
The third theoretical lens is mindset. Originally theorized by Dweck [2], it can be thought of
in terms of how “students may hold different theories about the nature of intelligence” [4]. In this
sense, mindset is a “continuum,” where on the one end students believe their intelligence is “an
unchangeable, fixed entity,” and on the other end it is “a malleable quality that can be developed” [4],
but in reality students might hold views anywhere between these points. Mindset has also been
shown to change over time and vary depending on context [106, 121]. As a framework, mindset
has been used in education research to show how students’ mindsets relate to how they respond
to their experiences in educational settings [106, 107, 108, 121, 122]. In computation-integrated
STEM, mindset has been used in limited ways so far. It was used once in a two-week intervention
where physics students did a computational project and experienced no significant changes in their
mindsets [123]. It was also used by Little et al. [122], who applied mindset to develop a mindsetbased
coding scheme and describe students’ interview comments about the challenges that they
faced in class. While three of the 21 interviews that compiled Little et al.’s [122] coding scheme
were from courses that had computational projects integrated into them, the majority of interviews
were from contexts outside of computation-integrated courses. The potential presence of mindset
in students’ perspectives in Chapter 6 and with the lack of mindset-based studies in primarily
computation-integrated physics contexts provide an opportunity to address mindset in my data.
Based on the literature outlined above, it makes sense that some of the students’ perspectives
around computation ended up relating to the theories of self-efficacy, self-concept, and mindset. In
truth, it is not surprising that these theories emerged from students talking about their experiences
with physics and computation, but I did not intend to investigate any particular theory at the outset
of the research. I instead aimed to build a broad understanding of the challenges students experienced
from their points of view. As the area of integrated computation is relatively unexplored,
providing a broad perspective makes sense as the first step towards deeper investigations into how
these theories manifest individually and relate to each other in a computation-integrated learning
environment. I provide a base of literature here on the landscape of research in computation and
physics relating to these theories to scaffold a discussion in Chapter 6 about how these theories
12
could be related to each other in our context. From there, further research can build on more
specific constructs with a stronger theoretical base, like the use of mindset in Chapter 7, which I
discuss in the next section.
2.4 Building onmy previouswork to highlight and address a curricular need in computationintegrated
physics
In Chapter 6, I found that students’ statements were related to the constructs of mindset, efficacy,
and self-concept. This study highlighted the need for more detailed investigations in the future to
understand how they impact students in the computation-integrated environment. In Chapter 7, I
addressed one aspect of this need by designing a study that listens to student perspectives with both
theory and context in mind. This study carries implications that are based on student perspectives
and that have the potential to improve student affect and learning in computation-integrated physics.
I chose to focus Chapter 7 in part on the theoretical lens of mindset (one of the theories I
applied in Chapter 6) for two reasons. First, mindset is connected closely with improving students’
outcomes, specifically through interventions [124, 125, 126, 127, 128]. The aims and outcomes
of interventions range among improved mental health [125], increased motivation [124, 127], and
boosted academic performance [126, 128]. Using the connection between mindset and the benefits
to students provides a motivation for understanding students’ perspectives with a mindset lens with
the hope to ultimately use those findings to potentially infuse mindset-fostering practices into a
computation-integrated physics curriculum.
The second reason I focus on mindset is that it is connected to the theoretical Computational
Thinking (CT) Dispositions Framework [1]. CT itself is a widely sought learning goal for instructors
in STEM contexts [129, 130, 131, 132]. Historically CT has had a variety of definitions, but
one widely supported operational definition of CT [132] splits CT into a framework with two
connected categories: practices included in CT (such as using algorithmic thinking to automate
a solution or representing data with a model) and dispositions that support and enhance the
practices (such as persistence through challenges). There is a wide array of research focused on
13
CT practices [40, 133, 66, 134, 76], but little focused on CT dispositions [1]. Pérez designed the
CT Dispositions Framework, using aspects of mindset and mindset-based research, to address this
need. He named three dispositions as central to the framework: tolerating ambiguity, persisting
on difficult problems, and collaborating with others [1]. For each disposition, Pérez argued that
there are three aspects of the disposition that interact with one another: “Inclination refers to
an individual’s tendency toward a particular way of thinking or acting. Sensitivity denotes an
individual’s attentiveness to opportunities to engage in that particular thought or action. Ability
refers to being able to actually produce that thought or action when one notices an opportunity
(sensitivity) and feels drawn to act (inclination)” (pages 434-435) [1]. By dividing each disposition
into its corresponding inclinations, sensitivities, and abilities and describing each aspect in detail,
Pérez organized the CT Dispositions Framework [1]. However, Pérez’s framework at this point is
a theoretical construct developed in a math context. There is a need to explore the aspects of CT
that Pérez pointed out in his CT Dispositions Framework and how they apply to different contexts.
This is especially true because CT is desired as a learning goal and dispositions are understudied
in comparison to practices.
Because of this need and CT dispositions’ connection to mindset, Chapter 7 examines how
students in a computation-integrated physics classroom express aspects of CT dispositions and how
those expressions relate to mindset as well. By connecting the newly developed CT Dispositions
Framework [1] with the well-established theory of mindset [2] and by building on a foundation
(Chapter 6) of exploring how mindset shows up in computation-integrated physics, I intend to
show the applicability of the CT dispositions framework in a computation-integrated physics context
and how it might be used to meaningfully and productively impact computation-integrated
physics curricula. The opportunity to do this rides on the history of listening to and incorporating
students’ perspectives into curriculum in research-based ways, the need for such work in
computation-integrated physics, and a context-heavy, student-centered foundation (Chapter 6) for
the more theory-driven approach of this particular opportunity (Chapter 7).
14
2.5 Summary of literature review
In summary, there is a precedent for incorporating students’ perspectives into curriculum design.
Either using students’ perspectives to build theory or using theoretical frameworks to process
students’ perspectives. It is clear that keeping contextual factors in view when attempting to
interpret and use student perspectives are crucial to make meaningful, productive changes to
curriculum that benefit students. The qualitative research on this topic provides some footing for
my own research, which aims to provide in-depth understanding of what students can tell us about
their experiences with computation-integrated physics. Computation-integrated physics is a widely
spreading context for which student perspectives have not yet been used in curriculum design. My
research in this area is situated by calls to investigate students’ experiences and perspectives. I
orient my research as developing a foundation from which students’ perspectives can be further
explored in the context of computation-integrated STEM courses. CT dispositions and mindset
provide an opportunity to build on this work by gathering students’ perspectives on computational
integration and translating their perspectives into implications for future curricular change and new
implementations of computation-integration physics.
15
CHAPTER 3
METHODOLOGY
In this dissertation,my focus is on howthe perspectives of students can lead to improving curriculum
and pedagogy. Much of the work I do is to elicit, observe, and report on their feelings about
computational modelling (in the case of a computation-integrated physics class) or physics pedagogy
(in the case of learning assistants). The research itself is qualitative and detailed because the goal
is not to generalize across contexts (using induction) but to build understanding of behavior in
particular situations (abduction) [135]. I cannot distill the experiences of these students into
parameters through which results can be interpreted and generalizations can be induced, but I
can gain an in-depth understanding of a small group of students, which can then be used to
make connections and gain insights to other narratives about computational modeling in high
school physics or physics pedagogy at the introductory university level. Everyday experiences that
persist longitudinally—such as being a high school science student or an undergraduate learning
assistant—cannot just be factored into components and reapplied elsewhere. This has driven my
choice of a qualitative methodology. With this dissertation, I aim to offer a deep understanding of
one context that others may use to make sense of their own.
The methodology I use in the following chapters is qualitative case study [46, 47, 48, 49, 7, 50,
51, 52, 53, 54, 55, 56]. This is a way of constructing truth from the perspectives of participants
and from the artifacts they produce. With case study, one can test out ideas about how things work
to see what is really going on in the lives and everyday settings of the student participants. This
is similar to how Stake [46] describes case study’s main functions: (1) creating explicit formal
knowledge about the phenomenon and (2) creating a vicarious experience for the reader. Below I
define case study and flesh out the purposes that it can serve.
16
3.1 Definition and purposes of case study
Case study is defined by the natural setting of the research context and multiple data sources [46,
48, 50, 51, 54]. The natural setting ensures that the phenomenon of interest plays out authentically,
and the multiple data sources ensure that evidence can be triangulated together for strong claims
when carrying out analysis on the data [7, 48, 47]. A case study is characterized by the case and
the phenomenon. The case is the site of the research, sometimes a specific place, individual, or
organization/institution [46, 47, 48, 49, 50, 51, 52, 54]. This can align with the context but does not
have to (e.g., if the case were an individual person). The phenomenon is an aspect of the case that
the researcher focuses on, like an action/practice, event, individual, or idea [46, 47, 50, 51, 52, 54].
Sometimes, the case is just an instance of the more general phenomenon. Together, the case and the
phenomenon form the research question in a case study [46, 50, 51, 54]. As an example, Ozen [17]
studied a physics classroom in which the teaching primarily took place online. The case was the
online class, the phenomenon was how students perceived their experiences in the online class, and
the research question was, “What are the students’ perceptions of an online College Physics course
as taught through the Internet?” [17].
Afeature of the case studies in this dissertation is that their data sets are bounded [47, 48, 46, 49].
It is important when designing a case study to specify what belongs in the set of studiable data
because the context can have drastic impacts on what and how data are generated. Since the context
ties the case to the phenomenon, and this context can include both time and space, the context plays
an important role in the planning stage of a case study. When deciding how to generate data, an
important question is, what in this context is relevant to the case? In designing for data generation,
it can help to understand how data comes together and how different types of data can strengthen the
validity of a case study. Understanding what data is used for in a case study is crucial for designing
where and how to generate data. In Figure 3.1, I provide a photocopied figure from Erickson [7]
that shows how multiple sources of data and evidence (e.g. field notes, interview comments, and
site documents) can be organized in a case study, specifically noting how multiple sources of data
17
Figure 3.1: Diagram from Erickson [7] about the organization of the links between data and
assertions in a qualitative case study.
are used together to create subassertions (or claims) and ultimately construct the general assertions
in the study.
While all case studies fall under the umbrella of gaining an in-depth understanding of a phenomenon
in a particular setting, they can serve a variety of purposes in research. Sometimes, a
case study is an existence proof [47]. An existence proof documents a phenomenon to show that
it can be done, and wherever it’s not done it is because of choices that have shaped the context,
not because it is impossible. Chapter 5 is in part an existence proof of LAs participating in a
student partnership on the development of curriculum and pedagogy. On the other hand, a case
study that serves as a falsification [55] shows that some general idea is not always true (e.g., all
cultures develop arithmetic). Case studies can also provide counter-narratives [48, 136, 137], much
like falsification, but specifically showing that the dominant idea does not necessarily describe the
phenomenon accurately (e.g., a study countering the narrative that Danish Muslim girls often do
not play sports because of their religion [138]). The difference between these two functions is
subtle: falsification proves a generally accepted idea wrong, whereas a counter-narrative broadens
understanding of a phenomenon by presenting a case that portrays the phenomenon in a new way,
often in contrast with dominant ideas.
Case studies can also generate theory about how a phenomenon works [46, 47, 49, 56]. This is
18
often groundbreaking and provides a new way of understanding a particular phenomenon. Building
on that, case studies can also substantiate or test the generated theory in other contexts to see howthe
theory extends to other situations [55, 139]. This parallels the theory-generating work of Pérez [1]
on CT dispositions and my extension of his theory into the context of computation-integrated high
school physics in Chapter 7. At other times, case studies can generate hypotheses [47, 56], which
in turn could motivate quantitative work. Designing a quantitative study beforehand often misses
data collection on the important factors that case studies could reveal [140].
All case studies contribute to a repertoire of stories [141, 136]. This means that for a given
phenomenon, there are many case studies of that phenomenon in a variety of contexts. The cases
one can access via the repertoire of stories can help provide insight to the phenomenon at large or
to deal with unexpected issues related to the cases. The next four chapters all serve this purpose
along with their other aims.
3.2 Traditions of case study
Case study can be categorized more formally than by its purpose. Over the history of case study’s
use, different traditions have emerged that align and differ in how the research is designed and how
claims are constructed from evidence. Contemporary use of qualitative case study can generally
be described with three categories: interpretivist, realist, and comparative. Below, I describe what
each tradition offers, how they differ and align with one another, and how I used them in this
dissertation.
3.2.1 Interpretivist case study
The hallmark of interpretivist case study [46, 50, 51] is the importance of language in framing
the case. Language is what constructs the case itself and gives insight into how the case is being
interpreted by the relevant characters. For example, what makes a place function as a classroom
is how people interact in that place and what they say. There are artifacts that reify and facilitate
the interactions, such as whiteboards and desks and lab equipment, but what matters most is how
19
the teacher and students come together and collectively do school-based learning. Interpretivist
case study believes in studying what participants say and do, focusing on how they interpret the
phenomena that play out in their setting.
Interpretive case studies investigate how a phenomenon is socially constructed and what it
means to its participants, hence the focus on interpretations. Phenomena of interest in interpretivist
case studies are often commonplace ideas that do not have fixed meanings, instead they mean
different things to different people. Researchers aim to unpack how others interpret that thing. We
seek to enter the participants’ “imagined world” [50] and understand the phenomenon from their
point of view.
The mechanism used in interpretivist case study to connect data to claims is called “anchor
points” [50]. These are the perspectives of the participants that emerge from data generation.
Each perspective serves as an anchor to the phenomenon, which is understood by generating
multiple anchor points and making inferences based on how participants interpret it. After all, the
interpretations of participants socially construct the phenomenon in the first place, so in the frame
of interpretivist case study, the closest data sources to the phenomenon itself are the perspectives
of those who participate in it.
Interpretivist case study emphasizes the social construction of the phenomenon. Within a single
context or activity, the participants might be focused on several different aspects of the activity,
and they might interact in several different ways in pursuit of several different goals. Even within
a single context or activity, many cases or phenomena could be constructed, because according to
the interpretivist stance, everything that the participants view as meaningful is meaningful [50].
This orientation towards participants’ perspectives is what led me to use interpretivist case study
as much as possible in my work.
The way I incorporated interpretivist case study into my research designs was to formulate
research questions, data generation methods, analytic methods, and assertions with the perspectives
of my participants at the center. For example, Chapter 5 [31] was designed with interpretivist case
study. The research questions were aimed to explore how LAs interacted with one another and
20
with aspects of the course (e.g., “how has the practice of feedback been shaped by the LAs?”). The
questions in the interview protocol were designed to be more open-ended [45] so as to allow LAs to
tell me what they thought was important (e.g., “How do you know if a student is struggling?”). The
data generation methods were to allow LAs to provide their perspectives through a variety of means
(e.g., interviews, feedback excerpts, email correspondences). The analysis and findings were also
shaped around what LAs said and did in the data—I treated each LA-centered data source as an
“anchor point” [50] through which to construct an understanding of the meaning that LAs assigned
to the phenomenon. Due to my focus on students’ perspective throughout the dissertation, not just
in Chapter 5, all four of the following chapters take up this interpretivist stance to some degree.
3.2.2 Realist case study
In contrast to the interpretivist approach, a realist case study takes more of an inductive approach to
case study research. Yin [47] provides a realist point of viewon case study, arguing that the difficulty
of case study research lies in there being many more variables of interest than data points. Rather
than focusing on the interpretation and experience of the phenomenon through the participants,
realist case studies focus on distilling the holistic and meaningful features of the phenomenon itself.
While subtle, there is a distinct difference in the goals of these methods. Interpretivist case study
focuses on the participants’ interpretations of the phenomenon, whereas realist case study focuses
on the phenomenon, using participants’ interpretations (among other data) to describe it. In a realist
case study research design, there are five components [47] that shape the study: research questions,
propositions, units of analysis, logic models, and rival explanations.
First, research questions are targeted by using the literature to narrow the topic of interest,
identifying interesting questions stemming from that literature, articulating some potential questions
of one’s own, and sharpening and supporting those questions with more literature on the same or
similar topics. Second, propositions are ideas for how the relationships of interest within the
research question come about [47, 49]. Propositions help create ideas for where to generate data,
and what components of the case are most interesting for research.
21
Third, units of analysis compose the case [47]. Units of analysis refer to the data sources as
characterized by how they are grouped in the research design. For example, in an interview study,
one could frame each conversational turn as holding valuable information, whereas someone else
might argue that it would be more valuable to analyze the interview in larger chunks based on the
interview questions. Both approaches may hold merit in different ways, but in one case the unit of
analysis is a conversational turn, whereas in the other case the unit of analysis is a larger episodic
chunk. The research questions and propositions help with deciding how to generate data from the
unit of analysis. Realist case study delineates between different structures for the unit of analysis.
Fourth, logic models [47] are what link the data to the propositions. They are a series of logical
steps that map analytic techniques onto the generated data. The logic model, which describes
how evidence will be constructed into claims, has a clear function during data analysis, but it can
also help when deciding how to generate and represent data in the first place. Hollweck [142]
provides a helpful interpretation of Yin’s [47] definition: “logic models...can help explain the
ultimate outcomes because the analysis technique consists of matching empirically observed events
to theoretically predicted events” [142]. Fifth and last, rival explanations are also important to realist
case study because they represent how others might conceive of the phenomenon of interest. When
building claims in a realist case study, one is often compiling a set of the most important variables
into an explanation of the phenomenon. By anticipating and addressing alternative groupings of
variables or alternative explanations, one can strengthen the findings of a realist case study.
Due to its focus on identifying relevant variables, realist case study can often employ techniques
like experimentation, correlation, statistical analysis, and quantitative and/or mixed methodology.
This is popular in science and sociology [47, 56, 139]. For my dissertation, which comprises completely
qualitative case studies, a realist stance does not make total sense, but I do use components
of realist case study in Chapter 7, taking precedent from prior combinations of interpretivist and
realist case study [143, 144, 145]. Specifically, I used propositions (realist) to shape the research
design around theory, and I used students’ perspectives as data (interpretivist)—these choices enable
a study that centers students’ perspectives and their role in expanding a theoretical framework.
22
This combination is described further in Chapter 7, but in short, the advantage of this approach is
that I can retain a focus on students’ perspectives (interpretivist) while applying structural features
of a realist case study that help organize assertions around a theoretical framework.
3.2.3 Comparative case study
Comparative case study focuses on the people, situations, and events that comprise a context and on
processes by which those objects interact [53, 54]. Though I did not use comparative case study in
any of the chapters of this dissertation, it is worth reviewing to compare it to interpretivist and realist
traditions and to give further context for what research design choices one makes when choosing to
employ a case study methodology. Comparative case study is a relatively recent development (last
ten years [53, 54]) in case study methodologies, and it provides some attention to aspects of a case
that interpretivist and realist case study do not attend to [53]. In this section on comparative case
study, we focus on the features of this methodology that contrast with interpretivist and realist.
One idea unique to comparative case study is that the research design is constructed along three
axes across which the context spans: horizontal (meaning across similar cases), vertical (meaning at
varying scales), and transversal (meaning back through history) [53]. By looking across these axes,
the researcher can gain insight from comparing across cases (e.g., two different schools), strengthen
findings by tracing them across different scales (e.g., classroom, institution, national curriculum
standards), and situate the work in the context’s history (e.g., school’s history). By expanding the
scope of research like this, comparative case study is resistant to bounding the sources from which
data will be generated [53]. This perspective on boundedness helps me point out that my case
studies, which draw more from primarily interpretivist traditions, are bounded in the sense that
they draw from a limited pool of data sources by design. Each approach has its merits: unbounded
research allows the researcher to examine everything that could be important for explaining the
phenomenon, whereas in bounded research, the researcher examines fewer data sources and so can
perform a more in-depth analysis of the phenomenon.
Comparison is the defining feature of comparative case study [53]. Comparison is what brings
23
significance to the research by addressing how insights drawn from different cases are related. This
relationship is what is most strongly applicable to other, unstudied cases. In the context of case
study, comparison can mean one of two different processes. Homologous comparison is across
similar (grain-size) sites [53]. Heterologous comparison looks at similar phenomena or policies
across different kinds of entities in terms of scope (e.g., looking at how bankruptcy affects a small
business versus a family) [53]. The focus on comparison contrasts with my non-comparative use of
interpretivist and realist case study. Comparative case study argues that insights mainly come from
comparing across cases and working outward to new data sources. I instead chose to investigate
with a focus on producing an in-depth look at the local, observable interactions that occurred within
each of my cases.
To be clear, my prioritization of depth did not prevent me from examining contextual factors or
new data sources. An alternative framing for this dissertation could be that Chapter 5 represents an
unbounding and re-exploration of Chapter 4, and Chapter 7 represents somewhat of a heterologous
comparison looking at how the CT dispositions framework operates when applied to a context
different from the one in which it was developed. The reality of my research designs was that I was
interested primarily in using students’ perspectives, and interpretivist case study made the most
sense for addressing that priority. Incidentally, realist case study made some sense for parts of the
design of Chapter 7, as described in Section 3.2.2.
3.3 Use of case study
Case study can be used for a variety of purposes and via different traditions as described above.
The chapters of this dissertation also use some of those purposes and traditions. It remains to
characterize the process of carrying out a case study. A key feature in carrying out a case study
is using the context and the research questions as a flexible guide rather than “an ideological
commitment to be followed whatever the circumstances” [47]. You spend a lot of time with the
generated data, writing narratives about it, attempting to explain what you are seeing, and getting
to the point where you know the data intimately. Often after researching a specific phenomenon
24
and playing around with the data, it can become clear that the research question could be better
explored by widening the angle of the research (see for examples, Chapters 5 and 6). This can turn a
straightforward research question into something new and interesting by exploring the phenomenon
from a new angle. This change in angle comes about from greater knowledge about a context, and
as a response, an informed shift in the approach. This shift in approach is a hallmark of case
study research which is also part of what makes it so powerful [50, 51, 53]. By allowing for
flexibility in research design—for example, by generating new data or applying theories to data
iteratively—one produces research that is especially well tailored to the phenomenon of interest.
Adding this strength to the depth that case studies provide makes qualitative case study an ideal
tool for researching misunderstood or under-understood phenomena.
Another language-based feature of the following chapters is that I use the term“data generation”
rather than data collection [48, 146] (e.g, Prasad [147], Bannerman [148]). This is a choice to
acknowledge the role of the researcher. I designed the studies, chose where to place the cameras and
microphones, chose which students to interview based on field notes that I took, created questions
for interview protocols, carried out the interviews, and created and/or edited the transcripts. The
idea of generating data implies someone is creating it, which is largely true—when I analyze
data, it is data that I played a role in creating. Without my intervention, much of the data (e.g.,
interviews, field notes) would not have existed. It would not be totally accurate to call this process
“data collection,” as if the data were already there and I simply swooped in to extract it, as if
the perspectives of human participants could be essentialized like that. Doing so would claim an
objectivity in this research that simply is not there [46, 50, 51, 53, 54].
There are also limitations of case studies. As Yin put it, case studies do not unearth causality,
but they are still useful in explaining how and why a relationship exists [47]. They do not provide
causal or correlative results, nor can be generalized to a broad audience. That is what quantitative
studies are for. Case studies are for studying interactions that exist within a phenomenon in which
humans participate. It is towards these interactions and participants that I attempt to orient the
following chapters.
25
CHAPTER 4
LEARNING ASSISTANTS AS CONSTRUCTORS OF FEEDBACK: HOW ARE THEY
IMPACTED?
This chapter represents my first exploration of students’ perspectives in a research context, which
was a flipped, introductory physics course at Michigan State University. Though the course had
computational activities integrated into it, this was not the focus ofmy research at time. Instead, this
was an initial exposure to a model of listening to students to gain insight into how a course impacts
its participants. This study was geared more towards the impact on a single undergraduate learning
assistant rather than a group of students, but it fostered in me an interest in qualitative case study
and students’ perspectives, which I pursued with a more in-depth study in Chapter 5. A version of
this chapter was published with second author Paul W. Irving and third author Daryl McPadden in
the proceedings of the 2018 Physics Education Research Conference [60]. My contributions were
research design, data generation, analysis, and writing.
4.1 Introduction
It is not a new idea that physics Learning Assistants (LAs) are impacted significantly by their
experience in the LA Program. The experience can be transformative with respect to their identities
as physicists [99], how they teach physics [149], and even their metacognitive development [150].
At Michigan State University, LAs are employed to work in the environment of Projects and
Practices in Physics (P-Cubed), a flipped section of introductory, calculus-based physics, which
is designed with a problem-based learning approach where students work in groups on complex
physics problems [63]. The LAs work ten hours per week and fulfill three duties: (1) Each LA
functions as a primary instructor for four to eight students by asking questions and prompting
discussion during class. (2) LAs meet twice weekly to prepare for teaching and once weekly to
debrief and reflect on how the week went. (3) LAs write personalized feedback for each of their
students on a weekly basis. This last expectation for LAs is uncommon at other universities, for
26
we could not find any published research mentioning such a requirement for LAs. The feedback
LAs provide is intended to be formative by giving students guidance for improving their scientific
practices within their in-class group work [151].
The intention behind individualized weekly feedback is to offer suggestions to the student for
how to improve their practices and provide the student with a justification for their in-class grade,
which comprises 20 percent of the total grade for the course. To this end, feedback is split up in two
parts to address both how the group performed and how the student performed within the group.
In each part, the LAs are to include something the student or group did well, something to work on
for next week, and a strategy for how to work on it.
There is precedent from existing research [99, 149, 150] to look at how the LA experience as a
whole impacts LAs, but reducing the grainsize to look at specific aspects of the LA experience is
much more rare. This presents us with an opportunity to look at the LA experience and report out
on it in a new way. This is especially interesting since the piece of the experience at which we are
looking—feedback construction—is distinctly a part of being an LA in P-Cubed.
As mentioned earlier, the feedback in P-Cubed exists to help the students improve their scientific
practices, and we postulate this impact extends in some way to the LAs who practice constructing
the feedback. The undergraduate LAs hired for P-Cubed were all once students in the class, so we
also intend to piece apart how receiving feedback plays into the impact of constructing it. With this
mind, we pose three research questions: (1) How does constructing feedback affect decisions LAs
make outside of P-Cubed? (2) How does receiving feedback as students affect LAs’s approaches to
constructing it? (3) How are the impacts of constructing and receiving feedback connected for LAs?
4.2 Methods
We selected three LAs to each participate in a recorded, semi-structured interview, which was
intended to probe at how the LA approaches feedback construction and how their experience as an
LA and as a student in P-Cubed might have impacted other areas of their lives (e.g., study habits,
working in groups in the workforce). We selected LAs to portray a broad range of approaches to
27
feedback. Alvin is in his second semester of being an LA. He is a sophomore physics major. Bella
is recent graduate who has been in the workforce for a couple months. She was an LA for three
semesters, and she studied biochemistry. Carly is in her fourth semester of being a P-Cubed LA,
and she is a junior majoring in biosystems engineering. All the LAs we interviewed are white.
We constructed [48] a pilot interview protocol meant to dig into three things: (1)We wanted to
learn about each LA as a student in other classes, so that they could reflect on experiences they have
had in contexts outside of P-Cubed—contexts in which feedback construction might have made an
impact. (2) We wanted to learn about how each LA interacted with the feedback when they were
a student in P-Cubed, because we thought that experience would play a big role in how the LA
interacts with feedback construction. (3)We wanted to find out how each LA approaches feedback
construction itself, because that experience is central to the impact feedback construction has.
The first two interviews were with Alvin and Bella. The protocols used were similar, the only
difference being some questions were rephrased to give Bella the opportunity to speak about her
experience in the workforce. The protocol was modified significantly for Carly’s interview, with
potential follow-up questions listed and pauses built in based on preliminary analysis and reflections
on the first two interviews. The goal of the modifications was to develop a more comprehensive
view of Carly’s experience with feedback construction than we were able to develop for Alvin or
Bella. These interviews are the first three among a larger ongoing investigation, for we intend to
interview additional LAs in the future to broaden the insight we make from our interviews with
Alvin, Bella, and Carly.
In our analysis, we focus heavily on Carly’s interview. The reason for this is hers has the
richest data. This reality is owed partially to her willingness to reflect deeply on her multi-year
LA experience, but also to the iterative development of the interview protocol, as described above.
Carly, as the third participant, was given the best opportunity to express how the feedback impacted
her, and more importantly she was prompted explicitly to think about and discuss ideas related to
this impact—Alvin and Bella were not asked to discuss their experiences in the same way. However,
Alvin and Bella did reveal enough to make us believe that there exist themes traceable between
28
LAs, and perhaps the feedback has impacted Alvin and Bella in personally meaningful and lasting
ways, even though they do not articulate the impact in the same way that Carly does.
We decide to align our research with case study [152, 47], the purpose being to generate theory
around how Carly interacts with her feedback practice. This choice was motivated by the richness
of her interview data and the limited theoretical background on the relationship between writers
and their written feedback. Also, the boundaries between the feedback’s impact and the impact
of the rest of Carly’s LA experience are not always clearly evident. Carly was chosen as a case
from which to investigate the effects of giving feedback on an LA and how those effects occur. We
specifically use logic models [152, 47] from the realist case study tradition to construct a theory of
how Carly’s experiences with feedback interact with one another. However, the main methodology
used was interpretivist case study [50], for this research is an investigation on how Carly perceived
her experiences and how those perceptions link together.
4.3 Results and Discussion
Alvin clarifies the feedback’s impact on his own life when he is asked to reflect on what feedback
construction means to him:
“Me thoroughly contemplating what advice to give somebody, is also me...really thinking
about good things to do...when I am in a group in the future and I have a similar
situation... So if I tell a student of mine that this is a good way to improve when you
have this sort of situation in your group, then, me having thought about that, and how to
write feedback, will help me in the future when I am in a similar situation in a different
group” – Alvin
Alvin’s quote demonstrates a theme that showed in Carly’s interview, too, which is that feedback
construction helps LAs think about their own group work outside of P-Cubed and respond in
thoughtful ways when difficulties arise. We explore this phenomenon in more detail in the following
case study of Carly. Her perspective on feedback is built out of a multitude of experiences, and she
29
is able to articulate this perspective clearly. As we will demonstrate, her own experience mirrors
Alvin’s reflection in the quote above, which suggests that the impact the feedback has had on Carly
is not anomalous.
4.3.1 How constructing feedback affects decisions Carly makes in other contexts
Carly’s take on how feedback construction plays into other areas of her life echoes Alvin’s:
“Writing feedback, it’s easy to look at a group of people working together and be very
objective about how everyone is behaving within that group. But then when you’re in
a group—to then be able to step back and reflect on your own group and how you’re
acting in that group—I think that’s what...I’ve taken away...If I’m in a group and I’m
getting frustrated, then it’s like, ‘Okay, what would I tell someone to be doing in this
situation?”’ – Carly
She highlights a strategy of relating her difficulties with group work to the same sorts of issues
that might come up for a group in P-Cubed. She also makes an important point that writing feedback
is not an isolated exercise—it involves observing behavior in class as it plays out. To understand
how Carly enacts this strategy in real life, we asked her to give an example, and we believe the
recollection she produced speaks richly to what the feedback’s impact can look like for an LA.
As part of a group, she came across a dilemma (which we will refer to as The Dilemma), and
she believes her reaction derives from her experience constructing feedback. We retell how The
Dilemma played out and use evidence from what Carly says to show that the feedback impacted
her response in the ways that she claims. First, it is helpful to know how Carly sets it up:
“I’m in a design group right now. Three of us get along really well. One guy is
very inconsistent as to when he’s there, but he puts a lot of work in, but it makes it
challenging because we’ll have done something and he will have missed all of it for no
apparent reason... He’ll come to the next meeting and be like, ‘Oh, look at everything
that I’ve done’. And we’re like, ‘Well we calculated that already, and we assumed
30
these numbers, and you assumed these [other] numbers...so this is what we’re gonna
go with.’ But, you have to be very tactful in how you say that.” – Carly
Carly’s dilemma is that one member of her group went off and did a lot of unnecessary work,
and now the group needs to figure out a way to tactfully bring him back into the fold. Carly admits,
“I tend to...kind of be the leader of the group”, and the decision is hers to make. Carly ended up
making the decision for the group to sit down for a couple hours, step through the calculations
everyone had done, and come to a consensus together about what approach the group should use
moving forward. The result was beneficial to the group—the group adopted some ideas that the
fourth group member had found while doing his own calculations, and maybe more importantly,
he was brought back into the group without feeling devalued.
Next, we intend to unpack the forces that played into Carly’s thought process in the way she
described it. The way she outlines her ability to “step back and reflect” when speaking generally
about The Dilemma in her first quote shows that she traces her tactful response to writing feedback
and helping P-Cubed students in class—this is represented with the relationship 1 ) 2 in Figure 4.1.
After recounting how the group responded to The Dilemma, Carly relates it back:
“As an LA I would never want to see someone’s work just completely dismissed. If
they...roll through a bunch of stuff, [and] someone [else] was just like, ’What are you
doing?...We’re doing it this way’... I would...have to find something...and validate both
sides.” – Carly
Carly relates the tact of her approach to how she might address a group of students in P-Cubed.
The Dilemma is relevant to her work in P-Cubed, for in both cases she wants students to feel that
their work is valued. It would be unrealistic to say feedback construction is the only aspect of the
LA experience Carly considered when formulating a respond to The Dilemma, so it is no surprise
that Carly includes her in-class work in this discussion. This quote is valuable in demonstrating that
Carly sees connections between her LA experience and how she conducts herself in other classes.
31
As we will show next, the way she approaches her LA duties is also strongly connected to the
feedback she received when she was a student in P-Cubed.
4.3.2 How receiving feedback as a student affects Carly’s approach to constructing it
Carly was student in P-Cubed two years ago, and its impression on her was indelible. In sorting
out what influenced her reaction to The Dilemma, we were hoping to separate her LA experience
from her student experience, but consistently Carly would bring up one when discussing the other.
It would be unfair at this point to say that for her they are not intertwined. One way to see the
connection is by comparing her description of what the feedback should look like with what it
looked like when she was a student. When Carly constructs feedback, she has a format in mind:
“The basic format [is]: highlight a positive, highlight something to work on, explain
why this will be beneficial to them, and maybe end on a positive if it works into your
feedback.” – Carly
She mentions three pieces: A highlight of what went well, a highlight of what to improve, and
some reasoning. At other points in the interview, Carly explains that the reasoning is a justification
for how the group will benefit from the improvement, and she sometimes includes an outline of
steps that can be taken to achieve the improvement. When she talks about helpful feedback she
received as a student, it matches up with the format she described:
“There was one week where the positive was essentially like, ‘You do a good job of
facilitating discussion within the group and asking people to pause and clarify what
they’re saying’...but then the follow-up was, ‘Sometimes though, you save questions for
me as the instructor when you could be asking these questions to your group, because
then that also can prompt discussion.’... For me then it was like, ‘...Okay, I can see this
thing that I’ve been doing well with, and this is a way for me to continue to improve
that. I’ve been facilitating discussion but now, like, I didn’t realize that I had been
32
saving questions just for the instructor, like I can now present those to my group as
well.”’ – Carly
All the pieces are there: a positive, a suggestion for improvement, and a justification. The
pieces of feedback she found valuable as a student are the same pieces she tries to emulate in her
own feedback. Further, all P-Cubed LAs are formally trained on how to give feedback, and the
format from the training has a slightly different structure: Carly seems to have appropriated it to
more closely match the feedback she received when she was a student. Seeing the parallels between
the feedback she received and how she arranges feedback in her mind today, we believe that Carly’s
experience receiving feedback shapes how she structures it today, which is represented with the
relationship 0 ) 1 in Figure 4.1. For Carly, there is still one more layer to the couplings described
between her many feedback experiences, which we will outline in the next section.
Figure 4.1: This shows direct and indirect influences of Carly’s experiences with feedback in
P-Cubed, as referenced in the discussion. The relationship 1 ) 2 is the focus of Section 4.3.1.
Section 4.3.2 focuses on the relationship 0 ) 1. Section 4.3.3 demonstrates the relationship
between the two influences, and argues that the first influence strengthens the second. The influences
and connections themselves are detailed in the discussion.
4.3.3 How the impacts of constructing and receiving feedback are connected for Carly
Carly’s reflections indicate that the practice of constructing feedback influenced how she chose to
respond to The Dilemma, as outlined in Section 4.3.1. Also, her experience receiving feedback
33
as a P-Cubed student influenced the way she goes about constructing it (Section 4.3.2). In this
section we will discuss the similarities in how she describes these two impacts, which make us think
that the significance of receiving feedback is twofold: (1) It helped Carly develop her practice of
constructing feedback (0 ) 1 in Figure 4.1), which we outlined above. (2) The ways she describes
the two influences are so in line with each other that we believe that the first influence (0 ) 1 in
Figure 4.1) played a role in facilitating the second (1 ) 2 in Figure 4.1). Perhaps the process of
pulling from her experience as a P-Cubed student to develop strategies as an LA is a practice that
Carly was able to refine and reuse to pull from her experience constructing feedback to develop
a strategy to solve The Dilemma. This relationship between the practices is represented with the
curved arrow in Figure 4.1.
The connection between the two impacts is best displayed by starting with a piece of Carly’s
first quote:
“To step back and reflect on your own group and how you’re acting in that group—I
think that’s what...I’ve taken away... If I’m in a group and I’m getting frustrated, then
it’s like, ‘Okay, what would I tell someone to be doing in this situation?”’ – Carly
We compare Carly’s explanation of how she reflects on feedback construction with a separate
quote on how she decides on what to say to her students as their LA:
“Part of [constructing feedback] is drawing on, ‘Okay, what was I feeling in class at
that point, what was I struggling with?’ ...having been a student prior to being an LA
for this class is really helpful... I think it just gives you a better understanding of the
students themselves.” – Carly
In both quotes, Carly is pulling from her past experience to recall how to solve a group-related
issue, applying a learned lesson to the situation at hand. In the first quote, she imagines herself
inspecting the group, constructing advice to help them overcome The Dilemma. In the second
quote, she imagines herself as the struggling student, remembering what feedback she heard in the
34
past that helped her overcome a similar difficulty. The reflection processes in each quote imitate
each other down to the questions Carly asks herself, which exhibits that they are in some way the
same process practiced twice.
4.4 Conclusions and Future Work
We now circle back to our three research questions listed at the end of Section 4.1. Our findings in
relation to those questions are as follows: (1) Constructing feedback helps Carly think critically and
make better decisions when faced with group-related difficulties in contexts outside of P-Cubed.
(2) Receiving feedback as a P-Cubed student was an experience that shaped how Carly thinks about
and constructs feedback today. (3) The process of pulling from old feedback to help her think about
how to construct it [finding (2)] is a process that Carly has practiced and refined in pulling from
constructing feedback to help her respond to dilemmas outside the context of P-Cubed [finding
(1)]—the third finding is the connection itself.
This research highlights the positive impact, in the context of P-Cubed, of hiring LAs who have
experienced the exact class they will be teaching—Carly alludes to this herself: “having been a
student prior to being an LA...is really helpful... I think it just gives you a better understanding of
the students themselves.” This might seem like an obvious conclusion but this matching of LAs
with their prior coursework is not always the case at institutions running physics LA programs [96,
97, 98]. The degree of this positive impact could be investigated further by interviewing LAs who
have taught in P-Cubed but not taken it. Currently this would describe one student. Alternatively,
this finding could highlight the need to investigate further how important it is for LAs to have had
prior experience in the same learning environment, especially when it is a transformed classroom
with a lot of innovations.
We acknowledge that we only fully represent one LA’s perspective in this chapter, but the insight
we were able to make into how Carly has interacted with the feedback as a P-Cubed student and
as an LA makes us optimistic for the investigation that will build from this work. We expect to
conduct and analyze more interviews with LAs. A preliminary analysis has been completed on one
35
such interview, and we believe that it will showcase interesting features of the feedback’s impact in
the same rich, personal way that Carly’s interview did.
36
CHAPTER 5
LEARNING ASSISTANTS AS STUDENT-PARTNERS IN INTRODUCTORY PHYSICS
This chapter represents an expansion of the research study in Chapter 4, with more learning assistants
(LAs) in my data sources and considerable attention to the theory of communities of
practice [62] and the model of students as partners [28]. This chapter builds on Chapter 4 incorporating
more LAs into the research design, situating their perspectives within the community of
LAs by attending closely to context and including the perspective of a faculty member who worked
closely with the LAs in the course. This chapter also represents a development of my understanding
of how research on students’ perspectives can unfold, understanding that I can carry with me to
subsequent research study designs. A version of this chapter was accepted for publication with
second author Paul W. Irving and third author Daryl McPadden in Physical Review Physics Education
Research [31]. Mycontributionswere research design, data generation, analysis, and writing.
5.1 Introduction and Background
The curriculum design process traditionally exists in the hands of expert instructors and practitioners,
outside the influence of students themselves. Recently there has been a push to acknowledge
and encourage the development of partnerships between students and instructors, where students
are consulted, for example, on improving teaching practice and curricular materials. The recent
focus of education research on Students as Partners (SaP) is marked by the launching of a journal
dedicated to the topic as recently as 2017 [153]. SaP has been conceptualized by Healey et al. [28]
as a “partnership learning community” that can take on four different, overlapping forms: (1)
students facilitating the learning, teaching, and assessment, (2) students conducting subject-based
research and inquiry, (3) students consulting on curriculum design and pedagogy, and (4) students
learning about and enhancing the quality of teaching and learning. This study will focus on a
specific case of the third type of partnership—namely how a community of practice of Learning
37
Assistants (LAs) in an introductory physics course led to a sustainable student-partnership that
influenced course structures.
Typically, the work of conceptualizing partnership learning communities involves comparing
them to the Communities of Practice (CoP) framework [62]. For instance, Healey et al. [28]
contrasted and aligned SaP with CoP. They argued that like in CoP, student-partnerships are
composed of apprenticeship-like relationships where newcomers (i.e., students) learn from oldtimers
(e.g., faculty) by engaging in practice together. Both models emphasize a shared enterprise
or goal that members of the partnership or community work towards together. The partnership also
involves a learning trajectory much like in a CoP, where the student learns what goes into teaching
(or other practice) behind the scenes, and comes to make their own contributions to the practice as
they gain expertise and offer their own input.
That said, there is a significant difference between CoP and SaP in howmembers are recruited. In
CoP, newmembers of a community join by forging relationships with existing members and aligning
their participation with the goals of the community. In a partnership learning community, “it may not
be enough to simply extend invitations for new partners to become part of existing communities.
In these new communities all parties actively participate in the development and direction of
partnership learning and working and are fully valued for the contributions they make” [28].
The contrast between who is responsible for facilitating the development of relationships in the
community indicates that faculty who wish to use student-partners must be willing to work hard at
forging partnerships.
Effort towards student-partnerships must also be focused. Matthews [37] highlights three tenets
that make a good student-partnership. First, the relationship between teacher and student must be
reciprocal, meaning all members of a partnership must give input and have their input valued. Sohr
et al. [29] warned against a troubling pattern where student voices get tokenized without being
incorporated, so that an institution might posture at having student-partnerships in place. Second,
the goals of a student-partnership must be good morally, meaning all parties benefit: faculty,
students in the partnership, and other parties impacted by the partnership. Third, the outcome must
38
strive for broad (beyond individualistic) improvement. For example, a partnership that changes the
structure of a course in a sustainable way would achieve this outcome, whereas a temporary impact
on a handful of students would not.
With these tenets in mind, SaP can still take many forms with varying levels of student involvement.
To illustrate the variety, we have adopted a visualization of the “participation ladder”
from Bovill and Bulley [8], as seen in Figure 5.1. The ladder was originally used to describe the
relationship between students and tutors in an active learning environment, but has been repurposed
[28] to describe the strength of impact in partnerships that are focused on curriculum design
and pedagogy. The bottom rung of the ladder represents a curriculum that is completely in the
control of the faculty member with no input from students. On the opposite end, the top rung of the
ladder represents students in complete control of the curriculum. The rungs in between represent
the different levels of balance between student and faculty control in the course.
The SaP literature reflects this variation. Bovill [154] emphasized that although authentic
student-staff partnerships are usually on the higher rungs of the ladder, “co-creation is not about
giving students complete control, nor is it about staff maintaining complete control over curriculum
design decisions.” She argued for reciprocal roles between faculty and students. As an example,
she described a course where the tutor/faculty guided the students in designing their evaluation
exercise for the course, gathering feedback, and compiling recommendations for the course from
the students themselves. In this case, the student-control over the evaluation process places this
partnership on the top two rungs. Flint and O’Hara [155] described a body of student representatives
who sit on university governing committees that incorporate the voice of students into institutional
decision-making. Because of the history of newer student members outnumbering other students in
the governing body, the students tended to have limited influence on what the committees oversaw,
placing this partnership on the third or fourth rung from the top. Sohr et al. [29] described students
who were recruited and interviewed to help redesign a quantum mechanics class. Due to the tutor
facilitation of meetings where students gave input, and the tutor-led synthesizing of feedback, this
partnership would likely exist on the fifth or sixth rung of the ladder, based on the iteration described
39
Figure 5.1: LA participation ladder: a visualization of the different levels of influence that LAs
can have on curriculum design and pedagogy. Borrowed from Bovill and Bulley [8] to conceptualize
student-faculty relationships instead of student-tutor relationships. The arrow represents
how student participation changes with the nature of the partnership. To be clear, this is not a
recommendation for classrooms to strive for the highest rung, rather a model for characterizing
partnerships based on the roles of students.
40
in the research. Mercer-Mapstone et al. [32] reviewed and analyzed 58 papers on SaP to show
that most of these partnerships focused on changing curricular materials in a course or altering
teaching strategy. These partnerships were often forged informally by professors who wished to
incorporate student perspectives but did not have the means to do so outside of asking students to
meet with them and provide their input. From these examples, the reality of researched partnerships
is that they tend to center students who do not teach and who operate within small-scale, unpaid
partnerships.
An alternate model for utilizing students in the classroom is the LA model. Over the last two
decades, LA programs have become an ever-growing feature in undergraduate programs across
the country, particularly in STEM disciplines. Initially conceived at the University of Colorado
Boulder in 2003 [96], LA programs have since spread widely in varying forms [96, 97, 98, 99, 156].
The premise behind an LA program is to hire undergraduate students as LAs to facilitate learning
in the classroom. The common goals of LA programs revolve around improving undergraduate
courses, helping LAs improve their teaching practices, and recruiting undergraduates into the
teaching profession. While these goals are distinct from those of student-partnerships, there are
some overlaps, especially in trying to improve undergraduate courses. In most of the LA programs
that have been discussed in prior research, LAs fulfill three duties: teaching students in a class,
attending meetings for class preparation and planning, and taking a pedagogy course or teaching
seminar with other LAs [34, 100, 101, 96, 98, 99, 156]. Within physics, many active learning
curricula, including SCALE-UP [157], University Modeling Instruction [158], Interactive Science
Learning Environment (ISLE) [159], and Washington Tutorials [160], have been implemented
in conjunction with LA programs to facilitate small group discussions and learning at the scale
required for university courses.
Previous research on LAs tends to focus on the benefit LAs bring to student engagement and
learning rather than to structural components of the course. For example, the presence of LAs in
STEM courses has been shown to improve student learning gains on conceptual inventories [96].
The same study demonstrated that LAs also improved students’ attitudinal gains compared to
41
non-LA courses, and they increased the instructors’ attention to student learning while planning
for class. An extensive study [102] on instructor effectiveness—a quantitative measurement of
collective student learning ascribed to instructors over multiple semesters—found that LAs helped
instructors maintain their effectiveness, whereas, without LAs, the instruction declined from year
to year (even when controlling for flipped- vs. lecture-style and previous teaching experience).
Several studies have confirmed the benefit that LAs have on the grades and passing rates in student
performance within the same courses [161, 162, 98], and in some cases especially for students
from underrepresented backgrounds [163]. Thus, there are strong motivations from research to
include LAs in the classroom, from improving student learning outcomes to improving teaching
effectiveness.
Despite the growing presence of SaP in education research and the overlap of goals with
LA programs, there is little research done on student-partnerships that involve LAs (LA-faculty
partnerships). Jardine [30] researched LA-faculty relationships in undergraduate biology courses,
where LAs were asked for feedback on course structures by the faculty members during meetings.
The goals of these questions were to redirect how course materials were drafted up and how exams
are graded, all ultimately at the discretion of the faculty. This reduced the amount of influence
held by LAs since the only mechanism for change was filtered through the faculty members. Other
studies highlighted LA-faculty relationships but did not analyze with a partnership lens. Sabella
et al. [34] demonstrated how LAs were used in a physics course to shape and improve curricular
materials on a semester-by-semester basis, but they did not use the SaP framework or extend to
broad, sustainable change to the course as outlined by Matthews [37]. Other research on the
impact of the LA experience on the LAs themselves [100, 101, 164, 165] highlights how LAs
grow as people during the experience. For example, Close et al. [99] described how these LAs’
identities shifted during their time as LAs using the CoP framework. However, that work focused
on the individual LAs and did not detail the impact of the LA-faculty partnerships on the course
structure. In some studies [164, 165], the impact was described on the basis of how the faculty
member benefited from working with LAs, but again the influence did not seem to extend beyond
42
the individuals directly involved.
This previous research highlights a disconnect between CoP, SaP, and LA programs. However,
it also offers the opportunity to reimagine LA programs at that intersection. We propose that
LAs in P-Cubed effectively occupy the “student” role in SaP, taking precedent from Jardine [30].
Undergraduate LAs have expertise as both teachers and experienced students, so why not leverage
this expertise to improve the underlying structure and teaching philosophy of future offerings of
the course? Through this study, we intend to contribute to this idea by conceptualizing a specific
LA program as a model for enacting SaP using the CoP framework. In the following subsection,
we outline the course context for the specific LA program.
5.1.1 P-Cubed
In examining an LA program with a student-partnership lens, we focus this study on one group
of LAs at Michigan State University (MSU) who work in a flipped, introductory, calculus-based
mechanics course offered in the Physics and Astronomy Department called Projects and Practices
in Physics (P-Cubed) [63]. Almost all LAs in P-Cubed have applied and been hired to the teaching
staff after taking the course as a student. Though they are not P-Cubed students anymore, they are
undergraduates, and they have a proximity to the P-Cubed student experience that other instructors
do not. LAs for the course have three primary obligations: (1) facilitate in-class group work and
problem-solving by acting as a tutor who guides but does not give answers, (2) attend pre- and
post-class meetings to prepare for class and reflect on how it went, and (3) write individualized,
weekly feedback aimed at improving students’ scientific practices. The third duty listed is different
from many other LA programs that research has covered in the past, which often do not have an
individualized written feedback component.
The problem-based design of P-Cubed means that students review material outside class and
solve small-scale problems for homework. When they come to class, students are arranged in groups
of four or five, and together they solve a single open-ended physics project that takes two hours to
work through. Projects are designed to give students exposure to scientific practices [24, 63] like
43
developing models, analyzing data, or arguing with evidence. Projects might ask students to create
a physical model for a situation where a new physics concept is at play, and subsequent parts of the
project might add complexity or require computational modeling. For example, one of the projects
asks students to model relative motion of hovercrafts with the goal of pulling off a rescue mission
for the occupants of the “runaway hovercraft.” The project builds in complexity by introducing
freefall and adds a computational component by having the students create a computational model
of the chase and freefall in order to communicate their findings. The projects tend to focus on
analysis and computation as tools for investigating physical phenomena, but experimentation is not
part of the course.
Each group of students has an undergraduate LA or other instructor assigned to it (75-80%
of the instructional staff is LAs), with each instructor responsible for 2-3 groups of students.
The instructor’s role is to guide the problem-solving process and to encourage collaboration and
creativity, as there are many ways to solve each project. During class, the instructors also make
observations for each of their students, which then can be used to construct feedback each week.
This written feedback allows the group’s instructor to reflect outside of class, address any issues
from the class, and make suggestions for improvement in the next week. Instead of grading each
problem for correctness, students are graded on their approach, process, and collaboration with
their group. In a given week, LAs attend two pre-class meetings to prepare for each class period’s
physics project and one post-class meeting after both class periods to reflect on the week together
and trade advice on teaching and/or writing feedback.
The P-Cubed course has several features that make it an ideal context for our investigation. First
and foremost, the course was originally designed from a CoP perspective [63, 166]. When the
course was conceived, the developers decided to focus the goals of the course on developing certain
practices aligned with the communities of undergrad physicists and engineers, who were most likely
to represent the students in P-Cubed. To make these practices authentic, the designers intentionally
made the problems ill-structured and under-defined so that students would be forced to engage
in solving these problems using authentic means. This means students would have to negotiate
44
the meaning of the problem and tackle complex and intricate issues collaboratively, which in turn
facilitates engagement in multiple scientific practices. While it is unclear howto design a community
of practice from the ground up explicitly, it is possible to create structures and opportunities that
would allow a community of practice to develop [167] among students and instructors. Second,
as a part of the CoP design, LAs were intentionally positioned as intermediate members of the
community of practice by the course developers [166]. Because LAs are simultaneously positioned
as both peers and experts, they offer a pathway into the center of the P-Cubed community. LAs
are a living bridge between the student experience and the instructor experience, representing the
learning trajectory that members of a CoP can travel to achieve centrality. This trajectory was
designed to be a guided experience through the use of the P-Cubed feedback mechanism (a direct
link for students to learn from more central members of the community). When we examine LAs
using an SaP lens, we view LAs as occupying the “student” end of the partnership, for we intend
to show how LAs influence the course from positions of less power than faculty, taking precedent
from Jardine [30].
Our first goal then for this chapter is to demonstrate that a community of practice has indeed
formed among LAs in the P-Cubed course. Though the broadest community of practice in P-Cubed
would encompass all students and instructors, as Irving et al. [166] proposed, we focus mainly on
the community of LAs. To be clear, classrooms can develop sub-communities within the larger
community, such as a small group of students, or the LAs, or the entire teaching staff including
graduate TAs and faculty. Demonstrating existence is an important first step since a community
of practice (of LAs) is not a guaranteed outcome based solely on design decisions. From there,
we will show how LAs have directed and influenced one particular practice in the course—namely
constructing feedback. We choose to focus on feedback rather than other practices for three main
reasons: (1) The practice of feedback is outlined and described in detail in the course materials
which were written upon inception of the course. This makes it easier to discuss how the LA
practice of feedback has evolved over time and taken shape based on LA experiences combined
with the original course design. (2) Due to the regular nature of feedback-writing and the ease with
45
which LAs can share their written feedback among one another, it is a practice on which LAs tend
to advise one another more heavily compared to other practices. (3) Feedback is an opportunity
for LAs to infuse their own expertise into the course because LAs have the freedom to write about
what they deem important to succeeding in P-Cubed. They also get to decide what it means to
write good feedback when they advise one another on feedback-writing throughout the semester.
For these reasons, the practice of constructing feedback exemplifies the community aspect of LA
duties and the trajectory that practice can take when under the influence of central members of the
LA community of practice.
Our second goal is to show how the LA community of practice can be viewed as a studentpartnership
in the course. This is a unique perspective and adds to the SaP literature because of
the unique trajectory LAs take within the P-Cubed community and the opportunities they have to
infuse their expertise into the P-Cubed curriculum. Their trajectories are special because P-Cubed
LAs are recruited when they are students in the course, so they have past experience as P-Cubed
students yet hold roles as undergraduate instructors.
To that end, we aim to answer the following research questions in this chapter: (1) Has a
community of practice developed around LAs in P-Cubed? (2) How has the practice of feedback
been shaped by P-Cubed LAs? (3) How can the LAs’ influence be characterized as a studentpartnership,
and what characterizes this partnership and its outputs?
To answer these questions, we will first dive into the details of what comprises a community
of practice and how the LAs in P-Cubed could be viewed and analyzed from this perspective in
Section 5.2. We will then describe our case study approach to the LA community and how this
approach helped us select and analyze our data sources in Sections 5.3 and 5.4. In Section 5.5, we
will present the results of the case study and answer the research questions above. In Section 5.6,
we will discuss the implications of having an LA model that begins with a CoP design and has
developed into a community-based partnership.
