THE EFFECTS OF COMMUNITY OF INQUIRY, LEARNING PRESENCE, AND MENTOR 

PRESENCE ON LEARNING OUTCOMES: A REVISED COMMUNITY OF INQUIRY 

MODEL FOR K-12 ONLINE LEARNERS 

 

 

 

By 

Yining Zhang 

 

 

 

 

 

 

 

 

A DISSERTATION 

Submitted to 

Michigan State University 

in partial fulfillment of the requirements  

for the degree of 

Educational Psychology and Educational Technology—Doctor of Philosophy 

2018 

 

ABSTRACT 

THE EFFECTS OF COMMUNITY OF INQUIRY, LEARNING PRESENCE, AND 
MENTOR PRESENCE ON LEARNING OUTCOMES: A REVISED COMMUNITY 

OF INQUIRY MODEL FOR K-12 ONLINE LEARNERS 

By 

Yining Zhang 

Amid an explosive increase in K-12 online education in the United States, the quality of 

online learning has become a primary concern for researchers, educators, and policymakers. Two 

theoretical frameworks, Community of Inquiry (CoI) and self-regulated learning (SRL) have 

provided especially insightful explanations of how students learn in online settings. The present 

dissertation proposes a revised CoI framework that incorporates learning presence (i.e., self-

efficacy and SRL strategies) and mentor presence and connects to learning outcomes in K-12 

online setting from a sample of 696 high-school level online learners. The study yields four key 

findings: First, in contrast to Shea and Bidjerano’s (2010) model, this study found a significant 

relationship between teaching presence and SRL, and a non-significant relationship between 

teaching presence and self-efficacy. Second, because learning under the supervision of a mentor 

is a unique feature of K-12 online learning, this study found that mentor presence significantly 

predicted students’ use of self-regulated learning strategies. Third, when using the revised 

framework to predict learning outcomes (i.e., satisfaction, perceived progress, final grade), it was 

found that the hypothesized model containing CoI and SRL can be related to learning outcomes, 

thus answering previous scholars’ calls for a realistic integration of learning presence and 

learning outcomes into the CoI framework. And fourth, based on a comparison of two groups of 

students with different primary online learning locations (i.e., at-home vs. at-school), the study 

found that at-school students showed significantly less feelings of isolation, higher ability in 

 

generating curiosity once the online learning starts, higher perceptions on mentor’s practice as a 

problem solver, and higher goal-setting and help-seeking strategies than those whose primary 

online learning location is at home. 

In short, this study is among the first to shed light on the relationships among CoI, self-

efficacy, SRL, and learning outcomes in a K-12 online-learning context. Its findings establish the 

capacity of the proposed theoretical framework to identify important components in K-12 online 

learning while striking a delicate balance between extensiveness and parsimony. This study also 

extends our understanding of the mechanisms of online learning among K-12 students, and thus 

has considerable practical implications for online educators as well as future researchers. 

Specific recommendations for future research projects are also provided.

 

To my late grandfather, who taught me the value of education. 

To my parents, who have loved and supported me unconditionally throughout my life. 

To my husband, who encourages me with steadfast love, care, and understanding. 

And to my daughter, who inspires me to be more than what I was the day before: May you 

understand the journey to knowledge never ends.

 

iv 

ACKNOWLEDGEMENTS 

 

First and foremost, I would like to express my deepest appreciation to my dissertation 

committee chair and advisor, Dr. Chin-Hsi Lin, for his continuous support during my doctoral 

studies. He stimulated my research interest in online learning, and instilled in me – by his own 

example – a love of and dedication to rigorous research and scholarship. Without Dr. Lin’s 

guidance and frequent help, this dissertation would never have been completed. I would also like 

to thank my other committee members: Dr. Dongbo Zhang, for his support and mentorship as I 

pursued my interest in self-regulated learning; Dr. Lisa Linnenbrink-Garcia, for her insightful 

suggestions on developing my research and conducting statistical analyses; and Dr. Rand Spiro, 

for his care and encouragement throughout my doctoral work. 

I would also like to extend my appreciation to Michigan Virtual Learning Research 

Institute for allowing me to conduct my dissertation research and providing me with assistance 

whenever I requested it. In this regard, special thanks are due to Dr. Joe Freidhoff, Dr. Kathryn 

Kennedy, Dr. Jungah Bae, Dr. Kristen DeBruler, and the members of the Blackboard team for 

their steady support, and to my participants for their enthusiasm. 

I would like to thank my editor, Dr. Dan MacCannell, for his reliable editing and 

proofreading of my papers throughout my doctoral studies. 