46
5.2 Theoretical Framework
We use the theory of CoP [62] throughout this chapter to describe and analyze the community of
P-Cubed LAs. By definition, a community of practice is a group of people who share common goals
and work together using shared practices to achieve them. The goals are communally negotiated
and evolve over time. Practices are patterns of activity that have been agreed upon over time and
developed as cultural norms among the group. Learning in a community of practice happens when
a member comes to participate in ways aligned with the shared practices and shared goals of the
community. Historically the “center” of the community, or the goals and practices to which new
members align their participation, shifts as central members leave (reduce participation) and new
members join and negotiate their participation in relation to their own experiences and personal
histories. We intend to use this theory of participation and learning to demonstrate that the LAs
have formed a community of practice and how that community of LAs took up the specific practice
of feedback-writing, made it their own, and wielded influence over other aspects of the course in a
partnership-like way. In the following subsections, we introduce CoP as laid out by Wenger [62]
and then show how we conceptualize the design of P-Cubed as an environment that encourages the
development of a community of practice.
5.2.1 Communities of Practice
EtienneWenger developed the learning theory of CoP [62] as a follow-up to Jean Lave and Etienne
Wenger’s Situated Learning [168], which expanded on the apprenticeship model to address the idea
of learning as legitimate peripheral participation. In CoP, Wenger drew primarily from Situated
Learning but provided more details on what it means to learn in a community of practice and framed
learning in terms of a duality between practice and identity. For this study, we focus primarily on
Wenger’s conception of practice, which he viewed in terms of five mutually defining and deeply
connected features: negotiation of meaning, community, learning, boundary, and locality.
Negotiation of meaning takes place in the duality of two member-driven processes: participation
47
and reification. Members participate in practice by directly engaging with other members and
actively carrying out the goals of the community. This participation ties to how they reify, or
“[give] form to experience by producing objects that congeal this experience into ‘thingness’ ” (p.
58) [62]. We can see the interlocked nature of these two processes when considering how reification
is brought about by historical patterns of participation. For example, physicists often draw a free
body diagram to help visualize the forces at play in an introductory mechanics problem. The setup
of the diagram is not by nature a representation of forces. However, it is widely interpreted that the
simplified free body and the straight arrows represent forces because of how participation patterns
over time in introductory mechanics have reified forces on an object into a free body diagram.
As Wenger puts it, “what is said, represented, or otherwise brought into focus always assumes a
history of participation as a context for its interpretation. In turn, participation always organizes
itself around reification because it always involves artifacts, words, and concepts that allow it to
proceed” (p. 67) [62].
Community refers to the members themselves and their relationships with one another [62]. It
also refers to their relationship with the context of the shared practice. A community consists of
three dimensions: mutual engagement, joint enterprise, and shared repertoire. Mutual engagement
in a community is marked by the togetherness of the practice and the relationships that exist between
members. Meaning is negotiated between members, not on an individual basis. Joint enterprise
in a community exists because members have mutual accountability to one another in carrying out
the practice in a way that advances the cause of the community. The enterprise is joint in that it is
mutually constructed and agreed upon together. Shared repertoire is (1) the set of routines, tools,
words, actions, or concepts that the community has reified over time and (2) the ways through
which these resources become a part of how community members engage in practice. We connect
these three dimensions through their mutual involvement in how members negotiate meaning. For
example, on a volleyball team, members need to interact constantly (mutually engage) in order to
convey where the ball should be hit and who should prepare to return the ball to the other side.
There might be compromises between varying interpretations of goals (joint enterprise)—some
48
members want to have fun while others focus more on winning. Even among these goals, there are
varying interpretations of how to achieve them. The shared resources (repertoire) of the team can
help facilitate pursuits of the enterprise and the gameplay such as recognizable shouts of “mine!”
between players to signal intent, techniques for setting the ball in a desirable spot, or announcements
of the score before each serve.
Learning refers to a trajectory that involves aspects of both negotiation of meaning and community
[62]. Members of the community traverse the trajectory by participating and reifying as
described above. When members participate, they remember and forget aspects of the experience,
and their memories change over time to embody how they view the relevant practice. In the same
way, they reify artifacts when they participate, and these artifacts preserve the history of practices.
Because of the choices that are embedded in reification and the selectiveness of memories, members
come to view practices in new ways and they gain experience with doing practices in such ways.
The practices themselves can change too, as more central members develop new perceptions and
ways of doing things based on the choices packed into reification and memory. This process defines
learning. Newcomers to the community invariably must learn the practices, and they embark on
this trajectory by participating (forming selective memories) and reifying (preserving their participation).
Importantly, this is a community process, because newcomers would not know how to
participate without mutual engagement, and they would not know how to reify without a shared
repertoire that they can dip into. As they move closer to the center from the periphery, their practice
transitions from learning from others to learning where the practices of the community are headed
(and influencing their direction).
Boundary emphasizes the ways that participation and reification can connect communities and
create a sense of continuity between their practices [62]. Members of multiple communities can act
as brokers by translating elements of practice across contexts through their participation in those
communities. The focus lies with participation of brokers because of the active role brokers play in
understanding how practices are connected and introducing and facilitating modes of participation
from one community to another. An example of a broker is a high school track coach who draws
49
on her experiences networking with other coaches at track meets to teach running techniques to her
student-athletes, thereby relaying participation from her coaching community to her high school
track team community. Reification can also serve as a robust inter-community connection, which
Wenger described using the term boundary object. We often refer to instances of reification as
“objects”, but when they belong to multiple practices, “they are a nexus of perspectives and thus
carry the potential of becoming boundary objects if those perspectives need to be coordinated”
(pp. 107-108) [62]. For example, the act of making possible and enabling the production and
distribution of a band’s music could be reified in a recording studio, but that studio is also used
as a place of work for sound engineers and technicians. The studio is a boundary object because
these different communities (band members, technicians) use it to come together and coordinate
their perspectives and practices.
Locality refers to the size and scope of a community of practice, especially in relation to larger
communities to which it belongs or smaller communities that it comprises. Multiple communities of
practices can sometimes be viewed as a “constellation” [62] of communities. This is meant to evoke
the image of communities grouped together by some measure of proximity, common participants,
or patterns in practice. The proximity of communities in a constellation can be officially reified,
in the case of a unified league of sports teams, or unstated, in the case of informal groupings of
skaters that might show up at the skatepark at given time. In the more structured communities, one
can even identify sub-communities and overlapping communities that exist simultaneously within
a constellation. For example, in a soccer team, there is a community of all players, a community
of midfielders, a community of coaches, a community of defenders and defensive coaches, and a
community of the entire team. The list could go on given the myriad skills and gatherings that
are important to the team. Each sub-community evokes a different level of locality. Together they
form a cohesive community and its practices, practices that reflect and refract the practices of the
sub-communities.
50
5.2.2 The design behind the development of a community of practice in P-Cubed
In thinking about how the burgeoning community of LAs in P-Cubed could embody the features of
a community of practice, it is helpful to discuss how negotiation of meaning, community, learning,
boundary, and locality are designed into the course. For the purposes of this chapter, we will focus
our examples on the practice of writing feedback; however, we recognize that this process could
also occur for the other practices designed into the course. First, we will address the negotiation
of meaning, which, at its core, is the duality of participation and reification. When we think of
how LAs are meant to engage in participation, we envision discussions that are encouraged during
post-class meetings about how to address student behaviors in feedback, interactions with students
in the class, and the processing of LA-jotted in-class notes into the feedback itself. When we
think of how LAs are meant to engage in reification, we envision how they use the assessment
guide in deciding how to frame their feedback. Phrases like “group understanding” (the title of
an in-class assessment category) can be used to communicate to students and other LAs about
in-class observations. LAs also need to interpret whiteboard scrawlings during class to understand
where their students got stuck. The ways that LAs are meant to participate and reify in their
feedback-writing practice are necessarily interlocked, and these processes together are how LAs
negotiate meaning—without them there would be no point in writing the feedback, and its contents
would not be meaningful to the students if not based on in-class observations and LA-written notes.
When extended out to other practices that LAs could have influence on, we look at negotiation of
meaning from an SaP perspective, which puts lasting, structural impact into focus. This highlights
that structural influence often takes the shape of reification, a process that produces artifacts for
future community members to shape their own practice around.
The LA program in P-Cubed is also designed to form a community (the second feature of practice).
We envision mutual engagement among LAs in discussions between LAs on feedback-writing
during post-class meetings, relationship-building through shared coursework and studenthood, and
the helping-out that happens when LAs ask one another to review their written feedback. For
LAs who write feedback, the joint enterprise would be focused on the goal of helping students
51
improve their scientific practices and group work through the process of writing individualized
feedback for them. The reasons for LAs to review one another’s feedback would include not only
the relationships that exist between LAs but also their understanding that the improvement of any
LA’s feedback is an advancement of the joint enterprise. The shared repertoire of feedback-writing
includes in-class note-taking, the act of writing the feedback itself, the norms of interaction during
post-class meetings, and the shared historical experience of having been a student in P-Cubed.
Because the design of P-Cubed encourages the collective experience described, we aim to explore
how LAs can have a powerful collective impact within the potential student-partnership that we
will investigate.
Third, the learning trajectories of LAs and of the community itself ideally begin when future
LAs are students in the class. The P-Cubed feedback and in-class teaching practices are meant
to onboard students with collaborative skills and an understanding of what it takes to be an LA.
Students who are recruited into the LA program would be positioned as “newcomers” in the
community. Newcomers learn by aligning their participation with more senior members of the
community, which in the case of P-Cubed would mean that LAs would collaborate by consulting
one another on feedback and by asking one another for help teaching during class when issues arise.
As LAs gain teaching expertise and travel along their own learning trajectory, they would also start
to gain influence over the direction of the practice. Over time, practices would evolve similar to
how we would envision the impact of a student-partnership. Learning trajectories are where we
would look to in order to examine whether the SaP model can be applied to P-Cubed.
Fourth, the boundary of the P-Cubed LA community can be illuminated by its brokers and
boundary objects. LAs could act as brokers by drawing on outside experience to bolster their
teaching and feedback practice—some LAs would have received vivid and helpful feedback when
they were students in P-Cubed, which they might translate into constructing feedback today. Others
study physics or engineering in upper-level classes, from which they can draw to provide a more
in-depth perspective on some of the physics concepts for their students. An example of a boundary
object is old feedback that an LA received as a student in the past. When they first received it,
52
it served to help them improve their scientific practices and group work, which advanced their
learning as a P-Cubed student. Now, as a P-Cubed LA, they can repurpose the old feedback to
help them understand what could be helpful for a current student to hear. Even though the LA’s
involvements in each community are separated by time, the boundary object (old feedback) reified
past participation (e.g., group work) in a way that has allowed the LA to access and translate it for
use in the current practice of feedback construction. This exemplifies one of the ways LAs have
personal influence on P-Cubed practices, which could be viewed as an instance of LAs leveraging
their power in an LA-faculty partnership.
Lastly, the locality of practice is also relevant, though mostly for clarifying how we use language
in this investigation. To summarize its presence in the design of P-Cubed, locality refers to the
unit of analysis of the community of practice. We could have expanded the scope (locality) of our
investigation to focus on the entire class of LAs and students, or we could have narrowed to a small
group of LAs that meet for pre-class meetings together. We chose to view the LAs as a single
community or “unit” because this is what allows us to most easily discuss the varied perspectives of
LAs and how those perspectives trace their roots and evolve. However, this community exists within
the community of teaching staff (including graduate TAs and faculty), which exists within the even
broader community of P-Cubed students and teaching staff combined. We will sometimes discuss
these other communities in our study, because movements and practices within the community of
LAs can sometimes reflect movements within the teaching staff or the whole class, especially when
conceptualizing the LA-faculty partnership.
5.3 Methodology
By focusing on the LA and faculty perspectives in this work, we choose to take up an interpretivist
lens [50] on the case study. This means we adhere to the idea that we use case studies for seeing
how participants socially construct a phenomenon and what it means to them as participants.
In this investigation, the phenomena encompass the relationships between instructors (LA-to-LA
and LA-to-faculty) and the relationships between LAs and teaching practices. The case itself is
53
the P-Cubed environment. The interpretivist stance is helpful for our study because it helps us
leverage the participants’ perspectives on the phenomena, which are centered around LAs and their
function—what matters to us is not the essence of P-Cubed’s LA program (if there even exists such
a distillation), but rather how LAs experience it.
We chose to generate most of our data from interviews with LAs and a faculty instructor. The
reason for this is that we treat each interview as a separate “anchor point” [50] from which we
can view the phenomena of interest. We use this terminology because the interviews serve as
anchors which ground the phenomenon, because the interpretations of participants are what give
meaning to the phenomenon. By accessing multiple “anchors,” or interpretations, we will examine
participants’ language to build our own understanding of the LAs’ perceptions of the phenomenon.
We describe the interviews and other data sources in greater detail in Section 5.4. Our interpretivist
stance motivated us to choose data sources that represented the LA/faculty perspectives on the
communities within P-Cubed, of which they are members.
In a prior study [60] on LA feedback in P-Cubed, we focused on how LAs transferred practices
and skills from their experience in feedback construction to other academic settings. A main
finding for one LA was that the feedback mechanism played a big role in making her LA experience
meaningful, as well as helping her manage her academic studies in other contexts. The depth
of the relationship between the feedback mechanism and this LA’s academic life pointed to the
boundary-crossing that happens when LAs learn to use their teaching expertise in other areas of
their lives. This process is part of what happens in and out of a community of practice, and our
research questions in this investigation focus on characterizing the LA community of practice,
highlighting how LAs have impacted the practice of writing feedback, and understanding how the
LA-centered partnership is reflected in other P-Cubed processes.
We chose to bound our case to the experiences made known to us through interviews, emails,
and written course artifacts like feedback and course materials, because we sought to compare the
experiences and artifacts that most closely related to the feedback construction process. These
glimpses into the lives of LAs and instructors gave us a unique view into the functioning of P-
54
Data Sources
Individual
interviews
Four semi-structured interviews:
three with LAs; one with faculty
Feedback Forty written feedbacks from each
LA (120 total)
Course
documents
Assessment guide; Presentation
from training
Notes One set of notes jotted during LA
group discussion during training
Emails Written email correspondence
with two LAs after interviews
Table 5.1: Five types of data sources: interviews, feedback, course documents, discussion notes,
and emails.
Cubed’s teaching staff that only the practitioners could give. By investigating how a feedback
mechanism like the one in P-Cubed functions in the hands of LAs, we intend to demonstrate how
the practice of constructing feedback has developed under the influence of LAs, and how that
influence points to the dual existence of an LA community of practice and a student-partnership
between LAs and faculty. When LAs allowed us to step into their world to see what they value
and practice as a community, we were able to demonstrate what this community of practice and
partnership looks like and how it functions.
5.4 Methods
Our analysis focuses on interviews with three P-Cubed LAs as the primary data source. We also
sought alternate angles on the feedback mechanism by gathering written feedback excerpts from
the interviewed LAs to cross-reference with their interview comments, interviewing a P-Cubed
teaching faculty member, and collecting artifacts from the semiannual LA training where new and
old LAs convene to be trained by Irving and McPadden (co-authors on the published version of
this chapter) on constructing feedback, among other things. We display the complete set of data in
Table 5.1. We found that the LA interviews provided the most profound insights, which is reflected
in how we showcase our analysis.
55
We conducted interviews in a semi-structured manner. The original protocol was developed
using Patton’s methods [45] to address earlier research questions about howthe feedback mechanism
influenced LAs’ academic lives. As the angle of our research shifted in response to interview
comments, so too did the interview protocol. In this way, we developed the protocol iteratively.
The artifacts and feedback excerptswere gathered directly from LA training and the course archives.
When the focus of this case study became the influence LAs wield upon P-Cubed teaching practices,
we generated additional data. First, we gathered email exchanges with the interviewed LAs where
they described how their roles changed over the course of their LA tenure. Second, we interviewed
a faculty member—Roland—to discuss how he had worked together with the LAs and leveraged
their expertise in a variety of ways. Roland has taught P-Cubed several times but was not involved
in its original curriculum development.
We interviewed a total of five LAs in our investigation (one of whom, Carly, was featured in
Chapter 4), though we focus on only three in this chapter. We chose these three LAs because they
portrayed the deepest reflection on their experiences and were able to articulate their relationships
with the feedback mechanism with nuance and clarity. We attribute this distinction in part to the
iterative development of the interview protocol, which did not give the first two interviewees (Alvin
and Bella) as much of an opportunity to discuss their feedback writing. The latter three (Carly,
Derek, and Erica) were also all seasoned LAs with multiple semesters of teaching experience to
reflect on, which has provided them with multiple, variegated perspectives on what the feedback
means to them and how they have helped the mechanism develop over time.
For these reasons we present the perspectives of Carly, Derek, and Erica. Carly is a biosystems
engineering major, who at the time of her interview was finishing up her fifth semester as an
LA. Derek was an LA for seven semesters, and he recently graduated and entered the workforce
as a mechanical engineer at a large manufacturing company. Erica is a physics major who was
a P-Cubed LA for four semesters and, at the time, was finishing up her second semester as an
Electricity and Magnetism P-Cubed (EMP-Cubed) LA. All three LAs are white, as is Roland. We
treat the interviews as individual anchor points [50] through which we can understand how LAs
56
came to perceive and influence the feedback mechanism during their time in P-Cubed.
5.5 Analysis and Findings
Our research questions were asking: (1) Has a community of practice developed around LAs in
P-Cubed? (2) How has the practice of feedback been shaped by P-Cubed LAs? (3) How can the
LAs’ influence be characterized as a student-partnership and what characterizes this partnership
and its outputs? In this section we will outline our findings with respect to this focus. First, we will
demonstrate how the LAs comprise a community through which new LAs can learn from older
ones and master teaching practices (specifically feedback) that P-Cubed LAs have taken ownership
over. Second, we will show how the LAs have taken ownership over the practice of feedbackwriting
and developed and changed it in a way that is aligned with LA-held values and experiences.
Third, we will show how the LAs have come to occupy influential positions within the P-Cubed
instructional staff and how they have forged an effective partnership with faculty in a way that gives
them influence on how the course is run beyond just normal LA duties.
5.5.1 The Learning Assistant community of practice
The first main finding was that the LAs experienced a learning trajectory within P-Cubed that
resembled attainment of central membership in CoP. Specifically, we found that LAs make up a
community in which newLAs learn from older ones and hone their teaching skills, eventually taking
on the roles of veteran LAs in a cyclic fashion. In Section 5.2.2, we demonstrated how the design of
the course was primed for a community of practice to develop, and here we will show that one has
indeed developed among the LAs. Since design decisions do not guarantee the development of a
community of practice, this finding serves as evidence that the design principles in P-Cubed [166]
did in fact lead to a community of practice developing among the LAs in the course. The LAs
share the joint enterprise of teaching this course, along with a shared repertoire of course materials,
problems, experiences, and training. Additionally LAs are continuously engaging with one another
(mutual engagement) through weekly pre- and post-class meetings. More importantly, we see that
57
LAs experience a learning trajectory in the community. This process, as we will demonstrate,
begins with being a student in the class, and develops over time as a student is recruited into
becoming an LA, and as that LA learns to hone their practice and exert their own influence and
perspective on how the course is run.
When we interviewed Roland about his teaching experience with LAs, he described this process
in detail. He started by talking about what it’s like for new LAs to adjust, which is an important
step that new members of a CoP go through when learning to take up practices at first.
“They come in and they worry about a lot of things. And they can then consult
somebody who isn’t, some guy that’s like their dad’s age or older. They can talk to
somebody who’s their peer about strategies for going through things. I mean, it’s one
thing for me to tell them...it’s another thing to have somebody who they see as their
peer say, ‘you know, this actually does work if we try this’...not necessarily how I’d
want to approach an issue, but [the older LA] might have tried things that might have
worked better for them.” (Roland Interview)
He first talked about how it’s easier for new LAs to learn from and consult other LAs as opposed
to Roland himself, who said he’s probably older than their dads. He highlighted this connection
between peers that might be more automatic and comfortable for these newer LAs, which puts the
older LAs in a perfect position to provide initial guidance into what it means to teach as a P-Cubed
LA. This is aligned with the CoP idea that newcomers learn best from more senior members of
the same community [62]. Roland also mentioned how LAs might provide suggestions that don’t
necessarily match up with how Roland would “approach an issue”, but he acknowledges that this
is a plus, because veteran LAs will have ideas about what worked for them, to which a newer LA
would likely relate much better than they would to Roland. This is also in line with the idea that
LAs learn from other LAs in the P-Cubed LA community of practice.
Roland also talked about how he noticed LAs helping each other out in a variety of settings,
like meetings and outside of school. One skill he said LAs learn is how to help each other out
58
with teaching duties. He commented on the disproportionate benefit it made when older LAs
exemplified this skill as opposed to Roland simply articulating it.
“I’d try saying this, ‘Try relying on your, your fellow LAs.’ But when you have senior
LAs...who were so willing to do this and so willing to go out of their way and help the
other LAs in the class, it was contagious.” (Roland Interview)
Roland highlights how the senior LAs model the behaviors that he wants to suggest to the new
LAs, and by doing so, they set the norms for the group. Though he was a member of the larger
teaching staff community, Roland witnessed these behaviors from outside of the tight knit LA
community. It was almost as if Roland’s mentoring duties as the faculty instructor were superseded
by the “contagious” behavior of veteran LAs who already exemplified what Roland hoped the new
LAs would learn to do. Rather than Roland teaching the new LAs, it was the LAs who taught
one another the practices of their LA community. Again, we see older LAs guiding the learning
trajectory of community newcomers and how LAs are able to mutually engage as a community.
The LAs themselves were also aware of their central role in maintaining the community of LAs
and their take-up of teaching practices. In reflecting on this process via email correspondence, Erica
provided an example of what this community process looks like on an everyday basis. In Figure 5.2,
we show a screenshot that Erica took of a conversation with a fellow graduate TA about how to
polish a feedback to be given to P-Cubed students. This occurred after Erica had been teaching
for seven semesters (as an LA, and newly as a TA) and her peer was a first-semester newcomer
to the teaching staff, which means this interaction offers a snapshot of the community that exists
across instructors with varying levels of experience. Though not an interaction from within the
LA community, this example serves as an insight into the LA-adjacent parts of the teaching staff
community, which are refracted into the LA community through the concept of locality.
In the exchange, Erica, writing in the blue text bubbles on the right-hand side, provided some
lighthearted comments about how to reword several sentences that the other TA (words in grey on
the left) had sought her advice on. It is obviously a friendly exchange as there are many exclamation
59
Figure 5.2: Text exchange between Erica and a graduate TA, exemplifying strong relationships
within the student teaching staff of P-Cubed. The blue bubbles on the right show Erica’s suggestions
for the TA’s feedback, while the grey bubble on the left shows the TA’s response to Erica’s
suggestions.
60
marks, laughter typed out in all caps, and use of emojis throughout the message. Erica and this other
instructor clearly have an easy rapport both as friends and fellow teachers. This provides insight
into what relationships between student-instructors can look like in the P-Cubed community. As
evidenced by the personal text message, the connections between instructors go beyond classroom
exchanges, and many see one another as friends and confidants. From a CoP perspective, this
extends the boundaries of the teaching staff community, and allows its members more opportunities
and places to share practices and help one another improve, such as through texting.
When we compared the feedback written by LAs with the perspectives they provided in interviews,
we found confirmation that the LAs accumulated expertise as they moved up through the
P-Cubed LA community. In Carly’s interview, she recalled how feedback she had received as a
student in P-Cubed helped her maintain confidence as she went through the course for the first time.
Specifically, she highlighted how balance can be helpful by talking about how criticism can be
connected to praise in ways that made it easy for Carly to see how she could improve.
“One thing that I found—I think that I usually found the most helpful was when the
positive thing that was being highlighted was connected as well to the thing that I
needed to improve on, because then that gave me a clearer idea going into class of,
‘okay this is one thing that I’m going to focus on today.’ ” (Carly Interview)
The reason we bring up this experience is because it demonstrates how Carly was thinking
about what was and was not helpful about the feedback from an early point in her LA trajectory,
even before she began constructing feedback herself. By design, new LAs are recruited from the
set of current and former P-Cubed students in part so that they can draw from experiences such
as these. Carly’s experience is proof that the learning trajectory of LAs can begin when they are
still students. Specifically, she was learning to construct feedback even before she became an LA,
which indicates that non-LA students can exist on the periphery of the P-Cubed LA community
just like Carly did.
61
In reading through excerpts of feedback written by Carly,we found several instances of balancing
praise and criticism. One instance is provided below.
“You showed a lot of initiative and did a good job of getting the group started on
working on the problem. I liked that you were thinking out loud and talking through
your work as you did it. However, there were several times where the entire group
was not on the same page or were not fully understanding what you were working on.”
(Carly Feedback)
Here, Carly pointed out that the student was promoting work within the group, which was good,
but the group members did not have a strong understanding, which was what needed to improve.
She tied the improvement to an already somewhat productive activity that was happening within
the group, thereby highlighting the connection between her praise of the group and her suggestion
for improvement. Her usage of this strategy and her acknowledgement of its helpfulness in her
interview show that her feedback practice is connected to her earlier experiences. This strengthens
the idea that students like Carly can find themselves on the LA learning trajectory even before they
are officially hired as LAs.
To show another example of what this trajectory-into-community can look like, we also demonstrate
how Derek gained expertise and used it in practice. During Derek’s interview, he provided
a detailed look into a time when feedback helped him improve his group work and start buying
into the class when he was a student in P-Cubed. The feature that he found so helpful was that the
suggestions in the feedback were justified. His instructor was trying to get Derek to interact with
his group more and generate discussions.
“ ‘You need to incorporate discussion with your other group members, because the goal
of this class is to work as a group, and you won’t be able to solve the problem and you
won’t be able to get your understanding better unless you start conversing with those
other group members.’ That was at the very beginning of the class, and that helped,
because once I started conversing with my other group members... it actually created
62
a better environment in our group, kind of almost trust, like ‘alright I know that you’re
asking this question about why I think it’s this way because that’s what we do in the
class.’ ” (Derek Interview)
The instructor suggested interacting more with group members during class. What stuck with
Derek was the fact that this interactive theme was aligned with the goals of the entire course. Derek
used this same justification to normalize question-asking within his group, and he says this process
led to “a better environment in our group.” There are two main takeaways from Derek’s reflection:
(1) Derek learned the importance of justifying critique in feedback, which was a major step for his
trajectory as a feedback-writer in the LA community. (2) He realized the importance of group work
in P-Cubed, which led him to improve his own group work. This is a key skill among members of
any community of practice, who collaborate frequently on mastering practices. Though Derek’s
trajectory began in the whole-class community, the skills and values he learned as a student were
replicated and refined as he joined the community of LAs within the larger P-Cubed community.
In reviewing the feedback written by Derek, we saw the same commitments carried out. Below,
Derek encouraged a group of students to work together instead of waiting for Derek to rescue them.
“Listen to each other’s ideas, don’t just wait around for me to give you the answer
because if you aren’t making an attempt to work with each other, I’m not going to be
much help. I know you can do this because at the end of class Tuesday everyone helped
each other when I asked those individual questions. It worked out really well when
you all worked together.” (Derek Feedback)
Derek justified his suggestion by writing that he would not be able to give away answers, which
means that the students would need to adopt a better strategy for working together. He referred to a
time when this group did work well together, and used it to back up his reasoning in the feedback.
In this example he is passing on and modeling practices that he came to value—justifying critique
and working together. This is Derek’s way of extending an opportunity to take up the learning
trajectory that many LAs and P-Cubed students have taken before.
63
Another nod to Derek’s high view of group work came during an interview comment when he
was reflecting on what it was like to be a P-Cubed student. In particular, he found it helpful to
get others involved and ask questions. These practices align well with his previous commitment to
“start conversing” with his group.
“Getting everyone involved, and asking questions... I valued those behaviors before I
became an LA, because when I would not do those behaviors, I would be like, ‘I’m
struggling in this class right now’, and when I would do those behaviors, I would go,
‘This class is really easy.’ ” (Derek Interview)
Again, Derek discussed how he learned to conduct himself in certain ways in order to be
successful—e.g. “getting everyone involved” or “asking questions.” It came as no surprise to see
elsewhere in Derek’s feedback statements like, “take a step back and talk to your group”, “ask
questions about what you are missing”, and “if you aren’t making an attempt to work with each
other, I’m not going to be much help.” The suggestions he provided in his written feedback stand
parallel to the behavior he adopted as a P-Cubed student and LA. Like Carly, this points to a
cohesive learning trajectory that Derek followed as he learned to construct feedback as a P-Cubed
LA. Each trajectory began when they were students in P-Cubed. As Carly and Derek grew from
students to LAs, practices from student-centered group work evolved into practices dear to the LA
community. This dual experience points also to a shared repertoire of practice that LAs develop
beginning with their time as students in P-Cubed.
We have argued that the LAs operated within a community of practice, and more specifically
that they underwent a learning trajectory as they progressed from P-Cubed student, to new LA,
to veteran LA. An important part of this process was the sense of togetherness, because the LAs
learned through their relationships with one another and their past experiences within P-Cubed.
Because of this closeness, they collaborated on many teaching efforts. Roland had a keen eye for
how these communities (LAs only and whole-class) came together while he was an instructor.
“[A senior LA] helped foster a willingness among the LAs to help each other...a
64
willingness to say, ‘let’s help each other.’ Because sometimes some of the other LAs
might have a solution and they rely on each other like that and that was really nice.”
(Roland Interview)
In this comment he described how a senior LA used her relationships with the other LAs to
encourage them to help one another with teaching issues or when a solution to a problem needed to
be shared. From the CoP perspective, this togetherness shows how members of the LA community
leveraged their relationships to build a shared repertoire of practices. In this excerpt and throughout
this subsection, we see community as the source from which LAs learn to grow and improve their
teaching.
Another way to conceptualize the trajectory towards central membership in the P-Cubed LA
community is by considering the student body of P-Cubed as its own community from which a
pathway leads to the LA community. Roland commented about the preparation LAs go through as
past P-Cubed students.
“They know the environment. They know the community of the class. They know how
the students within groups can interact...I think it just makes for a good community of
learners. And the LAs having done that—I think that also helps them help each other
as they teach the class.” (Roland Interview)
The feature he highlighted was the communal aspect of being a P-Cubed student. They learned
to help each other and collaborate as students in the past. Those same collaborative practices
continued to help them as they worked together on teaching the class. He went on to compare the
enterprises of each community: learning science and teaching it.
“The P-Cubed community is like, ‘how do we do science?’...While science educators
[are like], ‘what do we do as we teach science?’ We sort of can follow the same model:
what works, what doesn’t work. We collaborate with each other. And I think it does
transfer.” (Roland Interview)
65
He again highlighted the collaborative aspect of each endeavor and capped the discussion with
a reiteration that he “think[s] it does transfer”, meaning the LAs have transferred their model of
figuring out how to do science into a model of how to teach science. From a CoP perspective,
Roland is pointing to the broker-like nature of being a P-Cubed LA. Mastering science practices
together as students builds co-working skills—skills that new LAs have already learned to excel at
before they first stepped from studenthood into the LA position (from the whole-class community
into the LA community). This sets them on a course towards the center of the LA community just
like the LAs who came before. It’s notable to mention here that Roland did not participate in the
curriculum development process for P-Cubed, which further highlights that the class is seen as a
community even by those who did not design it into the course.
Through these quotes, we have provided evidence that there is a community of practice among
the P-Cubed LAs. Specifically, we showed Roland’s reflections on how LAs learn from one another
like a community of practice, we showed Erica’s account of how she guided another student
instructor in a similar fashion to central members guiding newcomers, and we showed evidence
from Carly and Derek that demonstrated how feedback practice developed in the hands of LAs
similar to the negotiation of practices from CoP. LAs shared the joint enterprise of helping their
students develop scientific practices and group work skills, and they learned to achieve these goals
by collaboratively figuring out best practices (or in CoP-speak, developing a shared repertoire
through mutual engagement). The existence of this LA community is owed in part to the learning
trajectory that LAs take as they move from students to senior LAs in the course.
5.5.2 Development of feedback practice
The second main finding was that the development of feedback practice among LAs resembles the
evolution of practice in a community of practice. In Section 5.5.1 we established the existence of
the LA community of practice and the LAs who participate. We dissected how Carly balanced
praise and critique in her feedback and how Derek leaned his feedback into group work and justified
his critiques. These were presented as evidence for the learning trajectories and memberships that
66
Carly and Derek have traversed and held within the LA community. Additionally, when analyzing
LA interviews and feedback excerpts, we noticed that the LAs had iterated on what feedback looked
like compared to its original presentation in the course materials in subtle but important ways. In
this subsection, we will show in greater detail how these LAs have developed the practice of writing
feedback. This influence on a practice in the LA community exemplifies how LAs participate
in negotiating a joint enterprise, which is an important part of any community of practice. It’s a
process that captures the trajectories of whole communities, because it indicates howa community’s
goals are changing.
Before we dive into what feedback has evolved into, we look to how it began. P-Cubed has
been offered every semester since Fall 2014, which means it has undergone six years of iteration
in its teaching. Its original developers penned a guide for constructing feedback, and one of the
course’s original developers and instructors (Irving) is still involved in the training of LAs. His
influence through the course materials, the LA training, and the management of LAs is important
when considering how the course has evolved. To see what the course materials look like, we
have analyzed the assessment guide meant to be used when LAs provide grades of in-class work
alongside the written feedback.
When describing how students will receive feedback, the assessment guide provides structural
details that students can expect to see. Based on their in-class performance, students receive a
numerical grade with written commentary that outlines something positive they did, something to
work on, and a suggestion for improving their in-class work.
“You will be provided with written feedback before the start of your next project based
on your performance on the previous week’s project that will focus on one type of
participation that you excelled at and one area we would like you to work on in the next
project and suggest how you might go about doing that.” (Assessment Guide)
This same structure is rephrased during a presentation to LAs at the beginning-of-semester
training. Again, we see praise, a critique, and a strategy for improvement.
67
Figure 5.3: The feedback structure described in the course materials is two similarly structured
paragraphs, one focused on in-class work by the group, one focused on in-class work by the
individual.
“Feedback has two parts: How did the group do? How did the individual do within the
group? Each part addresses three things: (1) Something the student/group did well, (2)
Something to work on for next week, (3) A strategy for how to work on it.” (Training
Presentation)
The training presentation further clarifies the structure of feedback; there is an explicit instruction
for LAs to write feedback that addresses both the group and the individual. We refer to this
portion of the feedback practice as “reified” because it is what has been baked into the materials
that have withstood cycles of LAs and in part guided the take-up of LA teaching practices. We represent
the feedback and its reified components (as portrayed by the instructor-designed materials)
in Figure 5.3. Though this does not capture the LA perspective, it provides a starting point which
will make it easier to show how the LAs elaborated and filtered the feedback mechanism in various
ways. When the course was conceived, this reified feedback represents a snapshot of the original
“enterprise”, upon which the LAs negotiated their own goals and best practices when they wielded
influence within the LA community.
To show how LAs have played a part in making feedback their own, we begin by reframing
the formative experience from the previous subsection that Carly talked about in her interview.
She asserted that balancing and connecting praise and critique was important to her when writing
68
feedback. The emphasis on connection is not part of the reified feedback, but it was part of Carly’s
P-Cubed student experience. She specifically remembered receiving a piece of feedback from her
time as a student that helped her come to this viewpoint.
“There was one week where the positive was essentially like, ‘You do a good job of
facilitating discussion within the group and asking people to pause and clarify what
they’re saying’...but then the follow-up was, ‘Sometimes though, you save questions
for me as the instructor when you could be asking these questions to your group.’ ”
(Carly Interview)
This clues us further into how Carly sees the balance—not just as a tally of positives and
negatives, but in a connected way, where the suggestions fit in alongside things that the student
is already doing well. In reading through excerpts of feedback written by Carly (when she was
an LA), we found several instances of balancing and connecting praise and criticism, such as the
excerpt in the previous subsection.
Another piece of feedback that Carly wrote exemplified this connective balance that she was
committed to providing for her students.
“You do a great job of working through the math problems that are involved within
these problems and I can tell this is an area you are comfortable in. If I had one
recommendation for you it would be to leave your work in variables for as long as
possible.” (Carly Feedback)
Carly used the same strategy as before. She praised a student for her proficiency with math and
then suggested a further improvement to use variables more often. Carly’s experience about feeling
reinforced by this connective balance was reflected in how she wrote feedback. The connection
between the positive and room-for-improvement aspects of the feedback was never outlined in the
assessment documents or discussed in the LA training around feedback. This connection, although
a subtle change, does significantly transform the direction of the feedback as being targeted around
69
one theme or practice as opposed to a divergent emphasis where positive and improvement aspects
are split in focus. Currently we have no way of evaluating whether a concerted focus or split
emphasis will have more of an impact on the students, however this is not the focus of this chapter.
Instead Carly, based on her experiences and what she believes to be beneficial, has added to the
feedback approach by deciding on the need for connectivity. In this way, she was able to influence
the enterprise of the feedback practice within the LA community.
When examining Derek’s and Erica’s feedback, we found similar patterns despite not hearing
about experiences from studenthood that reinforced this feedback-writing strategy. For example,
Erica’s feedback to one student highlighted his strength of putting in most of the group’s effort
alongside a caution that he should encourage other group members to try out their own ideas.
“You had many equations and drawings in front of you, something that your group
needed a lot. Don’t let yourself be the only one doing this, however, because it seemed
like your group was starting to become reliant on your work to get them through the
problem...If you see yourself being the only one doing writing or calculating, stop and
ask your group members what they think.” (Erica Feedback)
She connected the praise—supported his group by creating physics representations in front
of him—with the critique of suggesting that he encourage other group members to take the lead
sometimes. If Erica learned this connection-strategy from Carly (which would be in line with
how she shared practices with peers in the data presented in the previous subsection) rather than
from her own experience with feedback as a past P-Cubed student, then this suggests that LAs in
P-Cubed are learning to write feedback from both (1) their experiences as students and (2) their
collaboration with one another. Even the instructor who provided feedback to Carly years ago was
diverging slightly from the explicit, reified instructions in the course materials. This suggests a
gradual shift in how feedback is given in P-Cubed—the strategy of connecting and balancing was
already somewhat in practice when Carly took the class based on the feedback she was given, but
that strategy was never formally reified. Carly has now emphasized and centralized its importance
70
as shown in how she structured her feedback in the data above. This aligns with ideas from CoP
that would suggest LAs act as brokers who might transfer practices or values from outside the
community.
A second feature of feedback practice that we analyze here is the written justification of
critique. In the previous subsection, we demonstrated Derek’s commitment to this practice. When
we examined notes taken during a discussion among LAs at training, we noticed that older LAs
tended to suggest taking up this practice, despite its absence in the course materials: “Justify why
you are asking them to do something”, “Make sure you mention why their grades changed if they
did” (LA Discussion Notes). This focus on putting justification in the feedback is something that
all interviewed LAs agreed on. The fact that it surfaced during LA training in discussions between
LAs but not at all in the course materials suggests that this feature of the LA-filtered feedback has
emerged primarily from experience rather than course design.
We can point to Derek’s feedback excerpt in the previous subsection as a prime example of
critique being justified. An explanation for this commitment could be that LAs feel that they have
less authority than graduate TAs or faculty instructors, leading them to justify the feedback they
give to their students as a way to build credibility. Below we showcase some examples of how
Carly and Erica provided justification in a similar fashion to Derek.
For example, Carly told a student to use variables instead of numbers when doing math.
“Leave your work in variables for as long as possible. By only putting numbers in
at the very end, you will make it easier to catch simple mistakes and to add in other
variables as needed. This will also help you and your group to see the connections that
there are between various variables and equations.” (Carly Feedback)
The suggestion Carly gave was backed up with reasoning. Carly wrote that using variables
would make it easier to catch mistakes and see connections when carrying out the math. Her
feedback demonstrates a commitment to telling her students why a suggestion is being given.
71
For Erica, too, the feedback she wrote for her students exhibited a deep commitment to providing
her students with reasoning for her suggestions. Below, she wrote to her group about going through
the problem-solving process a second time.
“Take some time to explain the methods of what you’ve all done so far before I come
ask. It’s beneficial to do this because sometimes, the methods you all come up with
are not structurally sound or use equations that aren’t relevant. Sometimes, the group
needs to hear someone repeat what they’ve done so far as well because someone may
not have been following along. Hearing it repeated back can reveal the parts of it that
don’t make sense.” (Erica Feedback)
Her suggestion is simply, “take some time to explain the methods of what you’ve all done so far”,
but the feedback is far richer because Erica wrote several ways that this can be a helpful strategy in
class. One of the hallmarks of Erica’s feedback was these lengthy justifications for her suggestions,
which left no ambiguity around what Erica was trying to tell her students in the feedback and just
as importantly why she was making the suggestion.
For all three LAs, justification of critiques was a core feature of their feedback. This feature
of the LA-perceived feedback does not appear to originate from the course materials or training
presentation. The LAs have chosen to adopt this feature because of their own experiences and
values. The fact that they value justification so much suggests that the LAs have altered the
feedback structure from its original form. The fact that they have shifted practice like this suggests
that LAs truly have central membership in the P-Cubed instructional community (not just the LA
community), because they have made the step from learning to take up practices to dictating how
those practices are carried out at the highest level.
When we emailed the LAs to circle back to this theme of developing feedback, Erica provided
an explanation for how she evaluates her own feedback-writing practice.
“There is never a perfect way to have written feedback for a student. Knowing that
helped me realize that as long as it’s not daunting for the student to read and it conveys
72
the message I want them to hear for the week, then I’ll know I’ve written ‘good’
feedback by my standards.” (Erica Email)
By evaluating her feedback against her own standards, Erica expressed a part of the agency that
P-Cubed LAs have when carrying out their teaching practices. Though Erica was likely guided by
course materials, training, personal experiences, and her fellow LAs, she emerged with her own
criteria for her feedback. This is what P-Cubed was designed for, and it is why we claim that
LAs in this context have had their own, real impact on feedback practices while still remaining
grounded in the P-Cubed community and its traditions. The iterations that the LAs have made
to the feedback mechanism can also be viewed from the perspective that the LAs are identifying
crucial gaps in the curriculum design that need to be filled and are filling them. The need of
providing justification with feedback seems abundantly apparent and yet it was never formalized in
the training or documentation for the class. It is contributions like this to the curriculum design of
the class that leads us into our next finding about SaP.
5.5.3 Learning Assistant influence through student-partnership
The third main finding of the chapter is that the LAs in P-Cubed function as the student-end of
a student-partnership. To be clear, LAs are not students in P-Cubed, but they are undergraduate
students. They participate in the partnership by influencing the P-Cubed course alongside the
corresponding faculty instructor. Because student-partnerships are defined largely along the relationship
between students and a faculty member or university official, we foreground our interview
with Roland in this subsection, where he discussed his perspective on his relationship with LAs
when he taught P-Cubed and what he thought about the influence that LAs had. Since we saw in
the previous subsection that the LAs have had significant influence on the practice of constructing
feedback, we now explore how this partnership functions. We will show below that the breadth
of LA influence extends beyond feedback and suggests that the LA community of practice (within
the teaching staff community) in P-Cubed could be a model for employing LAs as partners in
curriculum design. In order to demonstrate that a partnership centered around curriculum design is
73
at work, we will show that LAs have a strong level of control over decision-making on curriculum
and pedagogy in P-Cubed. This control is particularly important for a specific type of partnership:
curriculum design and pedagogic consultancy, which requires significant participation and say-so
from LAs [28].
From Roland’s interview, it’s clear he learned early on how useful LAs could be to a leadinstructor’s
decision-making. He reflected that he learned to rely heavily on LAs for running the
course, in effect forging a partnership that gave LAs power as instructors that had major influence
on feedback practice, teaching practice, and shaping the LA community. In his own words, “I think
[LAs] bring a lot more to the class than any single instructor could possibly bring to the class, in all
those different experiences” (Roland Interview). In talking of the different experiences, Roland was
referring to the in-class experience LAs have as P-Cubed students in the past before they become
LAs and also the experience of accumulating expertise over several semesters of teaching.
The partnership that Roland went on to describe applied more so to the veteran LAs than the
newer ones. From his perspective, these seasoned LAs were often better suited to teach than even
the graduate TA assigned to the course.
“[The graduate TA] hasn’t done that teaching in that type of an environment before.
And if we get an undergrad LA, who has taught the class once before and was a student
in the class once before, they tend to be better than first-time grad students doing the
class.” (Roland Interview)
In making this observation, Roland referenced the environmental preparation that LAs have,
which makes them ideally suited to teach P-Cubed as instructors. We provide this quote to show
how Roland, as the faculty instructor, views the LAs—to him their teaching expertise is secondto-
none. This quote also highlights that Roland also views the LAs as more experienced members
of the teaching staff community than graduate students, even though graduate students might have
more content knowledge in the subject and/or more years in the broader physics community. This
sets up the partnership that Roland allowed to flourish by giving the LAs more responsibilities than
74
would normally be expected from undergraduates.
The LA influence on teaching practice was most apparent when Roland described the role that
senior LAs took up in his most recent semester of teaching P-Cubed.
“I see the more senior LAs as being responsible for the day-to-day running of the
class...they’ve done the class multiple times and they’ve seen a lot of the different
issues and things you could run into.” (Roland Interview)
Again we see Roland elaborating on the preparation these LAs have had by running into the
same problems many times. He saw them as co-managers of the course and entrusted them with
responsibilities that hewould not be able to oversee, because he knewtheywere experienced enough
to tackle issues on their own. This is one of the ways we are seeing the LAs have a level of control
over pedagogy.
One of the day-to-day runnings that he entrusted to LAs was twice-a-week meetings to prepare
for class. The meetings were held in separate groups to accommodate scheduling, and one set
of meetings was led by a senior LA, whom we will call Fiona. Roland recalled how Fiona used
probing questions, which he saw as reinforcements of good teaching strategies.
“She did a nice job of breaking down the problems and making sure everything that we
might conceivably run into in class was covered in these pre-class meetings, and asking
and modeling good probing questions for the junior LAs...she did a really good job of
modeling what good interactions with students would look like.” (Roland Interview)
By Roland’s account, these meetings were run well by Fiona. Not only that, but she was able to
model student-interactions, implying that she had an in-depth understanding of how students might
approach the relevant problem. By letting an LA take up a position of power like this behind the
scenes, Roland allowed for the LAs to take up central positions in the instructional staff as a whole.
This had the dual effect of leveraging LA expertise to improve teaching practices across the whole
staff and also encouraging a framing of P-Cubed instruction that centers LAs, which could be seen
75
by old and new LAs alike. Even the LAs who did not have these bigger responsibilities could see
that the partnership was at work.
The P-Cubed students also bore witness to this elevation of LAs because during class time,
Roland’s classroom was run by the same senior LAs that he talked about earlier in the interview.
The way the room was set up put Roland on one side of the classroom. The other side he left to
be managed by Fiona, the same LA who took charge of managing the class and helping other LAs
when hard-to-manage situations arose during group work. He trusted Fiona to manage problems
that arose among other instructors, and in his interview he commented on the peace of mind he had
during class.
“Sometimes what happens during class, [an LA] runs into something and they’re unsure
how to proceed with it. It was nice to have somebody who [LAs] could rely on on the
opposite side of the room.” (Roland Interview)
Through sharing management responsibilities at Roland’s discretion, Roland and a handful of
senior LAs forged a partnership where they all had their voices heard and their expertise appreciated
in how the class was run. As described by Matthews [37], this quote from Roland exemplifies a
“reciprocal partnership,” where LAs’ inputs are truly valued and not tokenized. This is an example
in which P-Cubed teaching practices stand on the second rung from the top of the participation
ladder in Figure 5.1.
The LAs themselves reflected in email correspondence that they felt their voices were heard on
course decisions and how the class was taught. This signifies that these veteran LAs had central
influence on how practices of the P-Cubed LA community were carried out, and it wasn’t just
Roland’s perception. Several examples of this follow. We begin with Erica’s email, where she
reflected on how she gained familiarity with P-Cubed’s in-class problems over time, and eventually
began making suggestions for improvements that would clear up sources of confusion.
“Over time, I became more familiar with what each problem was made for: each
problem had a concept it intended to convey through the story, and as that message
76
became clearer to me, I became more vocal about places that were routinely confusing
to me and in what places we could add more context or rephrase things to make them
clearer.” (Erica Email)
Erica only gave input on problem design after she felt that she had gained familiarity and
expertise on what the problems were meant to be about in the first place. This highlights another
benefit of having LAs participate in this partnership: their suggestions are grounded in the combined
experience of dealing with the course materials from a student perspective (as former P-Cubed
students) and an instructor perspective (as LAs).
The influence that LAs have in P-Cubed extends beyond the physics problems. In Carly’s email
correspondence, she discussed how she had an idea to change part of the structure for delivering
feedback: rather than providing grades according to a written rubric, she wanted for instructors to
input feedback into an app that mapped the rubric into a questionnaire that related more closely to
experiences instructors would have in class.
“I think that the professors and actual TAs had a lot of respect for the LAs and what they
had to contribute...My ideaswere taken seriously and either implemented or Iwas given
clear feedback about why they weren’t implemented. One thing that I contributed was a
different method for giving weekly grades to students. Although it wasn’t implemented
long-term, it was trialed for a semester and it felt like I’d been able to move the class
forward (even if it was more of a reassurance that the current method was still a good
one).” (Carly Email)
Carly’s app idea ended up on the back burner after a pilot semester, but it remains a testament to
the power that senior LAs are granted in steering the teaching practices and feedback practices of
P-Cubed. A common concern of student-partnerships is that the input from less powerful members
(LAs) is sometimes not taken seriously [29]. In Carly’s case, her ideas were encouraged until they
became full-on transformations of teaching practice and implemented broadly to test their efficacy.
77
Another, more direct example of an LA participating in decision-making around feedback
structure was when Erica had a chance to give input to the EMP-Cubed curriculum, which is a
P-Cubed-like course that covers introductory electricity and magnetism, first taught in Fall 2017.
“EMP-Cubed was being developed and Paul [Irving] was sitting at a table, thinking
about how to implement self-written feedback from students into the course structure.
I sat and I brainstormed with him, and my idea of dividing the self-feedback so that it
was slowly implemented in stages through the semester ended up being the structure
that was implemented.” (Erica Interview)
Though the context was not P-Cubed, Erica had forged a partnership with Irving in part from her
role as an LA in P-Cubed. This relationship made it natural for her to provide input on a new course
and reimagine what feedback practice could look like. In this way, the LA-faculty partnership had
a tangible impact beyond the course where it began.
The last partnership-like impact that we will describe in this subsection is the roles LAs play
when recruiting new LAs to the instructional staff. Roland described in his interview how LAs
provide special insight during this process.
“Like halfway through the semester, we’ll discuss recruiting new LAs and solicit input
from more senior LAs...the LAs might say, ‘yeah, the person might ought to be this,
but I’m not quite so sure about that.’ So we get LAs who would say, ‘I think this person
would make a really good LA.’ And having an LA approach one of the students in the
class and say, ‘you should really apply for this’, I think that helps with recruitment.”
(Roland Interview)
He described their input as a solicitation, meaning he has sought out their opinions because
he values what LAs have to say about potential applicants. The solicitation is another indication
that there was a relationship between faculty and LAs through which Roland felt he could consult
the LAs on the future of P-Cubed teaching. When he hears comments like “I’m not so sure” and
78
“this person would make a good LA”, this helps him direct the way he thinks about the recruitment
process, because he knows that many of his LAs knowthe current students much better than he does.
He admitted earlier in the interview that he really only gets to interact regularly with 25-percent
of the class over the course of the semester, which is why he relies heavily on LAs during the
recruitment of enrolled students. This reliance points once again to the negotiation of LA control
over the course. He also highlights the importance of having LAs encourage current students to
apply, the implication being that P-Cubed students might trust the suggestion of an LA who went
through that same process.
We used Roland’s commentary on the helpfulness of senior LAs to show how they had a
partnership with Roland wherein they were trusted to manage meetings and real-time in-class
issues without the intervention of a faculty or graduate TA. This pointed to the responsibility that
some P-Cubed LAs had, which rendered their class-wide influence akin to Roland’s. One product
of this LA-based power was that they learned to work together to reinforce learning strategies for
their students. As Roland recalled, LAs would identify broad needs in the classroom and work with
their students via feedback and in-class teaching to help them improve along those lines.
“Trying to reinforce [strategies], not just in feedback, but sitting down at the table with
their students face-to-face and reinforcing in two ways. You’d have multiple LAs sort
of reinforcing the same types of strategies...I think it just organically happened like
that.” (Roland Interview)
Because of how LAs worked together and collectively had influence over a large number of
students, they were able to impact in-class teaching and learning in a big way.
We also analyzed Erica’s journey with P-Cubed problem design: first learning to do the problems
and gaining familiarity, and eventually providing feedback on sources of confusion and improving
the LA solution guides. She seized similar opportunities to contribute to exam problems and
homework, which she elaborated on via email.
“I wrote exam problems fairly regularly since the beginning of my LA time, even up
79
until now. It feels like having a voice, because my ideas are directly implemented
in something a student receives and gets a grade on. Same thing for homework, like
deciding my own help room hours or choosing how I can run those hours. It’s like a
real-time judgement call.” (Erica Email)
She viewed these opportunities as “direct implementation” of her ideas onto the materials that
students would go on to use. An area where she had total control was her “help room hours” where
students would come to get help on homework, concepts, or studying. Erica was able to recognize
the ways she could leverage her strengths and have the most impact as an LA. She put it poignantly
in her reflection, comparing this impact to a historical, indelible influence on the trajectory of
P-Cubed.
“It’s sometimes scary, but it also feels very satisfying knowing that I’m putting a little
bit of myself in the history of the class.” (Erica Email)
Overall, we see many features of the course that demonstrate how these LAs have become
central members of the instructional community alongside the faculty instructors and graduate
TAs. Through the course design and the compliance of past instructors, LAs have been given
responsibility for managing students, opportunities to run meetings and shape the LA community
through recruitment, and in some cases seats at the table of curriculum development. And
through these myriad opportunities, LAs have stepped up. They ran the meetings, they shaped and
sustained the LA community through mentoring among their ranks, they took responsibility for
carrying out teaching practices in accordance with their experience, they grew the LA community
through recruitment, and they passed on these responsibilities to their protege LAs. The existence
of the opportunities listed above and the strength with which the LAs have used these opportunities
to wield control of the course are how we demonstrate the existence and characterization of a
student-partnership among the P-Cubed LAs.
80
5.6 Discussion and Conclusion
The goals of this investigation were (1) to demonstrate the development of a community of practice
among P-Cubed LAs, (2) to describe LAs’ influence on the development of a specific practice
(feedback) within that community, and (3) to demonstrate and characterize the partnership between
P-Cubed faculty instructors and LAs. Though our first and second findings could be described
as “outputs” of the partnership, we presented them separately to motivate the third finding and
demonstrate how the partnership functions in a more detailed manner.
This study highlights two specific design principles that encouraged the development of an LA
community of practice within the P-Cubed context: the feedback mechanism and the P-Cubed LA
program. According to Irving et al. [166], the feedback was designed to build trust between LAs
and students, offer explicit suggestions for improvement to help students take up scientific practices,
and legitimize student behavior when aligned with the goals of the class. The LA structure was
designed into P-Cubed as a way of providing a social “bridge” into physics, because LAs can be
seen both as experts and peers. In this way LAs were designed to be central members of the wholeclass
community. As we showed in our investigation, these design principles successfully set up a
community of practice among LAs—as evidenced in particular via the practice of feedback—in a
way that allowed P-Cubed students to follow a trajectory from physics-newcomer (just outside the
periphery of the LA community) to veteran LA. Although this study does not explicitly investigate
the student (pre-LA) part of the trajectory within the P-Cubed community of practice, the reflections
on the journey from student to LA from our participants do highlight that their student experiences
played important roles. This supports the notion that designing for the development of an LA
community of practice can be a fruitful way to orient a classroom. For P-Cubed in particular, the
LA program is a significant part of the manifestation of the CoP design.
For our context, an important part of the community-building process is that all theLAapplicants
are previous students from P-Cubed. Although it is somewhat typical for undergraduates to be
recruited to be LAs for classes they have taken, our study indicates that this style of recruitment
81
is essential for the P-Cubed LA program. It provides significant preparation for potential new
LAs, who often join the staff ready to operate in the collaborative P-Cubed environment. Our LA
interviewees often recalled how formative student experiences played into how they went on to
teach. Roland, too, commented on how he believes LAs in P-Cubed are well prepared for their role
because of their familiarity with the material.
Another benefit of drawing from the P-Cubed students as an applicant pool is that existing LAs
get to participate more authentically in recruitment. This feature in particular is a benefit to the LA
community, because LAs get to have a voice in who becomes more central to their community. They
do this by providing first-hand feedback on the character and preparedness of potential new LAs
based on their interactions with the applicants as students. If applicants came from outside the class,
the existing LAs would not have the personal relationships to draw from, and therefore would not
get to participate in community management as closely. The CoP framework has an apprenticeship
undertone to its set up, and LAs having a voice in the recruitment process allows them to choose
the next set of apprentices that they want a hand in guiding. This allowance reinforces to the
LAs that their voice matters with regard to the running of the class and maybe more importantly
who becomes more central members of the community. However, input on recruitment has to be
managed carefully as the culture of the community needs to place an emphasis on whether potential
LAs are demonstrating aptitude in the practices and values of the class and not letting a creep
towards a recruitment of LAs who are “similar” to them. For recruitment of new P-Cubed LAs,
the existing LAs will encourage students to apply and recommend potential candidates that align
with the community, as Roland described, but there is still an application form and an interview
process supervised by the course coordinators before any formal offer is put forth. This provides
an emphasis on creating an inclusive community within the P-Cubed classroom and maintaining
the goals (or, joint enterprise) of the community.
Despite the tight LA community that has flourished in P-Cubed, the CoP framework points to
new ways it can be improved. In particular, we examine participation and reification. Though LAs
have been able to change and direct what practices in the community looks like, their influence on
82
the course has gone un-reified. A prime example of this is in our more detailed exploration on the
practice of feedback in Section 5.5.2. The course materials around feedback still look the same
as they did when the course was first offered, despite the many contributions that LAs have made
to its structural components when they carry out the feedback practice and mentor other LAs in
it. Since the practice has evolved, CoP would suggest that these changes should be reified in the
course materials and shared repertoire of the LA community.
Furthermore, the process of onboarding new LAs through mentorship and expansion of the
LA community is still almost completely undocumented in curricular design materials. This
is potentially problematic because of the resulting instability around helpful strategies that LAs
have introduced into the course. For example, a new program coordinator or a series of new
faculty-instructors could completely change the enterprise of the feedback-writing practice, solely
because of the current enterprise’s heavy reliance on participation (without reification). The lack of
opportunities for LAs to reify their transformation of the feedback-writing practice is problematic
if the goal is to embrace LA-induced change. In order for this community to embrace the directions
that LAs appear to be pushing the practice, there needs to be some mechanism in place for LA
participation to be reified. Only through the duality of participation and reification can meaning be
negotiated by all members in the community. Such a mechanism would strengthen the existing LAfaculty
partnerships and allow the current LAs to contribute to the reifications of past curriculum
designers. This would then better satisfy the structural change needed in good student-partnerships
as outlined by Matthews [37]. In effect, LAs would be able to take part in negotiating and
documenting an enterprise that represents the collective experiences and values of feedback-writers
over time.
Currently, instead of integrating the adaptations formally, the class coordinators have instead let
the practice of feedback transform organically. Organic transformation versus imposed reification
opens up more possible research questions. For example, questions need to be asked about the
formalization process—should practices be reified after they reach a level of uniform use by
members of the LA community or is good practice just good practice and should be integrated
83
immediately? One of the realities of curriculum design is that there is no one “right” way to teach.
Maybe a level of uniformity being reached in how the LAs teach is an indicator of the utility of
a change in teaching practice and a point at which reification should occur. At the very least, the
feedback adaptations made by the LAs in this study and the lack of reification of those adaptations
highlight the need to listen and pay attention to the teaching approaches of LAs.
One way to address the current issues around reification in P-Cubed would be to update the
artifacts that exist in P-Cubed related to feedback, such as the assessment guide. By incorporating
LA perspectives into course materials that would be used in future semesters, we can strengthen
the positive influence that LAs have on the course structure. A more explicit strategy would be to
administer exit interviews with final-semester LAs that could be incorporated into the materials as
a way of preserving their legacy and the improvements that they made to the course during their
time. A shadow of this idea exists in pre-class meetings, when notes are gathered on the confusing
parts of the solution guide, which is then updated for future semesters. These strategies exist to a
degree in P-Cubed, but they could be leveraged in other areas of the course and expanded to be a
more explicit part of the curriculum development process.
Through our investigation, especially when examining how LAs have developed the feedback
mechanism, we demonstrated that in P-Cubed there exists a partnership centered around curriculum
design and pedagogic consultancy. In particular, this partnership is characterized by the long-term
tenure of LAs and the lasting influence they have on teaching practices. The three tenets of a good
student-partnership, according to Matthews [37], are at work: (1) Input from LAs is valued among
curriculum designers and faculty, meaning the partnership is reciprocal. (2) All parties benefit from
the partnership: LAs gain experience managing the community and bettering their teaching skills,
faculty get classroom management help and get to learn from peer-learning experts, and P-Cubed
students (though they are not members of the partnership) receive a more personal, relevant physics
education. (3) The outcome of the partnership is broad and sustainable in how it has a lasting effect
on the course pedagogy and the structure of the LA community among future generations of LAs.
Fulfilling these tenets is only possible because P-Cubed was structured for LAs to retain a
84
direct influence on the course for years and the LA-end of the partnership comprises undergraduate
student-instructors, as opposed to just students. The P-Cubed LAs have a special combination of
expertise and opportunity, which allows them to influence the course structure in positive, lasting
ways. Other curriculum-centered partnerships in publications are markedly different from the
P-Cubed model. For example, Cook-Sather [169] detailed a model that utilizes one-on-one studentfaculty
relationships to reform curricula. Unlike P-Cubed LAs, the students in this model had not
taken the course for which they advised. They instead learned about it by sitting in and gathering
observations. LAs in P-Cubed are special because of their closeness to the course, having spent
many semesters operating within the course. Also, the existence of a community of LAs helps
them build expertise via collaboration, which from a CoP perspective makes their advising all the
more valuable because it is more likely to be aligned with the values of the course and drawing
from a broader selection of experiences.
In another example, Bovill et al. [170] describe how students apply to course design teams for
courses they have taken before. In their findings they noticed that the partnerships suffered because
a lot of time elapsed before faculty in the teams noticeably ceded their authority and students began
to feel like they were being taken seriously. In contrast, the P-Cubed LAs have a long tenure where
they build trust with the faculty instructors (who often teach P-Cubed multiple semesters) and with
the LA program coordinator (Irving). Their voices are heard semester-after-semester, and taken
seriously, as shown in Section 5.5.3. The features that make the P-Cubed partnership unique are (1)
the LAs’ intimate experiential knowledge of the course, (2) the community of practice that exists
among LAs and influences the course as a collective, (3) the tiered nature of the LA community,
which allows for more senior LAs to take up significant course responsibilities and make their voice
heard on structural decisions without imposing the same pressure on more junior LAs, and (4) the
responsibility of LAs for carrying out the practices which they influence.
In most partnerships, students are recruited directly into partnership whereas in P-Cubed it
seems as though LAs gain credibility over time and are gradually consulted more and more on
course decisions and given more and more management responsibilities the longer they are an LA
85
with the class. Experience equating to credibility is one perspective, but an alternative framing
could be that new LAs do not feel equipped or have enough expertise to wield their voice related to
group decisions and instead defer to more senior LAs. The intertwined nature of experience and
credibility needs to be investigated further in order to understand how a student-partnership borne
out of an LA community of practice promotes and restricts the input of the LAs when it comes to
curriculum input.
The path towards centrality through experience could represent a more natural progression to
include student voices in curriculum development. The way P-Cubed is set up, LAs gain many
experiences with teaching the materials and operating within the LA community before being
offered some of the opportunities and responsibilities associated with the LA-faculty partnership
(more accurately associated with the slightly larger teaching staff community) that we described.
On the other hand, a potential problem with this model is that it privileges voices from more
experienced LAs. There is the potential for a form of institutionalization to occur as LAs spend
more time teaching the class with the possibilities of their inputs becoming more teacher-centered as
opposed to student-centered. At what point do the LAs stop being students and instead take on more
teacher-like perspectives, therefore losing the special influence of student-partners in curriculum
design? They will never be responsible for the running of the entire course, but an open question
becomes that for this SaP model, when do students become empowered enough that the source of
their influence is no longer authentic student experience? This also makes us wonder, what would it
look like for new LAs to infuse their voices into the course? We suspect because newer LAs are not
as central to the culture of P-Cubed, the course would change faster but perhaps with less overall
direction. The inputs of the newbie versus central member present an interesting future direction
for SaP research, and we are interested to see research from course contexts that have utilized this
more progressive approach to student-partnerships.
Overall, this investigation serves as a model for the fidelity of LA-driven student-partnership
leading to structural changes in a course. The lesson here is that student-partnership for LAs is
possible and can work well in the case of a course like P-Cubed that has been designed around
86
CoP. As we discussed, the features that make the P-Cubed partnership particularly effective are
the features that come from the LA community of practice that was designed into the course. By
learning to teach via the community of practice, LAs gain intimate knowledge of what works and
what doesn’t when teaching, they wield collective expertise when collaborating with their peers,
and they follow a natural progression towards a place where they have significant influence over
the direction of the course. The way this partnership is rooted in the LA community of practice
is what makes it as effective as it is. Re-conceiving LA programs as student-partnerships opens a
path to incorporate LAs into and reinforce sustainable curriculum change.
5.7 Acknowledgments
As written in the published paper: We would like to thank our generous participants for their time
and openness. For funding the curriculum development and staffing of P-Cubed, we thank the
CREATE for STEM Institute at MSU. We also thank the Department of Physics and Astronomy at
MSU for funding the hiring and support of LAs.
87
CHAPTER 6
STUDENTS’ PERSPECTIVES ON COMPUTATIONAL CHALLENGES IN PHYSICS
CLASS
This chapter builds on some of the research tools that I honed in Chapters 4 and 5: attention to
context, qualitative case study, and connecting students’ perspectives to theoretical frameworks.
What makes this chapter special is that it takes place in the context of a computation-integrated
high school physics class, a context that needs student-centered research and whose curriculum is
developing rapidly. Due to the gap in research on students’ perspectives in computation-integrated
physics, this chapter focuses primarily on analyzing and cataloguing what students say, and secondarily
on applying theoretical lenses to the data. A version of this chapter was submitted for
publication with second author Daryl McPadden, third author Marcos D. Caballero, and fourth
author Paul W. Irving to Physical Review Physics Education Research. My contributions were
research design, data generation, analysis, and writing.
6.1 Introduction and Background
There are increasing and wide-spread pushes to introduce computation to high school students [24,
25, 26]. Integrating computational practices with STEM classrooms gives learners a more realistic
view of what it means to do science, and better prepares students for pursuing careers in
a world where computation is ubiquitous [40]. These pushes are also associated with changing
standards [171] to teach our high school students how to “think computationally” [172]. As the
push for integrating computation into classrooms becomes more prevalent, we must reckon with
the problem that little is known about how students will take to computation-integrated science.
This research study contributes to the effort to find out more about the student perspective towards
computation when it is integrated into the science classroom. Here, we focus on a case of students
experiencing computational integration in their high school physics class. By detailing what
challenges and perspectives students face in this context, we can start to identify how to make
88
computation-integrated K-12 physics more equitable, enjoyable, and beneficial to learning.
For our purposes, we view computational integration as the act of altering the curriculum
of a STEM course to incorporate computational modeling, specifically as a tool to learn the
STEM subject. In this way, students don’t learn to program separately from learning science,
but rather they learn science in a new way, through computational modeling. This is a practice
that STEM professionals are intimately familiar with [33]; thus, integrating computation makes
STEM classes more authentic to future STEM careers. Authenticity is important in the sense that
computation provides a way for disciplinary science practices to be featured and learned in the
classroom [173, 174].
Computational modeling can be integrated in a variety of ways at the K-12 level. For instance,
at the high school level, teachers have created models for planetary motion in an attempt to help
students make predictions and discover Newton’s law of gravitation through experimentation on the
model [33]. This approach involved the teacher creating the computational model and the students
interacting with it. This integration focused on the practice of using computational models to explore
physical phenomena. Separately, a middle school chose to integrate computation into science
classes for fourth, fifth, and sixth graders [64]. The students used Scratch programming [175] to
create simple models of situations of their choice. For example, one student modeled a projectile
launched from a seesawand got real-time feedback from the computer as they constructed the model.
Because Scratch uses code-blocks rather than text, it was easier for students to interpret errors and
connect their computational choices to the model they made. Another example of computational
integration, at the college level, involved curricular transformation in an introductory undergraduate
lab-based course [65]. The labs in this coursewere redesigned to include one part traditional lab with
hands-on equipment, and one part computational modeling with VPython [176]. The integration
also included reflection questions to help students make connections between the programming and
the open-ended, hands-on experimentation. One benefit to the students was that by learning the
fundamentals of VPython, they were able to better visualize the relevant physics concepts in the
lab course [65].
89
Despite the increasingly widespread adoption, what we know about how students learn in
computation-integrated settings lags behind the speed of the changing curricula. As stated in a
recent report on the state of interdisciplinary computation-integration-based education, “We still
know very little about students’ thinking and learning as it unfolds with the use of computational
tools. At the very least, new tools for thinking and making sense of data call for curriculum
resources that consider students’ developing computational literacy. With the introduction of
this new competency, novel effects might emerge concerning student engagement, motivation, and
identity in computationally enhanced classrooms” (page 9) [33]. Essentially, Caballero et al. call for
researchers to develop an understanding of how computation impacts the experiences of students,
from the perspectives of students.
To date, there has been no in-depth qualitative research on the affective experiences of students
in computation-integrated STEM contexts in which to situate our study. We therefore looked to
similar work in other contexts. To start, studies on affect and investigations of students’ perspectives
have been a major focus in the last 30 years in broader STEM education research [77, 78, 79, 80, 81,
82, 83, 84, 85, 86, 87, 14, 88, 89, 90, 91, 92, 93, 94, 95]. In particular, previous research in math has
examined the affective impact on students when they engage in specific types of activities such as
problem solving [77, 78, 79, 80]. An example of this is a case study on affective responses during
problem-solving in a middle school math context [80]. Hannula demonstrated discipline-specific
connections between affect and student success, thereby suggesting that attending to student affect
in pedagogy offers a way to improve teaching and learning. In the discipline of chemistry education,
multiple studies have been carried out that examine student affect or constructs related to it, like
self-efficacy [81, 82, 83, 84]. In one study on student affect in an undergraduate chemistry lab [84],
the authors observed lab classes and asked students about their affective experiences. Galloway et
al.’s [84] findings and implications centered around students having complex, multifaceted affective
responses. The authors offered several suggestions for teachers to cultivate positive affect and imbue
meaning into the oft-rote manner of chemistry lab teaching. This study is important in that it was
the first to study affective experiences in chemistry labs with an in-depth, qualitative approach,
90
and the implications had the potential to make a significant impact on student-centered chemistry
lab teaching. In particular, the authors drew from Bretz [85] to demonstrate that affect-focused
research can provide insight into what students view as “meaningful learning”—an enterprise that
combines learning with relevance and represents part of students’ motivation to maintain effort in
school settings.
Similarly, in physics education, research abounds on students’ affective experience, beliefs, and
perspectives [86, 87, 14, 88, 89, 90, 71]. One study points specifically to a gap we are trying
in part to address—Gupta et al. [86] argued that there has been a lack of research in physics
education on the role of affect in modeling student learning, especially on fine-grain interactions.
They made the case that most research on student-centered physics learning focuses on the content
they know rather than their feelings about what they are experiencing [86]. To explore what role
affect can play in learning, Alsop andWatts [87] looked at how students approached a physics topic
(radiation and radioactivity) according to their attitude and perception towards it. Their study found
that it was possible to balance “impassioned knowledge and informed feeling” in the learning of
physics, which keeps students engaged but not off track. Some affect-based strategies for how to
achieve this balance of engagement and learning were explored by Häussler and Hoffman [14] and
Erinosho [88], who showed the importance (according to student perspectives) of linking physics
with non-traditional and/or out-of-classroom situations [14], providing materials that had concrete,
relevant examples [14, 88], and working on physics problems where students could collaborate
with peers [88]. This set of affect-based, student-centered physics education studies demonstrates
the relevance of affect to the field of physics education research, the need for deeper affect-based
work [86], and relevance of affect for exploring student perspectives.
Additionally, there have been a number of studies that center on students’ experiences in their
computer science classes. Gomes and Mendes [177] suggested that students struggle in computer
science because the necessary problem-solving strategies are new to students, especially in lowerlevel
undergraduate courses, where a lot of students have their first exposure to computation. On
top of that, students in these introductory courses are often experiencing the psychological stress
91
of their first year in college in tandem with developing new ways of problem solving and thinking.
From a broader perspective on computation, a study by Jenkins [178] highlighted specific barriers
associated with the computational tasks themselves. He described computational difficulties in
terms of a set of skills: coding (syntax, semantics, structure, and style), algorithms, and recipes for
translating ideas into code. He argued that the hardest part is the novelty of computation; compared
to other subjects, students need much more precision to achieve meaningful progress. This requires
mastery over coding skills and some degree of expertise with translating ideas into code, both of
which are hard to build when it is so easy to write imperfect code, to which the computer provides
convoluted feedback or outright rejects.
Much of the research on students’ experiences with computation, like the studies from Gomes
and Mendes and Jenkins, focuses on the challenges that students face rather than their reactions to
and perspectives on those challenges. Bosse and Gerosa [91] built a compilation of research studies
centered around learning difficulties in programming settings. Most of the results from their literature
review indicated students tend to be worried about learning syntax, variables, error messages,
and code comprehension. Students also generally experienced nervousness with unknown coding
concepts like functions and parameters, often resulting in students erecting affective barriers against
such challenges. For example, when a student realized their code contained a semantic error, they
were more likely to give up and not finish the programming activity because semantic errors take
a lot of time and effort to identify and fix [91].
In the last decade, computational education research has begun to explore the relationship between
affect and the challenges that students face in computer science courses. A relevant literature
review focused on qualitative research in computation education [179]. They identified self-efficacy
as a useful construct to examine students’ experiences in these contexts. However, much of the
existing qualitative literature on students’ affective responses exists in advanced, undergraduate
course contexts rather than more introductory levels. Additionally, they noticed that much of the
qualitative work was trying to develop theories about how learning happens in computational settings
rather than explore and explain computational difficulties from the perspectives of students.
92
According to this review, there is a need in computation education to research on how students
interpret their learning, especially at the introductory and/or K-12 levels [179].
A handful of studies address similar needs, though they are in short supply. Lishinski et
al. [92] studied students’ affective responses to computational challenges and how difficulties can
elicit self-efficacy judgments resulting in maladaptive learning strategies. They emphasize the
importance of attending to affect in programming environments, writing, “Emotional reactions
contribute to a feedback loop process in learning to program, and previous performance impacts
future performance both by virtue of the effect that past experiences have on learning, but also
via the effect that past experiences have on emotions” (page 8) [92]. A study from Kinnunen and
Simon [93] similarly found that students made assessments of their own self-efficacy throughout the
duration of computational tasks. Further, they found that affective experiences were the primary
feature of computational work that students remembered after class was over. This brought an
urgency to studying affect-based challenges in programming contexts.
The following year, Kinnunen and Simon [94] studied in more detail how students’ affective
responses were tied to their self-efficacy judgments. They found that self-efficacy was determined
early in the course when students had their initial failures or successes with computation. They
recommended that instructors should deliberately ensure that initial experiences with computation
should include several successes because it is so easy to “fail” by writing imperfect code if you
don’t know how to interpret feedback from the computer, which is often inadvertently masked
by confusing error messages. The same authors further studied the disconnect between affective
responses and self-efficacy with longitudinal interviews [95]. In their findings they attributed
the disconnect to a lack of reflective activities built into the course. They added to their previous
recommendations by suggesting that initial computational experiences should incorporate feedback
on the entire experience, not just the correctness of the result.
Studies like those from Kinnunen and Simon [93, 94, 95] and the recommendations that sprang
from them demonstrate the importance of exploring student affect in a given type of learning
environment. Eckerdal et al. [180] theorized about why computer science learning elicits in
93
students the affective responses that it does. They framed the initial experiences (where students
form their self-efficacy beliefs for the first time [94, 95]) as comprising a “liminal space.” In
everyday terms, they asked, how do computer science students cross the threshold to learning? If
it takes some persistence and confusion before students find their bearings in a computer science
course, what is helping them get over the hump? The authors examined affect and found that as
students crossed over the threshold, their feelings about learning computation transformed from
hate and fear to euphoria. This implies that teachers can take clues from affect about where students
are in the learning process, and even tailor instruction to help them cross the threshold to learning.
While there has been significant research into student affect and experiences in STEM courses,
including computer science, this research has traditionally been siloed into separate disciplines. As
computation becomes integrated into STEM courses [33, 64, 65, 66, 67, 68, 69, 44], it is important
to understand the effects of this integration. Recently, there has been some work that addresses
the challenges associated with computation-integrated STEM, though not from a student-centered
perspective. For example, one study investigated the ways that computational activities could
be difficult in a middle school context [181]. The authors justified doing this in a computationintegrated
STEM setting, writing, “learning a domain-general programming language and then
using it for domain-specific scientific modeling involves a significant pedagogical challenge.” They
found that certain features, such as the problem-solving process and the syntactic complexity of
programming languages, can be leveraged for learning by eliciting reflection on work or alleviated
by employing a simpler programming language like Python. Overall, they relied on identifying
challenges through observation of computational activities rather than through the perspectives or
affective responses of students. The samewas true in a study byVieira et al. [182], where the authors
evaluated a computation-integrated materials science and engineering course. They found that it
can be helpful to integrate computation with student-facing challenges in mind. For example, early
in the curriculum students performed poorly on framing and recognizing computational problems,
which could be addressed by providing extra scaffolding for problem-solving at the start of the
course. This study, like Basu et al. [181], based their investigation on performance metrics and
94
features of the computational activities that could be construed as difficult rather than centering
student perspectives or affect.
Several more studies in computation-integrated physics took up non-student-centered approaches
but did allude to students’ experiences at some point in their research processes. Weber
and Wilhelm [183] reviewed broadly the history of computational modeling in physics education,
and they identified several implementation-based hurdles, such as having students invest significant
time to familiarize themselves with the software. This is especially a hurdle in high school settings,
where there might not be time to learn a new programming language within an existing curriculum
and learning to program could be harder at that level. Leary et al. [72] focused on implementationbased
challenges from the perspectives of university faculty. They found several faculty-perceived
challenges: students being resistant to learning a new clunky tool, instructors not being able to
devote enough time for students to get used to a programming language, instructors not having
support from the department, instructors not being able to cover as much content, and instructors
not having time to prepare for the new material. The authors relayed from their participants that
it was hard as an instructor to prepare for computation because you must learn a lot about the
programming language, and it can be hard to make sure it will be accessible to students who have
not used it before.
Other studies highlighted the challenges and benefits to students of integrating computation
into a physics setting. Svensson et al. [184] viewed computation as a type of social semiotic,
meaning it can be used to describe many different phenomena and it can produce many different
answers to many different questions. In their view, becoming skilled at computation is like learning
to communicate with a new language. An example of this is when students comprehend how a
line of code that updates position is connected to the physical relationship between velocity and
position. The authors argued the challenge lay in students having limited use of computation: even
if students are aware of computation’s affordances, they might not be able to use computational
resources skillfully. On the other hand, with proper guidance or computational experience, students
can explore questions and build semiotic resources with code (e.g., conceptual connections and
95
syntactic understanding), and those resources can launch further inquiries. In the authors’ view, we
need to equip students to see the “affordances” of computational integration. We see a worrying
alternative, which is that without an understanding of computation’s benefit, students could adopt
the view that they have an inability to learn languages (like having a “fixed” mindset [2]), and this
could prevent them engaging with computation.
There are additional studies that highlight the student-perceived benefits that computation can
bring to STEM classrooms. In an investigation on the impact of a Python-based, university-level
computational integration [185], the authors reported that students were excited about learning
computation, though the integration didn’t have a significant benefit to learning until the second
year of physics, when students who had learned the computational tools were able to leverage their
proficiency with certain lab tools and data analysis techniques. Caballero et al. [68] highlighted
several other benefits that computation brings to physics. They focused their work on high school
settings where Modeling Instruction [158] was in use, and they argued that computation highlights
relationships between physics concepts, creates dynamic visual models, and can be used to explore
real-world, complex physics problems because of its computing power. Furthermore, they explained
that students who use computation are learning to use the tools that professional scientists use, which
makes physics learning more authentic.
Furthermore, Caballero [70] interviewed professional physicists and physics graduates about
how they use computation in everyday work, in an effort to paint a picture of what students should
be taught in a computation-integrated physics course. The relevant skills (based on the interviews)
were conceptual understanding of physics, writing pseudocode, computational thinking, connecting
ideas between math, physics, and computation, understanding the purpose of using computation
beyond analytic problems, and learning professional programming practices like writing comments
in your code. The interviewees in this study were self-taught programmers, which further shows
there is a need for these types of skills to be introduced into physics curricula.
Caballero et al. [33] summarized the research on computation-integrated STEM classes and
provided several recommendations for future research and implementations. They argued for the
96
need to (1) develop approachable computational models that reflect modern science so that students
can do science using the computation tools, (2) study how computation changes student
attitudes and problem-solving, (3) promote proven learning standards when implementing computational
integration, and (4) support teachers as developers of their own content and members of a
computation-integrating community.
Thus, we see that research into computation-integrated STEM classes has begun to address the
challenges of integration and the impacts on students; however, to our knowledge, there has not
been a study that focuses on students’ perceptions of the integration and the impacts on their affect,
despite its importance in other areas of STEM and multiple calls for research. We intend for this
study to begin to fill this gap and to focus specifically on the students’ perceptions, challenges,
and experiences in a computation-integrated physics course. With this setup in mind, we orient
our research question: What student-perceived, affect-based challenges do high schoolers face in
computation-integrated physics?
In Section 6.2, we describe the methodology that drives our use of the analytic tool and our
choice around research design which is followed by a description of the study context in Section 6.3,
including the teacher’s choices around computational integration. In Section 6.4 we describe our
methods for generating data, creating transcripts, and doing analysis. In Section 6.5 we outline and
describe our results, specifically around student-perceived challenges, and we connect our results
to affective literature in Section 6.6. In Section 6.7 we outline some of the student-perceived benefits
of computation, and in Section 6.8, we discuss our findings and implications of our research.
Finally, we conclude in Section 6.9.
6.2 Methodology
Our focus on student perspectives motivates us to use an interpretivist case study lens in this
research. We describe this work as a case study because of our variation in data sources and
because we aim to capture computational experiences of students in their natural classroom setting.
In particular, we take an interpretivist lens because of our focus on students and their perspectives.
97
The interpretivist approach [50] lends itself well to studies that focus on how people experience
and interpret a phenomenon, as opposed to the phenomenon itself. Because we are aiming to open
an exploration of how students experience computation in their physics class, an interpretivist case
study is ideal for exploring this in an in-depth, qualitative way. Using interpretivist case study, we
would describe the crux of this study as “how students perceive and react to” affect-based challenges
in computation-integrated high school physics with the case being a single physics class taught by
Mr. Buford (pseudonym).
In determining our data sources, we bounded the “reality” of our case to the students themselves
and classroom occurrences [47, 48, 46, 49]. For example, we did not study the home-life of any
students to see how they dealt with their physics obligations outside the classroom. The reason
for this bounding was to privilege data sources closest to the phenomenon: student interviews and
classroom observations. Though students occasionally mentioned out-of-classroom experiences
like school clubs or homework, we trusted the student’s account of the experience rather than joining
them for those experiences. Most of the discussion during class and during interviews revolved
around in-class activities, which was the main way Mr. Buford had integrated computation into his
physics class.
An important part of our methodology is to highlight the perspectives of students, who experience
computation-integrated physics firsthand. It is their perspectives on Mr. Buford’s curriculum
that this chapter is about. We intend for our emphasis on participant interpretation to be coupled
with a detailed discussion of the research context in which our participants operate. In the next
section, we will outline the context of our study and introduce the teacher in whose classroom we
generated our data. The rich contextual description we believe is important for practitioners to
relate their own experience to and for researchers to understand the setting in which our case study
played out.
98
6.3 Context
Mr. Buford teaches physics at Mulberry High School (pseudonym), a suburban, affluent, racially
diverse public high school. He has been teaching at Mulberry for 30 years. In an interview with
Mr. Buford, he commented that he tends to try to lean his teaching style towards problem-solving
and exploration while still covering the material for the AP physics exams, which he estimates
around half of his students elect to take for college credit. He said, “I like to try new stuff,” and he
confessed that he wishes he had more time to do wide-open, curiosity-driven activities in class: “I
think I don’t do enough of, ‘Okay, so here’s this principle that you’re responsible for. Today we’re
going to take some time, and you guys are going to brainstorm an experimental design.’ ”
One of the recent initiatives that Mr. Buford tried to introducewas computation. Hewas inspired
in part by an existing computation-integrated introductory physics curriculum at Michigan State
University (MSU) called Projects and Practices in Physics (P-Cubed) [63]. He began near the end
of the 2017-18 academic year by going through the major physics concepts after the AP Exam. For
each concept, he recalled, “I think about, does this one seem like it’s compatible with writing code
to illustrate. Then I try to come up with a scenario, and this is just piggybacking on the scenarios
that are used in P-Cubed.” For him, the computational activities were meant to be visual, and he
used the GlowScript programming language [186] along with a minimally working program to do
this. A minimally working program [187] is a piece of starter code that will compile without errors
and create a visual; however, there are lines of code that need to be edited or added by students
to create a realistic physical model. For example, Mr. Buford once introduced a program that
showed particles passing through an optical lens without refracting. The task was for the students
to break down their understanding of optics into steps so they could edit the computer program
accordingly and get the particles to refract. Mr. Buford would generally begin the computational
activities by explaining the minimally working program to the entire class. He would also explain
what the output of the code should look like when completed by either running a solution code or
drawing the output on the whiteboard. After Mr. Buford finished this explanation, he distributed
99
the program and students were free to work together to create computational solutions.
During the summer of 2018, Mr. Buford attended a workshop at MSU entitled Integrating
Computation in Science Across Michigan (ICSAM), funded by an NSF grant with the same name.
The weeklong workshop was designed to support high school teachers who wish to integrate
computation into their physics classrooms. During the workshop Mr. Buford collaborated with
other teachers and facilitators on learning to do physics with GlowScript, and by the end of the
week, he made a personalized plan for integrating computation into his curriculum for the upcoming
year. While Mr. Buford had begun integrating computation at the end of the previous year, he began
using it on a regular monthly basis in his AP Physics 1 and AP Physics 2 classes in Fall 2018.
Mr. Buford described in his interview how the computational activities would unfold in class.
Mr. B Grab a laptop and fire it up, and then I go through maybe five minutes—I try to
keep it as short as possible—a little explanation of what we’re doing, and tell [the
students] where to get the starter code and put it in GlowScript and start working.
Generally, Mr. Buford would project the minimally working program, or starter code, which he
wrote himself, up onto the whiteboard, so students could see as he read through the program’s code.
Then he explained how important bits of the program worked, ran the program to show the visual
at its minimally working stage, and described how it would need to change, occasionally drawing
parts of his explanation with diagrams on the whiteboard. Sometimes, he will take a couple minutes
near the end of class to project his solution on the whiteboard, so that he can explain a possible
solution path. Even though Mr. Buford was showing his own solution on the whiteboard, he would
always emphasize that many different solutions exist to the coding projects.
When designing the computational activities, Mr. Buford’s approachwas to build in checkpoints
that students can reach, even if their solutions depart from what he might have in mind. “The ideal to
strive for is, ‘Okay, now that you’ve done that, now do this,’ and actually have several of those in the
bullpen waiting.” When he says this, he is talking about progress students can see in the GlowScript
animation window. In the optics activity for example, students can reach these checkpoints first by
100
causing a light particle to move on screen, and then pass through the lens, and then refract, and then
add more particles to the animation. Mr. Buford’s aim is for students to progress along these steps
so no matter how far they go, they still have some sense of success. His main difficulty with this
approach has been, “students who struggle can still be working on that initial problem,” meaning
the first checkpoint that he described earlier. Some students are not even getting past that first step,
so they don’t get to experience the scaffolded nature of the activity, or even a little bit of tangible
progress.
The process by which Mr. Buford designs these activities is to first write the solution himself,
and then take out the bits and pieces that he thinks the students should be able to rewrite.
Mr. B I’ll try to think of a scenario that’s amusing, at least to me, but still is doable.
The physics is right in the ballpark of the physics they’re supposed to understand.
Then the part that I’m not very good at is how much code do I give them, because
I give them some starter code...I’ll write code that will do what I want it to do,
and then I have to try to pick the parts that I would take out and change... and
then have them try to figure out how to make it work.
Thus, Mr. Buford tries to address multiple concerns when writing these activities. He tries to
balance how much starter code to give students and how much to leave for the students to do, while
at the same time making sure that the difficulty and physics content of the problems are appropriate.
Mr. Buford also made some design choices around when the computational activities feature in
the curriculum.
Mr. B Those coding activities are culminating activities to studying a concept...It’s
usually after we’ve talked about something for a few days or worked on something
for a few days. We’ll do a coding activity if it fits.
Int Is that intentional, to have it be after they’ve learned the concept in part?
Mr. B Yeah...could you use it as a way of developing concepts? I think you probably
could. I just haven’t done that. I haven’t used it that way.
101
The computational activities in Mr. Buford’s class are designed to wrap up a unit. Students have
already spent several days learning about a concept, and then Mr. Buford inserts a computational
activity. He doesn’t use the computation activities to introduce new ideas, rather they are used to
reinforce what students have already learned and to apply those ideas in a new way.
When asked to expand on his views towards computation at the end of the unit, Mr. Buford
talked about the importance of visual modeling and coding skills:
Mr. B I hope it just enhances them thinking about the physics concept that we’re trying
to learn, ideally...I feel like when you’re writing the code for this, you have to
understand how projectile motion works, or you can’t write code that models
that very well...I guess my hope is that that’s what we’re doing is reinforcing the
concepts, and at the same time I just think writing code is just a skill that’s so
valuable in lots of other areas besides just physics.
Mr. Buford wanted the computation to serve as a way to enhance and reinforce conceptual
understanding of physics. His belief is that figuring out the computational activity entails figuring
out the physics within it.
On a separate thread, Mr. Buford wanted the computation to serve as a way for his students to
learn a skill that is widely applicable outside the realm of physics.
Mr. B This computational modeling is so appealing to me. It’s new. I’m not an expert
programmer. I have students that are really good at it. It’s cool to see what
they come up with and how they come up with it. From my perspective, the
problem-solving aspect of that I think is really valuable. The organization and
the logic behind it, oh, my gosh. I think those skills are fantastic to have.
From Mr. Buford’s perspective, these activities were about more than just physics; they were
about building new skills and letting his students’ creativity shine. Mr. Buford chose to not grade
the activities:
102
Mr. B It’s okay to not have a grade assigned to every activity in your class, especially
with students that are in advanced classes. You don’t have to get something for
every little bit of effort that you make, so it can be its own reward.
He believed that the opportunity to play with the program and create something intrinsically
rewarding was enough motivation for his students.
Overall, Mr. Buford designed the computational activities for the ends of units, when students
could reinforce their physics knowledge by applying it to something new and exercise creativity
by exploring the computational environment without pressure to turn in a solution. He viewed
the computational activities as reinforcements of conceptual knowledge but also opportunities to
build crucial computational skills for the future. The computational conditions that Mr. Buford
created in his classroom set up the environment that his students were working in and informed
the perspectives from students that follow in this study. We include Mr. Buford’s perspective here
to help readers understand some of the driving forces behind the development of this instance of
computational integration. In the sections below, we focus our investigation on the perspectives of
Mr. Buford’s students, who are the only ones that can tell us how these newly integrated computational
activities affect their feelings about themselves and their learning in this context.
6.4 Methods
We begin our methods section by introducing our student participants, who will be the main focus
of our study. The students were selected to represent a broad range of prior experiences (in terms of
physics classes and computational exposure) and in-class experience (determined through in-class
observations). The aim was not to generalize our results to any sort of population. Rather, we
chose a diverse set of research participants because we wanted to describe the variety of challenges
students faced in Mr. Buford’s class. The class we focused on in this study was Mr. Buford’s AP
Physics 2 in the 2018-19 academic year. To ensure we respected how the students wished to be
represented in this study [188, 189], we asked the students after data generation to self-describe
103
their gender identity, racial identity, preferred pseudonym, and preferred pronouns.
Otto (he/him) was a junior at Mulberry High School, and he took “regular” Physics 1 with a
different teacher before enrolling in AP Physics 2 with Mr. Buford. He always felt behind and
that this put him at a disadvantage when it came to the computational activities with GlowScript,
because he didn’t have any background with the language. While he did take AP Computer Science
the year before, Otto often felt frustrated that his computational background didn’t seem to help
rather than feeling prepared for GlowScript. Despite his difficulties with GlowScript, he did well
in the class, and tended to approach computational activities with the stance that he could just ask
Mr. Buford as many questions as it took to figure it out. He usually worked together with Blaine,
who also did not take AP Physics 1. Otto self-identified as a white man.
Circe (she/her) was a junior at Mulberry and took AP Physics 1 with Mr. Buford the year
before. She usually worked in a large group of six to eight other students who took AP Physics 1
together, including Beck and Ed, and felt a strong sense of community in the class. Often, Circe
felt that the computational activities were too hard to authentically engage in, so she usually ended
up copying someone else’s code toward the end of the period and passing on a working program
to someone else, calling it a “copy train.” Other than AP Physics 1, Circe had no prior experience
with programming, and she did not feel like she was “cut out” for programming or for physics.
Despite this, she gave a poster presentation with a couple other students at the state capital about
the cool things you can do in physics with GlowScript. Circe self-identified as a cisgender Central
Asian woman.
Beck (he/him) was a junior at Mulberry, and he took AP Physics 1 with Mr. Buford the year
before. He worked in the same large group as Circe, which was usually formed at the start of class
with students dragging three tables together. Beck was an avid coder, and he decided to learn more
GlowScript and do Khan academy physics over the summer after taking AP Physics 1. His dad
was a computer scientist. Beck felt that the computational activities helped him understand physics
concepts better because it was like “explaining it to the computer.” Because he could finish most or
all a computational activity without help and he liked to share his code and explain his thinking to
104
other students, Beck was often a resource for other students. Due to his relatively uniform positivity
with the computational activities, he did not discuss challenges with much depth. He did, however,
describe many positive aspects of computation. As a result, he does not feature in the next section
on challenges but does in later sections of the chapter. Beck self-identified as a white cisgender
man.
Blaine (he/him) was a junior at Mulberry, and he took “regular” Physics 1 together with Otto
before enrolling in Mr. Buford’s AP Physics 2. He took a helpless stance towards the computational
activities, and he was never able to finish an activity during the class period. During one class, he
threw his hands up and said, “what’s the point of learning code? I can draw this on a piece of paper
in fifteen seconds.” He often sat with Otto when doing computational activities and he frequently
expressed apathy towards programming. His only prior experience working with computer code
was when he spent a summer in middle school with his uncle, who worked at a university. Blaine
would try to work through programming tutorials while his uncle worked, but he felt like he didn’t
really understand any of it. Blaine self-identified as a cisgender biracial (Black and white) man.
Joyce (she/her) was a junior at Mulberry, and she took AP Physics 1 with Mr. Buford the year
before. She usually worked by herself but she also socialized with the larger table, especially after
she was done working and ready to share her solution or answer questions. Joyce always finished
the computational activity and was often the first in the class to do so. As a result, she spent a lot
of time explaining her ideas to other students after she was done. Despite this role, she viewed
herself as an average programmer, arguing that she couldn’t solve the problems “in five minutes.”
She was enrolled in AP Computer Science at the same time and thought that the conceptual ideas
from her computer science class helped her when she was using GlowScript. Joyce self-identified
as a cisgender Asian woman.
Ed (she/they) was a junior at Mulberry, and she took AP Physics 1 with Mr. Buford the year
before. She had some additional prior programming experience from participating in robotics club
competitions and writing instructions in code for the robots. Typically, she worked in the large
group with Circe and Beck, and she tried to figure out and understand the computational activities,
105
opting to ask for help from Mr. Buford or peers rather than join the “copy train” when she got
stuck. She said in her interview that she was able to figure out the computational activities around
one-third of the time, and this made her feel like she had the ability to successfully program every
time. She also felt a strong sense of community in the class. Ed self-identified as a Black agender
person. She clarified that she goes by she/they pronouns and suggested for us to pick one to use or
alternate between she and they. We opted to use she/her pronouns alone for consistency.
6.4.1 Data Generation and Transcription
We developed interview protocols and conducted semi-structured interviews [45] with the above
six students in Mr. Buford’s AP Physics 2 class. The interview questions were aimed to elicit
and discuss their feelings about physics class and computational activities in accordance with our
research question. The original interview protocol for students is provided in Appendix A. We
also interviewed Mr. Buford for the context in the previous section, we took field notes during
classroom observations, and we recorded two groups of students working on a computational
activity during one class period. The data sources are summarized in Table 6.1 In this study we
focused our analysis on the six student interviews, though we sometimes used in-class occurrences
to shape interview questions and prompt responses to things that students did or said during the
computational activities. Each student’s interview was treated as an “anchor point” [50] through
which to view challenges from a student’s perspective.
The interviews were transcribed for utterances. This choice was driven by a focus on what
participants said about their experience, which aligns with our choice to use interpretivist case
study. The interviews were conducted to ask about the perspectives of the research participants, and
their comments are taken to represent those perspectives. We understand that interview comments
can only represent how someone feels about their experiences [190], but still we foreground what
the participants said, because their responses were prompted verbally. We included non-verbal
communication in the interview transcripts when it added meaning on its own to what a student
said, such as a facepalm or eye roll.
106
Data Sources
Student
interviews
Six interviews and three follow-up
interviews (follow ups with Otto,
Circe, and Joyce)
Teacher
interview
One interview
Field notes Six class periods
Classroom
recordings
Two group recordings during one
class period, capturing all participants
except Circe and Joyce
Table 6.1: Four types of data sources: student interviews, a teacher interview, field notes, and
classroom recordings.
6.4.2 Data Analysis
To analyze the interview transcripts, we identified episodes from each interview where the discussion
centered around computation, physics, or feelings the student had towards the related
classroom activities. It turned out that each interview yielded ten to fifteen episodes of one to
two minutes each. The goal with chunking our data like this was to group utterances together into
comprehensive statements from the students about their experiences with physics. We carried out
analysis on these episodes by taking notes on the episodes one-by-one, and then tracing out patterns
across the different episodes and interviews, treating each interview as a separate data source from
which to view a given pattern. We named each pattern according to the common experience or
challenge that it represented for students. These names dictated our organization of the first findings
section (Section 6.5). After outlining and describing the student-perceived challenges, we discuss
how the challenges relate to affective constructs, such as mindset, self-efficacy, and self-concept.
107
6.5 Student-Perceived Challenges
We explore the question, What student-perceived, affect-based challenges do high schoolers face
in computation-integrated physics? by presenting the interview data in which our high school
student participants described their experiences and feelings around doing computation in their
physics class. In the results below, we describe patterns in the data that constitute different affective
challenges that students faced when doing computation in Mr. Buford’s class. The challenges listed
below are in no way exhaustive, nor are they necessarily confined to computation-based settings,
but instead represent an initial set of challenges experienced by students in this context. In the order
presented, we address each challenge: Stress/Frustration, Feeling Worse at Physics, Unbelonging
and Stereotypes, Repeated Confusion, Interpreting Code, and Interpretations of Implementation.
6.5.1 Stress/Frustration
One of the main challenges posed was the additional stress that computational activities brought
to students in Mr. Buford’s class. Stress often accompanies new experiences but what made this a
challenge was that students often saw the stress as uncalled for. They felt that they already knew
the relevant physics concepts, and computation was just forcing them to jump through hoops in
order to translate their physics knowledge into code. These experiences were often accompanied by
frustration when difficulty was unexpected. The unexpected frustration and the unnecessary stress
combined to make some students feel unprepared and inclined to give up.
When Circe talked about stress in the interview, she spoke more generally about the stress she
felt during all computational activities and coping strategies she employed.
Circe I feel like it’s just unnecessary stress, and I’m not about to put myself through
that. So I just kind of sit there with the people, and we just talk and wait for one
person to figure it out. Like I said, a copy train.
She felt stressed out during the computation, and her reaction was to not “put herself through
that.” Rather than confront the difficulty and “unnecessary stress” head-on, she opted to copy
108
answers along with the rest of the group. Her response was to disengage, indicating either that she
did not believe she could figure it out or that the stress of sticking it out was not worth it.
At another point in her interview, Circe talked about how the computational activities, or
“code” as she put it, frustrated her. During this discussion the interviewer asked a question to get
an explanation for what she meant.
Int What about the code frustrates you?
Circe It’s like, you think that you should do a certain thing, input a certain value, or a
new part of the thing, and you do that, and it’s just completely wrong. And you
sit there and you’re like, ‘okay, well, freak you, coding!
Circe felt that even when she made everything right in the computer program, or seemingly
right, it ended up being completely wrong. In this way, there was no middle ground when it came
to computation, and this made her feel that she couldn’t do anything right during the activities. Her
reaction was anger (“freak you!”) towards computation. There was no resolution, only frustration
and giving up.
Another student, Ed, also discussed experiencing significant stress, but she did not disengage
as readily as Circe did. Ed’s stress was also “undue” as she said below, and it had to do with a
tension between the computation and Ed’s perceived physics knowledge.
Ed I feel like [computation] causes me, sometimes, a lot of undue stress, which is
like ‘Oh, you don’t know this and this and this.’ So it’s like, ‘you do, you just
think about it in a different way, but that’s not a way that can be programmed on
the platform.’
She felt stressed because of how the computation challenged what she thought of her physics
knowledge. The stress was associated with the feeling of not knowing, and she had to coach
herself out of the difficult feeling essentially by saying, “you do know physics, it’s the computation
109
that’s confusing.” The “undue”-ness of the stress made it seem as if Ed viewed physics-throughcomputation
as inauthentic physics, because she did feel like she got it when it was just physics
without computation.
Ed also felt some unpreparedness for the computational activities. When asked about whether
she saw herself as “good at the coding activities,” she responded by commenting on the frustrations
of seeing the physics content being stripped of its familiarity.
Int Do you think you’re good at the coding activities?
Ed Not really, actually, which is kind of sad for me to be honest, because you have this
interest in something, but it’s back to why physics is so frustrating, because it’s
something that’s like ‘Oh, this is familiar, I know this,’ but then it’s just slightly
slanted a little and just becomes, because you expect it to be this way so much,
when it’s this way, it’s just, you can’t handle it.
She linked her negative self-evaluation to a frustration about physics in general. She compared
her computational frustration to the common experience of learning physics concepts that seem to
defy intuition about how the everyday physical world works. Computation made familiar material
confusing for her. Though the stress functioned in a different way than it did for Circe, the common
thread was that it came from the computation. Ed felt like she built expectations for how her ideas
would play out in GlowScript, but it never seemed to work out—she couldn’t “handle it.” From
this example, we see that Ed dealt with her frustration by separating physics, which was familiar
and understandable, from computation, which defied her expectations and caused her stress.
For Circe and Ed, computation added an extra, needless stress. Their reaction was to find ways
to avoid the stress. For Circe, this meant copying others’ solutions. For Ed, this meant separating
physics and computation mentally as a defense to preserve her self-view as a competent physics
student. Other students also experienced stress but did not articulate it in these terms, such as Blaine
becoming apathetic towards computation after repeatedly getting stuck or Otto feeling stumped and
110
behind because of his lack of previous GlowScript experience. Both of these accounts are described
further in the challenges below.
6.5.2 Feeling Worse at Physics
Another challenge students faced was the way that computation seemed to test and even diminish
the strength of their perceived physics knowledge. This isn’t necessarily a bad feature. After all,
Mr. Buford wanted the computation to “enhance them thinking about the physics concept that we’re
trying to learn... I guess my hope is that that’s what we’re doing is reinforcing the concepts.” For
some students, the “enhancement” of physics thinking instead meant that they had to reconsider
what they knew for the purposes of the computational activity, and this reconsideration often led to
feelings of incompetence at either physics or computation. An example of this challenge is when
Ed felt “undue stress” in the previous subsection. She recalled thinking, “ ‘Oh, you don’t know this
and this and this...You do, you just think about it in a different way, but that’s not a way that can be
programmed on the platform.’ ” She told herself that she did know the relevant physics, just not
in a computational way. In effect, she separated the two domains (computation and physics) in her
mind, so that her difficulty with computation wouldn’t affect her view of her physics competence.
Later in her interview, Ed reflected on how she viewed the connection between computation
and physics. She even suggested that computation changed her view of her physics knowledge.
Ed I think coding definitely affects my perception of my own knowledge about
physics... GlowScript especially, I feel like it caters to a very specific kind of
learner, a very specific way of learning physics...it just requires you to take apart
the numbers in a very strange way. Well, it’s not a strange way, it’s a strange way
for me.
She felt that being good at computation (especially GlowScript-based computation) was like
being good at learning physics in a special way. Ed felt unable to learn in this “strange way.” When
she struggled with computation, it felt like the class had been redesigned with a different type of
111
physics learning, and Ed’s physics knowledge did not line up with the “very specific way of learning
physics.”
For Joyce, getting stuck during computational activities is what made her question her physics
ability. Her self-doubts about her physics knowledge were rooted in not being able to translate the
formulas she knew into code.
Joyce Sometimes it’s made me think that I’m not as good at physics because when
you do everything that seems right on there, or if you use that equation, you get
the right answer on your own, but you can’t program it, then that made me feel
challenging.
Joyce linked her GlowScript-based struggles to feeling bad at physics. This happened when she
felt like she programmed everything right and she knew how to do the problem on paper, but it still
didn’t work on the computer.
The challenges that Joyce and Ed reference in the interview excerpts are not necessarily a bad
thing—in fact it might be a sign of growth and learning that they are being forced to reconsider
their physics knowledge in a way that aligns better with computational demands (assuming these
computational demands are part of an equitable learning environment). However, these experiences
are challenges all the same and must be addressed because they pose real concerns for students. For
both Ed and Joyce, computation forced them to reconsider their physics competency because they
felt incompetent when doing physics with computation. We do not have the data to say whether
or not these feelings of incompetence were temporary, but it is clear that they constituted real
affect-based challenges when doing computational activities. Some students who experienced such
feelings—or even stress and frustration like in Section 6.5.1—struggled with a tension between
their self-views of their computational competence and physics competence. For some students
who found computation to be unexpectedly hard, an appealing narrative could be, “I’m good at
physics already, this is just me being bad at computation.” It is much harder to swallow the pill
labeled, “I’m not as good at physics as I thought.”
112
6.5.3 Unbelonging and Stereotypes
The feeling of not belonging in computation and/or physics was also present in Mr. Buford’s
classroom. This challenge isn’t necessarily brought on by the implementation of a new curriculum,
but difficulty with the learning materials can exacerbate existing feelings of exclusion. Furthermore,
computation-integrated physics is the intersection of two STEM fields (computing and physics) that
have struggled to achieve diverse participation from people with different identities, such as women,
people of color, people with disabilities, LGBTQ+ people, and people of lower socioeconomic
class [33, 191]. As an example of a student feeling out place, we look to Circe, who talked at length
about this when she thought about the computation in Mr. Buford’s class. In the excerpt below,
Circe noticed patterns among her peers related to computation and physics. She used the word
“coding” to refer to the computational activities.
Circe I think I’ve noticed that there’s people who are really good at physics that are also
really good at coding. I think there’s a pattern there. I have a lot of friends who
are really good at coding, and they’re usually really good at physics, and vice
versa. It’s like, I don’t know. I guess it’s all the same kind of brain.
Circe compared being good at physics to being good at computation. She had noticed that a
lot of her friends were good at both, and there seemed to be a connection. It’s “the same kind
of brain,” she said, which indicates that she viewed those peers’ academic abilities as intrinsic
qualities that they had. The language she used suggested that she saw herself on the outside of
this peer-group: “I’ve noticed that there’s people,” “[my] friends,” “they.” By using otherizing
language, she positioned herself as not having the same type of brain, indicating that she saw herself
as not naturally cut out for physics and computation like a lot of her peers seemed to be.
Later in her interview, the conversation again turned to her sense of belonging in physics. Circe
had established earlier that she wasn’t interested in pursuing physics after high school, but she went
on to imply that computation somewhat confirmed her thinking.
113
Circe I don’t know if coding makes me feel like I don’t belong in physics. It doesn’t
make me feel like I do belong in physics.
She was sure that computation was not making her want to be a part of the physics community.
Even though computation might have been integrated into the course as a way of making physics
more authentic to students, its effect on Circe was not beneficial to her sense of belonging.
In naming and characterizing the challenge of not feeling cut out, we acknowledge that many
students choose to leave physics, and this choice can be in line with their interests and based on
a realistic understanding of what it means to do physics and be a part of the physics community.
However, many students can build views of physics or computation based on stereotypes of who
does physics and unrealistic views of what physicists do [192]. One possibility, based on Circe’s
views about the “kind of brain” that is made for physics, is that she bought into some of these
stereotypes, particularly to be good at physics you must be an innate “physics genius” [193].
Similarly, we see stereotypes of programmers and programming show up in the classroom. We
say “programming” here because often the students who adopt these stereotypes do not distinguish
between the computational activities (where student program physics) and more general programming.
An episode that encapsulates this view is when Joyce discussed why she felt like an average
student despite her repeated success at the computational activities.
Joyce I think I’m better than average, which is someone who doesn’t know how to code
at all. But I’m not... I can’t just look at the scenario and just code it in five
minutes. I’m definitely not that kind of person. I don’t know. Just average I
guess.
Joyce believed she was average compared to all programmers, implying that people who can
look at the problem and do it in “five minutes” are the good programmers. None of her physics
classmates were this fast, but she compared herself against this imagined programming genius
anyways. This led Joyce to feel average despite being one of the most competent programmers in
her class.
114
Stereotypes like the genius, five-minute coder can make computation feel inaccessible, and it
can make it hard for students to build a sense of belonging in computation and/or programming. The
challenge of stereotypes lies in this perception of unbelonging. The fact that some students must
overcome this perception and still performwell in class in order to see themselves as computationally
competent is a significant barrier.
The integration of computation into physics leaves the physics classroom open to stereotypes
about programming and computer science. Students have understandings of what it means to contribute
to computer code, and sometimes those understandings are built on unrealistic stereotypes
about who does programming, what programming looks like, and howpeople become programmers.
This is on top of the stereotypes of what it means to do physics, who gets to do physics [194], and
how one can succeed at physics (e.g. “physics genius”). The prospect of computation introducing
even more stereotypes into the physics classroom poses a significant challenge.
6.5.4 Repeated Confusion
Due to the open-ended nature of the computational problems in Mr. Buford’s class, many students
had difficulty working on them. For example, there were many places where students were
confused, encountered errors, or did not know how to proceed. How students reacted in these
moments could lead them to interpret their experiences as failures or could lead them to success
with the problem. The ways that students interpreted the successes and failures outlined below
constituted an affect-based challenge for some students.
From Otto’s experience, he often found success with the computational activities by working
through his difficulties and trying to simplify the problem. Even though the activity was confusing
to him, he felt like he could make sense out of it after thinking about it. He walked us through his
general approach to computation in Mr. Buford’s class.
Otto When I’m working through it, I’ll be like, ‘this is confusing.’ And I’ll start
working through it. I’ll try to simplify it to something that I can understand.
115
Then I’ll usually be able to think about it and be like, ‘Yeah, that makes sense. I
can implement that.’
Otto’s strategy to deal with confusion was to simplify the problem until he understood what
he needed to do. When he said he was “usually” able to figure it out, he indicated that there
was a pattern in his approach to computational activities. The phrase he told himself was, “I can
implement that.” Whether or not he succeeded, Otto usually came to a point during computational
activities when he at least felt like he could, even if he started the problem feeling confused. As a
specific example, he remembered getting stuck and eventually figuring out a complex computational
activity about the motion of charged particles in a magnetic field.
Otto There’s a part where you had to use vector cross products to show the direction in
which it would be moving, from like the direction of...the field and its movement
already. That clicked a little bit after I realized how that function worked.
Though he encountered a confusing function, he figured it out. The function in question was
the cross product function. His success in getting the function to work and understanding it is
evidence of Otto’s persistence in face of his typical computation-based confusion.
For Ed, experiences of success were more rare but not unheard of. When she did finish a
computational problem it made her feel like she could do any of them.
Ed On like one out of the three times we coded, each of those one times where
I’ve actually finished the whole thing, that always makes me feel like, ‘well you
finished that one, you can probably do all of these.’