I am also thankful to the entire EPET program at Michigan State University: all the 

faculty and staff that I worked with, took classes from, and met throughout my doctoral studies, 

all of my fellow doctoral students. Thanks for offering me an opportunity to start a wonderful 

journey, and for your encouragement and help along the way. 

 

v 

Last but not least, I would like to thank my family for supporting me with love, care, 

understanding, encouragement, and inspiration throughout my life. Thanks for always being 

there for me. Thanks for believing in me. Thanks for loving me.

 

vi 

TABLE OF CONTENTS 

 

LIST OF TABLES .......................................................................................................................... x 

LIST OF FIGURES ...................................................................................................................... xii 

CHAPTER 1 ................................................................................................................................... 1 
INTRODUCTION .......................................................................................................................... 1 

CHAPTER 2 ................................................................................................................................... 8 
LITERATURE REVIEW ............................................................................................................... 8 
Community of Inquiry ................................................................................................................ 8 
Teaching Presence. ................................................................................................................. 8 
Social Presence ..................................................................................................................... 10 
Cognitive Presence................................................................................................................ 12 
Self-regulated Learning ............................................................................................................ 13 
Mentors in K-12 Online Courses .............................................................................................. 17 
Mentor Presence.................................................................................................................... 17 
The Impact of Mentor Presence ............................................................................................ 18 
Learning Locations for K-12 Online Learners .......................................................................... 20 
Research Questions ................................................................................................................... 22 

CHAPTER 3 ................................................................................................................................. 27 
METHODS ................................................................................................................................... 27 
Context ...................................................................................................................................... 27 
Participants ................................................................................................................................ 27 
Procedure .................................................................................................................................. 29 
Survey ....................................................................................................................................... 30 
Demographic Information ..................................................................................................... 30 
Community of Inquiry .......................................................................................................... 31 
Teaching Presence ............................................................................................................ 31 
Social Presence ................................................................................................................. 31 
Cognitive Presence............................................................................................................ 32 
Self-efficacy .......................................................................................................................... 32 
Online Self-regulated Learning Strategies ............................................................................ 32 
Mentor Presence.................................................................................................................... 33 
Satisfaction ............................................................................................................................ 33 
Perceived Progress ................................................................................................................ 33 
Data Analysis ............................................................................................................................ 34 
Instrument Validation ........................................................................................................... 35 
Exploratory Factor Analysis ............................................................................................. 35 
Confirmatory Factor Analysis ........................................................................................... 36 
Structural Model ................................................................................................................... 38 
Testing Mediation ............................................................................................................. 38 

 

vii 

RQ1. Shea’s Model and the Alternative Model .................................................................... 38 
RQ2. Model with Mentor Presence ...................................................................................... 39 
RQ3. Model with Learning Outcomes .................................................................................. 39 
RQ4. Learning at school vs. Learning at Home .................................................................... 39 

CHAPTER 4 ................................................................................................................................. 40 
RESULTS ..................................................................................................................................... 40 
Exploratory Factor Analysis ..................................................................................................... 40 
Confirmatory Factor Analysis ................................................................................................... 44 
Composite Reliability ........................................................................................................... 46 
Convergent Validity .............................................................................................................. 47 
Factor Loading .................................................................................................................. 47 
Average Variance Extracted ............................................................................................. 47 
Discriminant Validity................................................................................................................ 48 
Descriptive Statistics ................................................................................................................. 48 
Measurement Model ................................................................................................................. 49 
RQ1. Shea’s Model and the Alternative Model ........................................................................ 52 
RQ2. Model with Mentor Presence .......................................................................................... 56 
RQ3. Model with Learning Outcomes – Satisfaction ............................................................... 59 
RQ3. Model with Learning Outcomes – Perceived Progress ................................................... 64 
RQ3. Model with Learning Outcomes – Final Grade ............................................................... 67 
RQ4. Studying at Home or at School........................................................................................ 70 