Approximately one out of three times, Ed could figure out the code, and it was a big confidence
boost. For her, it was the act of completing the program that made her feel the sense of attainment.
Though she usually didn’t finish, on the times that she did, it was a reaffirmation that she had the
ability to succeed at doing the computational activities. The intermittent successes sustained her.
116
Blaine, on the other hand, discussed how he had recently given up on engaging with computational
activities because of his failures to achieve anything that he perceived as progress.
Blaine I mean, I would try if I could literally get like anything. But since I literally can’t
get anything but a blank screen, I don’t really try to do any more cause I’ll put in
a hundred things and then I’ll just get a blank screen or I’ll get some error.
No matter what he tried, Blaine always got the same result: “a blank screen or some error.”
Both results are associated with a non-working animation, a fate to which Blaine had resigned
himself. Not only was this a wholly negative self-evaluation, but it was also a source of apathy
and disengagement for Blaine. He experienced repeated roadblocks, and he came to associate his
relationship to computation with incompetence.
Blaine I just, I just don’t even care. I’m like ‘whatever dude. I can’t do this shit.’
He felt like he couldn’t do the activities to the point that he just “[didn’t] even care” anymore.
He provided a sharply negative statement, saying, “I can’t do this shit.” He had no successes with
computation, and by the point of the interview he had given up entirely. Blaine was one of the two
students who did not take AP Physics 1, so his first exposure to computation-integrated physics was
Mr. Buford’s class. This points to the importance of having positive experiences and moments of
success when learning a new curriculum as suggested by Kinnunen and Simon [94]. Blaine had no
memories of success and articulated no hope that he would improve.
Some of these students experienced setbacks or confusion in the computational activities. While
Otto persisted through such a moment and eventually figured it out, Blaine interpreted his lack of
computational success with feelings of apathy and inability. Ed had enough positive experiences to
feel competent, but all the same itwas concerning that some of Mr. Buford’s studentswere not having
any positive experiences with computation. The prospect of students developing negative views
about computation after repeatedly failing at computational tasks presents a unique challenge,
especially when these failures are tied up with their first impression of computation-integrated
physics.
117
6.5.5 Interpreting Code
Another common challenge was brought on by the need to interpret code and errors in GlowScript.
This has been previously documented with students learning physics through VPython [195].
Students in Mr. Buford’s class often felt that they had a decent understanding of how to use the
relevant physics and apply it to the context in which Mr. Buford set up the computational activity.
The challenge came when they received an error message or had to interpret or write code to execute
their ideas. The elusive meaning of the error message or the challenge of using GlowScript syntax
was enough to derail the activity for these students.
For Blaine, the computational activity that he described involved modeling rays of light passing
through an optical lens. He had trouble with the first step because he couldn’t figure out how to use
GlowScript to animate a line to represent the light ray.
Blaine I feel like I’d like [the computational activities] if I knew what I was doing. I
literally wrote ((laughter)). I literally wrote ‘line’, just like ‘line period’, to try
and get a straight line. I don’t know anything!
When he talked about what it was like to troubleshoot after getting stuck, he laughed about
how little he understood GlowScript. He guessed at what the proper syntax would be because
he did not know any GlowScript commands for creating something that looked like a line. He
attributed the whole experience to his lack of knowledge: “I don’t know anything!” This admission
was reaffirmed below when Blaine described his inability to interpret an error message because
it referred to “line 17”, or the seventeenth line of the computer program, which he was unable to
interpret.
Blaine I’ll get some error. ‘Line 17.’ Well I don’t know! I don’t know what line 17 is,
man.
In this case, Blaine could not interpret the error message that the computer provided. His
responses about “not knowing what line 17 is” and “not knowing anything” indicate that Blaine felt
118
that he just did not know enough about the GlowScript language to do computation.
Otto had a similar, though less severe, reaction to getting stuck on using GlowScript. He
discussed the process of figuring out the relevant physics but not being able to translate his ideas
into code.
Otto The electron moving through the magnetic field... I know what direction it should
be moving and everything, how its velocity should be affecting everything. But
I don’t know how to put that into computer words... Even when I know what
should be happening, it just wasn’t happening, because I don’t know how to use
GlowScript that well.
He explained the roadblock: “I don’t not know how to use GlowScript that well.” Though his
programming inexperience prevented him from succeeding, Otto acknowledged that he did know
the ins and outs of the non-computational part of the physics problem. He contrasted what he
did and did not know, saying, “but I don’t know how to put that into computer words.” Otto’s
experience was different from Blaine’s because Otto was able to identify what he knew about the
problem and what exactly he got stuck on. This shows that the challenge of interpreting code can
present differently for different student and different contexts, but in each case it can present a
barrier all the same.
Circe also described her challenges with understanding the code. She recalled starting a
computational activity and immediately feeling lost.
Circe I feel like something like coding can’t help you understand physics better if you
don’t understand what the code means in general. He gives us the code to start off
with, but none of us really understand what that means. So we look at [the starter
code] and we’re like, ‘what does any of that mean?’ So then you add things to
that, but you don’t understand why.
She often felt that she did not understand the program, or starter code, which Mr. Buford
distributed to be worked on. This had the perceived effect of preventing Circe from learning
119
physics through computation. She even described attempting to engage with the activity and add
her own code but feeling confused and directionless. Her understanding of the computation was
that success depended on computational literacy of GlowScript and that some students did not have
the tools to engage on that level. Her use of “we” indicates that this experience of confusion was
shared among her peers and her.
Even for students who had seen programming before, using the GlowScript language, structures,
and syntax was still a challenge. For example, Otto had taken a physics class and a computer science
class before enrolling inMr. Buford’s physics class. Despite these experiences with the “ingredients”
of computation-integrated physics, Otto still felt like Mr. Buford’s version of computation was new.
Otto It’s a lot more physical in GlowScript because in the other class I took with
coding, it was more just data and lists and whatever. But this you’re having a
particle moving through whatever so you have to use like vectors and all that.
That’s new to me. I haven’t done anything involving movement and displays and
that.
He said GlowScript physics was unique because of the movement and the visual nature of the
activity, whereas computer science was about “data and lists.” Computation in physics felt totally
new to him, from the language (GlowScript) to the conceptual features (e.g., vectors, movement,
animation). Doing computation with GlowScript was different from both physics and computer
science in Otto’s view, and this unfamiliarity made it difficult for him. The difficulty manifested
when he had to combine physics with computation: “I know what direction it should be moving
and everything, how its velocity should be affecting everything. But I don’t know how to put that
into computer words.”
For Otto, who had prior experience with both physics and computer science, working with
GlowScript still felt totally new, and he found it difficult to put what he knew into “computer
words.” This indicates that interpreting code might be a significant challenge for all students to
some degree and that prior experiences with code do not directly translate to success at computation-
120
integrated physics activities. Otto pointed to the specific features of the integrated format (making
particles move, using vectors, making displayed simulations) that were still a challenge for him.
Just because he had the separate physics and computation pieces, it did not mean that Otto felt able
to combine them, and he still struggled with translating the ideas into the “computer words.”
Blaine, Otto, and Circe shared above howthey got stuck because of a difficulty with the computer
program, not the physics concepts. The impact was twofold. First, it stopped these students in their
tracks when they did not know how to deal with code during a computational activity. Second, it
caused negative affective responses, like Blaine’s self-evaluation (“I don’t know anything!”) and
Circe’s indictment of the activity itself (“coding can’t help you understand physics better if you
don’t understand what the code means”).
6.5.6 Interpretations of Implementation
There were also some implementation-based challenges that students faced in Mr. Buford’s class.
These were related directly to the students’ interpretations of the computational activities and
pedagogical choices made by Mr. Buford. We share these not as a critique of Mr. Buford’s
implementation but as a way to illustrate the variety of challenges that can arise for students and
how those can depend on the context.
6.5.6.1 Assessment and Motivation
In Mr. Buford’s class, the computational activities were intentionally not graded. Mr. Buford felt
that because the activities were new and not explicitly a part of the AP curriculum, they could
go ungraded and simply serve as opportunities for students to engage with physics concepts more
deeply than they normally would. He explicitly said in his interview, “You don’t have to get
something for every little bit of effort that you make, so it can be its own reward,” indicating that
he viewed the computational activities as intrinsically motivating.
In the interviews with students, we sawthat students understood this motivation and experienced
it for themselves at times. For example, Ed expressed a similar viewof computation, that the purpose
121
was to get a better grasp on programming concepts, which in turn helped her see the connection
between formulas and actual physics phenomena. We provide the excerpt below.
Ed And just seeing how just changing a couple of numbers could change the entirety
of the coding was interesting... That was helpful for me to get the whole concept
of coding.
However, at a different point in the interview, she articulated a much bleaker view of what
computation was all about, referencing the grading policy.
Ed [Coding activities] are just really tedious. When I’m doing it, I just feel like
there’s something else I could be doing... I feel like coding is like something you
kind of know... and it just feels kind of like busy work, but not busy work that
he’s going to grade, so it just feels useless.
The goal of computation, as Ed articulated here, was nothing! In her view, because it was not
graded, there was no point in engaging. The computation was “tedious...busy-work” which made
Ed want to disengage even more. Had the activities been graded, she might still have found them
tedious, but the fact that they were ungraded meant they were “useless”, at least in how Ed viewed
them in this moment.
Ed’s frustration at computation did not last throughout her interview, but the above excerpt
demonstrates that the ungraded nature of computation in Mr. Buford’s class can contribute to a
feeling that computational activities serve no purpose. Feelings like this can impact students’
motivation (“feels useless”), and given the open-ended, ungraded design of many computational
problems, motivation was important for students to want to explore the activities.
As Mr. Buford indicated, it was reasonable to not have every single activity be graded or
externally motivated. In fact, we can imagine several arguments for leaving computational activities
ungraded. For example, teachers might want to reduce the pressure and stress of grades while
students are doing a novel, unfamiliar task. However, as Ed’s response indicates, there is a need for
122
messaging about why students are asked to complete an ungraded activity, why the activity is not
graded, and why engaging in the activity can still provide benefits to students.
6.5.6.2 Solutions and “Right” Answers
When introducing the computational activities, Mr. Buford would explain the minimally working
program and show students what the output of the code should be when fully working (either by
drawing it on the whiteboard or showing the output from his solution code). He intended this
as a way to show students what the end product should be in an otherwise open-ended activity.
Mr. Buford was careful in his explanations to emphasize that there could be multiple right answers
or solution paths to the computational activities.
Despite his caution and explanation of multiple paths, knowing that Mr. Buford had a “correct
solution” posed an affective challenge for some of his students. For example, Circe was a student
who viewed “success” at the computational activity as “being right,” and she said that her own
ideas were always “wrong” when it came to computation. Below, the interviewer asked her about
this view.
Int How do you know it’s just wrong?
Circe Because you see the answers. I guess there’s multiple answers, so you might not
be completely wrong...but the one that we’re given, or the one that the smartest
kid in class figures out is different than the ones that we had.
She articulated that the goal was to get the answer that the teacher had or the smartest kid in
class had. Anything else she saw as “wrong.” She even acknowledged that there could have been
multiple solution paths, but she still interpreted a mismatch in her answers as “not completely
wrong” and set up this comparison for her work versus a “smartest” or “given” (teacher’s) solution.
Circe reasoned that “because you see the answers,” hers (which did not match) must be wrong.
From this perspective, showing the final output to the class might inhibit students’ ability to see
paths beyond the one they are shown and might pose an affective challenge for students who need
123
to reckon with the tension between being right and engaging openly with the problem. This desire
to be right also can prevent students from exploring the problem setting and making mistakes from
which they can learn important aspects of the problem.
That said, we do not know what would have happened if Mr. Buford did not provide the
output for the computation problems. Without knowing the output, students could potentially
struggle more with interpreting the code or might encounter more confusing moments as they work
through the open-ended problems. These implementation-based challenges are directly related to
choices that Mr. Buford made in integrating computation into his physics course; however, they do
not represent all the challenges related to implementation that students could face. More studies
should be done in a variety of contexts that look at students’ other implementation-based challenges.
6.6 Connection Between Challenges and Theory
From students’ interviews, we showed that they faced a variety of challenges when computation
was integrated into their physics class. While it was not the explicit focus of this study, the students’
statements point to theoretical constructs in education research that might help better understand
students’ experiences and how to help address these challenges in the classroom. Specifically, we
found ties between students’ comments, their mindset, self-concept, and self-efficacy.
Briefly, self-efficacy is a person’s belief in their own ability to complete a task [113, 112].
Within the context of a computation-integrated physics classroom, self-efficacy would address the
question of “how well can I do computation in this physics class?” Mindset, at its simplest, is
a person’s belief in their ability to change their own traits/competencies [2]; thus, mindset would
address the question of “how much can I improve at doing computation?” In contrast, self-concept
is “a person’s perception of self...inferred from their responses to situations” (page 411) [119].
Rather than being task related (as self-efficacy), self-concept is in relation to an entire subject area.
This would address the question of “how is doing computation related to me?” In the subsections
below, we define in more detail each of these constructs and how they are related to our data. We
then discuss the overlaps in these constructs and the implications for instructors and researchers.
124
6.6.1 Self-efficacy
Originally developed by Bandura, self-efficacy is “concerned with judgments of how well one can
execute courses of action required to deal with prospective situations” [112]. In discussing how
self-efficacy relates to students, Bandura suggested that it contributes to motivation and confidence
within a given academic subject: “The higher the students’ beliefs in their efficacy to regulate their
motivation and learning activities, the more assured they are in their efficacy to master academic
subjects” (page 18) [113].
Since its introduction, self-efficacy has been broken down into four sources: mastery experiences,
vicarious learning, social persuasion, and physiological state [113]. We looked at how
the four sources have been used in STEM education research to gain a deeper view of what they
could mean for a computation-integrated physics context [104, 105]. Mastery experiences refer to
the impact of successes and failure: “successes heighten perceived self-efficacy; repeated failures
lower it, especially if failures occur early in the course of events and do not reflect lack of effort
or adverse external circumstances” [113]. In our case, completing a coding task could count as
a mastery experience, or receiving an error message from the coding program could be seen as
a “failure.” Vicarious learning is when a student makes an adjustment to their self-efficacy after
witnessing a peer’s performance. For example, a peer’s success at a computational task can raise
self-efficacy if the student then thinks they can succeed too, but seeing a peer fail despite effort
can lower the observer’s self-efficacy for related computational tasks. Social persuasion is about
external appraisals of ability that a student then internalizes into their self-efficacy. Evaluations
can come from peers, authority figures, or other participants in the domain where the student must
perform. Social persuasion need not be verbal or direct, and its effect depends mainly on how the
student perceives it. Physiological state refers mainly to stress “as an ominous sign of vulnerability
to dysfunction” [113]. Students, when they are stressed, expect to perform worse, whereas when
they are calm and clear-headed they might feel a boost to self-efficacy.
A few examples from computation education research show how self-efficacy can be used in
computational settings and how it can reveal information about student learning. Self-efficacy
125
was employed by Lishinski et al. [92], who viewed self-efficacy as a reciprocal feedback loop,
where self-efficacy judgments based on affective responses can have a long term effect on learning
outcomes. The authors found that previous programming experiences impacted future performance
in part due the effect that past experiences had on self-efficacy, whether positive or negative. Kinnunen
and Simon [93] used self-efficacy to describe students’ affective responses to a computational
assignment in an introductory-level university computer science class. When students made an
affective self-assessment, the authors were able to describe it in terms of self-efficacy, indicating a
connection between self-efficacy and the act of affect-based evaluations of oneself. In a follow-up
study [94], Kinnunen and Simon used the four sources [113] to understand how self-efficacy was
tied to experiences that students had in the course. They also considered in their framework how
self-efficacy could evolve in response to experiences and what could set this evolution in motion.
A year later, the same authors [95] returned to self-efficacy, this time using it to describe emotionally
charged events they observed where students evaluated their own abilities and consequently
altered or reinforced their self-efficacy for programming. The evolution of how Kinnunen and Simon
[93, 94, 95] used self-efficacy to explore programming experiences demonstrates a precedent
for connecting self-efficacy (and its sources) to computation.
We can see these sources of self-efficacy in our data, with examples that might be either
contributing to or degrading students’ self-efficacy in computation. For example, in Section 6.5.4,
we saw Blaine, Ed, and Otto take on very different responses when faced with confusion and
uncertainty in the coding activities. Otto demonstrated a persistence in his approach to the problems,
experienced multiple successes (mastery experiences) with the computation problems, and often
said high self-efficacy statements like “I can implement that”. In contrast, Blaine experienced very
few mastery experiences. This connects to several of his statements which aligned with a lack of
computational self-efficacy. He said, “I can’t do this shit” and “I don’t really try to do any more
cause I’ll put in a hundred things and then I’ll just get a blank screen or I’ll get some error,” which
directly tie his lack of success (‘blank screen’ or “get some error”) to his belief that he cannot code
or cannot make progress. That said, Ed’s experience demonstrated that mastery experiences do
126
not have to be all or nothing. Ed had some moments of success with the code, but she indicated
that it was only one in three activities. However, even those moments of success made her feel like
she could code and contributed to her belief that “you finished that one, you can probably do all
of these”. All three of these students pointed to the importance of mastery experiences in building
views related to self-efficacy, especially Ed’s case, which highlighted that not all computational
experiences need to be successful.
There were also indications of the other sources of self-efficacy in our data. For example, Joyce
referenced the stereotype of a “fast coder” in her statements in Section 6.5.3, saying that she was
simply average because she could not “just look at the scenario and just code it in five minutes.”
Even though Mr. Buford never set any expectations about how fast students were expected to code,
Joyce still had this idea that the good coders were able to just look at the code and do it. Research
has shown that perceptions like these can come from societal stereotypes, media portrayals of
programmers, interactions with peers, and other forms of social persuasion [196, 194, 197, 198].
Social comparison of programming speed has been shown to reduce self-efficacy [196]. Ultimately,
this perception encompassed howJoyce sawherself and howshe evaluated her skill. We also showed
that Circe and Ed described computation as a stressful, frustrating activity in Section 6.5.1. This
outlines one of the physiological states that can contribute to self-efficacy. If a student’s experiences
of coding are all taking place in a highly stressful, tense physiological state, then that reduces their
self-efficacy and garners a feeling of inability to complete the task. We saw this with Circe, who
directly stated that she’s “not going to put [herself] through that” because the programming is “just
unnecessary stress.”
The sources of self-efficacy open questions for additional research in computation-integrated
classrooms. For example, what tasks and what grain-size lead to mastery experiences? Does interpreting
an error message successfully count as a mastery experience or does the whole program
have to be completed for students to feel successful? How can we as instructors and facilitators help
students see their success in each of these moments? How can we help students approach computation
without a stressful physiological response, while at the same time not seeing computation as
127
“useless” or “busy work”? At this point, we do not have answers to these questions, but our results
from the challenges students face would indicate that more research is needed in this area.
6.6.2 Mindset
Dweck [2] defined mindset in terms of self-beliefs about the mutability of abilities and delineated
between fixed mindsets and growth mindsets. She argued that a fixed mindset is detrimental to
learning because students who embrace this mindset lose motivation more easily and they are
harsher judges of self when faced with adversity. On the other hand, students who embrace a
growth mindset build motivation to improve when they experience failures. Blackwell et al. [4]
provided a review of perspectives a student would hold depending on how their statements and
actions aligned with mindset. The most fundamental perspective is that growth mindset aligns
with a belief that one can improve their intelligence through effort, whereas fixed mindset relates
to believing that intelligence is unchangeable. Growth mindset is about studying to learn, seeing
mistakes as learning opportunities, believing that effort is good because it makes you smarter, and
seeing knowledge as something that can be worked for [2, 4]. Fixed mindset is about studying
to prove smarts or superiority, avoiding mistakes for fear of being seen as stupid, believing that
too much effort signifies lack of intelligence, and seeing knowledge as something that comes from
authority figures [2, 4]. When students fail, some might react in ways aligned with growth mindset,
believing they need to change their studying strategies. Some students might react to failure in ways
aligned with fixed mindset, believing they failed because they are stupid or because the assessment
was unfair. As a disclaimer, the theory of mindset is flexible, meaning that reacting in a “fixed
mindset” way does not mean one will always react in that fashion [2]. Also, mindsets can vary
between contexts or even within a single context, meaning people can hold views related to both
growth and fixed mindset about different subject matters or even at the same time [2].
From the literature, mindset has been used in some initial studies to describe students’ approaches
to computation. In one study, Scott and Ghinea [106] set out to discover whether
programming-specific mindset could be differentiated from general mindset for school. They dis-
128
covered that the unique nature of programming activities led students to embrace a specific mindset
for programming, different from a more general, school-based mindset. To track learning in connection
with mindset, an intervention study was devised by Cutts et al. [107]. They intervened in an
introductory university programming class by having tutors teach mindset-related strategies. The
issue of stuckness was focal: the students’ mindset-related views hinged on whether they attributed
stuckness to internal factors (leading to an embrace of fixed mindset) or external factors (leading
to an embrace of growth mindset). These findings suggest that mindset-related views could change
or even develop anew when computation gets introduced into a physics curriculum. Lodi [108]
performed a similar study to Cutts et al. [107], but he focused on high school students and sought to
understand how the computer science curriculum impacted mindset-related views. He argued that
students with learning-oriented goals (e.g., aiming to learn and be challenged) aligned their views
with growth mindset, whereas students with performance oriented goals (e.g., aiming to score well
and avoid challenges) aligned their views with fixed mindset. These studies highlighted some of
the same features of mindset that emerged from Dweck [2] and Blackwell et al. [4], which gives us
precedent for applying these theories to a computational education setting.
In our data, we saw similar perspectives mirrored in how students articulated challenges in
Mr. Buford’s class. For example, in Section 6.5.6, Circe recognized certain answers as “right”, and
those answers came from the teacher or the smartest students in class. This aligns with the fixed
mindset tendency to look to authority/expert figures (like teachers) as the only trusted source of
knowledge. Tendencies of people to value accomplishments and grades because they signify high
intelligence align with aspects of fixed mindset, whereas tendencies to value learning because of
its connection to improving intelligence align with aspects of growth mindset. Circe articulated
a tendency to consult the teacher’s solution to see if hers was right, which represents a potential
challenge in other settings where computational activities are designed to have multiple solutions
and unanswered questions built into the learning process. For students who embrace a fixed mindset
at times, this design could present significant barriers to success.
Another example comes from Sections 6.5.3 and 6.5.2, where we observed Circe and Ed
129
provide similar views about feeling out of place or not knowing how to proceed when confronted
with computational challenges. For Circe, feeling out of place was tied with her belief that being
“really good at physics and coding” meant having “the same kind of brain.” When students take up
the view that they need to be built a certain way in order to succeed at physics and/or computation,
they align their views with fixed mindset, which at its core says that intelligence is an inherent
characteristic and impossible to change. For Ed, she felt that her understanding of physics was
questioned or alienated when she had to do physics with computational tools, to the point that
she believed she “just [thought] about [the material] in a different way,” and she emphasized the
computation was only strange for her. This distancing that Ed does indicates that the challenge
was related to fixed mindset, because she attributed her difficulties to her self-perceived faulty
way of thinking, and she viewed computational learning as “catered to a specific kind of learner,”
distancing herself from the opportunity for her to learn during those activities.
Lastly, we return to Section 6.5.4 to compare the mindsets that described what Otto and Blaine
said when faced with confusion. We focus on their difference in persistence. Both students
articulated a point of confusion or stuckness, but Otto’s response was to embrace the challenge
(“I’ll start working through it, I’ll try to simplify it”), whereas Blaine’s response was to give up
(“I don’t really try to do any more”). For Otto, the setback was an opportunity to learn, which
aligns with growth mindset, whereas for Blaine, the setback was paralyzing, which aligns with fixed
mindset. The contrast between how students respond to these challenges is closely aligned with
mindset theory, which indicates that mindset can be key in explaining whether students succeed at
overcoming challenges in Mr. Buford’s computational activities.
Our work suggests building on the premise that mindset is linked to how students respond to
computational challenges. For example, how do students develop views related to mindset theory in
their computational work? Are there pivotal experiences (like mastery experiences for self-efficacy)
that impact students’ mindsets in significant ways? Our data would also suggest observing how
students treat computational challenges differently in the wake of mindset interventions, similar
to many others’ recommendations [124, 125, 126, 127, 128]. We also recommend studies on
130
designing opportunities for students to embrace growth mindset could help students in other ways
in a computation-integrated physics context. We do not have the answers, but our results from the
challenges students face would indicate that more research is needed in this area.
6.6.3 Self-concept
Shavelson et al. [119] emphasized that self-concept is organized, or structured by domain, meaning
that a person can have a different self-view depending on the context (e.g., physics class) and focus
(e.g., computational activities). It is developmental, in that a person builds or develops a narrative
about oneself in a particular set of contexts. Though it was at first used to describe broad selfviews
(i.e., self-esteem), self-concept was only later used to examine academic realms. Marsh and
Craven [118] argued that what distinguishes academic self-concept is that students evaluate their
performance in comparison to their performance in other domains, their peers’ performances, and
their internal standards of performance quality. Though focused on evaluation, it is distinguished
from self-efficacy because the evaluation of performance is stabilized by previous evaluations and
exists broadly for an entire school subject, whereas a self-efficacy judgment has more to do with
prospective situations in a given academic domain. This would make the difference between selfconcept
and self-efficacy threefold: (1) domain-level versus task-level evaluation (2) evaluation of
past performance versus prospective performance, and (3) incorporation of evaluation into a sense
of self versus a sense of ability.
In a theory-building paper by Brunner et al. [120], they propose and evaluate the effectiveness
of a model for self-concept. The authors suggest using a first-order model (e.g., focusing broadly
on academic self-concept) or a nested model (e.g., considering broad academic self-concept and
math self-concept). They emphasize that self-concept can be split into separate self-concepts for
each academic domain when using the nested model. In our context, this would indicate that
this model of self-concept would be appropriate for the students who perceive computation as a
separate domain from physics (not integrated into the domain of physics as a learning tool). This
is in opposition to how Mr. Buford, the teacher, framed computation in his classroom.
131
While self-concept has not been used in computation research, there have been examples in
other areas of education research. For instance, Chen and Xu [111] studied self-concept for junior
high school English and its components: listening, speaking, reading, and writing. The qualitative
case study of multiple students demonstrated how students with different self-concepts for different
components can have drastically different trajectories in class, pointing to the complicated nature of
self-concept for specific academic domains and activities. Espinosa [109] produced a quantitative
study about cataloguing a variety of factors that build into academic STEM self-concept for college
students. The core of her methods addressed self-concept from its most basic definition: evaluation
of oneself. Mardiningrum [110] produced a case study on two participants in a university student
theater club. The collaborative nature of this environment made social interaction a focal aspect
of the participants’ self-concepts. In a learning environment that uses group-based computation
activities, we would expect social interaction to contribute to self-concept.
The studies above provide insight for how we might apply self-concept to a computationintegrated
physics setting. The construct has not been used in this type of environment before,
but we know that to apply it we need to focus on moments of self-evaluation [109], accounts of
social interactions [110], and nuances in how students see themselves in relation to computational
activities, computation, and physics as a whole [120, 111]. This construct adds to our study because
it can help us frame the way students discuss their feelings about computational experiences in
a way that involves perceiving their role, as opposed to perceiving their ability (self-efficacy) or
perceiving the malleability (or rigidity, in the case of fixed mindset) of their role and/or ability
(mindset).
For example, in Section 6.5.3, Circe articulated that computation “doesn’t make [her] feel like
[she] belongs in physics.” When students feel that they don’t belong in a computation-integrated
physics environment, they can also feel that they were not meant to belong there, as evidenced by
Circe’s later reflection on not having the brain for computation: “there’s people who are really good
at physics that are also really good at coding...I guess it’s all the same kind of brain.” This feeling
is related to computation and/or physics self-concept because it could be framed as a perception
132
of self in relation to a school subject. Feeling out of place in comparison to peers is part of
self-concept [119]. The challenge lies in the potential for students to feel this way and lose interest
in physics before gaining a realistic view of what it means to do physics.
Another challenge tied up with self-concept is Interpreting Code. Blaine lamented in Section
6.5.5 about his feeling of inability to understand what the code meant. For Blaine, it was about
feeling unable to make any progress on the activity and unable to interpret error messages. These
roadblocks produced an affective response: Blaine said, “I don’t know anything!” This evaluation
of self in relation to computation indicates a self-concept judgment. Blaine felt stupid when doing
computation.
Similarly, Blaine’s made statements related to low self-concept in Section 6.5.4. Here, he
outlined accumulation of negative experiences. Accumulations and patterns of experience are part
of how a student builds self-concept for a school subject [119, 118]. Blaine is a student who
identified a pattern in his computational experiences: “I don’t really try to do any more cause I’ll
put in a hundred things and then I’ll just get a blank screen or I’ll get some error.” The repeated
roadblocks with no success at overcoming them led Blaine to believe he “literally can’t get anything
but a blank screen.” He suggested that he had experienced computation enough already to develop
and hold this belief. Self-concept is tied to this challenge because the way Blaine’s statements align
with negative self-concept is tied to this pattern of experiences. It is important to acknowledge the
ramifications when students deal with challenges unsuccessfully like this, one consequence being
a development of self-views aligned with low self-concept.
As a final example, we look at Ed’s delineation between physics and computation in Sections
6.5.1 and 6.5.2. In these sections, Ed said that the way she thought about physics “can’t be
programmed.” This sends the message that not all physics knowledge is meant for a computer program,
in particular Ed’s physics knowledge was not meant for a computer program. One possible
theory-based explanation for this belief is that when Ed encountered a new, difficult type of physics
(i.e., physics through computation) Ed protected her physics self-concept by building a separate,
low self-concept for computational endeavors (or “GlowScript”, “coding”, etc.). This separation
133
can mean that some students do not let themselves develop as doers of computation, and it can
prevent them from learning on days when this is an aspect of their physics class.
Self-concept suggests that students can develop a view of themselves in physics that is different
from the view of themselves when doing computational activities, which validates the possibility
of Ed’s experience with separating the two domains. Because self-concept has not been applied to
computation-integrated physics before, ourwork indicates it might be a viable lens for understanding
how students are internalizing their experiences in computation. For example, future work could
point to the process by which self-views related to self-concept develop in these settings, how
students reconcile their views of the two different domains (physics and computation), and how
that fits in with the theory of broader academic self-concept.
6.6.4 Intersection of Self-efficacy, Mindset, and Self-concept
In talking about the challenges that they faced, the students in our data made statements that point to
their views related to the theories of self-efficacy, mindset, and self-concept. While we previously
discussed these constructs as separate ideas, we want to emphasize that these are not independent
theories or constructs. In fact, the overlap between these constructs illuminates avenues for future
research, curriculum design, and pedagogy.
For example, we can see aspects of all three constructs in how Blaine faces the Repeated
Confusion challenge. Blaine described how he experienced a series of failures related to doing
computation: “I literally can’t get anything but a blank screen...I’ll put in a hundred things and then
I’ll just get a blank screen or I’ll get some error.” These failures fit narratives about the reduction of
self-efficacy and negative impact on self-concept, and the way Blaine articulates them aligns with
the language of fixed mindset. Each framing provides a different insight into Blaine’s experience.
The self-efficacy framing shows the impact of serial mastery experiences on views related to selfefficacy,
as shown when Blaine described how he felt that he “literally can’t get anything but a
blank screen” after repeatedly failing to make progress in the computational activity. The selfconcept
framing shows how a pattern of negative experiences can come to define what computation
134
means to a student, as shown when Blaine expressed apathy when describing his relationship with
computation: “I just don’t even care. I’m like ‘whatever dude. I can’t do this shit.” The mindset
framing brings focus to the parts of Blaine’s behavior related to aspects of mindset, specifically
the reduction of effort in response to his failures, which relates to fixed mindset: “I don’t really try
to do any more cause I’ll put in a hundred things and then I’ll just get a blank screen or I’ll get
some error.” From this one example, we can see that the three frameworks overlap and build into
one another. Blaine’s repeated failures to make progress with the computation led to a reduction of
effort and no other change in strategy, aligning with aspects of fixed mindset. This accumulation
of failures also ties to his view of his work and of himself—views which align with having a lack
of self-efficacy and/or a low self-concept for computation.
This illustrates how the theoretical lenses can overlap and provide a fuller picture of the impact
that the affect-based challenges can have on students. We use all three to highlight different views
on the same individual experiences, but they provide varied angles from which to understand what
is going on with the students in our study. That said, this study only provides an initial window
into how these frameworks relate to one another, and we suggest future research specifically focus
on how each framework fits with one another in this context, how theory-based interventions might
impact students’ perceptions, and how these frameworks might be leveraged to better understand
computation-integrated classrooms. We view the presence of many angles as a way to identify
jumping-off points for further research on affect-based learning and challenges, which is sorely
needed and which we highlight in the discussion section. However, we first highlight some positive
experiences that students recounted in their interviews. These did not fit in with our challenges,
but still provide a unique perspective on what students experience and how computation can be
beneficial, according to students.
6.7 Positive Student Experiences
Along with the challenges students faced and recalled in their interviews, there were also indications
of positive experiences brought on by computation. In this section, we outline a handful of beneficial
135
impacts of computational integration that students interpreted. Afterwards, we discuss how they
relate to some of the goals that Mr. Buford set out to achieve by introducing computational activities
to his class.
We begin with a comment from Ed that demonstrates how she learned about using computation
to see physics. She describes getting “the whole concept of coding” through engaging in a
computational activity about collision physics.
Ed We were doing momentum, and we were looking at elastic and elastic collisions,
and we actually coded something where two blocks had to collide. And just
seeing how just changing a couple of numbers could change the entirety of the
coding was interesting... That was helpful for me to get the whole concept of
coding.
Ed came to understand how changing numbers in the program is connected to seeing the
physical consequence in the animation. Computation allowed her to make small changes to the
program and to see the relationship between momentum and the actual movement of objects. Ed’s
articulation of this and engagement at this level suggests an orientation towards learning physics
through computation rather than just trying to get through the activity. Though she outlined
many challenges in the previous section, this comment shows that students also see benefits to
computation, and one of those benefits is the visualization and strengthening of physics concepts.
Joyce expressed a similar perspective, which was that the process of translating ideas into code
was a way of learning physics concepts. While Ed focused on the benefits of interacting with the
dynamic, completed code, Joyce discussed how creating code was constructive for her.
Joyce By actually coding the formula and what variables go in, I think it helps in learning
the concepts. It’s just you might not catch [an error] at first and you might mess
up because we were supposed to put other stuff in [the program].
Joyce shared how she felt like she learned the physics concepts better by coding the formulas
and variables. In the second part of her quote, Joyce talked about the experience of accidentally
136
letting a bug, or coding error, get into the program (“we were supposed to put other stuff in
the program”) and prevent it from running properly. By relating learning physics concepts to
the debugging process, Joyce demonstrated that she understood there was value in meticulously
translating physics formulas into code and incorporating the computer’s feedback. This awareness
allowed her to engage with the activities in a way where she felt that they helped her learn physics.
Finally, we found computation can help some students build interest in physics. Beck discussed
at length how he viewed computation as an opportunity to connect with physics in a more authentic
way. Below, he talked about how a visual world of physics opened up when he used GlowScript.
Beck GlowScript provided even more visuals and stuff to actually connect with, which
is what made me understand physics and like it even better. The visuals, the
demonstrations, that ability to see the things in real life... they just helped provide
even more for that, and they even strengthened my liking for physics even more.
He connected with the visuals and felt as though hewas seeing the phenomenon in real life. Beck
went on to say more about the benefit of computation, describing how it provided an opportunity
to do some of the same activities that physicists do professionally.
Beck [Coding] allows you to apply stuff that you’ve learned in a way that’s different
from just solving a problem on paper, because you actually get to see the result
of what you’ve solved in real life. I mean it’s a computer, but you get to see it
actually work. It gives you a view of what physicists do, I suppose. Like you
get a problem and you use physics to solve the problem, then you see it actually
work... I like the coding in physics because of that.
In this excerpt, Beck saw the purpose of computation as seeing a physics problem at work in
a simulation of the real world. It was a way for him to connect what he was learning to what was
relevant to him. It was also a way to understand the type of work that actual physicists do. In Beck’s
case, this engendered an interest in him saying “I like the coding in physics” and “[it] strengthened
137
my liking for physics even more.” This shows that computation has the potential to help students
build an interest in authentic physics as well as help with learning.
The benefits that Ed, Joyce, and Beck described are similar to some of the goals that Mr. Buford
had for his computational integration. In particular, he wanted students to strengthen their understanding
of physics concepts through computation, saying, “I hope it just enhances them thinking
about the physics concept that we’re trying to learn, ideally....I feel like when you’re writing the
code for this, you have to understand how projectile motion works, or you can’t write code that
models that very well.” Both Ed and Joyce described the benefit to conceptual understanding,
though it is not clear whether Mr. Buford envisioned the same mechanisms of learning. For Ed,
she learned through interacting with the completed code, and for Joyce, learning happened through
creating the code itself and working through bugs. The benefit that Beck described goes beyond
what Mr. Buford said, namely the computation helps him do real physics and builds his interest in
the subject.
There were also some ideas that were missing from student interviews, benefits that Mr. Buford
envisioned but that did not seem to bear out in our data. Mr. Buford hoped that the open-ended
nature of the computational activities and the choice to not grade them would spur students to be
more creative, given that a lot of the constraints on traditional physics projects were stripped away.
Students did not seem to latch onto the creative freedom in their interviews, so it is unclear to what
degree this goal was realized in the actual implementation. Also, there remains the question of
what benefits could exist in other implementations. For example, Mr. Buford wondered whether
“you could use [computation] as a way of developing concepts” rather than just reinforcing. With
different design goals and in different contexts, this might be entirely possible, which could chain
into students seeing different benefits to the computational integration.
6.8 Discussion
From students’ interviews, we see that they faced a variety of challenges when computation was
integrated into their physics class. Some of these challenges were related specifically to code (e.g.,
138
Interpreting Code, Repeated Confusion), while others were related to the pedagogy and culture
of the classroom (e.g. Interpretations of Implementation), but many of them were unique or had
unique components due to the integrated physics classroom context (e.g. FeelingWorse at Physics,
Unbelonging and Stereotypes, Stress/Frustration).
The challenges that we found specifically related to code (Interpreting Code and Repeated
Confusion) are similar to the student challenges reported from computer science contexts. Jenkins
[178] highlighted barriers in introductory level computer science learning, mainly focusing on
the extra skills that students need to learn to engage with computation, such as syntax, semantics,
and algorithms. He argued that what made computation hard was chiefly the novelty of it. This
aligns with what we found in Mr. Buford’s computation-integrated physics class. For example,
a part of Interpreting Code is understanding syntax and how it pieces together as well as error
messages and strategies for addressing them. These are new skills that students did not encounter
before unless they took a computer science class. Even then, we found that students who had taken a
computer science course still struggled with the syntax and idiosyncrasies of Glowscript. Previous
research by Bumler et al. [199] found that students with prior computational experiences did not
view minimally working programs using the GlowScript platform as authentic computation. The
conflict between their previous experiences and the lack of utility of students’ previous experiences
in the context of this research implies there are difficulties transferring practice to the GlowScript
platform. The basis for this disconnect between platforms and contexts needs to be studied in
greater detail. This also speaks to the Repeated Confusion challenge because the process of learning
a programming language (especially debugging) requires persisting through many mistakes
and learning from them. This parallels another study, in which Bosse and Gerosa [91] catalogued
some of the main worries that students tend to have in programming settings, including trouble
with syntax, variables, error messages, and code comprehension. The worries were sometimes so
overwhelming that when a student realized their code contained an error, they were more likely to
give up. We saw a similar case with Blaine, who gave up after encountering numerous errors and no
longer proceeded with the activity. From another perspective, Svensson et al. [184] viewed compu-
139
tation as a social semiotic, or a way of communicating about and exploring phenomena. They saw
challenges emerge when students had limited skill with using the semiotic resources, even when
students did see the benefit of communicating and exploring through computation. This mirrors
the experiences of Ed and Otto, who both saw the usefulness of computation and often even knew
the relevant physics concepts, but they ran into roadblocks because they had limited experience
and comfort with computation itself and/or GlowScript. The fact that we saw the same challenges
and barriers in the computation-integrated environment that are seen in computer science contexts
indicates that students’ interpretation of code is a broader challenge for any type of coding activity.
Given the common challenge between contexts, this would indicate a place where computer science
educators and physics educators can learn from one another about how to best support students.
However, we also found several challenges that were unique to the computation-integrated
physics environment. For example, in Section 6.5.2, Ed separated the domains of computation
and physics, so that her difficulty with computation would not affect her view of her physics
competence. She reassured herself, “You know [the physics], you just think about it in a different
way, but that’s not a way that can be programmed on the platform.” This challenge was unique to
the computation-integrated physics environment, specifically because the curriculum merges two
subjects that for these students can sometimes be viewed as two separate domains. In a separate
computer science course, students’ perceived physics competencies and self-views would typically
not be threatened or involved at all. However, because of the integration, some students protected
one view of themselves, potentially at the cost of the other. In the case of Circe, the integration of
computation led to statements of unbelonging and a distancing of herself from physics as a whole.
In the case of Blaine, we saw that multiple failures at the computational activities led to statements
aligned with lack of self-efficacy and low self-concept when he said, “I don’t know anything!”. This
is similar to what Lishinski et al. [92] found during computational activities (and what Caballero et
al. [33] warned against), namely that judgments aligned with lack of self-efficacy can lead students
to use tactics that harm their learning rather than help.
The integration of computation into STEM is strongly motivated, including arguments about
140
preparation for students’ future careers and making STEM courses more relevant. However,
we showed that students intentionally separated the domains at times. This would indicate to
teachers, researchers, and curriculum developers that more attention needs to be directed to how
this integration occurs. For example, as part of the ICSAM workshop, Mr. Buford was altering
an existing curriculum. He already had lesson plans to teach all the necessary physics content,
and perhaps it made more sense to introduce computation at the transition points in the curriculum
rather than potentially disrupt the material mid-concept. Additionally, ICSAM teachers learned
how to program with GlowScript during a summer workshop. They were already physics experts
when they arrived, but many were novices at computation, meaning they learned to program as a
way of modelling and exploring what they already knew about physics. This process could have
transferred to how their students would go on to learn computation in their classrooms: physics
first, computation later. Ultimately this could have contributed to the separation of computation
and physics as separate domains. That said, there is certainly a precedent for integrating STEM
domains. After all, physics and math have been closely tied since the foundation of the field. We
don’t think twice about whether formulas and calculations are a part of physics, and for students,
learning to use math as a tool and learning physics go hand-in-hand. In the same way, we envision
a future where computation is also treated as an everyday tool for learning physics in classrooms
and viewed as such by students, but we need to learn more about what is happening in these
integrated classrooms. However, the math and science domains are blended at a much earlier point
in a student’s schooling. Students perceiving computation and physics as two different domains
highlights the need to investigate whether integrating at an earlier point in a student’s science careers
would impact their perceptions of computation being a tool for doing science.
Another challenge that is unique to the computation-integrated contexts is the balancing of
content between computation and physics. Given the other constraints that teachers are under (time
limitations, science standards that must be met, etc.), it can be difficult to add computation to an
already packed schedule. Mr. Buford commented in his interview that despite his natural curiosity
for new ideas in physics, it was hard to try new things when he had to cover all the content on the
141
AP Test. The year before he attended ICSAM, he simply saved computation until after the test
was over in the last month of the school year. When he tried to integrate computation into his
curriculum throughout the year, he was not able to let the students slow down enough to wrestle
with the computation and figure out how it could help them learn physics. For some students, the
purpose of doing computation in physics did not stick with such little time. Furthermore, there
might be some influence from the AP curriculum on what counts as “doing physics.” Without
changing the national expectations and standards to include computation, it will be near impossible
to create fully integrated courses.
We also saw several challenges that were related to pedagogical choices from Mr. Buford. For
example, Mr. Buford intentionally chose to not grade the computational activities, which led to Ed
commenting that the activities felt like “busy work.” He also chose to show the final output to the
class, and when Circe’s answers did not match, this made her feel like her answers were “wrong.”
In Section 6.5.1 we highlighted how the students felt as though the activities being after the concept
had been covered had framed them as causing undue stress. However, from Mr. Buford’s perspective
this was intentional because he thought introducing concepts via a computation activity would be
too stressful. This catch-22-like outcome highlights the struggle that teachers face when making
curriculum design decisions around integrating computation into their classrooms and highlights
a desperate need for research focused on curriculum design for such environments. None of the
above discussion points around pedagogical choices is intended as a critique of Mr. Buford (in fact
he had strong pedagogical reasoning for his choices), but this highlights that there might be unique
challenges depending on the specific implementation of computation-integrated physics and the
classroom structures that a teacher employs.
For example, Beck described a positive structure in his interview from Mr. Buford’s class. Beck
came upon a roadblock and had to ask for help from Mr. Buford, who pointed out to him a built-in
GlowScript function that did exactly what he needed. In fact, having students ask for this function
was part of Mr. Buford’s plan—he confirmed after class to the first author that part of the activity’s
purpose was to discover the need for a new function. The challenge lies not in what we generated in
142
the data, but in what was absent: the students who did not think to ask for help or who did not arrive
at the point in the activity to realize the need for a special function. Students might struggle to ask
for help for a variety of reasons; they might feel intimidated by asking questions to an authority
figure (their teacher in this case), they might feel too embarrassed by their “lack of progress” on the
problem to ask for help, or they might struggle with social anxiety. Alternatively, and especially
in a less collaborative context, students might have the impression from classroom norms or social
stereotypes that they are supposed to be coding alone. Any of these reasons might prevent students
from asking for help, and in turn, increase their frustration and perpetuate a negative view of
computation.
As Mr. Buford confirmed, a teacher might let students struggle with an idea intentionally or
might want students to discover an idea as part of the computational activity. With Beck, this
worked well, and he was able to learn about the unit vector from Mr. Buford. However, for this to
happen it was critical that Beck felt comfortable asking Mr. Buford for help and that Mr. Buford
promoted that in his classroom. In another classroom context, with a different classroom culture,
we could envision “Asking for Help” to be a challenge for students. This only points to the work
that needs to be done to build on this study and examine the contextual challenges in other implementations
of computation-integrated physics and other STEM courses.
6.9 Conclusions and Future Work
In this chapter, we have described the student-perceived, affect-based challenges that high schoolers
faced in a computation-integrated physics class: Stress/Frustration, Feeling Worse at Physics,
Unbelonging and Stereotypes, Repeated Confusion, Interpreting Code, and Interpretations of Implementation.
We also found connections between students’ descriptions of the challenges and
the theories of self-efficacy, mindset, and self-concept. This work is laying the foundation for
identifying affective barriers and those unique to computation-integrated STEM contexts, serving
as the first study in this context to examine affective challenges from students’ perspectives. While
this study is an initial step, more work needs to be done to understand the affective challenges
143
students face and how to best support them.
An example of the importance of student perspectives from our data was when Joyce said she
felt “just average” at coding when we were fleshing out the Unbelonging and Stereotypes challenge.
She appeared to be one of the most competent programmers in class, but she didn’t feel that way
about herself. It is only through asking students about their experiences that we can find out how
they feel about the challenges they face in class, and sometimes their answers can be unexpected.
To researchers, this study is a call to action. Computation-integrated physics courses continue
to grow as computation becomes synonymous with doing STEM. With it come the complexities
and difficulties of new curriculum and the need to understand the experiences of students in this new
environment. We have found that the lenses of mindset, self-efficacy, and self-concept might offer
meaningful insight into many student-centered processes, yet there is a need for more exploration,
particularly in how the integration takes form, how the protective separation of computation from
physics can be minimized, and how the difficulties and frustrations of learning a new programming
tool affect students. We need studies on affect, self-beliefs, and perceptions in computationintegrated
contexts where computational learning is supported by design, where the curriculum is
less constrained institutionally (e.g., “regular” instead of AP), where computational tools are the
focus of the course, and where features of implementation support underrepresented students.
To practitioners, this study is a call to consider many factors when designing or altering
curriculum for computational integration. We call for attending to the affect of students who take
part in the curriculum, the tools being used to integrate computation, the pedagogical strategies for
teaching computation, what it means to redesign existing curriculum, the curriculum’s potential
effect on students’ perceptions of computation and physics, and the role computation can play
pedagogically. We acknowledge that figuring out how computation best fits into the context of
one’s physics course is an immense task. We need to teach students authentic physics by using
computational tools, but we also need to find ways to ease the burden on physics teachers who are
often saddled with altering curriculum to meet new educational demands, of which computational
integration is the latest [72].
144
In conclusion, we highlight that the computational challenges raised in this chapter need to be
studied in more depth in the computation-integrated context as opposed to trying to understand them
by only applying knowledge from physics or computer science education research. This type of
curriculum is unique enough to warrant further studies, especially when considering the issues that
arose when students had to deal with computation and physics at the same time in the same context.
Computation in our physics courses is essential for the next generation of scientists, and it is imperative
that we learn how to best apply computation as an educational tool to the benefit of our students.
6.10 Acknowledgments
As written in the submitted paper: We acknowledge and give thanks to the National Science
Foundation (DRL-1741575) for funding this research, to Mr. Buford (pseudonym) for providing
access to his classroom and eagerly participating in the research, and to the students who gave their
time and insights to us during the interview process. We also thank Jacqueline Bumler for assisting
with data generation and providing additional insight during discussions about the data. Lastly, we
thank David Stroupe, Vashti Sawtelle, and Tyce DeYoung for feedback on the manuscript.
145
CHAPTER 7
DISPOSITIONS AND MINDSET IN COMPUTATION-INTEGRATED PHYSICS
This chapter builds on Chapter 6 by exploring the theory of mindset in more depth in the same
context. Because of the breadth in students’ perspectives and applications of theory in Chapter 6,
I was able to design a study that focused on fewer theories but went more in depth, in part by
incorporating more data sources. Specifically, I connected mindset to a theory that is more native
to computation, computational thinking dispositions, and I devoted this chapter to fleshing out how
these two theories show up in what students say and do in Mr. Buford’s class. This is the last
study in the dissertation, and it represents an operationalization of the research foundation I built
in Chapter 6 and the research tools I honed throughout all the research described thus far in the
dissertation.
7.1 Introduction
In the last fifteen years, there have been multiple calls and reports for the integration of computation
into both the high school and undergraduate physics curricula [33, 24, 183, 41, 200]. With
each of these frameworks and reports, a strong emphasis has been placed on students developing
Computational Thinking (CT) practices. Computational Thinking [129] is widely viewed as
an important learning goal in STEM settings [130, 131]. It encompasses the “practices” and
“dispositions” [132] associated with “solving problems, designing systems, and understanding
human behavior, by drawing on the concepts fundamental to computer science” (page 33) [129].
Though it is based on ideas from computer science, CT is meant to be applied to situations
with and without computers [129], designed as a set of developmental skills based on computing
principles rather than actual computer skills. CT has been a part of the STEM education zeitgeist
ever since Wing’s conception and publication of it in 2006 [129]. Since then, there has been a
widespread push to teach CT in STEM contexts, including at the K-12 level [171, 201, 172]. There
146
is wide agreement on the importance of CT, but there is no clear consensus on how to support
the development of these practices, especially in contexts where computation has been integrated
into a STEM discipline [27]. One of the main focuses of curriculum designers and researchers has
been CT practices [40, 133, 66, 202, 203], but there has been less of a focus on how students can
approach activities in ways that can help develop their CT practices.
The focus on practices in many of the instances of computational integration [40, 133, 66, 202,
203] fails to consider “dispositions,” which is the other half of the widely supported operational
definition of CT from the International Society for Technology in Education (ISTE) and the
Computer Science Teachers Association (CSTA) [132]. Dispositions is a termused in computationintegrated
STEM research [132, 1, 76] to describe how students perceive and approach activities in
which computational thinking is used. This aligns with efforts to incorporate student perspectives
and affect into computation-integrated physics curriculum design [204, 70, 205]. The argument has
been made that for students to develop and evolve their computational thinking practices then they
need to develop productive dispositions towards computation [132]. The argument has been made
in CT literature that a significant step in the successful incorporation of CT practice development
in a curriculum is fostering CT dispositions. To date, dispositions have been understudied in
comparison to the widespread research and implementation of CT practices, but they have been
argued to be just as important.
The importance of CT dispositions and not just CT practices can be traced to ISTE and CSTA’s
definition of CT [132], but it was recently incorporated into a detailed theoretical framework by
Pérez [1]. Originally developed during a workshop series for secondary mathematics teachers,
Pérez’s theoretical CT Dispositions Framework was situated in the context of a mathematics
curriculum, which leaves the question open as to how this framework could be applied beyond
mathematics contexts. Nonetheless, Pérez’s CT framework represents the most comprehensive
published research on CT dispositions. We intend to extend Pérez’s framework into the context
of a computation-integrated physics classroom, with the goals of (1) demonstrating how the CT
Dispositions Framework applies in a high school physics setting and (2) calling attention to the
147
importance of CT dispositions in the widespread movement to integrate computation into physics
curricula.
Previous research in computation-integrated physics classrooms has demonstrated that an important
learning goal for teachers is for their students to develop positive dispositions towards
computational thinking [206]. In their study, Weller et al., produced a collection of teacherarticulated
learning goals for computation-integrated physics learning. They found that teachers,
although not using specific terminology of dispositions, wanted their students to have a positive
experience with computation, experience a reduction in the intimidation of physics, and have a
sense of accomplishment when they work through computational activities. Building on this work,
Weller et al. [76] developed an expansive computational learning goals framework, where they analyzed
teacher-articulated learning goals and pointed out the importance to teachers of developing
CT dispositions. The idea is that helping students develop CT dispositions can foster positive affect
towards computation, making computation more accessible, less intimidating, and overall more of
a positive learning experience for students.
We focus on CT as a learning goal because it closely aligns with disciplinary practices. This
focus is in line with a push to make school learning more “thickly authentic” [207], in effect
grounding learning goals in ways that connect students more directly to practices from the relevant
disciplines or communities. Currently, teachers have no way of gauging whether their students are
given opportunities to develop dispositions in their curriculum’s computational activities and in
turn their computational integration as a whole. This study lays the groundwork for considering
dispositions when integrating computation, but first, we need to understand how Pérez’s framework
applies to a high school physics context and howdispositions might provide insight into the activities
of such a classroom.
Thus, our primary research question is, how do CT dispositions apply to the context of a
computation-integrated high school physics class?
Answering this research question entails taking the theoretical framework from Pérez [1] and
applying it to a new context. The CT Dispositions Framework was developed in a mathematics
148
setting through observing teachers and collecting reflections from teachers. We intend to take the
same framework and apply it to a physics setting through observing students and asking them about
their perspectives. As this is a new context (computation-integrated physics) and new data sources
(student perspectives), this will likely stretch the untested framework, which is part of the appeal of
this study. By extending the CT Dispositions Framework to a new setting, we hope to make it more
robust for other applications. If CT dispositions are as important as has been argued previously,
then we anticipate the CT Dispositions Framework will build in usage and usefulness as it is applied
to more contexts and used in curriculum development around computational activities.
Another consideration for our study is the relationship between CT dispositions and more
established constructs that address students’ affect towards and perceptions of computation. We
turn to one such construct, mindset [2], in part because Pérez drew on it to develop the CT
Dispositions Framework [1]. Additionally, mindset has been used in the past as a bellwether
of pedagogical effectiveness in computational classroom settings [107, 108]. Because the CT
Dispositions Framework is new in computation-integrated physics settings, we hope to tie our
application of it to mindset, which is much more well established in research on affect. Mindset
also varies based on context and method [2, 208], which provides further motivation to study both
constructs in the same investigation and to explore how the Pérez framework applies in different
contexts.
With this in mind, our secondary research question is, how are CT dispositions connected to
mindset in the context of a computation-integrated high school physics class?
We intend to answer this question by coding our data sources for dispositions and mindset
simultaneously and using overlaps and patterns to point out aspects of the relationship between the
constructs. We begin this endeavor in Section 7.2, where we outline the theoretical framework of
CT dispositions, its connection to mindset through Pérez’s work, and a coding scheme for mindset
that we used alongside the CT Dispositions Framework. We proceed in Section 7.3 to outline our
case study methodology for investigating dispositions and mindset. In Section 7.4 we describe the
context of our case study and the features of one teacher’s computational activities that make them
149
an appropriate setting for CT dispositions to develop. In Section 7.5, we describe our methods,
including participant selection, data generation, transcription, and data analysis. In Section 7.6, we
show the results of our study. The first part of the results, Section 7.6.1, addresses the first research
question and includes each how CT dispositions applied to each student’s data set, which combines
evidence from interview data and in-class recordings to show how the framework interacts with
different types of data. The second part of the results, Section 7.6.2, addresses the second research
question and shows evidence of mindset relating to students’ data sets in different ways, comparing
mindset to the dispositions results from Section 7.6.1. Finally, in Section 7.7, we discuss the
results and their implications for future work in CT dispositions. We include recommendations for
applying the framework and suggestions for potential changes to it for research studies where the
participants are high school students in a computation-integrated physics class.
7.2 Theoretical Framework
Developed by Arnulfo Pérez [1], the CT Dispositions Framework is an attempt to operationalize the
full definition of CT from ISTE and CSTA [132]. The ISTE and CSTA definition is split into two
categories: practices of CT and dispositions of CT. Due to the broad research and attention that has
been placed on CT practices to date [40, 133, 66, 202, 203] and the lack of focus on dispositions
(only recommendations to take dispositions and perspectives into account [204, 70, 205]), Pérez
developed a framework for researchers and teachers to use to understand how CT dispositions relate
to what happens in the classroom.
The framework was originally developed in the context of a workshop series for K-12 teachers
where they learned how to integrate CT into their mathematics curricula. Pérez and other institute
facilitators made observations of the teachers as they worked on CT activities, and they collected
reflections from the teachers about their learning. Using the observations, reflections, and a
synthesis of relevant literature, Pérez wrote the CT Dispositions Framework [1]. One of the key
features of the framework’s development was the acknowledgment that it needed to be applied
to other contexts: “the usability of the framework [increases] through examples of classroom
150
behaviors that may accompany developing or higher levels of a given disposition” (page 442) [1].
The wording of “developing or higher levels” refers to the degree to which an action or statement
aligns with the language of the framework, and it theoretically refers to alignment with opportunities
to develop CT practices. For example, one of the dispositions in the framework is persistence.
Having a high level of persistence is another way of saying that when a student is faced with a
challenge in the computational activity, they will continue to engage with it and as a result will
be more likely to take up CT practices. However, having a developing level of persistence could
translate into disengaging in the face of a challenge and not learning the CT practices associated
with the activity. That said, the word “developing” is, by definition, ripe for change and when
applied to a classroom can be indicative of a student actually in the process of undergoing that
change. Granted, the categories of “high dispositions” and “developing dispositions” are not strict
categories; instead, we view “high” and “developing” as two sides of a continuous spectrum. In
this way, the CT Dispositions Framework can be used as a tool for identifying how actions and
statements align with the dispositions spectrum and can potentially allowteachers to support growth
in different areas of CT dispositions.
In the CT Dispositions Framework, there are three dispositions identified by Pérez: tolerance
for ambiguity, persistence, and willingness to collaborate with others. Their definitions and characteristics
are listed in Table 7.1, which are synthesized directly from Pérez [1]. The characteristics
of each disposition are categorized into key inclinations, sensitivities, and abilities, which were
adapted by Pérez [1] from Perkins et al. [209]. The categories work together when a person acts
in a way that aligns with a particular disposition: “Inclination refers to an individual’s tendency
toward a particular way of thinking or acting. Sensitivity denotes an individual’s attentiveness to
opportunities to engage in that particular thought or action. Ability refers to being able to actually
produce that thought or action when one notices an opportunity (sensitivity) and feels drawn to act
(inclination).” (pages 434-435) [1].
The way the key inclinations, sensitivities, and abilities are tied to behavior and thinking make
them ideal for observing through group work and reflective assignments, which is exactly how
151
Pérez observed them at the original workshop series for teachers. This observable quality also
makes these categories ideal for our study, which is why we use Table 7.1 as an analytic tool later
in this chapter. The degree to which a student displays inclinations, sensitivities, and abilities for a
particular disposition is the degree to which that student displays that disposition as a whole in the
data set.
To be clear, the qualities described in Table 7.1 align with high dispositions, not the developing
side of the spectrum. When using this framework later in the study, we interpret the qualities of
developing dispositions to be opposite from the descriptions in Table 7.1. For example, the opposite
of an interest in exploring unfamiliar situations (high) could be characterized as an apathy towards
unfamiliar situations or an avoidance of unfamiliar situations (developing). The terminology is not
ideal given that theword “developing” implies a trajectory towards high dispositions, butwe use this
language all the same because we wish to align our application of the CT Dispositions Framework
with Pérez’s theorization of it. In Section 7.7, we provide a critique of the framework’s application,
but for now we adhere to the language provided. We also acknowledge that although the framework
accounts for developing and high dispositions, there is a spectrum between developing and high.
Even in the same excerpt, for example, a student might exhibit a high tolerance for ambiguity by
expressing a desire to discover new meaning while exhibiting a developing tolerance by insisting
there can only be one solution.
Teacher observations were not the only criteria by which Pérez [1] developed the framework
for CT dispositions. He also conducted an extensive review of literature on which to base the
framework. Several times when describing the features of CT dispositions, he cited work on growth
and fixed mindset [2, 5] or work foundational to mindset theory [210]. Pérez cited Dweck [2] when
discussing “the malleability and potential growth of positive dispositions,” and again he cited her
when he said, “tolerance for ambiguity...is malleable” and “all are capable of becoming increasingly
tolerant of ambiguity, a form of growth” [1]. This suggests that Pérez connected the malleability of
intelligence and skill (from mindset theory) to the malleability of dispositions, especially tolerance
for ambiguity.
152
Key Inclinations Key Sensitivities Key Abilities
Tolerance for
Ambiguity:
A tendency
to experience
ambiguous
situations or
stimuli as enriching
and
engaging
1. An interest in exploring
unfamiliar situations
2. The desire to discover
meaning or possibilities
that are not yet
apparent
3. A tendency to avoid
rigid categories and
take a flexible view of
categorization
4. An accepting view of
variance
1. Awareness that engaging
with uncertain
situations can lead to
growth
2. Alertness to opportunities
to clarify
what is known and
unknown
3. Responsiveness
to approaches for reframing
ambiguous
situations or stimuli
1. Acknowledging multiple
possible solutions or
explanations
2. Finding value in undertaking
“messy” tasks
3. Navigating incomplete
data and uncertain trajectories
toward a solution
Persistence:
A tendency
to continue
working or
to maintain
effort when
dealing with
a challenging
task
1. The tendency to
value extended effort
2. The desire to complete
difficult tasks
3. An interest in what
may be discovered even
in an attempt that is not
successful
1. Alertness to the
characteristics of a
given task
2. Awareness of the
satisfaction that will
be felt when efforts
eventually yield fruit
3. Attentiveness to
opportunities to shift
tactics when needed
1. Sticking with a task
for an extended period of
time
2. Trying a new approach
after considerable effort
3. Pursuing resources that
increase the effectiveness
of effort
4. Framing significant
effort as likely to produce
significant outcomes
Willingness
to Collaborate
with
Others: A
tendency to
coordinate effort
and negotiate
meaning
with peers to
accomplish a
shared goal
1. A willingness to have
one’s course changed by
interactions with others
2. A tendency to invite
and value perspectives
different from one’s
own
3. Curiosity about
multiple possible approaches
to solving a
problem
1. Alertness to interpersonal
dynamics
that may enhance or
impede effective interactions
2. Responsiveness to
the contributions of
peers
3. Attentiveness to the
unique insights that
emerge from interactions
1. Listening to and having
one’s actions shaped by
others
2. Articulating or justifying
the benefits of a
particular approach
3. Clarifying, questioning,
or negotiating the
group’s understanding
and/or course of action
Table 7.1: Definitions, inclinations, sensitivities, and abilities for each CT disposition. Features of
the table are copied from different parts of Pérez [1]. The inclinations, sensitivities, and abilities
as described here align with “high” dispositions.
153
Pérez also referred to mindset when discussing persistence, saying tasks that rely on rehearsed
procedure “can reinforce the belief that success in mathematics amounts to completing problems
quickly and easily” [1]. He cited Yeager and Dweck [5], an article that connected resilience
to growth mindset in mathematics contexts. This suggests that an activity designed around a
rehearsed procedure that does not require deep thinking can reduce persistence (and can foster
fixed mindset, according to Yeager and Dweck [5]). Pérez clarified this point, saying, “if students
do not believe that their efforts matter in a particular context, they are unlikely to persist” [1].
Again he cited Dweck [210] here, pointing to her study that demonstrated that teaching children to
take responsibility for failure and attribute failure to lack of effort can help children improve their
persistence. This is a direct connection between aspects of growth mindset (taking responsibility for
failure and responding to failure with increased effort) and one of the CT dispositions (persistence).
Pérez did not draw explicit connections between mindset and willingness to collaborate, though
he did for the other two dispositions and for dispositions as a collective. We find the use of mindset
compelling because it ties the dispositions framework to a well-established [211] social cognitive
theory often applied in educational contexts [211, 212]. There are several proven interventions [124,
125, 126, 127, 128] that can help foster growth mindsets among students, and drawing the connection
to CT dispositions could mean that these same interventions might impact CT dispositions as well.
In deciding how to operationalize mindset for our investigation, we turn to multiple works by
Dweck: her original theory [2], a subsequent study where she and colleagues turned mindset into
a questionnaire with explicit categories [4], a later book where she addressed in more depth the
behavioral patterns that give rise to growth and fixed mindset [3], and a study that that connected
resilience to growth mindset for adolescents in mathematics contexts [5]. We draw from these
sources in the following descriptions and in Table 7.2. The foundation of mindset lies in two
theories of intelligence [4, 2, 3]. The first is called “entity theory,” or more commonly, fixed
mindset. This is the idea that intelligence is a fixed entity that is bestowed upon someone, and
it cannot be changed by effort. The second is “incremental theory,” or more commonly, growth
mindset. This is the idea that intelligence is incremental, or changeable, especially when effort
154
is applied to it. Dweck [2] argued that for students, views aligning with fixed mindset can be
detrimental to learning because they can lead to a loss of motivation in the face of adversity and
harsher judgments of self when making mistakes. Conversely, students with views aligning with
a growth mindset might respond to failures with increased effort and motivation to learn from
mistakes.
There were several other aspects of mindset that we summarized in Table 7.2. Growth mindset
aligns with a desire to learn, whereas fixed mindset relates to proving one’s own intelligence and/or
superiority [4]. Growth mindset aligns with the belief that thinking hard and making mistakes
are worth it because they can help you learn, whereas fixed mindset aligns with the belief that
thinking hard and making mistakes should be avoided because they show a lack of ability [2].
Growth mindset aligns with the belief that effort is valuable and the path to success, whereas fixed
mindset aligns with the belief that effort is not valuable, and too much effort is a sign of inability
(i.e., successful students should not have to try) [2, 4, 5]. Failure can have different implications
depending on mindset [2, 4]. For growth mindset, failure can be interpreted as a need to study
harder and/or better, whereas for fixed mindset, it can be interpreted to mean you are stupid, you
are bad at the subject area (physics or computation in our context), or the assessment itself is unfair.
Failure also holds different opportunities depending on mindset [2, 4]. For growth mindset, failure
can be interpreted as a learning opportunity, whereas for fixed mindset, failure can lead to losing
interest in the topic. Growth mindset aligns with interpreting setbacks as roadblocks to overcome,
whereas fixed mindset aligns with experiencing setbacks as paralyzing progress on an activity [4].
Growth mindset aligns with taking responsibility for successes and failures, whereas fixed mindset
aligns with attributing success and failure to external factors [4]. Mindset also splits along the
interpretations of trying different strategies or getting outside help when stuck: growth mindset
aligns with valuing these tools, whereas fixed mindset does not [5]. Lastly, growth mindset aligns
with embracing challenges, whereas fixed mindset aligns with avoiding them [3].
Though we categorize mindset into “fixed” and “growth” columns in Table 7.2 and in our
descriptions above, we also acknowledge that mindset can exist on a spectrum between the two
155
Growth Mindset Fixed Mindset
Intelligence/skill is growable Intelligence/skill is fixed
Learning is important Proving intelligence/superiority is important
It is better to learn even if you have to
think hard or make mistakes
It is better to avoid thinking too hard
or making mistakes
Effort is valuable because it makes you
smarter and/or more skilled
Effort is not valuable because putting
in too much effort means you aren’t
very smart, and it won’t make you
smarter
Failure can mean: you need to study
harder, you need to study in a better
way
Failure can mean: you are stupid, you
are bad at physics/computation, assessment
is unfair
Failure is a learning opportunity Failure makes me less interested in
physics/computation
Setbacks are opportunities to overcome
a challenge
Setbacks are paralyzing
Success/failure is one’s own responsibility
Success/failure is not one’s own responsibility
Getting outside help and trying different
strategies are valuable tools
Different strategies and outside help
aren’t valuable
Challenges are to be embraced Challenges are to be avoided
Table 7.2: Indicators of growth and fixed mindset, to be used as a coding scheme for our data, and
developed from Dweck [2, 3], Blackwell et al. [4], and Yeager and Dweck [5].
156
columns. Though research on mindset has not focused much on what lies in the middle of spectrum,
it has pointed out that students can exhibit different levels of growth or fixed mindset at different
times, for different subject areas, and for different aspects of mindset [2, 3, 4]. For example, a
student might articulate that they believe effort is the path to success (growth mindset), but they
also have a strong desire to prove they are smarter than other students (fixed mindset). They also
might sometimes feel motivated to overcome setbacks and at other times feel paralyzed and unable
to continue working after facing a setback (growth and fixed mindset). Because our coding scheme
comes from such literature, we do not have codes for a mid-level mindset, but we remain attentive
to the possibility for fixed and growth mindset to describe the same piece of data in different ways.
We expect nuances like these to appear in our data, because nobody is a perfect encapsulation
of growth or fixed mindset (or any of the dispositions, for that matter). Additionally, mindset can
appear differently depending on context and circumstances [2, 208]. We supply the categorizations
in Table 7.2 simply as a tool to discuss the different aspects of mindset as they appear in the data.
The same is true for the descriptions of the dispositions in Table 7.1—no student will embrace
every inclination, sensitivity, and ability of a disposition. Those categories are simply there to help
us show how a student’s actions and statements can reflect certain aspects of dispositions in the
data.
We also consider in this study that there are several critiques of mindset theory [213, 214, 215,
216] that could potentially be extended to the CT Dispositions Framework. These interpretations
of mindset theory view it as a way of shifting focus to whether students have the qualities needed
to succeed in educational environments as opposed to whether the features of those environments
are structurally unjust [213]. Framing the “deficits” [217] of students instead of the shortcomings
of educational structures is also a way of promoting meritocracy [214]. We still choose to use
the frameworks of mindset and CT dispositions in this study, but we attempt to avoid this deficit
framing and provide a critique of their applications in Section 7.7.
157
7.3 Methodology
Our methodology is case study, with the purpose being to see how the dispositions theory extends to
students in a high school, computation-integrated physics classroom. We expect for some parts of
the theory to align with the behavior of our participants given the original context of the theory and
other parts to require rethinking for our setting. Case studies do not provide causal or correlative
results nor can they be generalized to a broad audience. Case studies are for studying interactions
within a phenomenon [46], and it is through studying human interactions thatwe intend to illuminate
the application of the dispositions theory in our context.
To clarify, a case study comprises a phenomenon and a case, around which the research
is designed. Our phenomenon is behavior that aligns with CT dispositions, and our case is a
collection of students in a single high school physics class taught by Mr. Buford (pseudonym).
Together, these form our main research question: How do CT dispositions apply to the context of a
computation-integrated high school physics class?
When we think about how we intend to pursue this research question, we employ aspects of
two different traditions of case study: realist [47] and interpretivist [50]. A realist case study is
characterized by several factors: the presence of propositions upon which the research is designed
and the results are validated, “logic models” [47] that describe how claims will be constructed
from data, embedded units of analysis (or multiple grain-sizes at which data is generated and
analyzed), and comparison between cases. In contrast, interpretivist case study is characterized by
the centrality of human interpretation in the data generation, the use of participant’s viewpoints
(called “anchor points” [50]) in the data to understand the “interpreted” phenomenon, and the focus
on a single case (as opposed to comparing multiple cases).
The perspectivewe take on our case study lies in the overlap between the realist and interpretivist
views. Both traditions attempt to “bound” the case in order to focus data generation on a handful of
in-depth data sources. We do this by focusing primarily on interview data and in-class recordings of
Mr. Buford’s students. There are also aspects of each tradition that can coexist—we use propositions
158
(e.g., high school students have CT dispositions that can be inferred from their behavior in physics
class), we loosely use a logic model (described in methods section), we use embedded units of
analysis (described below), we center human interpretation in our data generation, and we use
anchor points (described below). Where the traditions contrast (comparison of cases versus a
single case), we take a middle ground—we compare the findings between data sources as part of
our discussion to see how the theory applies, but we don’t compare cases, instead favoring the
in-depth look at our single case of embedded units.
Our case (the students in Mr. Buford’s class) is split in our research design into smaller embedded
units of analysis. In this case, the embedded units are the individual students. We organize our
findings based on these units in order to present how each student made statements and actions that
aligned in unique ways with CT dispositions. However, we do not keep these units separate—they
serve the collective purpose of testing the theory of CT dispositions and connecting that to mindset
theory as a well-established learning construct. By looking at the case as a whole (the collection of
students), we also return to the phenomena of interest, which are howCT dispositions apply to a new
classroom setting and how mindset is connected to CT dispositions in this context. As a parallel,
we also view these students as “anchor points” when analyzing the data associated with them. In
this sense, the students each provide a different view into the application of CT dispositions, and
all together they provide us with a triangulated understanding of how CT dispositions can connect
to what happens in a classroom.
In the methods section, we will describe in detail how we use our data sources to construct
claims around CT dispositions and mindset (sometimes referred to as our “logic model” [47]).
We also will describe how we link data to propositions and lay out our criteria for interpreting
findings. For the methodology section, it suffices to articulate our propositions. Propositions are
ideas for how the relationships of interest within our phenomenon come about [47]. We have two,
formulated with respect to our two research questions. First, we propose that CT dispositions are
present in what students say and do. Second, we propose that context-based connections between
CT dispositions and mindset can be drawn when both constructs describe the same piece of data.
159
The first proposition helps mainly with data generation and analysis of dispositions. The second
proposition helps mainly with analysis of the connection between dispositions and mindset.
7.4 Context
In this section, we describe Mr. Buford’s class as the context for our case study. Mr. Buford taught
physics at Mulberry High School (pseudonym), a suburban, affluent, racially diverse public high
school, where he had been teaching for 30 years. The course we studied was a section of AP
Physics 2. Initially, Mr. Buford integrated computation into AP Physics 1 in Spring 2018. After
attending a summer workshop at Michigan State University for high school physics teachers who
wanted to integrate computation into their physics curriculum, he then fully integrated computation
into AP Physics 2 for the 2018-19 academic year. The 2018-19 academic year is when this study
took place. For a detailed description of how Mr. Buford chose to integrate computation in his
class and the process behind his curriculum design, see Chapter 6. In this chapter, we focus on the
features of Mr. Buford’s implementation that promoted or provided opportunities for demonstrating
CT dispositions.
First, Mr. Buford’s computational activities contained ambiguity in several forms. The computational
activities were designed in the form of a minimally working program [187] using the
GlowScript language [186], which meant that Mr. Buford would provide students with the beginning
of a program that would fully compile without presenting errors. However, the minimally
working program was missing lines of code to properly model the physical phenomenon, which
students were expected to fill in. At the start of class on a computation day, Mr. Buford would
explain the minimally working program to the students. He would outline the physics concepts that
they were supposed to model, which was always a concept the students had seen before either in a
demo or drawn out on paper. While he explained the final output and the initial code, Mr. Buford
did not provide steps or instructions on how to complete the code. This made it so there were
multiple solution paths, and the students had the freedom to add whatever they wanted to, in any
order they saw fit. In addition, even though the end result was typically the same visual for all the
160
students, there were always several configurations of code that would produce the results students
were aiming for. Furthermore, often students would need to look up functions or objects in the
GlowScript library, which meant they would sometimes search online for ways to implement their
ideas. This open-ended searching, together with the multiple solution paths and solution configurations,
provided several opportunities for students to demonstrate a tolerance for ambiguity during
the computational activities.
The activities also afforded opportunities for students to exhibit and develop persistence. For
example, Mr. Buford would design checkpoints throughout the activity. These checkpoints were
not explicit milestones for the students; instead, they were measures of progress that Mr. Buford
used to check in with his students. In one activity, students were asked to model a ray of light using
small spheres to represent the photons in that ray of light. The minimally working program for
this activity provided a single, stationary ball and a rectangle that represented the lens. Mr. Buford
expected the checkpoints of this activity to be: causing a single light particle (sphere) to move on
the screen, and then make the light particle pass through the lens, and then change the angle of the
particle’s path to correctly represent refraction through the lens, and finally to add more particles
to the animation. These checkpoints highlighted the difficulty of the task and students’ need for
persistence. There were several steps a student had to get through in order to complete the activity,
and they were only going to succeed if they were willing to put effort into each step. Additionally,
the students had most of the class period (45-55 minutes) to work on the activity, which meant
there was an extended period of time over which they could work through the challenges of the
computational activity.
Finally, collaborationwas inherent in Mr. Buford’s computational activities. First, the classroom
was arranged into tables surrounded by four to six chairs each (though seats were not always filled),
which meant students usually sat facing each other, as shown in Figure 7.1. Additionally, the
surfaces of the tables were whiteboards, which meant work done on the whiteboard could be seen
by other students at the table. The only non-collaborative aspect of the activity was that students
worked on their own code on their own laptops. However, Mr. Buford encouraged the students
161
Figure 7.1: Mr. Buford’s classroom setup. This snapshot shows students gathered around tables in
groups of around three or four.
to work together and share ideas throughout the class period. Collaboration was not required for
success like tolerance for ambiguity and persistence were, but it was encouraged by Mr. Buford’s
messaging and the design of the classroom. Additionally, the design of the activities made them
challenging and ambiguous, making it easier to get stuck. This translated into opportunities for
students to ask one another for help and work together on the activity.
Thus, we expected students to express aspects of CT dispositions whenworking through Mr. Buford’s
computational activities because the activities are designed with ambiguity, persistence, and
collaboration in mind. This is important because ambiguous stimuli and opportunities for exploration
are necessary for students to embrace a tolerance for ambiguity. Encountering challenges
and wrestling with them for an extended period of time are necessary for students to embrace
persistence. We also need opportunities for dynamic interaction and negotiation for students to
embrace collaboration. Even a student who embodies every disposition will not be able to express
those dispositions in an activity that constrains them to work procedurally, without challenges, and
alone. Given the design of Mr. Buford’s activities, his classroom provided an ideal context to look
162
for CT dispositions.
7.5 Methods
We selected students to include in our study based on the availability of data from Chapter 6 meant
to explore the landscape of students’ experiences in Mr. Buford’s class. As stated in the prior
chapter, “participants were selected to represent a broad range of student prior experiences (in
terms of physics classes and computational exposure) and current in-class experience (determined
through in-class observations).” To make such determinations, we observed the students working
in class over several weeks. From these participants, we included the students for whom there
existed interview data and in-class recordings from the initial data collection: Otto, Blaine, Ed,
and Beck (pseudonyms). For our study design, it was important to have access to both data sources
for each student because we are testing the theory of CT dispositions. We want to be able to
capture students’ experiences from multiple types of data sources because the best way to identify
CT dispositions is not well established. All four students were juniors at Mulberry High School
at the time of data generation. To ensure we respected how the students wished to be represented
in this study [188, 189], we asked the students after data generation to select a pseudonym and
self-describe their gender identity, racial identity, and preferred pronouns.
Otto (he/him) took “regular” Physics 1 with a different teacher (Mrs. Carrera) before enrolling
in AP Physics 2 with Mr. Buford. He usually sat at a table with the one other student (Blaine) who
also jumped straight from Physics 1 with Mrs. Carrera to AP Physics 2 with Mr. Buford. Otto often
had difficulties doing the computational activities because of his unfamiliarity with GlowScript.
He tended to consult Beck, who sat at the neighboring table, for help during the computational
activities. Otto took AP Computer Science in the prior academic year but he felt that it was hard to
transfer what he learned to GlowScript. Otto self-identified as a white man.
Blaine (he/him) took the same path through “regular” Physics 1 as Otto did. Blaine consistently
had difficulties getting started on the computational activities. After a few minutes of trying
unsuccessfully to make progress on the computation, he usually shifted his attention to joking around
163
and spent minutes at a time stringing together jokes and lighthearted observational commentary to
whoever would listen. He had a prior experience with computation when he did some computer
science activities informally over a summer in middle school, but he felt that he still had not learned
anything about computation. Blaine self-identified as a cisgender biracial (Black and white) man.
Ed (she/they) took AP Physics 1 with Mr. Buford the year before. She usually worked in a large
group with three to five other students, including Beck. During class when she was working, she
often talked out loud to herself and asked questions to herself. She felt a strong sense of community
in the class, and she often checked in with her group mates to see how they were doing. She had
some prior computational experience from AP Physics 1 with Mr. Buford and from her time as a
programmer in robotics club. Ed self-identified as a Black agender person. She clarified that she
goes by she/they pronouns and suggested for us to pick one to use or alternate between she and they,
with no preference among those options. We opted to use she/her pronouns alone for consistency.
Beck (he/him) also took AP Physics 1 with Mr. Buford the year before. He worked in the same
large group as Ed, which was usually formed at the start of class with students dragging three tables
together. Beck was an avid coder, and he decided to learn more GlowScript and do Khan academy
physics [218] over the summer after taking AP Physics 1. His father was a computer scientist.
Beck was often a resource for other students because he could finish most or all of a computational
activity without help, and he liked to share his code and explain his thinking to other students. Beck
self-identified as a white cisgender man. We placed Beck’s portion of our results into Appendix B
for the sake of brevity—the results section is already quite lengthy, and Beck did not provide many
new insights in comparison with the other students in this study. Where appropriate, we describe
and discuss the data from Beck and the meaning we interpreted from it, with a detailed analysis
provided in Appendix B.
Our methods were guided by a set of propositions [47, 52]. These serve to motivate the data
generation, transcription, and analysis that follow. First, we propose that CT dispositions are
present in what students say and do. This motivates us to collect and analyze interview data and
in-class recordings of students working, as well as construct our results around how CT dispositions
164
Data Sources
Student interviews Four interviews
Teacher interview One interview
Field notes Six class periods
Classroom
recordings
Two group recordings during one class
period, capturing all participants
Table 7.3: Data sources used in this study (more were generated, detailed in Chapter 6).
connected to the student data sets. Second, we propose that context-based connections between
CT dispositions and mindset can be drawn when both constructs describe the same piece of data.
This motivates us to analyze our data for both constructs and present a discussion on their potential
overlap in the data. Both of the propositions come from the development and use of the theories of
the CT dispositions [1] and mindset [2]. It is with these propositions that we flesh out our methods
below.
7.5.1 Data Generation and Transcription
We developed interview protocols and conducted semi-structured interviews [45] with the above
four students. The interview questions were written to explore the students’ feelings about physics
class and computational activities in accordance with an exploratory research design that investigated
dispositions. The original interview protocol for students is provided in Appendix A. We
also interviewed Mr. Buford and took observational field notes [219] for further contextual understanding.
Additionally, we recorded two of the student groups completing one of the computational
activities in Mr. Buford’s class. In this study, we focused our analysis on both the student interviews
and the in-class recordings in order to construct a triangulated understanding of how CT dispositions
connected to each student’s statements and actions. The choice to focus on both interviews
and in-class data aligns with our first proposition.
The interviews were transcribed for utterances alone without non-verbal communication. This
165
choice was driven by a focus on what participants say about their experience. The interviews were
conducted to ask about the perspectives of the research participants, so their comments during the
interviews are taken to represent this perspective. We understand that interview comments can only
represent how someone feels about their experiences [190], but rather than try to capture every
piece of information by analyzing non-verbal communication, we focus on what the participants
say because their responses were prompted verbally. We only included non-verbal communication
in the interview transcripts when it added meaning on its own to what a student said, such as a
head-slap or eyeroll.
On the other hand, for the in-class recordings, we wanted to highlight non-verbal communication,
such as gaze, gesture, and body position, because a student is much more likely to communicate
non-verbally in significantways when they are not being guided by interviewprompts. We represent
non-verbal communication with double parentheses. We also use special symbols for intonation,
cadence, emphasis, and other speech patterns, drawing from Jefferson [6]. We provide a legend for
the in-class transcription in Table 7.4.
7.5.2 Data Analysis
We used our two propositions to guide our data analysis and our interpretation of our findings. For
the first part of our analysis, we coded the interviews and in-class recordings for CT dispositions,
using the inclinations, sensitivities, and abilities from the theoretical framework as a coding scheme.
This coding was motivated as a way of exploring our first proposition: CT dispositions are present
in what students say and do. Using the results from our coding, we built our results around how
the CT dispositions showed up in the data based on the patterns across each students’ coding.
We provide the patterns and examples of coding in the results section. This serves the purpose
of demonstrating how the dispositions framework can be extended to a setting different from its
original context.
For transparency, we often coded excerpts or utterances for CT dispositions, even when the
excerpt was not about computation. This is because CT dispositions are more general than the
166
Transcription Key
: Prolongs the sound immediately preceding, more colons for a longer
prolongation (about 0.5 seconds for each colon)
(1.0) Indicates time elapsed in seconds
word Indicates emphasis
°word° Encloses quieter speech
WORD Indicates louder speech
?!,. Indicates the usual intonation or continuation of speech (in-class
and interview transcripts)
(inaudible) Indicates an utterance that I couldn’t make out
ˆwordˆ Encloses speech spoken with the cadence of reading out loud
[ Indicates the simultaneous start of overlapping utterances
] Indicates the simultaneous end of overlapping utterances
((word)) Encloses descriptions of non-verbal actions (in-class and interview
transcripts)
$word$ Encloses speech uttered while suppressing laughter
*word* Encloses speech with a creaky quality as if feigning being on the
verge of tears
word- Indicates where speech has been cut off
word< Indicates the end of an utterance that came to an abrupt stop
word= Indicates no elapsed time between equals signs
word Bold text is used only for the write-up to draw the reader’s attention
(in-class and interview transcripts)
Int Abbreviation for Interviewer (interview transcripts)
Table 7.4: Transcription Key. Some transcription symbols borrowed from Jefferson [6]. Symbols
are used only for in-class transcripts unless otherwise specified.
167
subject of computation, an important step to embracing the “thick authenticity” [207] of ongoing
curricular change in STEM. For example, one can exhibit a tolerance for ambiguity in a variety
of situations, and this builds into CT dispositions regardless of whether those situations involved
computation. We still focused much of the data generation on computation-related interview
prompts and computational tasks in class, but everything that students said and did mattered to us
in cataloguing their CT dispositions.
For each data source, we examined utterances one at a time, or as a small set of utterances
if a student stayed on the same topic. Each time, we checked to see if any of the inclinations,
sensitivities, or abilities would describe what the student was saying or doing. To show this
process, we include two examples of how we coded high and developing tolerance for ambiguity.
As an example of coding for high tolerance for ambiguity, Otto spoke out loud during class about
what he knew and did not know about the task at hand: “I know, its velocity, is gonna have to stay
the same. But I don’t know how to change that into like, an x and y, like separate components.” He
noted the velocity was constant but contrasted that with his uncertainty about how to split it into
components. This recognition of what he knew and what he lacked lined up with a key sensitivity
for tolerating ambiguity: alertness to opportunities to clarify what is known and unknown. The key
sensitivity described what Otto said, so we coded his statement for a high tolerance for ambiguity.
In contrast, we also show an example of coding for developing tolerance for ambiguity in Blaine’s
data. Blaine described how he tried to get an answer that mimicked the teacher’s in the interview
setting: “I just do stuff until I can get an answer similar to his.” He revealed a tendency to try and
put together a solution that appears correct by the standards of the teacher’s solution. In saying this,
Blaine indicated an adherence to one type of solution, which opposes a key ability for tolerating
ambiguity: acknowledging multiple possible solutions or explanations. The contrast between this
aspect of tolerating ambiguity and what Blaine said caused us to code his statement for a developing
tolerance for ambiguity.
Though the examples above show single instances of an inclination, sensitivity, ability, or
respective opposite, there were also utterances or excerpts to which we assigned multiple codes.
168
Every coding choice depended on how we interpreted what a student said or did. Depending on
how students described their experiences or feelings, we sometimes even coded a high aspect of
a disposition right after a developing aspect of a disposition (a few times, the same disposition!),
with both codes captured in the same excerpt of student talk or behavior. This mix of codes is
represented in our results section.
For the second part of our data analysis, we coded the data sources for mindset, using our
synthesis of mindset literature provided in Table 7.2. This choice was made not to compare mindset
to dispositions, but to ground our extension of the CT Dispositions Framework in a related but more
well-established construct. We did this coding to explore our second proposition: context-based
connections between CT dispositions and mindset can be drawn when both constructs describe the
same piece of data. We framed this part of the analysis to see how CT dispositions connected with
mindset. We discuss this to understand the strength of the connection between dispositions and
mindset, as well as evaluate Pérez’s characterization of dispositions using mindset [1].
As an example of how we used the mindset coding scheme, Ed described in her interview a
recent success with a class project. She had made a working telescope with paper towel tubes, and
in her words, “it just worked...we had worked for a whole two days on it, because the first day we
were trying to put an actual model of something between, like a phone, and it was awful. That was
a bad day.” Ed described the initial day as bad because of her failures to get the project working, but
she also connected the rough start to her success on the second day of the project. In doing so, she
aligned her approach to the project with an aspect of growth mindset: “setbacks are opportunities
to overcome a challenge.” This is how we came to code Ed’s excerpt for growth mindset.
We note that in Table 7.2, we provided descriptions of both the growth and fixed mindset codes,
which is in contrast with Table 7.1, in which we provided only high (no developing) dispositions
codes. The reason for this has to do with how CT dispositions and mindset were presented in the
literature. CT dispositions, because of its limited use in research [1], provides only half of a coding
scheme (the “high” half of the spectrum), whereas we must infer codes for the other half based on
how developing dispositions are framed in opposition to high dispositions in Pérez’s framework.
169
We also acknowledge here that there is a discrepancy in the CT Dispositions Framework between
how “developing” is framed as the low end of the spectrum and how the word itself implies a
trajectory from low to high. Pérez describes this language choice as a way of highlighting the
dynamic nature of dispositions. However, this created a dissonance between the framework and
the data when we used “developing” in our coding scheme to describe behavior that indicated no
development whatsoever.
In contrast to how dispositions was framed, due to the extensive research and nuanced development
of mindset theory [2, 3, 4, 5], we were able to build a coding scheme that reflected both sides
of the mindset spectrum (fixed and growth) with enough detail to apply it to our data. This leads
to a nuance with our mindset coding scheme in that fixed mindset does not always equate to the
opposite of growth mindset, and vice versa. For example, in the second row of Table 7.2, growth
mindset aligns with seeing learning as important, whereas fixed mindset aligns with valuing innate
intelligence and/or superiority. These are not opposites. Not valuing learning (the opposite of the
growth mindset statement) is not the same as valuing intelligence/superiority. In analyzing for these
codes in our data, we coded only for instances of fixed mindset and growth mindset, as opposed
to opposite-of-growth mindset and opposite-of-fixed mindset, directly looking for statements that
lined up with either column of Table 7.2. As an example, Blaine made a comment in class which
got brought up in his interview: “what’s the point of learning code? I can draw this on a piece of
paper in fifteen seconds.” This is an example where it is tempting to code this quote for Blaine’s
devaluation of learning computation, a direct contrast to the growth mindset code, “learning is
important.” However, this statement was not coded. Instead, we coded a comment that Blaine
made later in the same excerpt: “Just have a line, make it curve. I can do that, real quick...I don’t
need to plug it into a computer to draw some straight lines.” In explaining his thinking, Blaine
revealed a desire to avoid the challenge of plugging what he knew into the computer. This avoidance
of challenges aligns directly with an aspect of fixed mindset from our coding scheme, so we were
able to analyze the excerpt and describe how certain features of Blaine’s statement aligned with
fixed mindset.
170
We interpret and evaluate our findings on a couple of criteria (our “logic model”). First, we
intend to see how well the dispositions framework describes what students said and did, which
would indicate the degree to which the first proposition describes our case. Second, we intend to
ascertain the nature of the connection between mindset and dispositions. We have already outlined
how Pérez connected the constructs [1] in Section 7.2, but we also dedicate part of our analysis
to evaluating their connection in our data. Some relevant questions are: How is mindset reflected
in dispositions in our data? How is it not reflected? How do these results compare to how Pérez
connected mindset and dispositions in his theory of CT dispositions? Exploring these questions
will indicate the degree to which the second proposition describes our case.
The last part of our methods is about addressing the inter-coder reliability [220] of the results
below. The process for coding involved the first author applying the CT Dispositions Framework
to the data iteratively. The three co-authors met between each of these iterations of applications of
the coding scheme to discuss how the first author had applied the framework, adding robustness to
the coding process each time by converging on interpretations of students’ statements and actions
via discussion. This convergent process reflects the process of computing a value for inter-coder
reliability, just without the numerical value [220]. We used the same process to refine our use of
the mindset coding scheme. This analytic procedure also reflects widely consulted measures of
reliability for qualitative analysis of discursive data [221]. Specifically, what matters for reliability
in interpretive work is not consensus, because consensus cannot be taken to mean the phenomenon
“exists independently of the speakers” (page 165) [221]. Instead, what matters in work like this is
“exploration” of the different interpretations that these three co-authors brought together, with the
acknowledgment that “there is always the possibility of a new interpration” (page 165) [221]. In
this way, we assert that the claims made in the results below hold a degree of robustness such that
at least three researchers agree on their interpretations. The claims do not amount to a construction
of objective truth, but the exploration of interpretations from our analytic process affords a degree
of “trustworthiness” [221]. We also include evidence abundantly because results from spoken data
are necessarily interpretive [220, 221]. That said, we assert reliability, interpretive consistency, and
171
trustworthiness as described above.
7.6 Results
We organize our results into two subsections. First, we apply the CT Dispositions Framework to
each participating student’s data set. Second, we analyze that same data for mindset and show how
the frameworks connect in the data. In this section, we first present the results from Otto, whose
behavior in the recorded class period and whose responses in the interview often aligned with
high dispositions and growth mindset. Then, we discuss Blaine’s data set, which contrasted with
Otto’s: Blaine’s actions and words during the in-class recording and interview were more often
coded for developing dispositions and fixed mindset. Finally, we present Ed’s data set, in which we
observed many statements and behaviors aligned with both sides of the dispositions and mindset
spectrums. Ed’s analysis revealed complexities related to the constructs. We do not present the
results from Beck’s data in detail here because his analyzed actions and words also aligned with
high-level dispositions and growth mindset. Given the overlap between the codes applied to Otto’s
and Beck’s data, we did not gain much additional insight into the framework from Beck’s results,
so we summarize his perspective at the end of the results in Section 7.6.1.4 and include a fuller
analysis of Beck’s data in Appendix B for those interested.
7.6.1 Dispositions Results
We separate results by each student, using data from their interview and in-class behavior to
construct an analysis of howCT dispositions align with what one student said and did in Mr. Buford’s
class. We compare and connect the results from each students’ data set later in Section 7.7.
7.6.1.1 Otto (Dispositions)
Otto’s statements and actions aligned with high levels across all dispositions. In Table 7.5, we
show how we coded Otto’s data from both the interviews and in-class data. The codes in the table
are split between dispositions to show variance among dispositions and between data sources. One
172
Tolerance for
Ambiguity
Persistence on
Difficult Problems
Willingness to
Collaborate with Others
Otto Interview 21 high
1 developing
11 high
1 developing
13 high
2 developing
Otto In-class 9 high
0 developing
9 high
0 developing
15 high
3 developing
Otto Total 30 high
1 developing
20 high
1 developing
28 high
5 developing
Table 7.5: Coded instances of alignment between CT dispositions and Otto’s data, separated by
data source.
aspect of Otto’s table, which we will return to, is that there were more instances of tolerance for
ambiguity in his interview than in his in-class data.
7.6.1.1.1 Tolerance for Ambiguity
Otto made many statements in his interview that aligned with high tolerance for ambiguity. Below,
he described his thought process when working through a complex physics problem (noncomputation).
Otto.Interview.1:
Otto I just kind of look at what I have and then, I just- I think about it. I just try to go
look at something and try to go off how that’s related... I’ll just try to go through
the process of how things work, see how- where different values appear. Yes,
if I’m looking at a sheet of equations, whatever we already have. And just try
to find a way that makes sense for me in my head. Try to find some solution that
logically makes sense to me.
His general approach was to survey what information or equations were available to him and
see what he could find from there. His tendency to see “where different values appear” and
173
look at equations “we already have” demonstrates an alertness for opportunities to clarify what
is known about the problem (interview commentary aligned with key sensitivity #2, tolerance for
ambiguity). After clarifying, he describes trying to find a solution that makes sense, in effect
navigating incomplete data towards a solution (interview commentary aligned with key ability #3,
tolerance for ambiguity).
When describing the computational activities, Otto again embraced a high tolerance for ambiguity.
In the example below, he chose to talk about the example of electrons in a magnetic field
when discussing his approach to computational problems.
Otto.Interview.2:
Int What process do you go through when you have to do a GlowScript problem?
Otto ...We were doing particles, like electrons moving through magnetic fields and
how they move. See where the forces were and everything. I guess with that
specifically, you just think about which direction things go and what kind of
vectors and how strong they all are, I guess. To break it up into each individual
little piece and just figure out the order and everything that goes together.
How they’re tied together.
His approach was to break down the information in the problem into vectors and figure out
how it all went together. Specifically, he described it as, “breaking it up into each individual little
piece and just figuring out the order and everything that goes together.” As indicated previously,
Mr. Buford framed the computational problems as inherently ambiguous, so this act of reorganizing
and tying different pieces of the problem together is an approach to reframe an ambiguous situation
(interview commentary aligned with key sensitivity #3, tolerance for ambiguity). Otto did this
to work towards a solution to the computational problem, which points to navigating through
uncertainty towards a solution (interview commentary aligned with key ability #3, tolerance for
ambiguity).
174
Otto also exhibited curiosity and open-mindedness towards computation when it was first
introduced into the class. This was his response to a question about what that was like:
Otto.Interview.3:
Int Was this the first time you used Python?
Otto Yeah. Yeah. He was just kind of like, ‘We’re doing coding today.’ I was like,
‘Oh, I guess I’ll figure it out a little bit.’
Otto was interested in figuring out the unfamiliar activity he was about to embark on. Otto could
have easily had a negative reaction and decided that he would be unable to do the day’s activity
since he had no previous experience with GlowScript or Python. Instead, his reaction to learning
about the new computation was, “Oh, I guess I’ll figure it out a little bit.” This willingness to figure
out something he had never done before indicates an interest in exploring unfamiliar situations
(interview commentary aligned with key inclination #1, tolerance for ambiguity).
In class, most of what Otto displayed in line with tolerance for ambiguity was when he talked
about what he did and did not know about a physics problem. For example, the excerpt below shows
Otto talking about his options for specifying how a particle would move in his computer program.
Leading up to this conversation, Otto was trying to tell Beck how he wanted to write the condition
of his while loop. Otto was weighing the options of using the lens (represented by a “box” in the
code) versus using the x-coordinate of a moving particle as it reached the lens.
Otto.In-class.1 / Beck.In-class.6:
Otto Yeah. And I don’t really wanna use like the actual, [box
Beck [Now you just have to
Otto I don’t wanna use the actual like lens as the thing that triggers it I just wanna
use the coordinate as the thing that triggers it
Beck Well yeah
Otto But
175
Beck Whenever its x posi- Once its x position reaches= the x position of the lens=
which is zero, then you can make the change
Otto =Zero =Yeah
Otto understood there could be multiple triggers in his program for the particle motion, and he
indicated a preference for one solution path without implying that there was only one answer (only
that he “wants” to use the coordinate). This indicated an acknowledgment of multiple possible
solutions to this aspect of the problem (in-class behavior aligned with key ability #1, tolerance for
ambiguity).
Later during class when the researcher was sitting at Otto’s table, Otto started explaining
unprompted where he was at in the problem. The “it” below refers to the moving light particle,
whose path Otto was trying to refract through the lens in the code.
Otto.In-class.2:
Otto I know, its velocity, is gonna have to stay the same. But I don’t know how to
change that into like, an x and y, like separate components. Gah:
Otto demonstrated that he could reiterate what he knew about the problem (“velocity is gonna
have to stay the same”) while identifying the parts that he was not sure about yet (“how to change
that into...separate components”). This awareness indicated alertness to an opportunity to clarify
what he knew and did not know (in-class behavior aligned with key sensitivity #2, tolerance for
ambiguity).
Overall, we sawevidence for Otto embracing a high tolerance for ambiguity in both his interview
and his in-class conduct. He took up several of the key inclinations, sensitivities, and abilities:
an interest in exploring unfamiliar situations, an alertness to opportunities to clarify what he
knew, acknowledging multiple possible solutions, a responsiveness to approaches for reframing
ambiguous situations, and navigating incomplete data and uncertainty to find a solution.
176
7.6.1.1.2 Persistence
Otto also embraced a high persistence on difficult problems. However, he once expressed in his
interview a statement aligned with developing persistence for the computational activities, which
was when we asked Otto about how he felt upon completing the computational problems. His
response indicated that he did not always feel satisfaction after completing a challenging activity.
In the conversation preceding the excerpt below, Otto was discussing how he often felt like he could
not succeed at Mr. Buford’s computational problems because he did not know the programming
language very well.
Otto.Interview.4:
Int Is there a point when you get [the code] to work or is it just like class ends and
you’re still like, don’t know what you’re doing?
Otto I got it to work, eventually, but it was still where it’s like, ‘finally’ type thing,
not like, ‘Okay, yeah.’
Int It wasn’t. Okay. It wasn’t like, satisfying or anything?
Otto No, it was just kind of like, ‘Yeah, I know it should’ve been working that entire
time.’
For the computational activity that was in Otto’s mind at this moment, he “got it to work
eventually,” indicating that he stuck with the task for an extended period (interview commentary
aligned with key ability #1, persistence). However, he expressed some feelings of exhaustion,
saying “finally,” and “it should’ve been working that entire time.” He said that he was not satisfied,
despite the success that his efforts yielded. This showed that Otto experienced this act of sustained
effort as exhausting rather than rewarding (interview commentary opposed to key sensitivity #2,
persistence). This demonstrates that Otto did not always articulate high persistence, even though
he did exert a great deal of effort towards solving the computational problems.
In contrast to his interview statement, Otto acted in accordance with high persistence during
class. He continuously ran into computational roadblocks during class and each time persisted by
177
asking for help or diving back into the work to see where he could fix the issue. For example,
the excerpt below shows Otto asking for help. Leading up to this conversation, Otto was working
consistently and had just run his code, which gave him an error message for an undefined variable.
Otto.In-class.3:
Otto Ah, mm. Why is that wrong? Mm. BECK. Why is it undefined? [No. I’ve
been doing-
Beck [°Yeah.° (inaudible). °It’s pretty cool°
Otto Um:. It said that x was undefined here for some reason. Like °I dunno what’s
wrong°. See?
Beck Um.
Otto So: it says right here in line 17.
Beck ˆLine 17ˆ at or °(inaudible)° okay. Uhm.
(3.5)
Beck ˆVel dot position dotˆ Okay, well vel, you just need vel dot x, not vel dot position
dot x, cause vel [is (inaudible)
Otto [Oh:: (2.0) Thanks:
(7.0)
Otto ((laughs uncontrollably)) $To be honest I don’t [ra-$
Blaine [No dude just ah- y- you got it.
Otto That’s a good step though.
Blaine Now just hit a negative sign.
Otto ((lifts up laptop and turns it into camera’s view)) IT WENT UP, I NEEDED IT
TO GO DOWN. ((laughs quietly through nose, and then a hard exhale)) So I just
need to switch<
178
(2.5)
Otto (inaudible)
(9.0)
Otto Oh, [because it needs to be. (1.0) Negative. Ah: that’s why
His reaction was to call out for Beck and ask him why the variable was “undefined” as the
error message indicated. Otto’s recognition of his need for Beck’s help showed an attentiveness to
the opportunity (presented by the error message) to shift tactics in order to move forward (in-class
behavior aligned with key sensitivity #3, persistence). Beck proceeded to attend to Otto and help
him handle the error. After this, Otto ran into another problem related to the animation (“It went
up, I needed it to go down”). He continued to expend effort by working on his own for the next few
moments, eventually arriving at an explanation for the issue (“it needs to be negative, that’s why”).
Otto’s perseverance through multiple roadblocks constitutes sticking with the task for an extended
period (in-class behavior aligned with key ability #1, persistence). He also pursued a resource
(Beck) who could help translate Otto’s efforts into progress (in-class behavior aligned with key
ability #3, persistence).
Later during the class period, Otto turned to Mr. Buford for help. When Ottowas communicating
what part he got stuck on, he reviewed all the steps he had taken so far.
Otto.In-class.4:
Otto ((briefly raises hand)) Mr. Buford. Okay. Okay.
Teacher ((takes chair with one hand, lifts it up and moves it forward so he can stand to
Otto’s left and look at laptop screen))
Otto So what I’ve got right here.
Teacher Yeah. That works.
Otto °Yeah°. But, this is gonna- it’s supposed to be the focal point.
Teacher Oh. [Alright, so, °ah°]
179
Otto [So that’s why, it’s weird.]
Teacher ((takes glasses out of shirt pocket and puts them on)) So how did you defi:ne.
((squats and folds arms on corner of table next to Otto)) How did you define
which way it would go after it cot- got to the lens? [What did you say?
Otto [Alright, so. I took the speed, and I did a bunch of: trig stuff. So I found the
angle right here. That it would need to go at. And I tried to turn that into
a: like x: components and y components of a vector. And I just changed the
velocity there.
From the beginning of the interaction, Otto was raising his hand and calling over the teacher to
help: “Mr. Buford.” The seeking of help at this point (with about eight minutes left in class)—after
working at his program for most of the class period—indicates that Otto wanted to continue working
at the problem, a sign that he could stick with a task for an extended period of time (in-class behavior
aligned with key ability #1, persistence). As they began interacting, Otto specified the source of the
issue for which he was trying to get help: “this is gonna- it’s supposed to be the focal point.” His
awareness of where this issue came into play, his knowledge of how it should have been different,
and his ability to point it out to Mr. Buford all indicate that he was alert to the characteristics of the
task (in-class behavior aligned with key sensitivity #1, persistence).
Overall, Otto demonstrated actions that aligned with high persistence on difficult problems,
though he once indicated that the computational problems did not make him feel satisfied for his
efforts, and that statement aligned with developing persistence. In terms of the key aspects of
persistence, in the above excerpts Otto demonstrated an alertness to a task’s characteristics, an
attentiveness to opportunities to shift tactics when needed, an ability to stick with a task for an
extended period of time, and a pursuit of resources that increased the effectiveness of his effort.
On other hand, Otto’s interview also once showed that he was unaware or unable to experience
satisfaction from the fruits that his efforts yielded.
180
7.6.1.1.3 Collaboration
Otto also demonstrated in his data several statements and actions aligned with a high willingness
to collaborate with others. According to Otto, this was something that was designed into the norms
of the class:
Otto.Interview.5:
Int What about the people you sit near? Do they kind of expect you to need help
during the coding or- ?
Otto I wouldn’t say they expect it, but they’re not surprised when I do. See what I
mean? It’s more of just an accepted thing that you help people.
In his last utterance, Otto identified a norm of in-class work: “It’s more of just an accepted thing
that you help people.” Earlier in the interview, Otto indicated that he had taken up these norms,
too. When he got stuck, he turned to his peers for help before considering asking the teacher:
Otto.Interview.6:
Int That makes sense. What role does Mr. Buford play when you’re working with
your classmates?
Otto Usually most problems can be understood just by talking to other people like
Beck and those people that are good at it. But if you, nobody really gets it at all,
you can just ask him and he’ll come over and explain it and help walk you through
the process of what’s happening.
This quote demonstrates Otto’s willingness to ask for help, and it also demonstrates that he saw
Beck’s smartness as a benefit to him rather than a threat/competition. The connection Otto made
between understanding and collaborating (“most problems can be understood just by talking to other
people”) indicates a tendency to value different perspectives (interview commentary aligned with
key inclination #2, collaboration) and an attentiveness to the insights that come from interactions
(interview commentary aligned with key sensitivity #3, collaboration).
181
Furthermore, Otto demonstrated his awareness of his peers and the value he assigned to their
explanations. Otto explained how he viewed the “smart” students, saying that the best indicator of
intelligence was the ability to explain concepts to peers.
Otto.Interview.7:
Int Is it like everybody’s on equal footing, contributing the same thing?
Otto It’s pretty egalitarian, yeah. I feel like, at least I personally, tend to take more of
an explainer type role. I think I have a little bit of aptitude for physics. Like the
dude that sits behind me, Beck, he’s probably like- If you could say one person
was an explainer type guy, it’s him. He’s really smart. Joyce too... She’s
really smart.
Int It sounds like you’re equating smartness with explaining.
Otto Well, ability to explain. There’s some people that are about as good as [Beck and
Joyce] are in terms of just getting problems right and understanding the concepts.
But [Beck and Joyce] tend to be the ones who are able to express that to other
people.
Otto set up a relationship between explaining well and being smart. He identified Beck and
Joyce as the best explainers and smartest students in the class: “If you could say one person was
an explainer type guy, it’s Beck. He’s really smart. Joyce too...She’s really smart.” By recognizing
the merits of explanation, Otto demonstrated an alertness to an interpersonal dynamic (explanation)
that enhanced effective interactions (interview commentary aligned with key sensitivity #1,
collaboration).
In the above excerpt, Otto recognized that explaining to others was a good thing. However,
he did not always take this view. Later in the interview when talking about his strongest class,
calculus, he admitted that he did not like to work together as much.
Otto.Interview.8:
Int Do you work in groups in [calculus] class or is it by yourself?
182
Otto That one’s a lot more solitary, I’d say. We get work time, but usually it’s just
trying to figure out the problem yourself.
Int Do you like that more?
Otto Yeah. I’d say so. Groups are fun, but I think I tend to work better by myself.
Especially in something like Calc where I feel like I have a stronger base and
everything.
He admitted that he preferred to work alone in calculus: “Groups are fun, but I think I tend to
work better by myself.” This indicates a resistance to having his course of action influenced by
interactions with others (interview commentary opposed to key inclination #1, collaboration). The
justification he gave for his reservation was that he had a “stronger base” in calculus. Otto’s extra
strength in calculus might have indicated that his peers had even more to gain from his help than
they would in physics, but in contrast he was more reluctant to collaborate. An open question is
whether Otto might see himself in calculus similar to how he sees Beck and Joyce in physics. If so,
this could create a conflict between his desire to work alone and his self-perceived competence at
explaining calculus to peers.
From the in-class data, Otto frequently collaborated with peers. Below, we analyze an example
of Otto collaborating with Beck on implementing an animation for a moving light particle. In the
lead-up to the excerpt, Otto asked Beck to help (“Beck help me”) and invited Beck to provide his
own perspective on how to edit the code (“How do I make it move?”). Below, their collaboration
played out after Beck had helped for a little while but there were still errors to deal with.
Otto.In-class.5 / Beck.In-class.3:
Otto So run that and it’ll just, ((pointing)) straight
Beck Let’s see what happens, should do (inaudible). Straight to the right. ˆInconsistent
indentation one fullˆ- let’s see, see that’s why I didn’t- Alright so, light- I’m just
gonna
Otto Just retype it
183
Beck ˆWhile light dot position dot x less thanˆ, °what was it?°
Otto Light- I mean um
Beck Focal point?
Otto Uh, yeah. Focal point dot pos: x
Beck °Position dot x°
Otto Hundred
Beck °Velocity one hundred°
Beck Er::, oh! Got it. Oh, colon
Otto Oh you need a colon? Ah!
Beck One hundred. Yeah, that’s a thing you do need. It should- Yeah! And that just
travels straight to the right. Until it gets to there
The sequence of contributions followed the pattern of Otto making a verbal contribution (e.g.,
“just retype it”) and Beck reading or adding to the code (e.g., “while light dot position dot x
less than”). The overall trajectory of the interaction moved from an initial error (“inconsistent
indentation one full-”) to an eventual solution (“It should- Yeah! And that just travels straight to
the right”). Otto’s utterances along the way pointed to his ability to clarify and negotiate the shared
understanding and course of action (in-class behavior aligned with key ability #3, collaboration).