CHAPTER 5 ................................................................................................................................. 75 
DISCUSSION ............................................................................................................................... 75 
Comparing Shea’s Model against the Alternative Model ......................................................... 77 
Teaching Presence and Self-regulated Learning Strategies .................................................. 77 
Teaching Presence and Self-efficacy .................................................................................... 78 
Mentor Presence in K-12 Online Learning ............................................................................... 80 
Mentor Presence.................................................................................................................... 80 
Mentor Presence, Self-efficacy, and Self-regulated Learning Strategies ............................. 81 
Connecting the CoI Models with Learning Outcomes ............................................................. 83 
Incorporating Learning Presence into the CoI Model ........................................................... 83 
Distinctive Learning Outcomes ............................................................................................ 85 
Learning At-home vs. Learning At-school ............................................................................... 86 
Feeling of Isolation ............................................................................................................... 86 
Triggering Event in Cognitive Presence ............................................................................... 88 
Perceived Mentor’s Help ...................................................................................................... 89 
Goal-setting & Help-seeking Learning Strategies ................................................................ 90 
Limitations and Recommendations for Future Research .......................................................... 90 
Implications for Practitioners .................................................................................................... 92 
Learning Presence ................................................................................................................. 92 
Mentor Presence.................................................................................................................... 93 
Online Learning at School .................................................................................................... 94 

CHAPTER 6 ................................................................................................................................. 96 

 

viii 

CONCLUSION ............................................................................................................................. 96 

APPENDICES .............................................................................................................................. 98 
APPENDIX A: Parental Consent Form .................................................................................... 99 
APPENDIX B: Student Assent Form ..................................................................................... 101 
APPENDIX C: Survey ............................................................................................................ 103 

REFERENCES ........................................................................................................................... 107 

 

ix 

 

LIST OF TABLES 

Table 1. Descriptive Statistics....................................................................................................... 28 

Table 2.  Test of the Validity of the Latent Constructs – Eigenvalue of the Exploratory Factor 
Analysis (N = 348) ........................................................................................................................ 40 

Table 3.  Test of the Validity of the Latent Constructs - Exploratory Factor Analysis (N = 348) 41 

Table 4.  Test of the Validity of the Latent Constructs – Confirmatory Factor Analysis (N = 348)
....................................................................................................................................................... 44 

Table 5.  Test of the Validity of the Latent Constructs - The Standardized Factor Loading, SSI, 
SEV, AVE, and Composite Reliability of the Latent Constructs (N = 348) ................................. 46 

Table 6.  Test of the Validity of the Latent Constructs - Discriminant Validity for the 
Measurement Model (N = 348) ..................................................................................................... 48 

Table 7.  Correlation Table ........................................................................................................... 49 

Table 8.  Measurement Model ...................................................................................................... 50 

Table 9.  Fit Indices of Shea's Model and the Alternative Model ................................................ 52 

Table 10.  The Structural Model for the Alternative Model ......................................................... 53 

Table 11.  The Structural Model for the Modified Alternative Model ......................................... 55 

Table 12.  The Structural Model for the Model with Mentor Presence ........................................ 56 

Table 13.  The Structural Model for the Modified Model with Mentor Presence ........................ 58 

Table 14.  The Structural Model for the Hypothesized Model with Satisfaction ......................... 60 

Table 15.  The Structural Model for the Modified Model with Satisfaction ................................ 62 

Table 16.  The Structural Model for the Hypothesized Model with Perceived Progress ............. 64 

Table 17.  The Structural Model for the Modified Model with Perceived Progress .................... 65 

Table 18.  The Structural Model for the Hypothesized Model with Final Grade ......................... 68 

Table 19. The Structural Model for the Modified Model with Final Grade ................................. 69 

Table 20.  The Demographic Differences between “At-school” Students and “At-home” Students
....................................................................................................................................................... 71 

 

x 

Table 21.  Group Comparisons with Means and ANOVAs.......................................................... 72 

Table 22.  Summary of the Tested Hypotheses ............................................................................ 76 

 

xi 

 

LIST OF FIGURES 

Figure 1. Shea's Model .................................................................................................................... 3 

Figure 2. The Hypothesized Alternative Model ........................................................................... 23 

Figure 3. The Hypothesized Alternative Model with Mentor Presence ....................................... 24 

Figure 4. The Hypothesized Alternative Model with Mentor Presence and Learning Outcomes 25 

Figure 5. The Path Coefficients of the Alternative Model ............................................................ 54 

Figure 6. The Path Coefficients of the Modified Alternative Model ............................................ 55 

Figure 7. The Path Coefficients of the Alternative Model with Mentor Presence ....................... 57 

Figure 8. The Path Coefficients of the Modified Alternative Model with Mentor Presence ........ 59 

Figure 9. The Path Coefficients of the Alternative Model with Mentor Presence and Satisfaction
....................................................................................................................................................... 61 

Figure 10. The Path Coefficients of the Modified Alternative Model with Mentor Presence and 
Satisfaction .................................................................................................................................... 63 

Figure 11. The Path Coefficients of the Alternative Model with Mentor Presence and Perceived 
Progress ......................................................................................................................................... 65 

Figure 12. The Path Coefficients of the Modified Alternative Model with Mentor Presence and 
Perceived Progress ........................................................................................................................ 66 

Figure 13. The Path Coefficients of the Alternative Model with Mentor Presence and Final Grade
....................................................................................................................................................... 69 

Figure 14. The Path Coefficients of the Modified Alternative Model with Mentor Presence and 
Final Grade.................................................................................................................................... 70 

 

xii 

CHAPTER 1 

 INTRODUCTION 

Enrollment in K-12 online courses has increased dramatically over the past decade. 