Otto did not just hand his computer to Beck and say “fix it”—he was working together with Beck,
suggesting paths (e.g., “just retype it”), contributing chunks of code (e.g., “focal point dot pos
x”), and clarifying the known quantities (e.g., “hundred”). The eventual solution to Otto’s coding
problem represented Otto’s willingness to let the interaction with Beck shape his course (in-class
behavior aligned with key inclination #1, collaboration).
Overall, Otto’s words and actions aligned with a disposition for high collaboration with others.
This was clear throughout his interview and in-class behavior. He displayed several inclinations,
sensitivities, and abilities aligned with high collaboration: a willingness to have his course changed
184
by interactions with others, a tendency to invite and value perspectives different from his own, an
attentiveness to the unique insights that emerge from interactions, an ability to listen to and have
his actions shaped by others, and an ability to negotiate the group’s understanding. A contrasting
instance was Otto’s hesitancy to collaborate with peers in calculus class, to which we return in
Section 7.7. This particular excerpt was coded for an resistance on Otto’s part to have his course
changed by interactions with others.
Otto’s data consistently embraced high CT dispositions, meaning that his actions and views
could potentially be characterized for their contribution towards a development and take-up of
CT practices, as theorized in Pérez’s description of the framework [1]. There were a couple of
exceptions to this general trend in Otto’s data, most notably the lack of satisfaction he felt after
persisting through a computational activity and his tendency to preferworking without collaboration
in an environment when he does not need help. However, the vast majority of Otto’s data that we
analyzed was coded for high tolerance for ambiguity, high persistence, and high collaboration.
7.6.1.2 Blaine (Dispositions)
In contrast to Otto’s data, Blaine’s statements and actions were coded for developing dispositions
more often than high. The coding of Blaine’s in-class data and interviewis summarized in Table 7.6.
There did not seem to be any major differences between the data sources for Blaine. The mix of
high and developing codes for collaboration compose a pattern in Blaine’s behavior that does not
align well with the descriptions of developing or high collaboration alone from Pérez’s framework.
We detail this finding further in Section 7.6.1.2.3.
7.6.1.2.1 Ambiguity
Blaine’s data aligned with a developing tolerance for ambiguity. He often made statements in
his interview and in class that opposed some of the key inclinations, sensitivities, and abilities
associated with tolerating ambiguity. At times, Blaine expressed distaste towards a lack of clarity,
185
Tolerance for
Ambiguity
Persistence on
Difficult Problems
Willingness to
Collaborate with Others
Blaine Interview
3 high
9 developing
3 high
6 developing
4 high
7 developing
Blaine Inclass
0 high
12 developing
1 high
10 developing
3 high
3 developing
Blaine Total 3 high
21 developing
4 high
16 developing
7 high
10 developing
Table 7.6: Coded instances of CT dispositions in Blaine’s data, separated by data source.
or he refused to engage with complex problems. For example, he recalled a time when he had to
create a magnetic motor as part of a physics project.
Blaine.Interview.1:
Blaine We had a motor project where he just gave us a wire and a magnet and he was
like, ‘Do it.’
Int Okay. Can you describe that for me?
Blaine Well, he just gave us a wire and the magnet and he was like, ‘Come back with
a motor and explain how you did it.’ I waited until the last night. I was
rummaging through the kitchen cabinet trying to pull out some stuff. But it was
really stressful because it was hard to get it to continuously work.
Int The motor?
Blaine Yeah, the motor because I can get the- I think what he wanted was- He didn’t
ever say what he wanted but you had to put the wire into a loop and then get
it to keep spinning for twenty seconds.
Blaine interpreted the teacher’s statement of the project to be just “do it,” indicating that Blaine
found a lack of clarity around the assignment. In the last line, he again emphasized that Mr. Buford
186
“didn’t ever say what he wanted,” yet Blaine also had an understanding that he had to get the motor
“spinning for twenty seconds.” This discrepancy indicates that the uncertainty for Blaine lay in
how to get the wire spinning properly. He was focused on this open-ended request from the teacher,
and only described the other features of the project when we asked for elaboration. These signs
indicate that Blaine was not keen to navigate the incomplete data or the uncertain trajectory of the
project (interview commentary opposed to key ability #3, tolerance for ambiguity).
At a later point in the interview, Blaine described his general approach to problem-solving in
class:
Blaine.Interview.2:
Blaine I’m just like, ‘I’m going to take every equation on this equation sheet and
we’re going to see what I can make happen.’ He usually puts his answers at
the front, so I just do stuff until I can get an answer similar to his.
When Blaine was not sure, he just took “every equation on the equation sheet” and saw “what
[he] can make happen.” This stood in contrast to Otto.Interview.1, when Otto said, “looking at a
sheet of equations...to find some solution that logically makes sense to me.” Blaine’s approach had
little to do with making sense. Instead, he tried each and every equation until he “gets an answer
similar” to the teacher’s, meaning Blaine was concerned with the appearance of correctness over
conceptual understanding. Hewas not interested in discovering meaning not yet apparent (interview
commentary opposed to key inclination #2, tolerance for ambiguity) or even considering multiple
possible solutions (interview commentary opposed to key ability #1, tolerance for ambiguity).
Blaine indicated a primary interest in having an answer that looked correct by superficial standards.
Given Blaine’s avoidance of ambiguity, we asked in his interview if there was anything at all
from the open-ended computational activities that he saw as beneficial. His response was one of
the few times that Blaine’s data could be interpreted in terms of a high tolerance for ambiguity.
Blaine.Interview.3:
187
Int Is there anything new that [computation] brings to the class: new material or new
understanding, new ways to see the physics?
Blaine I guess if you can actually do it, it gives you visuals on what would actually happen.
Because most of them it’s stuff we can’t- an electron going into something
and we can’t see that. It gives us real examples of what’s going on.
He acknowledged that there were some physics concepts you just cannot see, like electron
motion, and the code helped “give you visuals on what would actually happen.” This was one of
the only times in the data that Blaine expressed an embrace of uncertainty related to computation
or physics. When he acknowledged, “we can’t see [an electron]. It gives us real examples of what’s
going on,” Blaine was displaying an understanding that computation helps reframe the ambiguous
physics concept to make it more accessible (interview commentary aligned with key sensitivity #3,
tolerance for ambiguity).
Despite his statement above, during the in-class work, Blaine often displayed a negative stance
towards ambiguity in the computational activity. Blaine demonstrated frustration that the computation
was not more straightforward and literal. For example, when Otto articulated a roadblock he
encountered in the code, Blaine became frustrated and defensive about the pattern of roadblocks
that he encountered all the time:
Blaine.In-class.1:
Otto I have like this velocity vector saying that it’s going to the right, at that- but I
don’t know how to turn that into just a, you know. Down and to the right. Or up
and to the right, or whatever
Blaine ˆDown, parentheses ninety degrees.ˆ That’s how it should be. If I put in line, a
line should appear. I don’t understand why it doesn’t, you know?
Blaine expressed disapproval with how the code did not work in response to his commands
(“that’s how it should be”). This indicated that Blaine had little intention in this moment to
understand how computation worked and by extension no interest in discovering the meaning
188
associated with the computational task (in-class behavior opposed to key inclination #2, tolerance
for ambiguity). His input-output stance (“If I put in line, a line should appear”) reveals that he
has rigidly categorized the way he thinks GlowScript should work (in-class behavior opposed to
key inclination #3, tolerance for ambiguity). This straightforward view also indicates no interest
in navigating an uncertain trajectory toward a solution when dealing with computation (in-class
behavior opposed to key ability #3, tolerance for ambiguity).
Blaine again demonstrated a resistance against engaging with uncertainty when faced with
errors in his program. Instead of deciphering the error or reworking his program, Blaine searched
online for sample code to copy and paste into his program.
Blaine.In-class.2:
Blaine and then look. dude I did it look at how good I am. you see that? error
unexpected? what? ((taps six times on mouse)) I don’t know how to code
(4.0)
Blaine ((while searching online)) ˆcode. for. l<. straight. line. in glow script. in glow
s:cript. glow: s:cri:ptˆ
(8.0)
Blaine ˆSample code. Code glowscriptˆ
(19.0)
Blaine ˆGlowscript lightˆ (inaudible). ˆGlow. Glowscript. Glow script drake sample?
Glowscriptˆ
The episode began with Blaine trying to run some code that he thought would work. He got
an error (“error unexpected? what?”). His response to this surprise roadblock was to express,
“I don’t know how to code,” which demonstrated that he had little interest in the opportunity to
grow by engaging with an uncertain situation: the error message (in-class behavior opposed to
key sensitivity #1, tolerance for ambiguity). He went on to search the web for “sample code” that
189
he could use in his program, a strategy that he returned to three minutes later after making no
progress (“Sample code! ˆGlowscript. Glow:scriptˆ...I don’t want a tutorial I just want sample
code.”). His choice to copy code can be described as a legitimate alternative approach for creating
computational models [205], but Blaine’s adherence to the strategy was not constructive.
Given the abundance of times Blaine avoided engaging with ambiguity, it was clear that Blaine
preferred to find an answer, even one that was not right or he did not understand, than to explore
the problems on his own. He seemed to recoil from situations that presented uncertainty. Overall,
Blaine was resistant to engaging with ambiguity. When examining the key inclinations, sensitivities,
and abilities present in the above excerpts, we saw that Blaine at times avoided exploring
unfamiliar situations, had little interest in discovering meaning not yet apparent, rigidly categorized,
was unaware that engaging with an uncertain situation could lead to growth, was unaware
of an opportunity to reframe an ambiguous situation, and was unable to navigate an uncertain
trajectory toward a solution. On one instance when reflecting on the big picture of computation
in Blaine.Interview.3, Blaine displayed an appreciation for computational opportunities to reframe
ambiguous situations, but this embrace of tolerance for ambiguity was rare in his data. Overall,
Blaine’s data aligned with a developing tolerance for ambiguity.
7.6.1.2.2 Persistence
Blaine’s data also aligned closely with developing persistence. After describing the motor project
in Blaine.Interview.1, Blaine went on to discuss other stressful and time-sensitive features of the
project. Afterwards, we asked whether he felt that he had learned anything:
Blaine.Interview.4:
Int Did you feel like you learned something when you did that?
Blaine I learned that YouTube can teach you a lot of things.
Int It’s a resource, is it?
190
Blaine Yeah. It explained how the motors were working because we had to figure out
how to get them to work and use the magnet to simulate the current and stuff. It
put some of the things we were learning in class and physically applied them
to make sense of it.
In the end, he felt that consulting YouTube helped him contextualize the concepts he had been
learning about (“physically applied some of the things we were learning in class”). The way
YouTube helped Blaine put the physics concept into action indicates that he pursued a resource
(YouTube) that could increase the effectiveness of his efforts (interview commentary aligned with
key ability #3, persistence). When compared to other excerpts in Blaine’s data set, this was one of
only ones coded for high persistence.
Despite occasional instances of high persistence codes, the vast majority of what Blaine said and
did in the data aligned with developing persistence on difficult problems. Below, Blaine described
how he came to adopt a tendency to avoid effort during computational problems.
Blaine.Interview.5:
Blaine I would try if I could literally get anything. But since I literally can’t get
anything but a blank screen, I don’t really try to do any more because I’ll put in
a hundred things and then I’ll just get a blank screen or I’ll get some error.
It’s like, ‘Line 17.’ Well, I don’t know what line 17 is, man.
This demonstrates his lack of engagement in extended effort for the computational activities
(interview commentary opposed to key inclination #1, persistence). He indicated that he would
try “a hundred things,” only to “get a blank screen or some error,” but we did not code this for
high persistence because he indicated that he has given up on trying anymore (“I would try if I
could literally get anything”). The sequence of error messages and blank screens after trying so
many times shows that he was unable to change his approach after considerable effort (interview
commentary opposed to key ability #2, persistence), though the effort itself shows that Blaine
was persistent before losing hope (interview commentary aligned with key ability #1, persistence).
191
When thinking about what Blaine meant when he said that he tried so many times, it helps to
consult his in-class attempts at progress, such as the one in the next excerpt.
During class, Blaine behaved in alignment with developing persistence. At one point in class,
Blaine displayed relatively consistent engagement with the problem. Below, he ran into an issue
where he was not sure how to represent a variable as a vector in the code. He spoke to himself
throughout the excerpt.
Blaine.In-class.3:
Blaine How do you make position a vector?
(3.5)
Blaine It says position °must be a vector°
(6.5)
Blaine ((stands up to gaze over Beck’s shoulder))
(26.0)
Blaine I cede
The error was new to him, but within ten seconds he decided to seek answers by looking at
Beck’s code. This might at first seem like an effort to leverage Beck as a resource, but Blaine did
not interact with Beck as Otto did. Instead, Blaine opted to try absorbing or copying the answers
by looking at Beck’s code over his shoulder. In this way, we also see Beck ignoring an opportunity
to collaborate with Blaine on tackling this error (for Beck, this was opposed to key sensitivity
#1, collaboration). Less than half a minute later, Blaine announced that he “cedes,” or gave up.
This demonstrates a devaluation of extended effort (in-class behavior opposed to key inclination
#1, persistence) and an instance of giving up on a task without spending much time on it (in-class
behavior opposed to key ability #1, persistence). His disengagement with the activity continued for
the rest of the class period, about ten minutes.
Blaine was quick to give up when encountering challenges in the computational activity. In
terms of key aspects of persistence, Blaine’s data showed an example of devaluing extended effort,
192
giving up on a task before much elapsed time, and a resistance to changing approaches after
considerable effort but no progress. On one occasion, in Blaine.Interview.4, he demonstrated an
ability to pursue a resource (i.e., YouTube) that increased the effectiveness of his effort. On the
whole, Blaine’s data represented a pattern of codes for developing persistence.
7.6.1.2.3 Collaboration
When it came to collaboration, Blaine’s data consistently encompassed both high and developing
codes. In his interview, he described a tendency to work solo, and when he did ask for help,
the request was for answers, not to develop a shared solution with meaningful contributions from
multiple people. He described his approach to in-class physics problems below.
Blaine.Interview.6:
Blaine I just usually sit next to Otto. Then Otto will know most of the stuff. Then he’ll
ask Beck how to do other stuff and I just watch what they’re doing. I’m like,
‘All right.’ Then I try to do it because if I were to ask questions for every
problem I need help with, I’d ask questions on every problem. I just don’t even
really...I just try and figure out how they got there.
Blaine described observing Otto and Beck and then trying to do the problem based on what he
saw, rather than participating in the collaboration. On the one hand, this represented an ability to
listen to and have his actions shaped by others (interview commentary aligned with key ability #1,
collaboration) and an attentiveness to the insights derived from interactions (interview commentary
aligned with key sensitivity #3, collaboration), even if on the other hand, the interactions did not
involve Blaine. His silent observation also pointed to a hesitation to clarify, question, or negotiate
the understanding that Otto and Beck were building (interview commentary opposed to key ability
#3, collaboration). In fact, Blaine expressed a trepidation to ask questions, out of fear that he would
“ask questions on every problem.” This is an excerpt where Blaine aligned his behavior with both
developing and high codes for collaboration.
193
In all, Blaine articulated a preference for working on his own and trying to replicate what he
saw from other students rather than working together to co-create a solution from which all parties
could benefit. Below, Blaine explained an inclination against asking the teacher for help:
Blaine.Interview.7:
Int Do you ever get Mr. Buford to help?
Blaine No. Nobody gets Mr. Buford to help...You basically have to teach yourself physics.
His immediate negative response (“No”) indicates that Blaine tended against inviting Mr. Buford’s
perspectives during class (interview commentary opposed to key inclination #2, collaboration).
He did not see asking questions to the teacher as an option. From the in-class data we saw
that Beck, Ed, and Otto often engaged with Mr. Buford and ask him for help, but yet Blaine still
said, “nobody gets Mr. Buford to help.” This could indicate that he did not view the interactions
as helpful, which would mean that Blaine often did not recognize the insights that emerge from
interactions like these (interview commentary opposed to key sensitivity #3, collaboration). The
last utterance he gave in his answer, “you basically have to teach yourself physics,” confirmed that
he tended to approach learning physics as a solo endeavor where he could not ask for help, similar
to his avoidance of question-asking in Blaine.Interview.6.
In Blaine’s in-class data, codes for developing collaboration were not as common. There were
only a few instances where we coded at all for collaboration (six times compared to 16 times
for Otto, with whom Blaine sat together at a table), perhaps explained by Blaine’s solo work, in
which he did not “do” much to demonstrate a key inclination, sensitivity, ability, or lack thereof.
We first examine an excerpt in which Blaine’s non-verbal actions aligned with his avoidance of
collaboration. We used it earlier in Section 7.6.1.2.2.
Blaine.In-class.3:
Blaine How do you make position a vector?
(3.5)
Blaine It says position °must be a vector°
194
(6.5)
Blaine ((stands up to gaze over Beck’s shoulder))
(26.0)
Blaine I cede
The half-minute when he gazed over Beck’s shoulder in search of an answer to his question
indicates that Blaine’s actions were oriented towards absorbing an answer. He displayed no interest
in negotiating a shared understanding (in-class behavior opposed to key ability #3, collaboration).
At other times, when he did say or do something that represented an effort to collaborate, other
students did not take Blaine up on his offer to work together. In particular, his tablemate Otto
sometimes ignored what Blaine had to say (which for Otto was opposed to key sensitivity #2,
collaboration). An example of Blaine trying to engage with Otto about the activity is below.
Blaine.In-class.4:
Blaine What are you trying to do? You trying to find the angle that you’re gonna need
to, refract it by?
Blaine’s question represented an effort to collaborate by clarifying and questioning Otto’s
course of action (in-class behavior aligned with key ability #3, collaboration), to which Otto gave
no response. Otto’s non-response was coded for developing collaboration, but some behavioral
patterns throughout class help explain Otto’s ignoring of Blaine in this moment. Before Blaine
asked the question above, Blaine spent 20 of the previous 25 minutes telling jokes to Otto. Three
times during this period, Otto told Blaine, “shut up.” The final time, Otto cut off a three-minute-long
joke about sombreros, saying “Shut up! I’m trying to think through this,” indicating that he viewed
Blaine’s jokes as distracting. Blaine’s effort to collaborate in the excerpt above stood in contrast to
his prior behavior, meaning Otto might not have viewed it as an opportunity to collaborate, even if
that is what Blaine was trying to do.
195
As shown above, Blaine’s reputation could sometimes come at a detriment to Blaine’s credibility
when he made reasonable attempts at collaboration. Below is an example where Blaine’s
recommendation for Otto’s code went completely unheard.
Blaine.In-class.5:
Otto $To be honest I don’t [ra-$
Blaine [No dude just ah- y- you got it.
Otto That’s a good step though.
Blaine Now just hit a negative sign.
(30.0)
Otto Oh, [because it needs to be. Negative. Ah: that’s why
Blaine Isn’t that literally what I said?
Otto [I don’t know.
Blaine [I told you just make it negative.
Otto No. No the y needs to be negative.
Blaine That’s what I said.
Otto ((coughs)) *I wasn’t listeni:ng*
This excerpt involved Otto stuck on a coding issue. He had just figured out how to get a
particle to move on screen, but it was going in the opposite direction than he wanted. Blaine made
a surface-level suggestion for a fix: “just hit a negative sign.” Later in the excerpt, we find out
that Otto “wasn’t listening.” Blaine did not specify where or how the negative sign should be
applied, nor did he provide any justification for the benefits of his suggestion (in-class behavior
opposed to key ability #3, collaboration). Otto, too, demonstrated an unresponsiveness to Blaine’s
suggestion (for Otto, opposed to key sensitivity #3, collaboration). The disagreement blew over,
but it highlights an instance where Otto was not listening to Blaine, possibly because Blaine was
not taken seriously as a potential collaborator given his previous behavior in class.
196
This points to the collaborative dilemma in which Blaine had ended up. Blaine’s tendency to not
take the computational activities seriously was potentially tied to the tendency for his peers to not
take Blaine seriously. Blaine demonstrated in his statement and actions several oppositions to some
key aspects of collaboration, namely a resistance to have his course changed by interactions with
others, neglecting to articulate or justify the benefits of a particular approach, and an avoidance of
negotiating a shared understanding. Blaine’s occasional efforts to collaborate were almost always
ignored by his peers, even though these efforts sometimes represented instances where Blaine
seemed aware of the unique insights that emerge from interactions, able to listen to others, and
able to negotiate a shared understanding. His data leaned more towards developing codes for
collaboration, but there were complexities embedded in Blaine’s collaborative moves as described
above, complexities that did not fit neatly into the dispositions framework.
Overall, Blaine exhibited behavior that aligned with developing tolerance for ambiguity, developing
persistence, and a considerable mix of high and developing codes for collaboration.
7.6.1.3 Ed (Dispositions)
In contrast to both Blaine and Otto, Ed’s data aligned with a mixture of high and developing codes
for each disposition, with considerably more high codes for persistence and collaboration. The
codes for her interview and in-class data are summarized in Table 7.7. Notably from this table, she
had far more codes for tolerance for ambiguity in her interview, and these codes were much more
often developing than they were for her in-class data. Also, we coded for collaboration much more
often in her in-class data. We address these features and more as we present the results from Ed’s
data below.
7.6.1.3.1 Ambiguity
Ed’s data was coded for both developing and high tolerance for ambiguity. In her interview, she
made many statements aligning with developing tolerance of ambiguity, but in class, her behavior
tended to align with high tolerance. For example, there was a moment in her interview when she
197
Tolerance for
Ambiguity
Persistence on
Difficult Problems
Willingness to
Collaborate with Others
Ed Interview 6 high
8 developing
12 high
4 developing
6 high
3 developing
Ed In-class 4 high
0 developing
8 high
2 developing
20 high
3 developing
Ed Total 10 high
8 developing
20 high
6 developing
26 high
6 developing
Table 7.7: Coded instances of CT dispositions in Ed’s data, separated by data source.
discussed a challenging example from the optics unit. In describing her difficulties, she framed the
ambiguity of optics as a source of confusion.
Ed.Interview.1:
Ed [Optics] was just incredibly confusing for me. Literally just the sign convention,
it was so- I don’t know why, there was just something weird to me about how
if you got closer or farther from a lens, the image could literally be flipped
upside down, depending on what kind of lens it was. And what you would
mark that as for the focal point, is it negative or positive, or where? And like
mirrors and lenses and how, I think, if an image is on the same side for a mirror,
it’s a positive image whereas if it’s on the same side for a lens it’s negative. It
was just too much.
Ed could not make sense of the new material. Specifically, the sign convention of the focal
length equation tripped her up. She went on to describe the different rules of the sign convention,
e.g., “flipped upside down, depending on the lens” and “if an image is on the same side for a mirror,
it’s a positive image, whereas...” The complicated rules that Ed seemed to have committed to
memory indicated that she was putting optical situations into rigid categories (interview commentary
opposed to key inclination #3, tolerance for ambiguity). She was overwhelmed by the task to
198
remember all this: “it was just too much.”
When it came to computation, Ed displayed a similar stance towards ambiguity. The excerpt
below was a reflection she made on her relationship with computation in Mr. Buford’s class.
Ed.Interview.2:
Ed GlowScript especially, I feel like it caters to a very specific kind of learner, a
very specific way of learning physics that’s like oh, if you- it just requires you to
take apart the numbers in a very strange way. Well, it’s not a strange way, it’s a
strange way for me.
In Ed’s view, “taking apart the numbers” during computational activities was not something she
was cut out for. She said that learning physics through GlowScript “caters to a very specific kind of
learner,” indicating a rigid, inflexible categorization of who benefits from the new computational
activities (interview commentary opposed to key inclination #3, tolerance for ambiguity). When
she acknowledged that computation might not actually be that strange, just “a strange way for me,”
she framed computation as something not for her. The reason, in Ed’s view, was computation’s
strangeness. This indicates a lack of interest in exploring and undertaking the strange parts of
computation (interview commentary opposed to key ability #2, tolerance for ambiguity).
On the other hand, Ed could also see the benefits from the ambiguous parts of computation. In
class, we tended to code her behavior for high tolerance for ambiguity, in contrast to many of the
statements in her interview. For instance, the excerpt below shows Ed making some considerations
about changing the variable that represented velocity in her code. She talked to herself and asked
questions to herself throughout the excerpt, only once directing a question to Beck, which he
answered promptly. At multiple times, she demonstrated an ability to navigate the uncertainty of
the situation.
Ed.In-class.1:
Ed =Should I just make the velocity a scalar? Possibly
(4.5)
199
Ed Should I make my velocity maybe a scalar and just do ah- °I can like°, I don’t
know how that would work though
(8.0)
Ed You have velocity dot x?=
Beck =Velocity’s a vector so this should (inaudible)
(8.0)
Ed Actually, maybe I might have something (11.0)
Ed ((hums and sings to self))
She began by wondering out loud whether velocity could be represented with a scalar quantity
in her program. She admitted, “I don’t know how that would work,” which indicates there was
significant uncertainty in this situation. Once Beck confirmed that “velocity is a vector,” Ed
appeared to be able to figure out a path forward (“maybe I might have something”) and seemed
committed to implementing the new idea, as indicated by the eleven seconds that passed and the
humming to herself, which was a marker throughout the class period that she is focused on the
code. Altogether, this demonstrated a navigation via an uncertain trajectory towards a solution
(in-class behavior aligned with key ability #3, tolerance for ambiguity).
Overall, the contrast between Ed’s interview and in-class data showed that the sides of a
disposition’s spectrum were not enough to fully characterize how Ed dealt with ambiguity in
Mr. Buford’s physics class. Ed had some moments in her interview that showed a resistance against
tolerance for ambiguity, such as her rigid categorization of optical situations and the way she was
put off by the strangeness of computational tasks. However, in class she showed that she was
capable of navigating incomplete data and uncertain trajectories toward a solution. This suggests
that Ed might have had a different understanding of her conduct than she displayed in class. Either
way, she was capable of approaching computational activities with a high tolerance for ambiguity
even though she at times articulated in her interview an intolerant view towards ambiguity.
200
7.6.1.3.2 Persistence
Ed embraced persistence on difficult problems. There were several moments during her interview
where she came off as persistent in nearly all the endeavors she took up. For example, she was “the
only remaining programmer on my robotics team,” and her mom encouraged to persevere through
tough initial experiences with physics (last year) and violin (eight years ago). When analyzing
for key aspects of persistence in her interview and in class below, the findings based on the codes
for persistence confirmed this interpretation. For example, Ed discussed the satisfaction that came
from getting a class project to work.
Ed.Interview.3:
Ed Sometimes when we’re doing projects and in just the rare moment that it just
goes okay, that feels good. And you feel it.
Int Nice. Can you describe a moment in a project like that?
Ed Yesterday. We recently got projects like to make a telescope, basically, which
is two converging lenses, and it just worked. We just made it, and it just felt
good. It’s like we found the- It’s just two paper towel tubes, basically, and we just
put them on either end and just focus it, and it just worked. And that was a good
moment. Even though- Oh, gosh. But even though we had worked for a whole
two days on it, because the first day we were trying to put an actual model of
something between, like a phone, and it was awful. That was a bad day.
She first nodded at the good feeling before describing an example: “the rare moment that it just
goes okay, that feels good.” She said this again in reference to completing the telescope project: “We
just made it, and it just felt good.” These utterances demonstrated Ed’s awareness of the satisfaction
derived from success after significant effort (interview commentary aligned with key sensitivity
#2, persistence). We know she put in significant effort because she described “working for a whole
two days on it,” which pointed to her ability to stick with a task for an extended period of time
(interview commentary aligned with key ability #1, persistence). This excerpt also demonstrates
201
Ed’s ability to try a new approach after considerable effort (interview commentary aligned with
key ability #2, persistence)—she described working for the first day “putting something between
like a phone,” which was an “awful” approach that failed, and then she shifted to the paper towel
tubes method.
That said, there were a couple points where Ed seemed to distance herself from persistence,
including when she described giving up in the midst of confusing aspects of computation.
Ed.Interview.4:
Int Do you think you’re good at the coding activities?
Ed Not really, actually, which is kind of sad for me to be honest, because you have
this interest in something, but it’s back to why physics is so frustrating, because
It’s something that’s like ‘Oh, this is familiar, I know this,’ but then you just-
It’s just slightly slanted a little and just becomes, because you expect it to be this
way so much, when it’s this way, it’s just- you can’t handle it.
The confusion she described in this excerpt was unexpected confusion: “you expect it to be
this way so much, when it’s this way, it’s just- you can’t handle it.” Ed perceived a “familiarity”
with the computational tasks, and then that familiarity was betrayed, leading her to give up in the
moment: “when it’s this way, it’s just- you can’t handle it.” Instead of taking the opportunity to
shift tactics or reframe the problem, Ed was unable to change her approach during the confusion
(interview commentary opposed to key sensitivity #3, persistence).
The occurrence of the above confusion and disengagement was rare given Ed’s statements
coded for high persistence in her interview. In the in-class data, Ed was consistently considering
new ideas to implement in the code or new ways to deal with a roadblock. The first instance of this
is when she was not sure how to model a light particle with a visual object in GlowScript, so she
entertained an idea to model it as a sphere:
Ed.In-class.2:
Ed What the $fu:ck$
202
(10.0)
Ed °I’ll just do::° Okay I’ll just do a sphere, trail, okay I got this, it’s fine, it’s cool
(41.0)
Ed °Make the trailer true:° ((clears throat))
(7.0)
Ed °So we’ll do lines°
She began with a statement of confusion (“what the fuck”) and then moved on to consider an
option for modeling the particles with a sphere. Over the next 41 seconds, she worked, and her next
utterance indicated she was still on the same task, as “trailer” referred to the sphere’s trail that she
mentioned earlier. The initial consideration to implement a “sphere” after being stuck indicates that
she was attentive to an opportunity to try a new tactic (in-class behavior aligned with key sensitivity
#3, persistence).
At a later time about halfway through the class period, she arrived at the need to implement a
while-loop in her code. She proceeded to engage Beck in one turn of conversation and then worked
on the loop herself, once looking at Beck’s code for additional help.
Ed.In-class.3:
Ed How do I do a LOOP. ((laughs one exhale)) °time to just (inaudible) Beck’s
(inaudible) things°. So velocity::
Beck Huh. Why don’t you put a [focal point (inaudible)?]
Ed [(inaudible) like]. V:ector
(4.5)
Ed In the: x direction and not in any other direction cause we not about that bullshit
(9.0)
Ed ((glances at Beck’s laptop)) I always forget d t too, alright. And then, °let’s
move°
203
In the first line, her intentions were not clear due to the inaudible speech, but she announced
her need to create a loop and then acknowledged that it was time to do something involving Beck.
We infer that Beck’s code or expertise was desired, because Beck answered her comment in the
next line, and later on Ed glanced at his code while trying to implement the while-loop. This
represented a pursuit of resources (Beck’s insight and Beck’s code) that increased the effectiveness
of her efforts (in-class behavior aligned with key ability #3, persistence). This glancing at code was
different from Blaine.In-class.3 (when Blaine gazed over Beck’s shoulder) because Ed’s glancing
was an enhancement (reminding her of “d t”) of the effort and interaction that she was already
engaged in. In contrast, Blaine’s gaze was the only activity he was engaged in, and he was looking
for answers as a substitute for engaging in the activity in other ways.
Despite her behavior aligned with high persistence throughout most of the class period, when
Mr. Buford asked Ed a question about her progress near the end of class, her response aligned with
developing persistence.
Ed.In-class.4:
Mr. B Did you get something going?
Ed Not really, to be honest. I was just, staring at it in the hopes that it would
make sense
Her summary of what she did during class was about trying to absorb information and indicated
no overall change in strategy or attentiveness to shift tactics when needed (in-class behavior
opposed to key sensitivity #3, persistence). “Staring at” the problem also indicated no desire to
apply extended effort to the computational activity (in-class behavior opposed to key inclination
#1, persistence). This account did not match with her conduct throughout the class period, which
meant she was not accurately representing her work with this statement, though this statement could
be how she was interpreting events.
Overall, Ed’s words and behavior aligned with high persistence, as shown by her awareness of
the satisfaction of effort paying off, her attentiveness to opportunities to shift tactics when needed,
204
her ability to stick with a task for an extended period, her ability to try a new approach after
considerable effort, and her pursuit of resources that increased the effectiveness of her effort. At
times she either did not recognize her own persistence, or she wished to represent her workflow
more modestly or more in line with how she was feeling in the moment. This was the case in
Ed.Interview.4 and Ed.In-class.4, where she demonstrated an inclination against extended effort
and did not seem attentive to opportunities to shift tactics. These examples aligning with developing
persistence contrasted with and were overshadowed by a deeper and more sustained pattern of high
persistence in Ed’s data.
7.6.1.3.3 Collaboration
Ed also embraced a high willingness to collaborate with others, though infrequently she also
aligned her behavior with a developing willingness. Below, she described her tendencies to work
with others but also to trade answers transactionally.
Ed.Interview.5:
Int So in general in class, do you tend to work by yourself, or like in a group of
students?
Ed I don’t think I’ve ever once worked by myself, to be honest.
Int Okay. Maybe a test, yes?
Ed Yeah, a test.
Int When you work with your classmates, what role do you play in doing group work?
Ed I feel like I kind of just leech off people, to be honest. Like if I just don’t know
an answer, I’m just going to ask around until someone gives me the answer.
Just being completely honest.
Int Okay. Are there times you do know the answer, or you know part of what you
need to do?
Ed Yes. Beautiful, happy times.
205
Int Okay. And is your role different when that’s the case?
Ed Yes. Then I get to tell people the answer.
She started with a strong statement: “I don’t think I’ve ever once worked by myself, to be
honest.” This indicated that Ed often invited the perspectives of others into what she was doing
(interview commentary aligned with key inclination #1, collaboration), and she was willing to
let these interactions change the course of her work (interview commentary aligned with key
inclination #1, collaboration). She went on to describe how she felt that her role was to “leech”
answers from her peers. The use of the word “leech” indicated that Ed viewed this practice
negatively. Taking answers without contributing to them indicated simultaneously a willingness to
let others shape her actions (interview commentary aligned with key inclination #1, collaboration)
and an inability and/or unwillingness to negotiate the understanding that the group was building
(interview commentary opposed to key ability #3, collaboration). Even when she could contribute,
it was still just answer-giving: “I get to tell people the answer.” The implication of “giving” the
answer was that she did not articulate the explanation behind the solution or justify its benefits
(interview commentary opposed to key ability #2, collaboration). However, when we asked her
to clarify this peer group dynamic, it turned out that the answer-transfer practice was a result of
attempted-but-failed explanation of answers:
Ed.Interview.6:
Int Is it ever like, explaining how the answer is, or explaining how it works, or is it
just like ‘this is the answer’?
Ed I feel like we all attempt to explain. I’ve noticed some people try to explain to
me and like ‘I didn’t get that, but I believe you,’ and it just works. And I’ll try to
explain it to them and they’ll be like ‘I don’t understand what you’re saying.’
So we try, it doesn’t really work, though.
The “attempt to explain” indicated that Ed and her peers actually did try to articulate and justify
the approach behind the answer they were trying to share with the group (interview commentary
206
aligned with key ability #2, collaboration). However, these attempts fell flat—responses included
“I didn’t get that” and “I don’t understand what you’re saying.” This failure to communicate
understanding indicated that Ed (and her peers, according to her) did not have an alertness to
the interpersonal dynamics she might have been able to leverage to make these interactions more
effective (interview commentary opposed with key sensitivity #1, collaboration), even though she
did show awareness that such collaboration would be worthwhile.
When we looked at the in-class data to further understand Ed’s group work, it seemed that she
was much more collaborative than she gave herself credit for, and the “leeching” relationship did
not play out so transactionally. Ed collaborated openly and often. For example, shortly after the
beginning of work time, she checked in with her tablemates (“So how’s everyone doing?”). Later
in class, she engaged in a conversation with Beck about visualizing rays of light (“lines”, below).
Ed.In-class.5:
Ed You know what? It’s a dot, we’re getting there, we’re doing okay
Beck That’s good
Ed Thank you
Beck If you don’t want your lines to be so dark, you can give em a thickness. (inaudible)
((points at Ed’s screen))
Ed Oh yeah
Beck If you just go to size and make that one, the: zero
Ed Make the zero one=
Beck =Yeah=
Ed =Oh: true you’re right!=
Beck =For, for the lens [and the optical axis]
Ed [So it’s not transparent]
Beck And then, same here with the lens- no not there, not there, [°not there°
207
Ed [Right
The conversation revolved around Beck’s suggestion to thicken some lines in her visual so
they would be easier to see. Ed accepted Beck’s suggestion with excitement (“Oh: true you’re
right!”), which indicated a responsiveness to the contributions of peers (in-class behavior aligned
with key sensitivity #2, collaboration). She also made comments as Beck explained (“make the
zero one”, “so it’s not transparent”) to follow along, and this demonstrated tendencies to both clarify
the understanding he was providing (in-class behavior aligned with key ability #3, collaboration)
and to value his perspective on this matter (in-class behavior aligned with key inclination #2,
collaboration).
At another time, Ed checked in with a struggling group member, Brian.
Ed.In-class.6:
Ed How are you doing Brian?
Brian Huh, what?=
Ed =I said how you doing?
Brian Terrible
Ed Amazing
(2.0)
Ed Is that just your response to anything that- I thought this was gonna be a &deep
conversation about our shared struggle with, writing this glow script&
Her question (“how are you doing Brian?”) demonstrated an interest in the well-being of her
peers. When Brian responded negatively and with only one word (“terrible”), Ed explained that
she was open to sharing in the struggle of the computational activity. This indicated an alertness
on Ed’s part to the interpersonal dynamics that could make interactions more worthwhile (in-class
behavior aligned with key sensitivity #1, collaboration). In this case, the implication was that
checking in with others could benefit interactions within the group. Her offer to share in Brian’s
208
struggles also showed that Ed was not always just giving or taking answers like she indicated in
her interview. Ed had a connection with her classmates, and her collaboration in class went beyond
just the moments when Beck helped her figure something out.
When reviewing the key inclination, sensitivities, and abilities of collaboration, Ed displayed
several in the excerpts above: a willingness to have her course changed by interactions with
others, a tendency to invite and value perspectives different from her own, an alertness to effective
interpersonal dynamics (in the case of checking in with Brian), a responsiveness to the contributions
of peers, and an ability to clarify and negotiate a shared understanding and course of action. She
also demonstrated, in describing how she shared answers with peers in Ed.Interview.6, an earnest
attempt to articulate and justify the benefits of a particular approach. This answer-sharing that
Ed engaged in also related to some developing aspects of collaboration: an unawareness of how
to enhance the effectiveness of interactions, an inability (even through her earnest attempts) to
articulate and justifying the benefits of a particular approach, and an inability to negotiate a shared
understanding. These developing codes did not appear much in the in-class data, where the vast
majority of Ed’s coded data aligned with high collaboration.
Overall, Ed embraced high dispositions, with several notable exceptions where she described
her behavior in alignment with developing dispositions, especially when describing how she treated
ambiguity in her interview. We interpreted these momentary discrepancies as evidence that she
was slightly modest or not fully aware of how much her behavior aligned with high dispositions.
7.6.1.4 Summary of Dispositions Results
To summarize the results above, we coded examples for both high and developing dispositions,
representing a variety of views and situations in Mr. Buford’s class. Otto’s interview comments
and in-class behavior aligned overall with high dispositions, with some exceptions. For example,
he sometimes did not recognize or experience a sense of satisfaction after persisting through a
difficult computational task. He also indicated that he was less willing to collaborate in a setting
where he held more expertise: calculus. In contrast, Blaine’s interview comments and in-class
209
behavior aligned with developing tolerance for ambiguity, developing persistence, and a mix of
high and developing codes for collaboration. His in-class interactions seemed overshadowed by
his reputation for not taking the computational activities seriously, which made it difficult for him
to collaborate with others even on occasions when he did have questions and suggestions. Ed
sometimes talked about her behavior in a way that indicated more of a developing disposition for
her tolerance toward ambiguity, whereas her behavior in class was more closely aligned with a
high disposition. Overall, Ed’s data contained a mix of high and developing codes for tolerance
for ambiguity, and she more clearly embraced high persistence and high collaboration. Similar to
Otto, Beck embraced high dispositions with few exceptions (discussed in detail in Appendix B).
One feature of Beck’s tolerance of ambiguity was that he mostly embraced a high tolerance but
also articulated that he preferred when activities were more clear-cut and oriented towards a single
solution.
We also use this summary section to provide Table 7.8, which shows how the disposition codes
were split between data sources and between key inclinations, sensitivities, and abilities. In short,
tolerance for ambiguity was more likely to show up in the interview setting, key inclinations for
persistence were rarely observed at all, and key abilities for collaboration were overwhelmingly
coded in the in-class setting. When we examine Table 7.8 in more detail, we notice that we were
more likely to code for tolerance for ambiguity in interview comments, indicating that coding for
this disposition should involve data in the formof some sort of reflective activity (like the interview)
for students rather than trying to observe their behavior directly. We also noticed that instances
of inclinations for persistence were comparatively lower than other key aspects, which could mean
there is another data source we did not consult that could have provided better insight into how
inclinations for persistence relate to how students interact with computation. Lastly, the in-class
data (instead of the interview setting) was where students were far more likely to exhibit key abilities
for collaboration, indicating that this aspect of collaboration can be related to direct observations,
as long as students have opportunities in class to collaborate with one another.
210
High Disposition Developing
Disposition
Total (High and
Developing)
Interview In-class Interview In-class Interview In-class
Ambiguity Inclinations 11 5 16 5 27 10
Ambiguity Sensitivities 21 8 2 4 23 12
Ambiguity Abilities 16 11 9 3 25 14
Ambiguity Overall 48 24 27 12 75 36
Persistence Inclinations 2 1 3 4 5 5
Persistence Sensitivities 17 15 4 5 21 20
Persistence Abilities 18 16 4 3 22 19
Persistence Overall 37 32 11 12 48 44
Collaboration Inclinations 11 11 4 0 15 11
Collaboration Sensitivities 4 11 4 7 8 18
Collaboration Abilities 10 32 4 4 14 36
Collaboration Overall 25 54 12 11 37 65
Table 7.8: Codes for inclinations, sensitivities, and abilities, separated out among the dispositions
for comparison. The codes in this table are binned by data source. The left four columns
are separated between high and developing dispositions, and the right two columns simplify the
information by providing counts that combine high and developing codes for each disposition.
7.6.2 Mindset Results
In this section,we explore howthe mindset coding scheme related to theCTDispositions Framework
in our data for each student. We also show how mindset was present in the data independent of the
dispositions, which highlights how mindset can help characterize what students say and do even
when dispositions cannot. This combination allows us to discuss for each student’s data set how
mindset connected to dispositions.
211
Otto Blaine Ed Beck
Interview 20 growth
3 fixed
4 growth
26 fixed
10 growth
11 fixed
14 growth
1 fixed
In-class 10 growth
1 fixed
0 growth
19 fixed
7 growth
8 fixed
6 growth
0 fixed
Total 30 growth
4 fixed
4 growth
45 fixed
17 growth
19 fixed
20 growth
1 fixed
Table 7.9: Coded instances of mindset in each student’s data, separated by data source. The bottom
row provides a total count of fixed and growth mindset codes for each student’s data.
7.6.2.1 Otto (Mindset)
In Section 7.6.1.1, we showed how Otto generally embraced high dispositions, though there was an
exception with respect to persistence because he indicated little satisfaction at completing especially
difficult computational activities at times. Building on this analysis, we now show how mindset
could describe aspects of his data (as shown in Table 7.9).
The first excerpt we return to is Otto.In-class.3, when Otto recruited Beck’s help to interpret
and deal with an error message. In the dispositions analysis, we used this to show that Otto had
displayed a sensitivity and ability for shifting tactics as well as an endurance for sticking with the
task at hand. These codes signaled high persistence. In the same quote, we also coded for growth
mindset. Otto had an immediate reaction to the error message: “Why is that wrong? Beck. Why
is it undefined?” The choice to draw attention to the mistake and bring Beck into the fold was both
an opportunistic tactic shift and an indication of a desire to learn. Otto’s treatment of the setback
as an opportunity to learn and overcome aligned with part of the growth mindset coding scheme.
In Otto.Interview.7, Otto explained a view that the “smartest” students in class were also the best
explainers of physics concepts. In the dispositions analysis, we coded this for Otto’s sensitivity to
how explaining presented an interpersonal dynamic that helped students work together, which also
signaled high willingness to collaborate with others. In the quote, Otto talked about two students,
212
Beck and Joyce: “If you could say one person was an explainer type guy, it’s [Beck]... [Beck
and Joyce] tend to be the ones who are able to express [problems and concepts] to other people.”
This showed that Otto valued understanding over answers because of how he attributed Joyce’s and
Beck’s competence to their ability to explain, not to their high grades. Otto’s high valuation of
understanding aligned closely with an aspect of growth mindset, “learning is important.”
Even outside of the excerpts that had dispositions codes, there were instances of Otto embracing
growth mindset. For example, the excerpt below was from Otto’s interview, and precedes
Otto.Interview.3. The interviewer asked whether Otto felt he was good at the computational
activities.
Otto.Interview.Mindset:
Int Do you think you’re good at the coding activities?
Otto I’ll get better. I’m not very good at it right now.
Int Have you noticed yourself getting better even in the last couple of months?
Otto Yeah, I’d say so. I’m starting to understand like the- well, Python syntax for
one is weird. I like Java more.
Otto did not express a high self-evaluation, but what matters here is that he expressed a belief
that he would “get better.” This encapsulates a belief that his skill at computation was temporary
(“not very good at it right now”) and could be grown, a core aspect of growth mindset. Even though
he admitted that what he was learning was “weird,” he also could see that he was “starting to
understand” it. Otto’s comments here showed that he could sometimes make a comment explicitly
in line with growth mindset without displaying dispositions for tolerance for ambiguity, persistence,
or collaboration.
Overall, Otto’s data was coded for dispositions, mindset, and sometimes both at the same time.
His statements and words more often aligned with aspects of growth mindset. The times when
both constructs described Otto’s words or actions serve to show how mindset and dispositions can
213
overlap. The times when mindset was coded without dispositions serve to show that mindset can
function on its own in this data set and is separate from dispositions in some instances.
7.6.2.2 Blaine (Mindset)
Different from Otto, Blaine’s behavior was aligned with developing tolerance for ambiguity, developing
persistence, and a mix of high and developing codes for collaboration. When we looked at
how mindset presented in his data, we found that Blaine’s behavior often aligned with aspects of
fixed mindset (see Table 7.9).
For example, in the Blaine.In-class.2 excerpt, Blaine ran some code, received an error, exclaimed
that he “[didn’t] knowhowto code,” and then searched online for answers to copy. In the dispositions
analysis, we showed that this was an example of Blaine’s resisting an opportunity to grow by
engaging with an uncertain situation and an indication of his disinterest in reframing ambiguous
stimuli, both aligned with developing tolerance for ambiguity. In this same excerpt, there were also
some aspects of fixed mindset. When Blaine evaluated himself as “not knowing how to code” after
receiving an error message (something that happens to everyone who codes), he was interpreting
this small mistake as a message that he was bad at computation. This type of interpretation was a
characteristic of fixed mindset from our coding scheme from Table 7.2: “failure can mean you are
bad at computation.”
Similarly, in the Blaine.Interview.5 quote, Blaine described howhe tried to code with GlowScript
correctly countless times in the past without any success, and he did not try anymore because of
his continued failure. He always seemed to get “a blank screen or...some error” whenever he did
computation. In the dispositions analysis, we interpreted this comment to mean that he was unable
to shift his approach in the face of constant failure, signalling an aspect of developing persistence.
There was also alignment with aspects of fixed mindset from Table 7.2; his choice to not try
anymore showed that his failure had made him less interested in computation, and his inability to
proceed after receiving the error showed that he was paralyzed by setbacks.
There were also comments aligned with fixed mindset independent of the dispositions analysis.
214
For instance, when we prompted Blaine to reflect on a statement he made near the end of class, he
responded:
Blaine.Interview.Mindset:
Int I’m going to read you a quote that you said on Monday. I want you to unpack it
for me a little bit. The quote was, ‘What’s the point of learning code? I can
draw this on a piece of paper in fifteen seconds.’
Blaine I did say that...I could draw it on a piece of paper in fifteen seconds. Okay? All
right? That’s how I was feeling at the time. Just have a line, make it curve.
I can do that, real quick. I didn’t mean learning code in general, but doing this
code, this code’s wackeroonie. Whatever.
Int Just this specific project that you were working on at the time?
Blaine Yeah, I’m sure code could be applicable in a lot of places but I don’t need to
plug it into a computer to draw some straight lines.
In the excerpt, Blaine defended his judgment of the computational activity. He emphasized that
the point of the activity was, “just have a line, make it curve,” which was something he could easily
do on a piece of paper. He went further into his evaluation of the activity, saying, “this code’s
wackeroonie,” which we took to mean that he saw the activity as convoluted and/or pointless, an
evaluation of the assessment as unfair (and an indicator fixed mindset, as shown in Table 7.2).
Blaine’s insight into his comment showed that he had no desire to engage with the challenge of
computation because the same visual could be achieved by drawing it: “Just have a line, make it
curve. I can do that, real quick...I don’t need to plug it into a computer to draw some straight lines.”
This avoidance of taking up the opportunity to try plugging what he knew into the computer signals
an avoidance of the challenge, which is another indicator of fixed mindset (see Table 7.2).
In class, Blaine’s behavior also aligned with aspects of fixed mindset. The excerpt below happened
directly after Blaine.In-class.1. In it, Blaine was commenting on how he thought GlowScript
should work and highlighted his inexperience with computation compared to his tablemate Otto.
215
Blaine.In-class.Mindset:
Blaine That’s how it should be. If I put in line, a line should appear. I don’t understand
why it doesn’t, you know?
Blaine I personally think whoever made this &glow script& didn’t know what they
were doing
(11.5)
Blaine &Just a note for the record, Otto has taken a, coding class, at this school. It’s
AP computer science&
Otto I was bad at it
Blaine &He might’ve learned a few things, he got an A in the class&
Otto I was bad at it
Blaine &Most kids failed it. But um, that’s why he has a little bit of an advantage on me.
You know I’ve never even seen a code before, what is a code?
The first line in this excerpt was the last line in Blaine.In-class.1, and it represented an instance
of Blaine expressing his disapproval of how GlowScript worked. He thought it should have drawn
a line when he typed “line.” His next statement was what we coded for mindset: “whoever made
this &glow script& didn’t know what they were doing.” His emphasis on “they” implied a switch:
earlier, Blaine admitted, “I don’t understand,” and then later he emphasized “they” to indicate that
the designers of GlowScript were the ones who actually messed up. This indicated that Blaine was
avoiding responsibility for his recent failure, which is an aspect of fixed mindset from Table 7.2. It
is possible that Blaine was being facetious here, but fixed mindset still sat at the center of the joke
that he was making. He went on to contrast his preparation against Otto’s. Blaine had no prior
academic experience with computation, and he leaned into this narrative: “I’ve never even seen a
code before, what is a code?” He framed this disclosure as “a note for the record,” which indicated
that he did not want to come across as stupider than Otto, it was just that he had less experience.
216
Blaine’s choice to highlight his lack of experience was different from when Otto did so in his
interview (“I’ll get better. I’m not very good at it right now”) because Otto framed his situation as
something to improve on, whereas Blaine framed it as a lack of “advantage” that he could explain
their difference in performance. This foregrounding of experiential disadvantage on Blaine’s part
aligns with an aspect of fixed mindset: “success/failure is not one’s own responsibility.”
Overall, Blaine’s data was coded for dispositions, mindset, and sometimes both at the same
time. In the context of Mr. Buford’s computation-integrated physics class, Blaine more often
behaved in line with aspects of fixed mindset than growth mindset, based on our coding scheme.
Like Otto, there was evidence of mindset and dispositions overlapping and instances of mindset
functioning separately from dispositions in Blaine’s data set.
7.6.2.3 Ed (Mindset)
Ed’s data generally contained a mix of high and developing codes for tolerance for ambiguity,
and more clear alignment with high persistence and high collaboration. Despite many instances
coded for high tolerance for ambiguity (especially in class), she made comments in her interview
aligned with developing tolerance for ambiguity, and during class she summarized her work to
Mr. Buford in a way that indicated a misunderstanding and/or modest interpretation of the way
her behavior represented high persistence. Ultimately, there was some misalignment in how her
descriptions of her own behavior were coded for dispositions and how her behavior itself was coded
for dispositions, with the codes for high dispositions more common in her behavior. When coding
Ed’s data for mindset, we saw a similar story. Unlike Otto (whose data mostly aligned with growth
mindset codes) and Blaine (whose data mostly aligned with fixed mindset codes), Ed’s data was
coded to be fairly evenly split between growth and fixed mindset (see Table 7.9).
For example, in the Ed.Interview.4 quote, Ed reflected about how computation felt familiar at
first, but then it defied expectations and caught her off guard, leading her to feel like she “can’t
handle it.” In the dispositions analysis, we coded this for an hesitation to change approaches
when confronted with a confusing situation, signaling developing persistence. There was also
217
evidence of setbacks could be paralyzing for Ed, because she made the direct connection between
the unexpected and losing control: “when it’s this [unexpected] way, it’s just- you can’t handle it.”
This response to setbacks aligns with a code for fixed mindset: “setbacks are paralyzing.”
At other times, Ed embraced growth mindset in how she responded to computation. For
example, in Ed.Interview.3, she reflected on the satisfaction she felt from a difficult class project
when she had to build a telescope. We coded this excerpt for Ed’s ability to apply effort in the face
of setbacks and for her awareness of the satisfaction derived from success after significant effort.
These attributes of the excerpt aligned with high persistence, but they also conveyed a parallel to
growth mindset. In particular, her sustained effort through a day of no progress (“we had worked
for a whole two days on it, because the first day...was awful”) indicated that the setbacks of the first
day of the project were opportunities to overcome and succeed (an aspect of growth mindset from
Table 7.2), which is what Ed did.
There was more evidence aligned with aspects of mindset, both growth and fixed, in other parts
of Ed’s data. In her interview, we discussed one of her favorite subjects, music, and compared it to
physics.
Ed.Interview.Mindset:
Int What’s a subject that you really don’t like? If there is one.
Ed I don’t think there is one. They all have their ups and downs.
Int Okay. Is there one that you like significantly less or more than physics?
Ed I can’t really- Significantly less than physics, maybe- No, that’s not true. Okay,
nothing significantly less, but significantly more than physics, probably music,
orchestra.
Int Music? Okay. How does your experience in music compare to your experience
in physics?
Ed In physics, you kind of feel like- A lot of what I feel in physics class is being
almost kind of powerless as information is being fed to me, I’m kind of not
218
understanding it all the way, and in music you’re like completely controlling the
situation.
Int Okay. Yeah, that’s a big difference. Are there things that you can do in physics
that make you feel as if you have more control over the situation?
Ed No. Just learn to deal with having no control.
In this excerpt, a few features of physics class stood out compared to music. What Ed said
almost spoke for itself: “A lot of what I feel in physics class is being almost kind of powerless as
information is being fed to me.” The feeling of powerlessness indicated that Ed felt no responsibility
for her learning in physics class—an aspect of fixed mindset from Table 7.2. When asked if she
could do anything to change the power dynamic, she responded, “No. Just learn to deal with having
no control.” This underlines our interpretation that she did not see any control over her learning in
this moment and did not see physics as something she could learn to do.
However, when approached with difficulties during class, Ed oscillated between statements
aligned with fixed and growth mindset. She would often voice out loud that she was stuck or doing
a bad job at the task at hand (aligned more with fixed mindset), yet still encouraged herself to keep
going (aligned more with growth mindset). Below are three separate instances to demonstrate how
she talked to herself in these moments.
Ed.In-class.Mindset:
Ed I’m do it- this is so bad
(4.0)
Ed Girlie. Get- get a grip on yourself
...
Ed I can’t do this anymor:::e $heheher:$
(2.5)
Ed Yes you can, you’re doing fine, yes you’re /fine/
219
...
Ed °I can’t /do:/ this. *I can’t do this.* /Yes/ I can°
Joyce You can do this $Ed$
Ed I $can’t though:$. *Today’s not the day:* ((laughs one exhale))
Each time, Ed expressed a negative evaluation of her performance (“this is so bad”, “I can’t
do this”), and then encouraged herself to push past it (“get a grip”, “you’re doing fine”). This
presented an interesting pattern of responses. When she first evaluated herself, it seemed as if she
was interpreting the situation to mean that she was stupid or that the setback was going to stop her
from advancing (aligned with, “failure can mean you are stupid” and/or “setbacks are paralyzing”
from fixed mindset, Table 7.2). However, each time, she encouraged herself back into the task,
indicating that she believed that she just needed to keep working or that the setback was something
to be overcome (aligned with, “failure can mean you need to study harder” and/or “setbacks are
opportunities to overcome a challenge” from growth mindset, Table 7.2). This flip-flop between
codes for fixed and growth mindset represented how Ed often displayed features of both sides of the
mindset spectrum, and in many situations neither side on its own could relate to her behavior. The
flip-flop also reflected how she sometimes indicated a code for high dispositions in her behavior
and developing dispositions in a reflection on that behavior.
Like the other students, Ed’s data was coded for dispositions, mindset, and both at once. She
exhibited aspects of both fixed and growth mindset, sometimes rapidly switching between them.
7.6.2.4 Summary of Mindset Results
One focus of the analysis above was on the overlap between mindset and dispositions in the data.
In Figure 7.2, we show this overlap for each disposition. Notably, there was a significant amount
of non-overlap, suggesting the constructs showed up differently in the data. Also, the overlap for
collaboration was about half the size of the overlaps in the other two dispositions. We discuss the
implications of this difference in the discussion section.
220
Figure 7.2: Partially filled-in bars representing the overlap of mindset codes into how each disposition
was coded through all the data. Each bar represents the number of excerpts coded for a
particular disposition, and the diagonal shading represents the portion of the excerpts also coded
for mindset. The percentage of overlap, rounded to the nearest percent, was 27% for tolerance for
ambiguity, 30% for persistence, and 14% for collaboration.
Figure 7.3: Waffle diagram showing the number of mindset-coded excerpts in total, the portion of
those excerpts that were also coded for dispositions, and the number of those overlap-coded excerpts
that showed alignment (either between growth mindset and high dispositions or fixed mindset and
developing dispositions).
221
Growth
Mindset
Fixed
Mindset
Both Growth
and Fixed
High Dispositions 26 1 0
Developing
Dispositions
1 20 0
Both High and
Developing
0 3 0
Table 7.10: The way mindset overlapped with dispositions for excerpts in which we coded for both
constructs.
In Figure 7.3, we provide a waffle diagram for the entire collection of data (including Beck,
whose mindset analysis is shown in Appendix B) that shows how codes for dispositions and mindset
overlapped and aligned in the mindset-coded excerpts. Most excerpts coded for mindset were
coded only for mindset (74 out of 124). This substantial non-overlap with dispositions indicates the
constructs have significant differences in how they were embodied in students’ words and actions.
A notable feature of the waffle diagram is that of the 50 excerpts that were coded for both mindset
and dispositions, 45 were coded with aligning codes, meaning only high dispositions and growth
mindset, or only developing dispositions and fixed mindset. This breakdown is shown in more detail
in Table 7.10. This alignment between dispositions and mindset in the data shows that language
representing high dispositions seems to be tied to growth mindset language (and developing tied
to fixed), even though the two constructs can still drawfocus to different aspects of behavior and talk.
7.7 Discussion and Conclusion
In this study, we asked two research questions: (1) How do CT dispositions apply to the context
of a computation-integrated high school physics class? (2) How are CT dispositions connected to
mindset in the context of a computation-integrated high school physics class?
222
7.7.1 Application of Dispositions Framework
To answer the primary research question, we coded the in-class and interview data for Otto, Blaine,
and Ed using Pérez’s framework (summarized in Table 7.8), coding for tolerance for ambiguity,
persistence, and collaboration. Pérez postulated that these three dispositions could promote CT
practices in the classroom [1]. While originally developed in a math context with a cohort of
teachers, we were able to apply the CT Dispositions Framework to students’ data in a computationintegrated
physics classroom. The framework allowed us to code each student’s data set for
dispositions in detail, which showed that the framework could be extended to a context where
students are learning physics through computation. Additionally, the framework allowed us to
identify nuances in how we coded for dispositions, but it also raised further questions about the
framework and about computational pedagogy.
Some questions came about from differences in context between our study and Pérez’s original
theorization. For instance, when using the dispositions framework, can teachers be part of a
collaboration when the focus of the study is on students? This is a new question since the
framework itself was developed in the context of a workshop series for teachers with no students
involved. For example, in our data, Otto often received help from Mr. Buford in class and described
these interactions in his interview. In both data sources, we coded for collaboration whenever Otto’s
statements and actions aligned with some of inclinations, sensitivities, and abilities of collaboration,
even though he was just talking with the teacher. This is a small but possibly important point
that calls for clarification in the dispositions framework when applied to a classroom. In the
dispositions framework, the definition for collaboration is, “a tendency to coordinate effort and
negotiate meaning with peers to accomplish a shared goal” (page 449) [1], with “peers” identified
as potential collaborators. This would seem to exclude teachers from being potential collaborators.
However, wewould argue that this might be dependent on the individual’s perspective on the teacher
and the role that the teacher plays in the classroom, rather than a blanket rule that teachers cannot
be collaborators. For example, if an instructor asks a student guiding questions, does the student
not participate in the creation of the solution/answer as much as the teacher? The perspective
223
around this point becomes more complicated in a classroom with an array of power dynamics (e.g.,
undergraduate learning assistants, teaching assistants, and faculty) all combined together.
From our dispositions analysis, we also saw alignment of dispositions across data sources
with a couple exceptions, which has implications for computational pedagogy. When looking at
the data, students’ statements and behavior frequently aligned with codes for each disposition in
similar ways across their interview and in-class data, which suggests that the same dispositions
codes could be observed in class or ascertained by talking with a student about their perspective
on the computational activities. From a practice perspective, this might highlight multiple paths
forward for operationalizing the CT Dispositions Framework as a curriculum design tool. If the
aspects of CT dispositions are determined to be part of desirable learning goals, then teachers
might be able to use a version of the framework to identify behavior related to dispositions and
promote development of dispositions. Alternatively, the aspects of dispositions that we coded more
readily in the way students answered interview questions could be prompted and developed through
pre-course surveys or reflective assessments that allow students to articulate their views like in
the interview setting. The choice to pursue these possibilities depends on whether promoting the
development of dispositions is desirable in a given context. To this point, we provide a critique of
the framework later in the discussion.
In exception to the previous paragraph, we found that there were a couple of instances of
misalignment between data sets (interview and in-class) and that there can be nuance within the
coding of a single disposition that might not be apparent from one data source alone. For example,
most of Otto’s statements related to ambiguity came from the interview rather than the in-class data,
as seen in Table 7.5. A possible explanation is that in the interview, Otto had many opportunities
to explain how he viewed problems and how he liked to explore different features of them. On
the other hand, in the in-class data, we were only able to say that he was displaying an aspect of
tolerance for ambiguity when he talked about what he did and did not know about the problem.
This trend of tolerance for ambiguity codes in the interview also held for Beck’s data (Table B.1)
and Ed’s data (Table 7.7). This implies that relating words and actions to tolerance for ambiguity
224
might be better done via reflective surveys and essays rather than relying on pure observation within
the confines of the classroom.
An instance of imbalance in the application of the framework was Beck’s collaboration codes, as
seen in Appendix B. From our observations, Beck tended to give help far more often than he received
help. This was reflected in the data in that 15 out of his 20 codes for collaboration were two key
abilities: articulating or justifying the benefits of a particular approach, and clarifying, questioning,
or negotiating the group’s understanding and/or course of action. Beck’s case demonstrates a
stratification within the collaboration disposition between those two key abilities and the rest of the
codes in Table 7.1.
In addition to these structural discrepancies, we also saw some differences for some of the
students between what they said in their interviews with regard to a disposition and how they acted
in line with that disposition in the classroom. For example, what Ed said to Mr. Buford about
not persisting in Ed.In-class.4 is drastically different than what we observed. We analyzed this
excerpt because it showed Ed telling the teacher essentially that she hardly persisted at all during
the class period, yet we saw throughout the in-class data that she had acted out many aspects of
high persistence. This suggests Ed provided a harsher account (by the standards of CT dispositions)
of her behavior than we observed during the class period and indicates that some students might
understate their embrace of high dispositions when describing their own behavior. Ed’s discrepancy
between what she said and what she did could be due to any number of reasons. For example, Ed
might have taken her need to persist as a sign that shewas making many mistakes or that computation
did not come easy for her. Alternatively, she might have thought that Mr. Buford valued the right
answer (which she did get) over her ability to persist. Any of these reasons would prevent her from
seeing persistence as a strong positive attribute for computation, despite the positive framing of
persistence in the CT Dispositions Framework.
A parallel to Ed’s treatment of persistence is how she related her words and actions to tolerance
for ambiguity. Though we coded her data with a mix of high and developing codes, all codes for
developing tolerance for ambiguity came from her interview. She described computational activities
225
as if the ambiguity in them was usually intolerable, but when we observed her behavior in class,
she embraced the ambiguity in the computational activity. This was another instance indicating
students can sometimes align with different levels of a disposition depending on the context. For
Ed, some explanations could be: she did not like the ambiguity but knew how to navigate it in
class, she portrayed her embrace of tolerance for ambiguity more modestly when talking about it
with the researcher, or she perceived her in-class performance with an accentuation on the times
when she was less tolerant of ambiguity. Ed’s complex relationship with the dispositions offers
a word of caution to practitioners and researchers, namely that one source of data (be it in class
observations, interviews, reflections, surveys, etc.) might not tell the whole story. How Ed viewed
herself and what we observed were at times drastically different, yet no one data source in this case
is “correct.” Both how Ed felt and how she acted are equally valid, with both data sources offering
a more robust view of Ed’s relationship to dispositions than either one could provide individually.
Ed also gave a unique description for her collaboration, which raises another question about
applying the CT Dispositions Framework to the classroom context. In Ed.Interview.5, she told us
about the “leeching” interaction, where there was sometimes some explanation happening but no
co-creation of meaning. When she shared answers with her peers, nobody could ever fully explain
the solution being shared. In the current dispositions framework, a disposition for a “developing”
willingness to collaborate is described as, “the learner may see others merely as a ‘means to an end’
rather than as co-participants in a process or co-creators of meaning.” In Ed’s case, there seemed
to be a middle ground, which would be, “the learner may attempt unsuccessfully to co-participate
in the learning process.” Alternatively, we could see the focus of high collaboration shifted in the
framework to emphasize the intention behind the attempt over the success of the attempt, which
would make Ed’s attempts at explanation aligned with high collaboration. This shift would more
closely align the framework with the actual word “disposition,” which more closely resembles
attitudes than practices.
Another notable feature of our analysis comes from howBeck viewed ambiguity in his interview.
Although he could easily handle ambiguity in class, Beck liked when there was an equation to guide
226
the process, as we discuss in Appendix B. This was a source of comfort for Beck, and perhaps
there needs to be a place in the dispositions framework to acknowledge students who have maybe
not embraced the value that can be found in ambiguous problems but still demonstrate a high
tolerance for ambiguous problems. This is different from Ed’s description of ambiguity because
she articulated an intolerance for ambiguity and then embraced it in class, whereas Beck simply
articulated a preference for less ambiguity, which was still consistent with his actions of navigating
the ambiguity in class well. Adding a description to the framework, such as “preference for
clarity”, would allow for more nuance in describing students’ actions and views with respect to
the dispositions spectrum and identifying what might nurture a development of their dispositions.
Further work is needed to study whether these new categories of codes would be valid for other
students and whether other modifications might add robustness.
An interesting corollary is that Beck’s ability to strip away the ambiguity from computational
problems could indicate that if you reach a certain level of ability with the computation, you might
start to perceive less ambiguity in the computational activities. Beck’s preference for clarity might
be driven by Beck’s ability to address and handle ambiguity and derive clarity from it. This points
to the potential value of “hidden curriculum” [222] as well—if teachers communicate why it is
not only alright but a good thing to tackle ambiguous problems, then it can facilitate students to
embrace higher tolerance for ambiguity, in particular one of the key sensitivities: “awareness that
engaging in uncertain situations can lead to growth” [1].
There were also indications that students might differ in how much they embrace dispositions
depending on context. For example, Otto said he preferred to work solo in his calculus class
because he was highly skilled at calculus. His extra strength in calculus indicated that his peers
had even more to gain from his help than they would have had in physics, yet he was more reluctant
to collaborate in his calculus class. This might imply that some students are good collaborators
only when they benefit from the collaboration, which would indicate that students might be less
likely to collaborate (like Otto said he was in calculus) if they become better at working through
and understanding the material. This could also mean that Otto switched approaches/beliefs about
227
collaboration depending on the context of the activity (calculus vs physics). For example, the main
contextual factor for Otto seemed to be that he never needed help in calculus because he did so well
in the class by himself, whereas for physics he often ran into roadblocks, which predisposed him to
seeing the value of collaboration in the context of his physics class. Alternatively, the messaging
in his physics class might have better promoted collaboration as a tool over his calculus class. We
did not actually see Otto progress or switch his beliefs in our data in part because that was not the
design of the study. Our data represented a snapshot rather than a development in behavior over a
period of time or a contrast between contexts. However, this could be a focus of future work on CT
dispositions.
Another avenue for future work could investigate the role of the teacher in promoting the
development of CT dispositions. In our data, we found that the teacher’s role of building classroom
norms related to the dispositions that showed up in the data. For instance, we learned of the class
norm of “helping people” in Otto.Interview.5, which was something that Otto had come to expect in
Mr. Buford’s class only two months into the academic year. We coded Otto’s acceptance of this as
high collaboration, which indicates that there are actions instructors can take to support students in
developing dispositions, such as: making resources available and accessible [223], being proactive
with facilitating collaboration [223, 63], scaffolding curriculum to provide many opportunities for
accomplishment [177], acknowledging the normal computational experiences of frustration and
partial completeness [92, 180], and making the material relevant to students [224].
We summarized our dispositions analysis with Table 7.8, which pointed out howthe inclinations,
sensitivities, and abilities of the three dispositions were distributed among the data sources. Some
key points could prove useful to practitioners and researchers. We found that tolerance for ambiguity
was much more prevalent in the interview comments than the in-class behavior, which indicates
that teachers wishing to assess tolerance for ambiguity might find better results by assigning a
reflective essay (or something that somewhat mirrors the interview prompts) to students than trying
to observe it directly. We also found that abilities for collaboration overwhelmingly showed up in
the in-class data, which means that a teacher wishing to observe collaborative abilities could do
228
so by simply attending to what students do in class, as long as students are given opportunities to
collaborate in the first place.
7.7.2 Application of Mindset Coding Scheme in Connection with Dispositions
To answer the secondary research question, we also coded our data for mindset, drawing connections
between the mindset coding scheme and the CT Dispositions Framework. When it came to mindset,
we found that there were several examples of times when students expressed aspects of mindset
and dispositions at the same time and several examples of times when they expressed aspects of
mindset without dispositions. Altogether, we take this to mean that students sometimes told us
about their dispositions and about their mindset with the same action or utterance. This did not
mean that mindset is a combination of dispositions (or vice versa) because we saw many instances
of no overlap, indicating that mindset involved non-dispositional factors as well. By the same token,
CT dispositions involve factors unrelated to mindset. However, it does mean that the constructs
can often be related, and students can indicate both their dispositions and their mindset at the same
time.
One notable feature of the overlap between constructs in Figure 7.2 is that codes for persistence
and tolerance for ambiguity were on average twice as likely to align with mindset than collaboration
was. The difference between collaboration and the other dispositions (in that collaboration was
less likely to co-exist with mindset) could be explained by the design of the research study and the
development of the theoretical framework. To explain, collaboration was the only disposition not
explicitly required in Mr. Buford’s computational activities (see Section 7.4), though it was still
encouraged through messaging and the structure of the classroom. Also, collaboration was the
only disposition for which Pérez [1] did not explicitly cite mindset literature when developing the
CT Dispositions Framework (see Section 7.2), indicating that collaboration was not as closely tied
theoretically to mindset as the other two dispositions. These factors could explain why there was
less of an overlap between collaboration and mindset in our data.
In analyzing for mindset, we found several interesting occurrences in our data that have im-
229
plications for applying mindset theory to computational contexts and for how mindset relates to
the CT Dispositions Framework. For example, we saw that Ed flip-flopped between aspects of
fixed and growth mindset in Ed.Interview.Mindset. We related this flip-flop to an aspect of how
Ed’s behavior related to dispositions; she often reported in line with more aspects of developing
dispositions in her interview and then behaved in a way more aligned with high dispositions in
class. The fluidity in how Ed expressed dispositions and mindset points out that both constructs
are intended to be interpreted as a spectrum, and that our analysis of “high” and “developing” or
“growth” and “fixed” as two sides of a spectrum did not capture the full picture. This inconsistency
highlights two concerns when trying to utilize dispositions in one’s teaching practice. The first
is that when operationalizing, a teacher should remember that neither dispositions nor mindset
should be construed as being a rigid, unchanging dichotomy for a particular student, but rather
that students can exist and behave fluidly on these spectra with respect to time and context. The
second is reiterating the point that different data sources can provide different insights into a how
a student relates to dispositions and mindset. This suggests that a combination of approaches
(observations/surveys/reflective essays) might be needed to gain insight into a particular student’s
relationship with the constructs. Moreover, Ed’s case and her fluidity between aspects of fixed and
growth mindsets and developing and high dispositions further strengthens the argument that there
is a meaningful connection between the two constructs.
Another aspect of the relationship between dispositions and mindset is that fixed mindset codes
tended to correspond with developing dispositions codes, while growth mindset codes tended to
correspond with high dispositions codes, as seen in Figure 7.3 and Table 7.10. This alignment held
for 45 out of the 50 excerpts that were coded for both constructs. This suggests that when mindset
and dispositions are related in a student’s actions or statements, they are strongly tied. It remains
open whether there is any causation in this relationship, but given the widespread application of
mindset interventions [124, 125, 126, 127, 128], we recommend for researchers to measure shifts
in behavior related to dispositions in these settings in order to ascertain whether intervening on
behalf of mindset can also foster an impact for dispositions.
230
Despite the high alignment between mindset and dispositions, there were a few cases of nonalignment,
meaning fixed mindset and high dispositions coded in the same excerpt, or growth
mindset and developing dispositions. These instances are outlined in Table 7.10. We argue that
examining the times when there was misalignment can provide further insight into the relationship
between the two constructs. Focusing on when mindset and dispositions anti-align, we describe
two of the five non-aligned excerpts. In one excerpt, Blaine explained in his interview how he could
learn computation from a coding for dummies book if he wanted to, which we coded for an aspect of
growth mindset (given his stated belief that he could grow his computational skills) and developing
collaboration (given his rejection of the hypothetical opportunity to learn by working with peers,
instead opting to use the coding for dummies book in this scenario). In another excerpt, Blaine
copied and pasted a past project’s code into his GlowScript window (right before Blaine.In-class.2).
We coded this for an aspect of fixed mindset (given his avoidance of thinking about what he was
copying) and high persistence (given his pursuit of a resource that in principle could help his efforts
pay off, even though he was not using the resource, or copied code, effectively in the moment).
What made these excerpts special was that they represented moments where Blaine expressed an
idea or did something to advance his computational skill or progress in the computational activity,
but that idea or action was aligned with fixed mindset or developing dispositions in some way.
In the excerpt with the coding for dummies book, Blaine ignored opportunities to learn through
collaboration. In the code-copying excerpt, Blaine ignored opportunities to restrategize. Though
the excerpts represent earnest attempts to advance in the activity or improve a skill, dispositions
and mindset indicate to us alternative approaches that would enrich these attempts and make them
more productive.
Another mindset-based insight from Blaine’s results came from our analysis of Blaine.Inclass.
2, where Blaine reacted negatively to an error message and then avoided engaging with it by
searching for sample code online. We coded this excerpt for an aspect of fixed mindset, yet there
were some common computational experiences in what Blaine did: encountering an error and
searching for someone else’s code online. We should note, though, that copying code—or reusing
231
and remixing code as it is often referred to in CT terms [205]—is an accepted and legitimate strategy
to creating computational models. The commonality of this practice makes it an odd choice to be
coded for fixed as opposed to growth mindset. These instances of these experiences aligned with
fixed mindset because of how Blaine reacted to the error and the reason he searched for sample
code. This points to an opportunity to study mindset in computational settings, because there are
many common experiences when programming, such as dealing with errors and google-searching
for answers, but mindset entangles with the way a student responds to and/or brings about such
experiences.
7.7.3 Critique of Dispositions Framework
While we have demonstrated the applicability of the CT Dispositions Framework to Mr. Buford’s
computation-integrated physics classroom and its relationship to mindset, there were aspects of
the framework that did not apply or proved problematic when used in this context. Additionally,
there are some critiques of mindset theory that may also apply to the dispositions framework given
their overlap. For example, we pointed out in Section 7.2 that mindset theory can sometimes shift
responsibility off of structural faults in curriculum and onto students to conform their learning to
the faulty structure [213]. Mindset can be framed this way because it is theorized as something
that students can possess and develop, and not necessarily an aspect of curricular structure, except
in the form of mindset interventions. Like mindset, dispositions is also framed as a set of qualities
that describe how students orient themselves with respect to CT. This gives us pause when recommending
the use of CT dispositions in future research and future interventions. We provide a
critique here to point out the pitfalls of using CT dispositions and how to take care when applying
them, using lessons from our findings.
The main critique we highlight is the potential for centering the deficits of students (“deficit
thinking” [217]) when applying the CT Dispositions Framework. Though students are the subjects
of analysis when we use CT dispositions, this does not need to mean that the fault and responsibility
for improving dispositions lies with them. Improving dispositions indiscriminately is not a goal of
232
this work. We ultimately want to highlight what CT dispositions can look like and mean in one
context and discuss how teachers might be able to assess and impact dispositions if they see the
endeavour as valuable in their contexts. Whether or not this is a desirable goal we cannot yet say,
as we did not compare dispositions with learning goals or performance. We also wish to highlight
that a first step towards improving dispositions (if such development is desired) should be to assess
the structure of the curriculum, to see if it is affording opportunities to persist, collaborate and deal
with ambiguous situations. This focus on structure helps to direct curricular change away from
deficit thinking.
We took care to avoid framing students in terms of deficits by using language that tied students’
actions to codes from the framework, rather than tying students themselves to the dispositions
spectrum. For example, in Section 7.6.1.3.2, we analyzed Ed.Interview.3 in terms of the words she
said in connection with aspects of the framework: “We know she put in significant effort because
she described ‘working for a whole two days on it,’ which pointed to her ability to stick with a task
for an extended period of time.” When we summarized how Ed’s data was coded for persistence,
we described in terms of Ed’s relationship to persistence throughout her data rather than describing
persistence as a quality of hers: “sustained pattern of embracing high persistence on Ed’s part.”
One area of difficulty in applying dispositions directly to students’ behavior was when coding
interactions and miscommunications between students in class. An example of this is Blaine.Inclass.
5, when Otto did not listen to Blaine’s suggestion for typing a negative sign into Otto’s program.
The first issue was that the miscommunication could have relied on the context—Blaine’s pattern of
distracting Otto and Blaine’s statements about not knowing how to code could have reduced Otto’s
trust in collaborating with Blaine, leading to Otto tuning him out during this excerpt. The way we
coded did not account for these contextual factors, which meant that we would have needed a larger
grain-size on our unit of analysis in order to capture these details; we would have had to analyze
the data on the level of statements and actions like we did but also on the level of larger narratives
in the interview and overarching behavioral patterns in class. Studies utilizing CT dispositions
need to consider how much it matters to capture these complexities when designing the research.
233
Second, the term “disposition” seemed inappropriate to characterize this miscommunication. Otto
not listening to Blaine had nothing to do with Blaine’s “disposition” in the traditional sense of
the word, yet we coded this excerpt for dispositions nonetheless. This calls into question whether
dispositions should include subcategories like “key abilities” when the name of the framework
seems much more closely related to attitudes.
Another confusing aspect of the dispositions framework was the word “developing.” Though
it was meant to describe one end of the dispositions spectrum, the word itself implies a trajectory
towards high dispositions. At times, we coded behavior completely antithetical to high dispositions
with this word, “developing.” We agree in principle that it is worth pointing out that dispositions
are not fixed qualities and can change with time and circumstance, but this type of acknowledgment
belongs elsewhere, not in the naming of types of behavior. There was also not a description of
what developing inclinations, developing sensitivities, or developing abilities might look like in the
CT Dispositions Framework. These were the only aspects of dispositions from Pérez [1] that could
have been transformed into a coding scheme, which is what we did, but we were left to define the
developing aspects of dispositions in opposition to the high aspects. We propose an enlargement of
the descriptions of the dispositions to account for what this side of the spectrum can look like, apart
from brief descriptions of examples of how “developing” learners might act, as provided in Pérez’s
framework [1]. This also brings up a discrepancy between the framing of mindset and dispositions
as spectra and the framing of dispositions and mindset research in terms of the two ends of each
spectrum: high and developing and growth and fixed. This focus on the ends of spectra rather than
the middles makes it hard to treat mindset and dispositions as spectra when analyzing for them. The
middles of the spectra remain poorly defined both in the theory and in the data. We are instead left
to construct an incomplete picture of what the middle of the spectrum could look like by discussing
analyses that aligned with both ends, like Blaine’s collaboration, Ed’s tolerance for ambiguity, and
Ed’s mindset.
The last critique we outline is the structure of the aspects of dispositions: key inclinations,
key sensitivities, and key abilities. There is a contrasting presentation of these terms in Pérez’s
234
framework [1]. On the one hand, they are framed as “key,” which would imply that they are core
aspects of embracing the CT dispositions. This raises slight concerns when students fail to exhibit
certain key aspects. At the same time, Pérez emphasized an idea from Voogt et al. [225] that what
matters in building understanding is not “necessary and sufficient conditions” but instead “a more
graded notion of categories with an emphasis on the possible rather than the necessary” (page
719) [225]. When we apply this idea to the key aspects of dispositions, it would seem that they
are not “key” after all, but rather “possible” forms of taking up a disposition. This interpretation
also frames CT dispositions with less of a deficit focus, because it allows for students to approach
dispositions on their own terms rather than checking all the boxes that “key” aspects would seem
to imply are necessary.
We accompany this critique of the dispositions framework with a discussion of three limitations
associated with our use of it. First, the in-class data captured only one class session, whereas
there were multiple computational activities with different opportunities to interact with aspects
of CT dispositions. In this sense, the research design encapsulated a snapshot of CT dispositions
at play in the classroom rather than a trajectory over time. The emphasis on “development” of
dispositions by Pérez suggests that researching dispositions as a trajectory would adhere more
closely to the framework’s conceptualization. Second, we coded semi-structured interviews and
referred to the number of codes in our analysis. We tried to avoid making claims based solely on
the number of codes, but using the numbers at all could be seen as unreliable because much of what
is said in a semi-structured interview setting depends on the path of the conversation. We provided
numbers to point out how key aspects of the dispositions framework compared to one another,
but those numbers should not be taken to have significant meaning outside our data set without
further research. Third, the interviews were designed to explore how students perceived their
experiences in Mr. Buford’s class (see Appendix A). They were not designed as a tool for exploring
dispositions, though the topics that students discussed lent themselves well to an application of
the CT Dispositions Framework. It would have been even better to design the interview protocol
around key aspects of the dispositions.
235
7.7.4 Concluding Remarks
Though we have not been able to show conclusively that mindset and dispositions are tied together
causally or whether they develop together, we were able to show that they are correlated strongly
in instances when they both described a student’s words or behavior. Given that the origin
of persistence and tolerance for ambiguity in the Pérez paper [1] came in part from mindset
literature, a possible outcome of this study could have been a demonstration of dispositions being
a contextualized version of mindset for environments that emphasize computational thinking.
However, our results demonstrated that the constructs of disposition and mindset are related and
yet different. Given the correlation, mindset interventions [124, 125, 126, 127, 128] may have an
impact on students’ dispositions; however, it remains to be tested what the impacts would be or if
adaptations would be needed to address dispositions. We highlight again our suggestion to attend
to dispositions in settings where mindset is also developing or where mindset interventions are
taking place. More information about the relationship between dispositions and mindset could lead
to proven methods for improving students’ CT dispositions.
As we discuss the ramifications of the CT dispositions and mindset frameworks in Mr. Buford’s
class, we note that this was just one setting where this framework could be applied. As Pérez
said, “the usability of the framework [increases] through examples of classroom behaviors that may
accompany developing or higher levels of a given disposition” (page 442) [1]. We have provided
one case with a handful of examples, but other settings with other computation-integrations will
provide different perspectives and nuances to using this framework. Our study provides evidence
that aspects of the dispositions framework were applicable beyond the original context but the
uncertainty around how to interpret and apply “developing” dispositions codes is a concern. We
encourage future studies in the context of physics classrooms to continue to build on this framework
and account for more than just CT practices when examining computation in the classroom. We
also encourage attempts to build up and flesh out the CT Dispositions Framework to address the
features that we critiqued: a way to analyze data that falls along the middle of the spectrum, an
alternative to the wording of “developing” dispositions, and a rethinking of what it means for a
236
disposition to have “key” aspects.
Our work demonstrates an initial connection between the ends of the dispositions spectrum and
the mindset spectrum. We translated both theories into independent codes that could be perceived
separately but often appeared in the data with a great deal of alignment. Future research could
help understand whether dispositions are an aspect of mindset that has been ignored previously
or an aspect of mindset that manifests in this particular context. This connection also brings into
question whether the CT Dispositions Framework is altogether necessary since mindset, which is
much more grounded in past research, seems to be connected to it.
Another avenue for future work addresses CT dispositions and practices in the same setting.
Longitudinal studies should be conducted that would allow us to understand whether there is a
connection between CT practices and dispositions as posited by previous research but also whether
different computational activities elicit different dispositional responses. As we reviewed earlier,
there are many examples of research focusing on CT practices [40, 133, 66, 202, 203]. It would be
interesting to see how dispositions and CT practices coexist in the same setting so that the impact
and importance of CT dispositions can be articulated alongside and intertwined with the impact and
importance of CT practices, since these are the two sides of ISTE and CSTA’s definition of CT [132].
In terms of curriculum design, such work could entail dissecting a computational activity into parts
and connecting them to dispositions codes. This process could be akin to applying Vygotsky’s
zone of proximal development [226] to the experiences students have in a computational activity,
where they have to engage with the opportunities in the activity in order to overcome its difficulties.
Students may engage with CT dispositions and CT practices as they struggle through the activity
and in effect develop positive and/or negative affect-based views of computation.
Future work could also entail analyzing this chapter’s data set with a different focus. To address
the issue with the ill-defined low end of the dispositions spectrum, future work could take this
data set and use the examples that opposed high dispositions to characterize a new category. This
endeavor would also address part of our critique of the Pérez work, in that it would add robustness
to the framework and make it more readily applicable to other settings. A different approach to the
237
same data set could entail identifying interesting actions or descriptions from the students that were
not flagged when applying our coding schemes for dispositions and mindset. Such research would
serve to explore what the CT dispositions framework might be missing in terms how computational
activities can support students’ development in ways that encourage and enhance CT practices.
Depending of the direction of the research and findings, this could also serve to flesh out the low
end of the dispositions spectrum, because it is there that the CT Dispositions Framework is most
underdeveloped.
We conclude by returning to the main outcome of this work and the answer to our primary
research question—CT dispositions could be extended and applied to the setting of Mr. Buford’s
physics class using a research design that centered the perspectives of students. Though we had
many recommendations and questions earlier in this section for applying the framework to physics
students, we can say confidently that the framework is flexible to different contexts. Furthermore,
it showed some strong correlation with mindset theory for instances when both constructs could
describe what a student expressed. The relationship between dispositions and mindset opens up
avenues for future research. It is our hope that student perspectives will continue to be used to
ascertain both the effectiveness of computational integrations like Mr. Buford’s and the applicability
of learning theories like CT dispositions.
238
CHAPTER 8
DISCUSSION
In this dissertation, I have explored the use of students’ perspectives in curriculum development and
applied the idea of leveraging students’ perspectives in a computation-integrated physics context. In
the first three chapters, I laid the groundwork for the research studies that followed and demonstrated
the utility of qualitative case study for achieving the goals that I outlined. In Chapters 4 and 5, I
showed how the perspectives of LAs in an introductory physics course can function as voices in
curricular decision-making, showing how theoretical frameworks and attention to context can give
structure and meaning to students’ perspectives in research. In Chapter 6, I provided a catalog of
student-perceived challenges in a computation-integrated physics course and laid a foundation for
more focused studies by exploring how affect-related constructs (self-efficacy, mindset, and selfconcept)
related to students’ experiences. In Chapter 7, I built on the foundation from the previous
chapter by applying the theories of mindset and Computational Thinking (CT) dispositions to see
how they interact in a computation-integrated physics context, in effect extending the theory of CT
disposition and highlighting the potential for mindset to function as a window into how students
enact CT in a computation-integrated STEM context. Overall, this dissertation serves to amplify
the perspectives of students in a new and increasingly more widespread context—computationintegrated
physics—where there is a notable opportunity to infuse students’ perspectives into
curriculum development widely.
In more detail, we set up Chapters 4 and 5 by showing that student perspectives are consulted
broadly in physics education, but rarely have students’ voices had an explicit part in curricular and
pedagogical decision-making in the way that more recent efforts have shown [29, 31], in which
students’ perspectives factor directly into curricular and pedagogical decision-making. My research
showcased in Chapters 4 and 5 characterized a student-partnership in an introductory physics course,
and it showed that Students as Partners was applicable to learning assistants (LAs) in a course with
a Communities of Practice design. I also learned by doing this research the importance of paying
239
attention to theoretical perspectives and contextual factors when listening to students’ perspectives.
There is potential to build on this work in several ways. First, the specific course (P-Cubed) could
be improved by further reifying the contributions of LAs into the curriculum. Researchers could
re-envision other LA programs as student-partnerships, especially those designed with a leaning
towards fostering a community of practice. LAs’ voices need to be promoted and made louder
especially in cases where they experience classrooms both as a student and a teacher, giving them
unique, often untapped perspectives on the courses they teach. The LAs in P-Cubed also return for
multiple semesters and often garner more experience teaching the class than the TAs and faculty
empowered to teach the class. Developing a community and procedures that value their experiences
and amplify their voices is essential for the necessary continued evaluation of our classrooms. There
is also an opportunity to investigate other P-Cubed LA practices (other than the formative feedback)
to see how else to leverage LAs’ perspectives and incorporate them into the class’s design.
Building onmy work in Chapters 4 and 5, I applied what I learned about how to listen to students
productively to the context of computation-integrated physics. In designing and carrying out the
study showcased in Chapter 6, I identified and addressed a significant gap: computation-integrated
physics is a curricular setting where students’ perspectives have not been incorporated. Given
how recent computation-integrated initiatives are in education, research on students’ perspectives
in computation-integrated physics is surface-level and scarce [33, 72, 71, 70, 73, 74]. In response,
I designed and carried out a case study that explored students’ difficulties in their computationintegrated
physics classroom. In terms of findings applicable to the curriculum, the students in
Mr. Buford’s class struggled with several affect-related challenges: Stress/Frustration, Feeling
Worse at Physics, Unbelonging and Stereotypes, Repeated Confusion, Interpreting Code, and
Interpretations of Implementation. For curriculum developers,my findings highlight the importance
of communicating expectations when introducing computational activities, designing activities with
some easily attained successes in them, relating to students’ computational struggles, and discussing
the positive long-term impacts of learning computation. I also connected the students’ perspectives
to the educational theories of mindset, self-concept, and self-efficacy. The connection to theories
240
was an emergent part of the analysis, demonstrating howtheoretical lenses can showup and function
in a computation-integrated physics setting. My research in Chapter 6 calls for exploratory research
in other computation-integrated school contexts, especially those with different implementations
from Mr. Buford’s and with different curricular constraints than the AP curriculum. My work
also found initial connections to affect-based constructs; however, more work needs to be done
to explore how these constructs manifest in the computation-integrated physics context. I would
recommend such studies to apply theoretical lenses onto students’ perspectives, whether new lenses
or using mindset, self-concept, and/or self-efficacy in more depth. Ultimately, Chapter 6 can serve
as a jumping-off point for any study that examines students’ perspectives in computation-integrated
physics.
Chapter 7 directly built on the exploratory work of Chapter 6. This chapter was focused on
adapting a theoretical construct that related to the barriers that emerged from students’ perspectives
in Chapter 6 to the context of computation-integrated physics curricula. In this chapter, I outlined a
study that showed more in-depth how theory can enhance descriptions of what students experience
in computation-integrated physics, highlighting areas with the potential to instigate meaningful
and productive curricular change. In Chapter 7, I extended the utility of the CT dispositions
framework by showing how it applied in a new setting, and I investigated the relationship between
CT dispositions and mindset. I also demonstrated how different data sources in Mr. Buford’s class
provided different insights into students’ CT dispositions. This finding in particular could help
teachers and researchers select appropriate methods for examining CT dispositions in computationintegrated
STEM settings. I also included a critique of the CT Dispositions Framework. The
critique involved discussing the potential for CT dispositions to encourage a deficit-framing of
students’ relationships with computation, outlining the trouble with applying the framework to data
of different grain sizes, and highlighting language from the theorization of the framework that
proved confusing or ill-fitted when applied to student data. Future work could include applying
CT dispositions during mindset interventions as a way of further characterizing the relationship
between the constructs. CT dispositions could also be explored in other contexts (even other
241
computation-integrated STEM contexts) to further strengthen the framework. Furthermore, there
is an opportunity to study how CT dispositions and practices come together in a computationintegrated
STEM context. Lastly, while we focused on the construct of mindset, there are other
affect-based constructs that could be applied in meaningful ways in this context (as shown in
Chapter 6). Future work could also include applying these other theories to students’ perspectives
in computation-integrated STEM. In a sense, Chapter 7 serves as precedent for such further studies,
aswe have already pointed out the gap in computation-integrated STEM literature and demonstrated
that research can be carried out to address it.
I now synthesize across the chapters, first comparing Chapters 6 and 7, and then all four main
body chapters together. First, I noticed that CT dispositions touched on but did not fully address
the affect-based, computational barriers from Chapter 6. By this I mean that although dispositions
provided a fuller picture of howmindset could relate to students’ behavior and howa newframework
could apply to this new setting, the barriers themselves remain unaddressed. Chapter 7 could be
characterized as a deeper analysis of how some students reacted to those barriers, but the insights
gleaned using CT dispositions do not tell us how those barriers came to arise in the first place or
how we can help students traverse them. I was able to provide some suggestions for designing
activities with opportunities to embrace dispositions, but perhaps the main takeaway should be that
there could be another framework or connection of ideas that would better address the barriers from
Chapter 7, perhaps even a reorientation of the dispositions framework as discussed in my critique
of it in Section 7.7.3. Answering that call encompasses future work on computation-integrated
physics classrooms.
I alluded to future work on fleshing out the missing parts of the CT Dispositions Framework
in Section 7.7.4. Such work could be done by leveraging the findings from Chapter 6. Certain
challenges from Chapter 6 align with ideas from CT Dispositions, for example, Repeated Confusion
relates closely to persistence. Other challenges, such as Unbelonging and Stereotypes, do not fit
nicely into CT dispositions. Given that these challenges came from the same data set I used
in Chapter 7, they may point to aspects of students’ experiences that my application of the CT
242
Dispositions Framework missed. The example of Unbelonging and Stereotypes points to structural
forces that challenge students, forces that dispositions might not be able to address. Future work
in computation-integrated settings could involve reworking the dispositions framework in terms of
what it can mean for students in the face of affect-related challenges. Connecting the two studies like
that could potentially highlight pathways through some of the challenges that students identified.
Future work in computation-integrated settings could also involve leveraging students’ perspectives
more directly like in Chapters 4 and 5. This structure of students (inmy studies, undergraduate
LAs) playing a role in curricular decision-making is much different than the format of Mr. Buford’s
class, in which he designed and carried out the curriculum, and students were not given much
control at all over what they learned and how they learned it. Because my dissertation is about
centering students’ perspectives in curriculum, I can imagine several ways in which students could
have been given some decision-making power in Mr. Buford’s class, especially in how computation
became a part of their physics learning. This might not mean full-fledged students as partners in
a high school physics class, but it could mean climbing a rung or two on the student participation
ladder from Figure 5.1. One process could be that students build a repository of computational
tools together as they work through the computational activities throughout the year, a repository
that could be accessed by anyone in the class and remixed collaboratively. Another idea could be for
students to have an assignment to build computational projects to model ideas from other classes or
from outside school, ideas that they are interested in. One idea from another teacher’s classroom in
which I generated data (but did not analyze in this dissertation) was to have students pick a favorite
movie scene of theirs that contained “bad” physics and create a computational animation to show
how the movie scene was unrealistic. Lastly, I could imagine students gaining some control over
their computational learning by being given a space to share the strategies that they had built and
discovered during the computational activities—this could take the form of a five to ten minute
period at the end of class where students take the floor and share their struggles, successes, and
strategies for working through the computation.
Given the opportunity, there are a few aspects of this research that I would change. My focus in
243
the latter chapters lay in the experiences of students doing computation-integrated physics. I would
have liked to apply this focus to my earlier chapters where I presented my research on LAs. Much
of the research design in Chapters 4 and 5 was devoted to exploring the LA experience in relation
to the practice of giving feedback. However, those same LAs engaged with computation when they
were students in P-Cubed, and they taught computation-integrated physics as LAs. I am now much
more interested in this aspect of the LA experience, and I wish this practice had been my original
focus. Had I developed a deeper understanding of what that practice meant to LAs, I could have
related their experiences more closely and with more insight to the experiences of the high school
students in Chapters 6 and 7. In these latter studies, I generated data originally meant to explore
what computation meant to high school physics students. As discussed in Section 7.7.3, I could
have generated a richer data set by recording multiple computational activities and designing a new
interview protocol to explore dispositions. There are also many avenues of future work related to
enacting versions of these changes and/or re-examining the same data with a new lens or direction.
I described such work earlier in this chapter and in greater detail in Section 7.7.4.
The limitations of this work comprise the limitations imposed by my design of data generation
and analysis, the limitations of qualitative case study, and the limitations of doing research in a
somewhat unexplored context. First, the interviews in Chapters 6 and 7 were constrained to a
single free class period lasting 45 minutes, meaning the protocol had to be designed and carried
out with that in mind. This was a limitation in terms of the breadth and depth of experiences
students were able to tell me about. A similar limitation applies to the recording of only a single
class period in Chapters 6 and 7. Though the smaller data set allowed me to analyze more in depth,
there was no variance in the computational activities represented in the data, meaning I was only
able to capture how a student behaved during a single activity. Though these limitations do not
affect the validity of my claims, they do narrow the context in which my claims reside, discussed
in greater detail at the end of Section 7.7.3. The analyses, too, rely on my interpretations of data
and my fashioning of theoretical frameworks into analytic tools. My perspective, as well as the
perspectives of my participants, is infused into the findings of the chapters above. I am unable
244
to make claims about causality [47]. Though I have recommended pedagogical strategies and
certain features of curricular implementation, I cannot guarantee their effectiveness in any context.
Anyone who wishes to use this research and its recommendations needs to build an awareness of
their own context in order to come to reasonable conclusions about what they can expect based on
how their context relates to the cases I presented in this dissertation. This research is also limited
in the sense that computation-integrated physics remains a barely explored area. It is difficult to
ascertain how my cases relate to others due to the scarcity of student-centered research in this
area. Mr. Buford’s implementation of integrating computation could bear hardly any resemblance
to many other implementations, which would limit the immediate practicality of my findings. It is
hard to know without more research.
As computation continues to be integrated into STEM classrooms in schools around the world,
students’ perspectives continue to be an excellent (but underutilized) resource for curriculum
designers and researchers. There are opportunities to incorporate students’ perspectives into
curriculum with more depth than ever before catalogued in research [28]. There are opportunities
to leverage student input in computation-integrated contexts, which are growing in number and
ever changing as research calls for it [33]. Now is the perfect time, as computation spreads widely,
to design research and curriculum that centers students’ perspectives.
245
APPENDICES
246
APPENDIX A
MULBERRY HIGH SCHOOL INTERVIEW PROTOCOL
1. Tell me about yourself.
a) What year are you in school?
b) Why did you choose to take this physics course?
c) Have you taken a physics course before this one?
d) What do you want to do after high school?
2. Tell me about what you do in physics class.
a) Are there different sorts of activities you do? Can you describe them for me?
i. Do you always solve for a number? Do you have to design things?
ii. Do you ever work with equipment?
iii. Do you always work by yourself, or do you work with your classmates?
iv. How do you interact with your classmates?
v. How do you interact with the instructor?
b) How is this class different from prior physics classes?
c) Do you think you’re good at physics?
d) Are there times you struggle more than others?
e) Are there things you do in class that make you feel as if you can or can’t do physics?
f) Are there times in class when you feel more like a scientist/physicist?
3. About the computational activities in Mr. Buford’s class...
a) Why do you think Mr. Buford added computational activities to the class?
b) Have you done anything with computation before?
247
c) Was there anything new or exciting that you were able to do with computation? Can
you give an example?
d) Do you like the computational activities? Why or why not?
e) Do you ever get frustrated in class? What has frustrated you and why?
f) Do you think you’re good at computation?
g) Are there things you do in class that make you feel as if you can or can’t do computation?
4. When you get stuck with computation, what do you do?
a) Do you wait until Mr. Buford can help?
b) Do you try to consult with your group mates?
5. What do you learn/gain during the coding days?
a) What about regular days?
b) What do you learn? How do they differ?
6. How do you tend to participate in class?
a) When Mr. Buford is talking to the class?
b) When you are working together in a small group?
c) Does this change when you are doing computational activities as a group?
i. What role do you take on when the group doing computation?
7. What if you were told that computation is a big part of what you want to do in the future?
8. What is a subject you really like (or really don’t) and how does your experience in that class
compare to physics class?
9. Have you done computation before Mr. Buford’s class? How did you feel about it?
248
APPENDIX B
BECK RESULTS
B.1 Beck (Dispositions)
Beck’s statements and actions aligned with high dispositions. In Table B.1, we show the codes
for his interview and in-class data. One notable feature is that we only coded for collaboration
in Beck’s interview twice. For tolerance for ambiguity, we coded several times for developing
tolerance in his interview but not once in his in-class data. We provide a possible explanation for
this discrepancy below.
B.1.1 Tolerance for Ambiguity
Beck’s statements and actions tended to align with high tolerance for ambiguity, especially during
the in-class activities. He embraced ambiguity when it presented itself, though he did not seek it
out on his own; he preferred clarity and concreteness. In the excerpt below, he contrasted different
school subjects based on his “interpretation” of what he had to do in them.
Beck.Interview.1:
Tolerance for
Ambiguity
Persistence on
Difficult Problems
Willingness to
Collaborate with Others
Beck Interview
18 high
9 developing
11 high
0 developing
2 high
0 developing
Beck In-class 11 high
0 developing
14 high
0 developing
18 high
2 developing
Beck Total 29 high
9 developing
25 high
0 developing
20 high
2 developing
Table B.1: Coded instances of CT dispositions in Beck’s data, separated by data source.
249
Int You mentioned some subjects you don’t like, English, history, how do those
compare with physics and math?
Beck It’s mostly the thing I was talking about, a lot of it is interpretations stuff, things
like- poetry is one of my least favorite things. You have to interpret it and there’s
so many different ways to interpret it, and you’re like, yes, that’s correct. But
now you have to support your answer with everything. And I like something
that has a clear answer. I’ve come to realize that the physics conceptual things,
they obviously- they do have a clear answer. But at the beginning, since I didn’t
understand, I wasn’t able to figure out what the clear answer was. So I didn’t
really like it, the conceptual stuff that much. But now, I mean I understand
most of the things pretty well so I can see, ’Oh yeah,’ there is one clear answer.
Even if I don’t get it at first, there is something, that it has to be correct...
Int So I definitely have some things I want to follow up on... You mentioned earlier,
when you are able to explain a physics concept, that’s howyou knowthat you really
know it and you can explain it to other people. Is that in some way explaining
your interpretation of the problem?
Beck A little bit, but yeah, sort of I guess. But what I mean is there’s an equation.
For example, the thing I was doing the other day was refraction, when there’s a
ray of light that goes into a substance, like a glass. If the speed of light in them is
different than it’ll change direction and it’ll bend. That sort of stuff is, I feel like
it’s... There is of course some interpretation, but I feel like it’s more specific.
The first subject he brought up was poetry, where “there’s so many different ways to interpret
it.” This ambiguity did not sit right with Beck, who preferred “something that has a clear answer,”
indicating a preference for a single solution, though not an inability to recognize when there were
multiple solutions. He went on to describe his initial physics experience, when he “wasn’t able
to figure out what the clear answer was.” He said this caused him to “not like the conceptual
250
stuff,” indicating that he had a disinterest in exploring unfamiliar situations (interview commentary
opposed to key inclination #1, tolerance for ambiguity). He acknowledged that he had since grown:
“Even if I don’t get it at first, there is something, that it has to be correct.” His growth came from
warming up to the murky conceptual physics problems, in effect exploring an unfamiliar situation
(interview commentary aligned with key inclination #1, tolerance for ambiguity).
We noticed in his response to the follow-up question that he was focused on the presence of “an
equation” whenworking through physics concepts. Hewent on to describe the concept of refraction,
qualifying it with, “I feel like it’s more specific.” This focus on the “specific” nature of physics
concepts and the related equations pointed again to a preference for anchoring the physics problem
in a more concrete, rigid idea (interview commentary opposed to key inclination #3, tolerance for
ambiguity). The discrepancy between Beck’s ability to handle ambiguity and his preference for
more straightforward problems indicated Beck sometimes might not have been seeing the value in
the ambiguous problems.
Adding more nuance to Beck’s views on ambiguity, he described the complexities of applying
physics knowledge to computation and the benefits of interacting with a working, dynamic solution.
Beck.Interview.2:
Beck Well for the coding you have to actually apply what you’ve learned... understand
the math behind it and what’s actually going on. Because when
you’re coding, first of all, you actually get to see it happen in real time... And
also you’re able to implement the different things that you’ve learned and alter it
slightly, and can make huge changes and things like that.
Beck’s first comment about doing computation read like an instruction: “you have to actually
apply what you’ve learned.” Later, he clarified: “understand the math behind it and what’s
actually going on.” This entailed bringing together prior learning, digging into the underlying
math, and seeing the problem for what it “actually” was. This was a reframing of the seemingly
ambiguous features of the computation as an opportunity to clarify what was known about the
problem (interview commentary aligned with key sensitivities #2 and #3, tolerance for ambiguity).
251
In Beck’s view, this reframing was valuable: “you actually get to see it happen in real time.” This
achievement represented an awareness on Beck’s part that engaging with the uncertain parts of
computation could lead to a growing understanding of physics (interview commentary aligned with
key sensitivity #1, tolerance for ambiguity).
In class, Beck demonstrated a high tolerance for ambiguity in how he acted and talked with other
students about the computational activity. Below, he expressed some comfort with just picking a
number for his program, without regard to whether it was “correct” or not (or even whether there
was a correct answer).
Beck.In-class.1:
Brian Are we supposed to have like five balls?
Beck I have four. I don’t know if there’s a certain number we need
He did not know how many “balls” (representing light particles) were required, he just picked
a number. Unlike Brian, who was in search of specific guidelines, Beck seemed content with
choosing “four.” This may seem trivial but the ability to make a choice without being worried
about whether it was right indicated that he was okay with the presence of multiple possible
solutions (in-class behavior aligned with key ability #1, tolerance for ambiguity) and that he had
accepted the variance that may result from students picking different numbers (in-class behavior
aligned with key inclination #4, tolerance for ambiguity).
Beck’s data represented an overall embrace of high tolerance for ambiguity, as indicated by
the key aspects of this disposition that he displayed in the excerpts above: an interest in exploring
unfamiliar situations, an accepting view of variance, an awareness that engaging with uncertain
situations could lead to growth, an alertness to opportunities to clarify what was known and unknown,
and a responsiveness to approaches for reframing ambiguous situations. This differentiates
from his interview when he articulated a preference for less ambiguity. We coded aspects of this
preference for a disinterest in exploring unfamiliar situations, an unawareness that engaging with
uncertain situations could lead to growth, and an adherence to the idea of a single solution path.
252
Overall, this preference did not preclude Beck from embracing high tolerance for ambiguity during
class.
B.1.2 Persistence
Beck aligned his statements and actions high persistence in both data sources. When asked how
he dealt with stuckness in his interview, he explained a go-to strategy that demonstrated he did not
tend to give up right away.
Beck.Interview.3:
Int What do you do when you get stuck?
Beck I just try to write it out on a paper and say I would, I try to draw the thing that
we’re doing because a lot of it, most of it is visual for the coding. So I try to draw
the thing and see what sort of relationships I have. Like yesterday, I drew the
light and the lens, and I was like, Oh there’s a triangle here. Maybe I can find the
portion on the bottom, the vertical portion, and then I could do the Pythagorean
theorem on it to figure it out or something. Or use trig or something.
The first tactic he described taking up was “to draw the thing and see what sort of relationships
I have.” His interest in exploring relationships between aspects of the problem indicated that his
focus was on discovering new information, even in the midst of stuckness, without guarantee of
success (interview commentary aligned with key inclination #3, persistence). This also showed that
Beck had an alertness to the different characteristics of the task, since he was able to derive new
insight simply from sketching out its features (interview commentary aligned with key sensitivity
#1, persistence).
We followed up later in the interview to see what other strategies Beck might have turned to.
Beck.Interview.4:
Int Okay, so you’re drawing it on paper. Do you ever wait for Mr. Buford to get help?
253
Beck I try not to. I mean if I ever get really stuck I will just go up to him and ask
him because if I have no ideas whatsoever in my head. Like I draw it out and I
just don’t have any idea what to do, I’ll definitely ask him, yeah. I’ve done that a
couple of times.
At times, when Beck was really stuck, he would just go and ask Mr. Buford for help. This
indicated that he had a backup plan, or safety net, for when he could not figure out how to
overcome the difficulty at hand. This would constitute trying a new approach after considerable
effort (interview commentary aligned with key ability #2, persistence).
The in-class data further reflected Beck’s embrace of high persistence. One feature of his
workflow is that he often thought out loud about what he was doing, which provided a window
into his thought process. His think-aloud style of working through the computation showed that he
was consistently trying new things and hitting snags with the code, but he always persisted through
them without giving up. One example happened right at the start of class when he was searching
for how to correctly use a specific function in the code.
Beck.In-class.2:
Beck °So: how do I: ˆhelpˆ°
(3.5)
Beck °How do I-° Oo there we go, ˆadd an arrowˆ
(24.0)
Beck How do I- oh there we go, ˆattach arrowˆ
The carets (“ˆ”) indicate a cadence for reading text, which means Beck was likely reading off
options in the help menu as he searched for a function he could use. He started out with a question
indicative of stuckness (“how do I”), and he cut himself off to indicate that he had navigated to the
help menu, a popular GlowScript resource in Mr. Buford’s class. The same pattern happened twice
again over the next 30 seconds, indicating that he was using the GlowScript documentation as a
254
resource to help him carry out the task more effectively (in-class behavior aligned with key ability
#3, persistence).
Later in class, he ran into an error message while helping Otto. Beck’s willingness to try out a
new approach right away aligned with high persistence.
Beck.In-class.3 / Otto.In-class.5:
Otto So run that and it’ll just, ((pointing)) straight
Beck Let’s see what happens, should do (inaudible). Straight to the right. ˆInconsistent
indentation one fullˆ- let’s see, see that’s why I didn’t- Alright so, light- I’m
just gonna
Otto Just retype it
Beck ˆWhile light dot position dot x less thanˆ, °what was it?°
Otto Light- I mean um
Beck Focal point?
Otto Uh, yeah. Focal point dot pos: x
Beck °Position dot x°
Otto Hundred
Beck °Velocity one hundred°
Beck Er::, oh! Got it. Oh, colon
Otto OH you need a colon? Ah!
Beck One hundred. Yeah, that’s a thing you do need. It should- Yeah! And that just
travels straight to the right. Until it gets to there
While helping Otto get a particle to move in a straight line, Beck got an error message (“inconsistent
indentation one full-”). He immediately reacted to the error message and tried to fix
the mistake. His reaction was to process the error message (“let’s see...alright...I’m just gonna”),
255
indicating with his next few utterances a retyping of portions of the code (e.g., “while light dot
position dot x less than”). His constructive response to the error indicated an attentiveness to the
opportunity to shift tactics (in-class behavior aligned with key sensitivity #3, persistence). When
his efforts yielded success, he interrupted himself with a positive exclamation (“Yeah!”), indicating
satisfaction at the fruits (“that just travels straight to the right”) of his significant effort (in-class
behavior aligned with key sensitivity #2, persistence).
Throughout Beck’s interview comments and in-class conduct, he embraced high persistence
on difficult problems. When examining the excerpts above, we observed several key inclinations,
sensitivities, and abilities: an interest in what might have been discovered even in an unsuccessful
attempt, an alertness to a task’s characteristics, an awareness of the satisfaction that would be felt
when efforts paid off, an attentiveness to opportunities to shift tactics when needed, an ability to try
a new approach after considerable effort, and a pursuit of resources that increased the effectiveness
of his effort.
B.1.3 Collaboration
Additionally, Beck embraced a high willingness to collaborate, which was consistent across his
interview and in-class data. Beck’s view towards collaboration could be exemplified in what he
thought of “explaining,” shown below.
Beck.Interview.5:
Beck If I can explain it to somebody then I usually know I understand it pretty well.
Like I’ve explained a couple of things like that to my dad, I do that sometimes-
Because teaching things usually helps you learn it even better. For me at least. So
if I can explain something to somebody else, then that’s usually a sign that I
know it pretty well.
For Beck, when he explained something successfully it indicated that he understood it: “if I
can explain something to somebody else, then that’s usually a sign that I know it pretty well.”
256
Sometimes he even just explained stuff to his dad. This embrace of explanation indicated an ability
to negotiate meaning with others (interviewcommentary aligned with key ability #3, collaboration).
When asked about group work, Beck acknowledged that he participated regularly:
Beck.Interview.6:
Int Do you ever consult with your other group members at your table?
Beck Yeah. Yeah. Even with people not at my table, like there’s our table and there’s
a table behind me too. Kind of just one big table, I’m part of both of it. I ask
people if they have any ideas, or if they’re ahead of me or behind me.
This excerpt shows that Beck believed that sometimes peers could be sources of ideas, and also
it was nice to gauge where everyone was at: “I ask people if they have any ideas, or if they’re ahead
of me or behind me.” He liked to know how others were doing during physics class. This showed
his tendency to invite and value perspectives different from his own (interview commentary aligned
with key inclination #2, collaboration).
Examples of this idea-sharing and collaboration abounded in Beck’s in-class data. Below, we
showBeck’s collaborating with Otto. In the conversation, Beck helped Otto implement a while-loop
in his code to make some particles move on-screen.
Beck.In-class.4:
Otto We’re gonna, think about that later. So, how do I make it move?
Beck Okay, so. ((laughs)) $Pretend that never happened.$ So yeah you need a while
loop. So [you wanna s- you wanna set something
Otto [Just do control z there
Beck No you wanna set s- °I’m gonna type (inaudible).° You wanna have something
d t, change in time
Otto Okay
Beck Point one’s usually a good one. So you need a while loop. Uh:, for now we’ll
just do true you can go set the condition [when you want it to stop later
257
Otto [No I like this- I have, °I have a condition that I like. I have a condition that I
want to (inaudible)°
Beck Okay. Cool then. That’s a good condition
The sequence of interaction in the excerpt began with Otto asking for Beck’s help, Beck
suggesting a while loop, and then Beck showing Otto how to implement it. In the middle of the
excerpt, Beck spent time explaining features of the loop (“you wanna have something d t, change
in time” and “you can go set the condition when you want it to stop later”). When Beck explained
the time-step (“d t”), this represented a negotiation of the approach that Beck was implementing in
Otto’s code (in-class behavior aligned with key ability #3, collaboration). When Beck explained
that the “true” condition allowed Otto to set up a different stopping condition later, this was a
justification of the benefits of Beck’s suggestion (in-class behavior aligned with key ability #2,
collaboration). Otto responded to this point by saying he already had a condition in mind (though
the description was inaudible). Beck responded positively (“that’s a good condition”), which
indicated that he valued Otto’s perspective on this part of the code (in-class behavior aligned with
key inclination #2, collaboration).
We also examine an interaction that Beck had with Blaine near the beginning of class. It was
when Blaine indicated that he had an issue with his code, a code that Beck had tried to help him
with during a prior class period’s computational activity.
Beck.In-class.5:
Blaine It still doesn’t curve
Beck I don’t unde- I don’t understand what your problem is Blaine, okay? ((turns
back towards own table)) A- It literally in the end put my<
Beck ((turns abruptly around the other way)) How are you doing Otto?
Blaine’s complaint (“it still doesn’t curve”) was met with a response from Beck: “I don’t
understand what your problem is Blaine, okay?” The harshness of this response indicated a
negative interpersonal dynamic that Beck initiated and/or perpetuated with this comment (in-class
258
behavior opposed to key sensitivity #1, collaboration). Beck then began to comment on the issue,
but he cut himself off abruptly, as indicated by the less-than symbol (“<”). Beck then checked in
with Otto (“how are you doing Otto?”), indicating a contrast to what Beck said to Blaine—with this
check-in, Beck initiated a productive interpersonal dynamic with Otto (in-class behavior aligned
with key sensitivity #1, collaboration). The way Beck cut himself off could mean that he was
initially willing to engage with Blaine but then thought otherwise. Whatever the reason, we coded
this excerpt for developing collaboration in Beck’s interaction with Blaine, and high collaboration
in Beck’s interaction with Otto.
On the whole, Beck usually embraced a high willingness to collaborate with others. Though he
was more often on the helping or explaining side of a collaboration, he still recognized the value that
his peers brought to the table. We coded for collaboration much more often in his in-class data than
in his interview, but all the same we coded consistently for high collaboration in both data sources.
In the above excerpts, Beck demonstrated a tendency to invite and value perspectives different from
his own (with one exception in his interaction with Blaine), an alertness to interpersonal dynamics
that might have enhanced or impeded effective interactions, an ability to articulate and justify the
benefits of a particular approach, and an ability to clarify and negotiate a shared understanding and
course of action.
Through all three CT dispositions, Beck’s data aligned with high codes. The main pattern in
contrast with this was that for tolerance for ambiguity, Beck articulated a preference for clear-cut
answers in his interview, but this preference did not stop him from enacting a high tolerance for
ambiguity consistently during the computational activity.
B.2 Beck (Mindset)
Beck embraced high dispositions. When we looked at how mindset was present in his data,
both in and out of our dispositions analysis, we that Beck had a vast majority of growth mindset
codes (compared to fixed mindset) in his data.
For instance, in Beck.Interview.3, Beck described what he did upon getting stuck during
259
computational activities. In the dispositions analysis, we focused on his strategy to draw out
different pieces of the problem on paper and try to see relationships that might have helped him get
unstuck. This was evidence for his tendencies to look for ways to discover new information and
remain alert to different characteristics of the task, both aligned with high persistence. His list of
tactics and emphasis on learning more about the problem also pointed to a few characteristics of
a growth mindset: a view of setbacks as overcome-able, interpreting a mistake (or stuckness) as a
learning opportunity, and a view of effort as the path to success.
Much like the other students, we also found mindset codes that did not overlap with the
disposition codes. For example, Beck described the benefits of computation and why he liked
solving problems in this way, focusing on computation’s creative possibilities and the relationship
between computation and the real world, rather than its complexities and ambiguities.
Beck.Interview.Mindset:
Beck I mean GlowScript, it allows you to apply to stuff that you’ve learned in a way
that’s different from just solving a problem on paper, because you actually get
to see the result of what you’ve solved in real life. I mean it’s a computer, but
you get to see it actually work. And it gives you a view of what physicists do,
I suppose. Like you get a problem and you use physics to solve the problem,
then you see it actually work... I like the coding in physics because of that.
Beck highlighted a few different times the opportunities that he “gets to” have when he did
computation. He “gets to see it actually work.” He “gets to see the result...in real life.” His framing
of “getting” to have these experiences indicated that saw computation as an opportunity and he
wanted to learn via computation. He made this explicit at the end of the excerpt: “you use physics
to solve the problem, then you see it actually work...I like the coding in physics because of that.”
Finally, we saw Beck in a situation where he became aware of a mistake in class and said what
he did wrong earlier that led him to become stuck.
Beck.In-class.Mindset:
260
Ed Maybe you could low key just like, choose [a focal point and say goes towards
focal point=
Beck [Oh! =That’s literally what I’m doing
Ed $Yeah$ don’t try to, be smart about it
Beck I just, I wrote in the wrong variable is the problem
...
Beck Here he- this is what I have ((turns laptop towards Ed, then turns laptop back to
self)) Oh: no! It keeps not working. [I keep putting while and forgetting to
do anything after
In the first comment, Beck talked back and forth with Ed, where Ed suggested the straightforward
fix of making the particle go towards the focal point, and Beck said that he was already trying to
do that. The interaction did not lead to any change, but we did get to see Beck articulate the source
of the problem: “I wrote in the wrong variable is the problem.” In the next comment, Beck was
about to show his new animation to Ed when an error popped up, preventing the code from running.
He again said the issue out loud: “I keep putting while and forgetting to do anything after.” Both
admissions demonstrated that Beck was aware of the exact mistake that caused him to get stuck,
and he was not hesitant to say out loud to his peers what the mistake was. This indicated that he
was not trying to avoid or deny mistakes, and he was instead taking responsibility for his mistakes,
an indicator of growth mindset (see Table 7.2).
The takeaways from how Beck’s data was coded for mindset are not unique compared to other
students. Beck aligned his views and actions with growth mindset through many parts of the data,
as seen in Table 7.9.
261
REFERENCES
262
REFERENCES
[1] Arnulfo Pérez. A framework for computational thinking dispositions in mathematics education.
Journal for Research in Mathematics Education, 49(424), 2018.
[2] Carol S. Dweck. Mindset: The New Psychology of Success. Random House, New York,
2006.
[3] Carol S. Dweck. Self-theories: Their role in motivation, personality, and development.
Psychology Press, Philadelphia, PA, 2013.
[4] Lisa S. Blackwell, Kali H. Trzesniewski, and Carol Sorich Dweck. Implicit Theories of
Intelligence Predict Achievement Across an Adolescent Transition: A Longitudinal Study
and an Intervention. Child Development, 78(1), 2007.
[5] David Scott Yeager and Carol S. Dweck. Mindsets That Promote Resilience: When Students
Believe That Personal Characteristics Can Be Developed. Educational Psychologist,
47(4):302–314, 2012.
[6] Gail Jefferson. Glossary of transcript symbols with an introduction. In Conversation
Analysis: Studies From the First Generation, pages 13–31. John Benjamins Publishing
Company, Amsterdam, 2004.
[7] F. Erickson. Qualitative Methods in Research on Teaching. In M.Wittrock, editor, Handbook
of Research on Teaching, pages 119–161. MacMillan, New York, 1986.
[8] C. Bovill and C. J. Bulley. A model of active student participation in curriculum design: exploring
desirability and possibility. In C. Rust, editor, Improving Student Learning (ISL) 18:
Global Theories and Local Practices: Institutional, Disciplinary and Cultural Variations,
pages 176–188. Oxford Brookes University, Oxford, 2011.
[9] Charles Baily and Noah D. Finkelstein. Refined characterization of student perspectives
on quantum physics. Physical Review Special Topics – Physics Education Research,
6(2):020113, 2010.
[10] BenjaminWDreyfus, Edward F Redish, and JessicaWatkins. Student views of macroscopic
and microscopic energy in physics and biology. 1413(1):179–182, 2012.
[11] Susan Ramlo. Student perspectives on learning physics and their relationship with learning
force and motion concepts: A study using Q methodology. Journal of Human Subjectivity,
2008.
[12] Katherine K Perkins and Chandra Turpen. Student Perspectives on Using Clickers in Upperdivision
Physics Courses. pages 225–228, 2009.
263
[13] Tania S. Moraga-Calderón, Henk Buisman, and Julia Cramer. The relevance of learning
quantum physics from the perspective of the secondary school student: A case study.
European Journal of Science and Mathematics Education, 8(1), 2020.
[14] Peter Häussler and Lore Hoffman. A Curricular Frame for Physics Education: Development,
Comparison with Students’ Interests, and Impact on Students’ Achievement and
Self-Concept. pages 689–705, 2000.
[15] Dehui Hu, Benjamin M. Zwickl, Bethany R. Wilcox, and H. J. Lewandowski. Qualitative
investigation of students’ views about experimental physics. Physical Review Physics
Education Research, 13(2):020134, 2017.
[16] B¼aya Aryal, Marcia D. Nichols, and Aminul Huq. A qualitative case study exploring
student comfort with ambiguity in phyisics, math, and literature . The Online Journal of
New Horizons in Education, 8(1), 2018.
[17] Kadriye Ozen. A case study of students’ experiences in an online college physics course.
PhD thesis, 2000.
[18] Athanassios Jimoyiannis and Vassilis Komis. Computer simulations in physics teaching
and learning: a case study on students’ understanding of trajectory motion. Computers &
Education, 36(2):183–204, 2001.
[19] David Rosengrant, Alan Van Heuvelen, and Eugenia Etkina. Case Study: Students’ Use of
Multiple Representations in Problem Solving. AIP Conference Proceedings, 818(49), 2006.
[20] U. Wutchana and N. Emarat. Student effort expectations and their learning in first-year
introductory physics: A case study in Thailand. Physical Review Special Topics – Physics
Education Research, 7(1):010111, 2011.
[21] Simon Bates and Ross Galloway. The inverted classroom in a large enrollment introductory
physics course: a case study. The Higher Education Academy, 2012.
[22] Shelley Yeo, Robert Loss, Marjan Zadnik, Allan Harrison, and David Treagust. What do
students really learn from interactive multimedia? A physics case study. American Journal
of Physics, 72(10):1351–1358, 2004.
[23] Laura Lising and Andrew Elby. The impact of epistemology on learning: A case study from
introductory physics. American Journal of Physics, 73(372), 2005.
[24] NGSS Lead States. Next Generation Science Standards: For states, by states. The National
Academies Press, Washington, DC, 2013.
[25] Cynthia Bauerle, Anthony DePas, David Lynn, Clare O’Connor, Susan Singer, Michelle
Withers, CharlesW. Anderson, Sam Donovan, Shawn Drew, Diane Ebert-May, Louis Gross,
264
Sally G. Hoskins, Jay Labov, David Lopatto, Will McClatchey, Pratibha Varma-Nelson,
Nancy Pelaez, Muriel Poston, Kimberly Tanner, David Wessner, Harold White, William
Wood, and Daniel Wubah. Vision and Change. Technical report, American Association for
the Advancement of Science, Washington, DC, July 2011.
[26] Christine L. Borgman, Hal Abelson, Lee Dirks, Roberta Johnson, Kenneth R. Koedinger,
Marcia C. Linn, Clifford A. Lynch, Diana G. Oblinger, Roy D. Pea, Katie Salen, Marshall S.
Smith, and Alex Szalay. Fostering Learning in the Networked World: The Cyberlearning
Opportunity and Challenge. Technical report, National Science Foundation, 2008.
[27] AAPT Undergraduate Curriculum Task Force. AAPT Recommendations for Computational
Physics in the Undergraduate Physics Curriculum. Technical report, American Association
of Physics Teachers, 2016.
[28] Mick Healey, Abbi Flint, and Kathy Harrington. Engagement through partnership: students
as partners in learning and teaching in higher education. The Higher Education Academy,
2014.
[29] Erin Ronayne Sohr, Ayush Gupta, Brandon J. Johnson, and Gina M. Quan. Examining the
dynamics of decision making when designing curriculum in partnership with students: How
should we proceed? Physical Review Physics Education Research, 16(020157), 2020.
[30] Hannah Elizabeth Jardine. Positioning undergraduate teaching and learning assistants as
instructional partners. International Journal for Students as Partners, 4(1):48–65, 2020.
[31] Patti C. Hamerski, Paul W. Irving, and Daryl McPadden. Learning Assistants as studentpartners
in introductory physics. Physical Review Physics Education Research, in press.
[32] Lucy Mercer-Mapstone, Sam Lucie Dvorakova, Kelly E Matthews, Sophia Abbot, Breagh
Cheng, Peter Felten, Kris Knorr, Elizabeth Marquis, Rafaella Shammas, and Kelly Swaim.
A Systematic Literature Review of Students as Partners in Higher Education. International
Journal for Students as Partners, 1(1), 2017.
[33] Marcos D. Caballero, Kathi Fisler, Robert C. Hilborn, Colleen Megowan Romanowicz,
and Rebecca Vieyra. Advancing Interdisciplinary Integration of Computational Thinking in
Science. The American Association of Physics Teachers, College Park, 2020.
[34] Mel S. Sabella, Andrea G. Van Duzor, and Felicia Davenport. Leveraging the expertise
of the urban STEM student in developing an effective LA Program: LA and instructor
partnerships. In Dyan L. Jones, Lin Ding, and Adrienne Traxler, editors, Proceedings of the
Physics Education Research Conference, pages 288–291, Sacramento, 2016.
[35] Steve Cohen, Bárbara González-Arévalo, and Melanie Pivarski. Students as partners in
curricular design: Creation of student-generated calculus projects. 2018.
265
[36] Mick Healey, Abbi Flint, and Kathy Harrington. Students as Partners: Reflections on a
Conceptual Model. Teaching & Learning Inquiry, 4(2):1–13, 2016.
[37] Kelly E. Matthews. Five Propositions for Genuine Students as Partners Practice. International
Journal for Students as Partners, 1(2), 2017.
[38] Kelly E Matthews, Alexander Dwyer, Lorelei Hine, and Jarred Turner. Conceptions of
students as partners. Higher Education, 76(6):957–971, 2018.
[39] Anders Malthe-Sørenssen, Morten Hjorth-Jensen, Hans Petter Langtangen, and Knut
Mørken. Integrating Computation in the Teaching of Physics. 2016.
[40] DavidWeintrop, Elham Beheshti, Michael Horn, Kai Orton, Kemi Jona, Laura Trouille, and
Uri Wilensky. Defining Computational Thinking for Mathematics and Science Classrooms.
Journal of Science Education and Technology, 25(1):127–147, 2016.
[41] Norman Chonacky and David Winch. Integrating computation into the undergraduate curriculum:
A vision and guidelines for future developments. American Journal of Physics,
76(4):327–333, 2008.
[42] Jaime R. Taylor and B. Alex King III. Using Computational Methods to Reinvigorate
anUndergraduate Physics Curriculum. Computing in Science and Engineering, 2006.
[43] David M. Cook. Computation in undergraduate physics: The Lawrence approach. American
Journal of Physics, 76(4):321–326, 2008.
[44] Marcos D. Caballero, Norman Chonacky, Larry Engelhardt, Robert C. Hilborn, Marie Lopez
del Puerto, and Kelly R. Roos. PICUP: A Community ofTeachers Integrating Computation
into Undergraduate Physics Courses. The Physics Teacher, 57(397), 2019.
[45] Michael Quinn Patton. Qualitative Research and Evaluation Methods, chapter Qualitative
Interviewing, pages 339–428. Sage, Thousand Oaks, CA, 2002.
[46] Robert E. Stake. The art of case study research. Sage, Thousand Oaks, CA, 1995.
[47] R. K. Yin. Case study research: Design and methods. Sage, Thousand Oaks, CA, 1994.
[48] J. W. Creswell. Qualitative inquiry and research design: choosing among five approaches.
Sage, Thousand Oaks, CA, 2013.
[49] M. B. Miles and A. M. Huberman. Qualitative Data Analysis: An Expanded Sourcebook.
Sage, Thousand Oaks, CA, 1994.
[50] Anne Haas Dyson and Celia Genishi. On the Case: Approaches to Language and Literacy
Research. Teachers College Press, New York, 2005.
266
[51] S. B. Merriam. Qualitative research and case study applications in education. Jossey-Bass,
San Francisco, 1998.
[52] Pamela Baxter and Susan Jack. Qualitative Case Study Methodology: Study Design and
Implementation for Novice Researchers. The Qualitative Report, 13(4):544–559, 2008.
[53] Lesley Bartlett and Frances Vavrus. Rethinking Case Study Research. Routledge, 2016.
[54] Robert E. Stake. Multiple case study analysis. Guilford, New York, 2013.
[55] Bent Flyvbjerg. Five Misunderstandings About Case-Study Research. Qualitative Inquiry,
12(2):219–245, 2006.
[56] Kathleen M. Eisenhardt. Building theories from case study research. The Academy of
Management Review, 14(4):532–550, 1989.
[57] R. M. Sperandeo-Mineo, C. Fazio, and G. Tarantino. Pedagogical Content Knowledge
Development and Pre-Service Physics Teacher Education: A Case Study. Research in
Science Education, 36(3):235–268, 2006.
[58] StephanieV. Chasteen, Steven J. Pollock, Rachel E. Pepper, and Katherine K. Perkins. Thinking
like a physicist: A multi-semester case study of junior-level electricity and magnetism.
American Journal of Physics, 80(10):923–930, 2012.
[59] Benjamin T Spike, Noah D Finkelstein, Chandralekha Singh, Mel Sabella, and Sanjay
Rebello. Examining the Beliefs and Practice of Teaching Assistants: Two Case Studies.
pages 309–312, 2010.
[60] Patti C. Hamerski, PaulW. Irving, and Daryl McPadden. Learning Assistants as constructors
of feedback: How are they impacted? In Adrienne Traxler, Ying Cao, and Steven Wolf,
editors, Proceedings of the Physics Education Research Conference,Washington, DC, 2018.
[61] Westley James, Caroline Bustamante, Kamryn Lamons, and Jacquelyn J. Chini. Beyond Disability
asWeakness: Perspectives from Students with Disabilities. In Adrienne Traxler, Ying
Cao, and Steven Wolf, editors, 2018 Physics Education Research Conference Proceedings,
2018.
[62] Etienne Wenger. Communities of Practice. Cambridge University Press, Cambridge, 1998.
[63] Paul W. Irving, Michael J. Obsniuk, and Marcos D. Caballero. P3: A Practice Focused
Learning Environment. European Journal of Physics, 38(055701), 2017.
[64] Hilary A Dwyer, Bryce Boe, Charlotte Hill, Diana Franklin, and Danielle Harlow. Computational
Thinking for Physics: Programming Models of Physics Phenomenon in Elementary
School. In Engelhardt, Churukian, and Jones, editors, Proceedings of the Physics Education
267
Research Conference, pages 133–136, Portland, Oregon, 2013.
[65] Hayden W. Fennell, Joseph A. Lyon, Alejandra J. Magana, Sanjay Rebello, Carina M.
Rebello, and Yuri B. Peidrahita. Designing hybrid physics labs: combining simulation and
experiment for teaching computational thinking in first-year engineering. IEEE, 2019.
[66] Christopher J Burke and Timothy J Atherton. Developing a project-based computational
physics course grounded in expert practice. American Journal of Physics, 85(4):301–310,
2017.
[67] C. M. Orban, R. M. Teeling-Smith, J. R. H. Smith, and C. D. Porter. A hybrid approach
for using programming exercises in introductory physics. American Journal of Physics,
86(831), 2018.
[68] Marcos D Caballero, John B Burk, JohnMAiken, Brian D Thoms, Scott S Douglas, ErinM
Scanlon, and Michael F Schatz. Integrating Numerical Computation into the Modeling
Instruction Curriculum. The Physics Teacher, 52(38):38–42, 2014.
[69] R. Mansbach, A. Ferguson, K. Kilian, J. Krogstad, C. Leal, A. Schleife, D. R. Trinkle,
M. West, and G. L. Herman. Reforming an Undergraduate Materials Science Curriculum
with Computational Modules. Journal of Materials Education, 38(3–4), 2016.
[70] Marcos D. Caballero. Computation across the curriculum: What skills are needed? In
Alice D. Churukian, Dyan L. Jones, and Lin Ding, editors, Proceedings of the Physics
Education Research Conference, College Park, MD, 2015.
[71] Bernard G. David and Flávio S. Azevedo. A Feel for Physics: A Study of the Affective
Dimensions of Professional Physics Practice. In International Conference of the Learning
Sciences 2020 Proceedings, pages 805–806. International Society of the Learning Sciences,
2020.
[72] Ashleigh Leary, Paul W Irving, and Marcos D Caballero. The difficulties associated with
integrating computation into undergraduate physics. In Adrienne Traxler, Ying Cao, and
Steven Wolf, editors, 2018 Physics Education Research Conference Proceedings, Washington,
DC, 2018.
[73] Kathryn M. Rich, Aman Yadav, and Christina V. Schwarz. Computational Thinking, Mathematics,
and Science: Elementary Teachers’ Perspectives on Integration. Journal of Technology
and Teacher Education, 27(2), 2019.
[74] Alanna Pawlak, Paul W. Irving, and Marcos D. Caballero. Learning assistant approaches
to teaching computational physics problems in a problem-based learning course. Physical
Review Physics Education Research, 16(010139), 2020.
[75] Debbie A. French and Andrea C. Burrows. Inquiring Astronomy: Incorporating Student-
268
Centered Pedagogical Techniques in an Introductory College Science Course. Journal of
College Science Teaching, 46(7), 2017.
[76] Daniel P Weller, Theodore E Bott, Marcos D Caballero, and Paul W Irving. Developing a
learning goal framework for computational thinking in computationally integrated physics
classrooms. 2021.
[77] Douglas B. McLeod. Research on Affect in Mathematics Education: A Reconceptualization.
In Douglas A. Grouws, editor, Handbook of Research on Mathematics Teaching and
Learning, pages 575–596. Macmillan Publishing Company, New York, 1992.
[78] Gilah Leder and Peter Grootenboer. Affect and mathematics education. Mathematics
Education Research Journal, 17(2):1–8, 2005.
[79] Amy Schweinle, Debra K. Meyer, and Julianne C. Turner. Striking the Right Balance:
Students’ Motivation and Affect in Elementary Mathematics. The Journal of Educational
Research, 99(5):271–294, 2006.
[80] Markku S. Hannula. Emotions in Problem Solving. In S. J. Cho, editor, Selected Regular
Lectures from the 12th International Congress on Mathematical Education, pages 269–288.
Springer, 2015.
[81] Martina Nieswandt. Student affect and conceptual understanding in learning chemistry.
Journal of Research in Science Teaching, 44(7):908–937, 2007.
[82] Lloyd M. Mataka and Megan Grunert Kowalske. The influence of PBL on students’ selfefficacy
beliefs in chemistry. Chemistry Education Research and Practice, 2015.
[83] Ajda Kahveci. Assessing high school students’ attitudes toward chemistry with a shortened
semantic differential. Chemistry Education Research and Practice, 2015.
[84] Kelli R Galloway, Zoebedeh Malakpa, and Stacey Lowery Bretz. Investigating Affective
Experiences in the Undergraduate Chemistry Laboratory: Students’ Perceptions of Control
and Responsibility. Journal of Chemical Education, 93(2):227–238, 2015.
[85] Stacey Lowery Bretz. Novak’s Theory of Education: Human Constructivism and Meaningful
Learning. Journal of Chemical Education, 78(1107), 2001.
[86] Ayush Gupta, Andrew Elby, and Brian A. Danielak. Exploring the entanglement of personal
epistemologies and emotions in students’ thinking. Physical Review Physics Education
Research, 14(1):010129, 2018.
[87] Steve Alsop and Mike Watts. Facts and feelings: exploring the affective domain in the
learning of physics. Physics Education, 35(2), 2000.
269
[88] Stella Y. Erinosho. How Do Students Perceive the Difficulty of Physics in Secondary School?
An Exploratory Study in Nigeria. International Journal for Cross-Disciplinary Subjects in
Education, 3(3), 2013.
[89] Mustafa Fidan and Meric Tuncel. Integrating augmented reality into problem based learning:
The effects on learning achievement and attitude in physics education. Computers &
Education, 142:103635, 2019.
[90] Madelen Bodin and Mikael Winberg. Role of beliefs and emotions in numerical problem
solving in university physics education. Physical Review Special Topics – Physics Education
Research, 8(1):010108, 2012.
[91] Yorah Bosse and Marco Aurélio Gerosa. Why is programming so difficult to learn?: Patterns
of Difficulties Related to Programming Learning Mid-Stage. ACM SIGSOFT Software
Engineering Notes, 41(6):1–6, 2016.
[92] Alex Lishinski, Aman Yadav, and Richard Enbody. Students’ Emotional Reactions to Programming
Projects in Introduction to Programming: Measurement Approach and Influence
on Learning Outcomes. pages 30–38, 2017.
[93] Päivi Kinnunen and Beth Simon. Experiencing Programming Assignments in CS1: The
Emotional Toll. pages 77–85, 2010.
[94] Päivi Kinnunen and Beth Simon. CS Majors’ Self-Efficacy Perceptions in CS1: Results
in Light of Social Cognitive Theory. In Proceedings of the 2011 International Computing
Education Research Conference, pages 19–26, Providence, 2011.
[95] Päivi Kinnunen and Beth Simon. My program is ok – am I? Computing freshmen’s experiences
of doing programming assignments. Computer Science Education, 22(1):1–28,
2012.
[96] Valerie Otero, Steven Pollock, andNoah Finkelstein. Aphysics department’s role in preparing
physics teachers: The Colorado Learning Assistant model. American Journal of Physics,
78(11):1218–1224, 2010.
[97] Louis S. Nadelson and Jillana Finnegan. A path less traveled: Fostering STEM majors’ professional
identity development through engagement as STEM Learning Assistants. Journal
of Higher Education Theory and Practice, 14(5), 2014.
[98] Renee Michelle Goertzen, Eric Brewe, Laird H. Kramer, Leanne Wells, and David Jones.
Moving toward change: Institutionalizing reform through implementation of the Learning
Assistant model and Open Source Tutorials. Physical Review Special Topics – Physics
Education Research, 7(020105), 2011.
[99] Eleanor W. Close, Jessica Conn, and Hunter G. Close. Becoming physics people: Devel-
270
opment of integrated physics identity through the Learning Assistant experience. Physical
Review Physics Education Research, 12(010109), 2016.
[100] Jessica Conn, EleanorW. Close, and Hunter G. Close. Learning Assistant Identity Development:
Is One Semester Enough? In Paula V. Engelhardt, Alice D. Churukian, and Dyan L.
Jones, editors, Proceedings of the Physics Education Research Conference, pages 55–58,
Minneapolis, 2014.
[101] Eleanor W. Close, Jessica Conn, and Hunter G. Close. Learning Assistants’ Development
of Physics (Teacher) Identity. In Paula V. Engelhardt, Alice D. Churukian, and Dyan L.
Jones, editors, Proceedings of the Physics Education Research Conference, pages 89–92,
Minneapolis, 2014.
[102] Daniel Caravez, Angelica De La Torre, Jayson M Nissen, and Ben Van Dusen. Longitudinal
associations between Learning Assistants and instructor effectiveness. In Lin Ding, Adrienne
Traxler, and Ying Cao, editors, Proceedings of the Physics Education Research Conference,
pages 80–83, Cincinnati, 2017.
[103] Ben Van Dusen and Jayson M Nissen. Systemic inequities in introductory physics courses:
the impacts of learning assistants. 2017.
[104] Amy L. Zeldin and Frank Pajares. Against the Odds: Self-Efficacy Beliefs of Women
in Mathematical, Scientific, and Technological Careers. American Educational Research
Journal, 37(1):215–246, 2000.
[105] Vashti Sawtelle, Eric Brewe, Renee Michelle Goertzen, and Laird H. Kramer. Identifying
events that impact self-efficacy in physics learning. Physical Review Special Topics – Physics
Education Research, 8(020111), 2012.
[106] Michael J. Scott and Gheorghita Ghinea. On the Domain-Specificity of Mindsets: The
Relationship Between Aptitude Beliefs and Programming Practice. IEEE Transactions on
Education, 57(3):169–174, 2014.
[107] Quintin Cutts, Emily Cutts, Stephen Draper, Patrick O’Donnell, and Peter Saffrey. Manipulating
mindset to positively influence introductory programming performance. In SIGCSE,
pages 431–435, Milwaukee, 2010.
[108] Michael Lodi. Does Studying CS Automatically Foster a Growth Mindset? In Innovation
and Technology in Computer Science Education, pages 147–153, Aberdeen, Scotland, 2019.
[109] Lorelle L. Espinosa. The academic self-concept of African American and Latina(o) men
and women in STEM majors. Journal ofWomen and Minorities in Science and Engineering,
14:177–203, 2008.
[110] Arifah Mardiningrum. EFL Self-Concept in an English Drama Club: A Case Study of Two
271
English Language Education Department Students. Advances in Social Science, Education,
and Humanities Research, 353:15–21, 2019.
[111] Xia Chen and Hao Xu. Understanding Interaction Inside and Outside the EFL Self-Concept
System: A Case Study of Chinese Learners in Junior High School. Chinese Journal of
Applied Linguistics, 38(2):133–149, 2015.
[112] Albert Bandura. Self-Efficacy in Changing Societies. Cambridge University Press, New
York, 1995.
[113] Albert Bandura. Self-efficacy mechanism in human agency. American Psychologist,
37(2):122–147, 1982.
[114] Sarantos Psycharis, Konstantinos Kalovrektis, Apostolos Xenakis, Ioannis Paliokas,
Matthaios Patrinopoulos, Petros Georgiakakis, Paraskevi Iatrou, Paraskevi Theodorou,
Theodora Papageorgiou, and Vasiliki Ntourou. The Impact of Physical Computing and
Computational Pedagogy on Girl’s Self–Efficacy and Computational Thinking Practice.
2021 IEEE Global Engineering Education Conference (EDUCON), 00:308–315, 2021.
[115] Su Cai, Changhao Liu, Tao Wang, Enrui Liu, and Jyh-Chong Liang. Effects of learning
physics using Augmented Reality on students’ self-efficacy and conceptions of learning.
British Journal of Educational Technology, 52(1):235–251, 2021.
[116] Russell Feldhausen, Joshua Levi Weese, and Nathan H Bean. Increasing Student Self-
Efficacy in Computational Thinking via STEM Outreach Programs. In Proceedings of the
49th ACM Technical Symposium on Computer Science Education, pages 302–307, 2018.
[117] Todd R. Kelley, J. Geoffery Knowles, Jeffrey D. Holland, and Jung Han. Increasing high
school teachers self-efficacy for integrated STEM instruction through a collaborative community
of practice. International Journal of STEM Education, 7(1):14, 2020.
[118] Herbert W. Marsh and Rhonda Craven. Academic Self-Concept: Beyond the Dustbowl. In
Handbook of Classroom Assessment, pages 131–198. Academic Press, 1997.
[119] Richard J. Shavelson, Judith J. Hubner, and George C. Stanton. Self-Concept: Validation of
Construct Interpretations. Review of Educational Research, 46(3):407–441, 1976.
[120] Martin Brunner, Ulrich Keller, Caroline Hornung, Monique Reichert, and Romain Martin.
The cross-cultural generalizability of a new structural model of academic self-concepts.
Learning and Individual Differences, 19(4):387–403, 2009.
[121] Catherine Good, Aneeta Rattan, and Carol S. Dweck. Why Do Women Opt Out? Sense of
Belonging and Women’s Representation in Mathematics. Journal of Personality and Social
Psychology, 102(4):700–717, 2012.
272
[122] Angela J. Little, Bridget Humphrey, Abigail Green, Abhilash Nair, and Vashti Sawtelle.
Exploring mindset’s applicability to students’ experiences with challenge in transformed
college physics courses. Physical Review Physics Education Research, 15(1):010127, 2019.
[123] Jared Washburn. The Effects of a Coding Curriculum on Freshman Physics Students. PhD
thesis, 2017.
[124] Jeni L. Burnette, Michelle V. Russell, Crystal L. Hoyt, Kasey Orvidas, and Laura Widman.
An online growth mindset intervention in a sample of rural adolescent girls. British Journal
of Educational Psychology, 88(3):428–445, 2018.
[125] Jessica Schleider and John Weisz. A single-session growth mindset intervention for adolescent
anxiety and depression: 9-month outcomes of a randomized trial. Journal of Child
Psychology and Psychiatry, 59(2):160–170, 2018.
[126] Lisa Brougham and Susan Kashubeck-West. Impact of a Growth Mindset Intervention
on Academic Performance of Students at Two Urban High Schools. Professional School
Counseling, 21(1), 2017.
[127] Teresa K. DeBacker, Benjamin C. Heddy, Julianna Lopez Kershen, H. Michael Crowson,
Kristyna Looney, and Jacqueline A. Goldman. Effects of a one-shot growth mindset intervention
on beliefs about intelligence and achievement goals. Educational Psychology,
38(6):1–23, 2018.
[128] Angela Fink, Michael J. Cahill, Mark A. McDaniel, Arielle Hoffman, and Regina F. Frey.
Improving general chemistry performance through a growth mindset intervention: selective
effects on underrepresented minorities. Chemistry Education Research and Practice,
19(3):783–806, 2018.
[129] Jeannette M. Wing. Computational thinking. Communications of the ACM, 49(3):33–35,
2006.
[130] Stefania Bocconi, Augusto Chioccariello, Giuliana Dettori, Anusca Ferrari, and Katja Engelhardt.
Developing Computational Thinking in Compulsory Education. Technical report,
Joint Research Center, Seville, 2016.
[131] Mahsa Mohaghegh and Michael McCauley. Computational Thinking: The Skill Set of the
21st Century. International Journal of Computer Science and Information Technologies,
7(3):1524–1530, 2016.
[132] Operational definition of computational thinking for K–12 education. Technical report,
International Society for Technology in Education and the Computer Science Teachers
Association, 2011.
[133] Pratim Sengupta, John S. Kinnebrew, Satabdi Basu, Gautam Biswas, and Douglas Clark.
273
Integrating computational thinking with K-12 science education using agent-based computation:
A theoretical framework. Education and Information Technologies, 18(2):351–380,
2013.
[134] K. M. Rich, A. Yadav, and R. A. Larimore. Teacher implementation profiles for integrating
computational thinking into elementary mathematics and science instruction. Education and
Information Technologies, 25(3161), 2020.
[135] Gary Thomas. Doing Case Study: Abduction Not Induction, Phronesis Not Theory. Qualitative
Inquiry, 16(7):575–582, 2010.
[136] C. Reismann. Narrative methods for the human sciences. Sage, Thousand Oaks, CA, 2008.
[137] Robert Q. III Berry, Kateri Thunder, and Oren L. McClain. Counter Narratives: Examining
the Mathematics and Racial Identities of Black Boys who are Successful with School
Mathematics. Journal of African American Males in Education, 2(1), 2011.
[138] Sine Agergaard. Religious culture as a barrier? A counter-narrative of Danish Muslim girls’
participation in sports. Qualitative Research in Sport, Exercise and Health, 8(2):213–224,
2015.
[139] Robert Klonoski. The Case For Case Studies: Deriving Theory From Evidence. Journal of
Business Case Studies, 9(3), 2013.
[140] David L. Morgan. From themes to hypotheses: Following up with quantitative methods.
Qualitative Health Research, 25(6), 2015.
[141] Leonard Webster and Patricie Mertova. Using Narrative Inquiry as a Research Method: Introduction
to Using Critical Event Narrative Analysis in Research on Learning and Teaching.
Routledge, London, 2007.
[142] Trista Hollweck. Book Review: Case Study Research Design and Methods. Canadian
Journal of Program Evaluation, 30(1), 2015.
[143] Sarah C. Young. Reframing metalinguistic awareness for low-literate L2 learners: Four
case studies. PhD thesis, 2016.
[144] John R. Gallagher. A Framework for Internet Case Study Methodology in Writing Studies.
Computers and Composition, 54:102509, 2019.
[145] Michael Macaluso. Reading pedagogy-as-text: Exploring gendered discourses as canonical
in an English classroom. Linguistics and Education, 35:15–25, 2016.
[146] C. Wilkinson. Ethnographic methods. In Laura J. Shepherd, editor, Critical Approaches
to Security: An Introduction to Theories and Methods, pages 129–145. Routledge, London,
274
2013.
[147] Gail Prasad. Children as co-ethnographers of their plurilingual literacy practices: An
exploratory case study. Language and Literacy, 15(3), 2013.
[148] Julie Katherine Bannerman. Singing in School Culture: An Ethnographic Case Study of a
Secondary Choral Program. PhD thesis, Northwestern University, Evanston, IL, 2016.
[149] M. J. Volkmann and M. Zgagacz. Learning to teach physics through inquiry: The lived
experience of a graduate teaching assistant. Journal of Research in Science Teaching, 41(6),
2004.
[150] S. Sandi-Urena, M. M. Cooper, and T. A. Gatlin. Graduate teaching assistants’ epistemological
and metacognitive development. Chemistry Education Research and Practice, 12(1),
2011.
[151] P. W. Irving, V. Sawtelle, and M. D. Caballero. Troubleshooting formative feedback in p3
(projects and practices in physics). In A. D. Churukian, D. L. Jones, and L. Ding, editors,
2015 Physics Education Research Conference Proceedings, College Park, MD, 2015.
[152] M. Stjelja. The Case Study Approach: Some Theoretical, Methodological and Applied
Considerations. Land Operations Division DSTO Defence Science and Technology Organisation,
Edinburgh South Australia, 2013.
[153] A. Cliffe, A. Cook-Sather, M. Healey, R. Healey, E. Marquis, K.E. Matthews, L. Mercer-
Mapstone, A. Ntem, V. Puri, and C. Woolmer. Launching a journal about and through
Students as Partners. International Journal for Students as Partners, 1(1), 2017.
[154] Catherine Bovill. The Student Engagement Handbook: Practice in Higher Education. In The
Student Engagement Handbook: Practice in Higher Education, pages 461–476. Emerald
Group Publishing Limited, 2013.
[155] Abbi Flint and Mark O’Hara. Communities of practice and ‘student voice’: engaging with
student representatives at the faculty level. Student Engagement and Experience Journal,
2(1), 2013.
[156] Ying Cao, Christina Smith, Ben Lutz, and Milo Koretskey. Cultivating the next generation:
Outcomes from a Learning Assistant program in engineering. In American Society for
Engineering Education Annual Conference & Exposition, Salt Lake City, 2018.
[157] Robert J. Beichner, Jeffery M. Saul, David S. Abbott, Jeanne J. Morse, Duane L. Deardorff,
Rhett J. Allain, Scott W. Bonham, Melissa H. Dancy, and John S. Risley. The studentcentered
activities for large enrollment undergraduate programs (SCALE-UP) project. In
E. F. Redish and P. J. Cooney, editors, Reviews in PER Volume 1: Research-Based Reform
of University Physics, College Park, 2007. American Association of Physics Teachers.
275
[158] Eric Brewe. Modeling theory applied: Modeling Instruction in introductory physics. American
Journal of Physics, 76(12), 2008.
[159] E. Etkina and A Van Heuvelen. Investigative science learning environment – A science
process approach to learning physics. In E. F. Redish and P. J. Cooney, editors, Reviews in
PER Volume 1: Research-Based ReformofUniversity Physics, College Park, 2007. American
Association of Physics Teachers.
[160] L. C. McDermott. Physics by Inquiry. John Wiley and Sons, New York, 1996.
[161] EleanorW. Close, Jean-Michel Mailloux-Huberdeau, Hunter G. Close, and David Donnelly.
Characterization of time scale for detecting impacts of reforms in an undergraduate physics
program. In Lin Ding, Adrienne Traxler, and Ying Cao, editors, Proceedings of the Physics
Education Research Conference, pages 88–91, Cincinnati, 2017.
[162] Ben Van Dusen and Jayson M. Nissen. Systemic inequities in introductory physics courses:
The impacts of Learning Assistants. In Lin Ding, Adrienne Traxler, and Ying Cao, editors,
Proceedings of the Physics Education Research Conference, pages 400–403, Cincinnati,
2017.
[163] Nadia Sellami, Shanna Shaked, Frank A. Laski, Kevin M. Eagan, and Erin R. Sanders.
Implementation of a Learning Assistant program improves student performance on higherorder
assessments. CBE–Life Sciences Education, 16(ar62):1–10, 2017.
[164] G. Begley, R. Berkey, L. Roe, and H. Schuldt. Becoming partners: Faculty come to appreciate
undergraduates as teaching partners in a service-learning teaching assistant program.
International Journal for Students as Partners, 3(1), 2019.
[165] F. Daniello and C. Acquaviva. A Faculty member learning with and from an undergraduate
teaching assistant: Critical reflection in higher education. International Journal for Students
as Partners, 3(2), 2019.
[166] Paul W. Irving, Daryl McPadden, and Marcos D. Caballero. Communities of Practice as a
curriculum design theory in an introductory physics class for engineers. Physical Review
Physics Education Research, 16(020143), 2020.
[167] E.Wenger, R. A. McDermott, andW. Snyder. Cultivating communities of practice: A guide
to managing knowledge. Harvard Business School Press, Boston, 2002.
[168] Jean Lave and Etienne Wenger. Situated Learning: Legitimate Peripheral Participation.
1991.
[169] A. Cook-Sather. Students as learners and teachers: college faculty and undergraduates
co-create a professional development model. To Improve the Academy, 29:219–232, 2011.
276
[170] C. Bovill, A. Cook-Sather, and P. Felten. Students as co-creators of teaching approaches,
course design, and curricula: implications for academic developers. International Journal
for Academic Development, 16(2):133–145, 2011.
[171] National Research Council. A Framework for K-12 Science Education: Practices, Crosscutting
Concepts, and Core Ideas. National Academies Press, 2011.
[172] Cynthia C. Selby and JohnWoollard. Computational Thinking: The Developing Definition.
2010.
[173] Operationalizing relevance in physics education: Using a systems viewto expand our conception
of making physics relevant. Physical Review Physics Education Research, 15(020121),
2019.
[174] Shulamit Kapon, Antti Laherto, and Olivia Levrini. Disciplinary authenticity and personal
relevance in school science. Science Education, 102:1077–1106, 2018.
[175] Scratch. https://scratch.mit.edu, 2007.
[176] David Scherer. Vpython. https://vpython.org, 2000.
[177] Anabela Gomes and A. J. Mendes. Learning to program – difficulties and solutions. In
Proceedings of the International Conference on Engineering Education, Coimbra, 2007.
[178] Tony Jenkins. On the Difficulty of Learning to Program. In 3rd Annual LTSN-ICS Conference
Proceedings, pages 53–58, Loughbourough, UK, 2002.
[179] Lauri Malmi, Judy Sheard, Päivi Kinnunen, Simon, and Jane Sinclair. Theories and Models
of Emotions, Attitudes, and Self-Efficacy in the Context of Programming Education. In
Proceedings of the 2020 International Computing Education Research Conference, pages
36–47, Virtual Event, New Zealand, 2020.
[180] Anna Eckerdal, Robert McCartney, Jan Erik Moström, Kate Sanders, Lynda Thomas, and
Carol Zander. From Limen to Lumen: Computing students in liminal spaces. pages 123–132,
2007.
[181] Satabdi Basu, Gautam Biswas, Pratim Sengupta, Amanda Dickes, John S. Kinnebrew, and
Douglas Clark. Identifying middle school students’ challenges in computational thinkingbased
science learning. Research and Practice in Technology Enhanced Learning, 11(13),
2016.
[182] Camilo Vieira, Alejandra J. Magana, Anindya Roy, Michael L. Falk, and Michael J. Reese
Jr. Exploring Undergraduate Students’ Computational Literacy in the Context of Problem
Solving. Computers in Education Journal, 2016.
277
[183] Jannis Weber and Thomas Wilhelm. The benefit of computational modelling in physics
teaching: a historical overview. European Journal of Physics, 41(034003), 2020.
[184] Kim Svensson, Urban Eriksson, and Ann-Marie Pendrill. Programming and its affordances
for physics education: A social semiotic and variation theory approach to learning physics.
Physical Review Physics Education Research, 16(010127), 2020.
[185] Ruxandra M Serbanescu, Paul J Kushner, and Sabine Stanley. Putting computation on a par
with experiments and theory in the undergraduate physics curriculum. American Journal of
Physics, 79(9):919–924, 2011.
[186] David Scherer and Bruce Sherwood. Glowscript. https://www.glowscript.org, 2011.
[187] Shawn A. Weatherford. Student Use of Physics to Make Sense of Incomplete but Functional
VPython Programs in a Lab Setting. PhD thesis, North Caroline State University, Raleigh,
NC, 2011.
[188] Ruth E.S. Allen and Janine L. Wiles. A rose by any other name: participants choosing
research pseudonyms. Qualitative Research in Psychology, 13(2):149–165, 2015.
[189] Maria K. E. Lahman, Katrina L. Rodriguez, Lindsey Moses, Krista M. Griffin, Bernadette M.
Mendoza, and Wafa Yacoub. A Rose By Any Other Name Is Still a Rose? Problematizing
Pseudonyms in Research. Qualitative Inquiry, 21(5):445–453, 2015.
[190] Marja Kuzmanic. Validity in qualitative research: Interview and the appearance of truth
through dialogue. Horizons of Psychology, 18(2), 2009.
[191] National Center for Science and Engineering Statistics. Science & Engineering Indicators
2020. Technical report, National Science Board, 2019.
[192] Lori Kendall. “White and Nerdy”: Computers, Race, and the Nerd Stereotype. Journal of
Popular Culture, 44:505–524, 2011.
[193] Ravinder Koul, Thanita Lerdpornkulrat, and Chanut Poondej. Gender compatibility, mathgender
stereotypes, and self-concepts in math and physics. Physical Review Physics Education
Research, 12(020115), 2016.
[194] Kevin D. Finson. Drawing a Scientist: What We Do and Do Not Know After Fifty Years of
Drawings. School Science and Mathematics, 102(7), 2002.
[195] Marcos D. Caballero, Matthew A. Kohlmyer, and Michael F. Schatz. Implementing and
assessing computational modeling in introductory mechanics. Physical Review Special
Topics - Physics Education Research, 8(020106), 2012.
[196] Mica A. Hutchison-Green, Deborah K. Follman, and George M. Bodner. Providing a voice:
278
Qualitative investigation of the impact of a first-year engineering experience of students’
efficacy beliefs. Journal of Engineering Education, 97(2), 2008.
[197] Colleen M. Lewis, Ken Yasuhara, and Ruth E. Anderson. Deciding to Major in Computer
Science: A Grounded Theory of Students’ Self-Assessment of Ability. In ICER, pages 3–10,
Providence, RI, 2011.
[198] Colleen M. Lewis, Ruth E. Anderson, and Ken Yasuhara. “I Don’t Code All Day”: Fitting
in Computer Science When the Stereotypes Don’t Fit. In ICER, pages 23–32, Melbourne,
VIC, Australia, 2016.
[199] Jacqueline N. Bumler, Patti C. Hamerski, Marcos D. Caballero, and PaulW. Irving. How do
previous coding experiences influence undergraduate physics students? In Ying Cao, Steven
Wolf, and Michael B. Bennett, editors, Proceedings of the Physics Education Research
Conference, Provo, UT, 2019.
[200] Marcos D. Caballero and Laura Merner. Prevalence and nature of computational instruction
in undergraduate physics programs across the United States. Physical Review Physics
Education Research, 14(2):020129, 2018.
[201] Shuchi Grover and Roy Pea. Computational Thinking in K–12. Educational Researcher,
42(1):38–43, 2012.
[202] B. V. Barr and C. Stephenson. Bringing computational thinking to K-12: What is involved
and what is the role of the computer science community? ACM Inroads, 2(48), 2011.
[203] A. Yadav, J. Good, J. Voogt, and P. Fisser. Computational thinking as an emerging competence
domain. In M. Mulder, editor, Competence-based vocational and professional education:
Bridging the worlds of work and education, page 1051–1067. Springer International,
Cham, Switzerland, 2016.
[204] Luke D. Conlin, Nicole M. Hutchins, Shuchi Grover, and Gautam Biswas. “Doing Physics”
And “Doing Code”: Students’ Framing During Computational Modeling in Physics. In
International Conference of the Learning Sciences, Nashville, 2020.
[205] Karen Brennan and Mitchel Resnick. New frameworks for studying and assessing the
development of computational thinking. In American Education Research Association,
2012.
[206] Daniel P Weller, Marcos D Caballero, and Paul W Irving. Teachers’ intended learning
outcomes around computation in high school physics. 2019 Physics Education Research
Conference Proceedings, 2019.
[207] David Williamson Shaffer and Mitchel Resnick. “Thick” Authenticity: New Media and
Authentic Learning. Journal of Interactive Learning Research, 10(2):195–215, 1999.
279
[208] Angela Little, Vashti Sawtelle, and Bridget Humphrey. Mindset in Context: Developing
New Methodologies to Study Mindset in Interview Data. In Dyan L. Jones, Lin Ding, and
Adrienne Traxler, editors, 2016 Physics Education Research Conference Proceedings, pages
204–207, Sacramento, 2016.
[209] D. N. Perkins, Eileen Jay, and Shari Tishman. Beyond abilities: A dispositional theory of
thinking. Merrill-Palmer Quarterly, 39(1):1–21, 1993.
[210] Carol S. Dweck. The role of expectations and attributions in the alleviation of learned
helplessness. Journal of Personality and Social Psychology, 31(4):674–685, 1975.
[211] Junfeng Zhang, Elina Kuusisto, and Kirsi Tirri. How Teachers’ and Students’ Mindsets in
Learning Have Been Studied: Research Findings on Mindset and Academic Achievement.
Psychology, 08(09):1363–1377, 2017.
[212] Jeni L. Burnette, Ernest O’Boyle, Eric M. VanEpps, Jeffrey M. Pollack, and Eli J. Finkel.
Mindsets Matter: A Meta-Analytic Review of Implicit Theories and Self-Regulation. Psychological
Bulletin, 2012.
[213] Eamon Tewell. The Problem with Grit: Dismantling Deficit Thinking in Library Instruction.
portal: Libraries and the Academy, 20(1):137–159, 2020.
[214] Ellen L. Usher. Acknowledging the Whiteness of Motivation Research: Seeking Cultural
Relevance. Educational Psychologist, 53(2):131–144, 2018.
[215] Victoria F. Sisk, Alexander P. Burgoyne, Jingze Sun, Jennifer L. Butler, and Brooke N. Macnamara.
To What Extent and Under Which Circumstances Are Growth Mind-Sets Important
to Academic Achievement? Two Meta-Analyses. Psychological Science, 29(4):549–571,
2018.
[216] Štepán Bahník and Marek A. Vranka. Growth mindset is not associated with scholastic
aptitude in a large sample of university applicants. Personality and Individual Differences,
117:139–143, 2017.
[217] Richard R. Valencia, editor. The evolution of deficit thinking: Educational thought and
practice. The Falmer Press, London, 1997.
[218] Sal Khan. Khan academy. https://www.khanacademy.org, 2007.
[219] R. M. Emerson, R. I. Fretz, and L. L. Shaw. Writing Ethnographic Fieldnotes. University of
Chicago Press, Chicago, 2011.
[220] Cliodhna O’Connor and Helene Joffe. Intercoder Reliability in Qualitative Research: Debates
and Practical Guidelines. International Journal of Qualitative Methods, 19:1–13,
2020.
280
[221] L. A. Wood and R. O. Kroger. Doing Discourse Analysis, chapter Warranting in Discourse
Analysis, pages 163–178. Sage, Thousand Oaks, CA, 2000.
[222] E. F. Redish. Introducing students to the culture of physics: Explicating elements of the
hidden curriculum. In Chandralekha Singh, N. Sanjay Rebello, and Mel Sabella, editors,
2010 Physics Education Research Conference Proceedings, Portland, OR, 2010.
[223] Casey E. Davenport. Evolution in Student Perceptions of a Flipped Classroom in a Computer
Programming Course. Journal of College Science Teaching, 47(4), 2018.
[224] Aman Yadav, Hai Hong, and Chris Stephenson. Computational Thinking for All: Pedagogical
Approaches to Embedding 21st Century Problem Solving in K-12 Classrooms.
TechTrends, 60(6):565–568, 2016.
[225] J. Voogt, P. Fisser, J. Good, P. Mishra, and A. Yadav. Computational thinking in compulsory
education: Towards an agenda for research and practice. Education and Information
Technologies, 20(4):715–728, 2015.
[226] Lev S. Vygotsky. Mind in society: The development of higher psychological processes.
Harvard University Press, Massachusetts, 1978.
281