During the 2007-2008 school year, an estimated 1.03 million K-12 students in the United States 

took at least one online course, 47% more than had done so in 2005-2006 (Wicks, 2010). By the 

2014-2015 school year, there were some 3.8 million online K-12 course enrollments by an 

estimated 2.2 million students (Watson, Pape, Murin, Gemin, & Vashaw, 2015). 

In addition to these overall increases in enrollment in K-12 online learning, the number of 

U.S. states providing it has also increased markedly: from 38 in 2006 (De Laat, Lally, Lipponen, 

& Simons, 2007), to all 50 states and the District of Columbia by 2011 (Watson, Murin, Vashaw, 

Gemin, & Rapp, 2011). Some states including Michigan, Virginia, Alabama, Florida, Idaho, 

Georgia, and West Virginia have even included online courses among their high-school 

graduation requirements (Corry & Stella, 2012; DiPietro, Ferdig, Black, & Preston, 2008; 

Watson et al., 2015). 

Amid this explosive increase in K-12 online education, its quality has become a primary 

concern for students, parents, researchers, educators, and policymakers (Black, Ferdig, & 

DiPietro, 2008; Borup, Graham, & Davies, 2013; Corry & Stella, 2012; Liu & Cavanaugh, 

2012). As enrollment expands both numerically and geographically, many institutions that offer 

online courses are facing challenges in terms of both high student attrition rates and low 

academic performance, as compared to brick-and-mortar schools (Bernard, Abrami, 

Borokhovski, Wade, Tamim, Surkes, & Bethel, 2009; Borup et al., 2013; Cavanaugh, Gillan, 

Kromrey, Hess, & Blomeyer, 2004; Smith, Clark, & Blomeyer, 2005). 

 

1 

To address these two issues, it is crucial to understand K-12 students’ online-learning 

processes and the factors that predict their performance. Two theoretical frameworks have 

provided especially insightful explanations of how students learn online in both higher education 

and K-12 contexts. The first is the Community of Inquiry (CoI) framework devised by Garrison, 

Anderson, and Archer (2001), and the second, Zimmerman’s (1986) self-regulated learning 

(SRL) theory. CoI proposes that three components – teaching, social, and cognitive presences – 

explain how learners construct knowledge in online settings. SRL, on the other hand, focuses on 

how learners regulate their own learning processes metacognitively, motivationally, and 

behaviorally. 

CoI reflects social constructivism, insofar as it focuses on a dialogic approach to 

knowledge construction, and postulates that when teaching and social presences are provided, 

learners will necessarily acquire knowledge. This framework, however, does not take account of 

learners’ confidence, goal orientation, beliefs, or systematic efforts to plan, implement, and 

monitor their learning (Cho, Kim, & Choi, 2017; Shea & Bidjerano, 2010); and when learners 

are not motivated, or do not monitor their learning progress, desirable learning outcomes are less 

likely (Pintrich & De Groot, 1990). 

To address these limitations, Shea and Bidjerano (2010) developed a revised CoI model 

that includes learning’ presence, which relates closely to the strategic learner role in SRL 

theory. Shea and Bidjerano’s model further suggests that teaching presence, social presence, and 

learning presence are all associated with cognitive presence (see Fig. 1). Support for Shea and 

Bidjerano’s approach was provided by Cho et al.’s (2017) findings that SRL played a significant 

role for students perceived CoI, and that developing students’ SRL is crucial to the creation of 

positive online communities of inquiry. 

 

2 

 

Figure 1. Shea's Model 

Though their model is more comprehensive than either CoI or SRL by itself when it 

 

comes to explaining online learning performance, Shea and Bidjerano (2010) conceded that it 

still had limitations with regard to the conception of learning presence. Specifically, their revised 

CoI operationalized learning presence as self-efficacy and effort regulation, but the latter 

construct only covered a part of SRL. In other words, self-regulation is a metacognitive, 

motivational, and behavioral process in which learners actively participate in their own learning 

(Zimmerman, 1986); yet Shea and Bidjerano’s revised CoI only included students’ self-efficacy 

and one aspect of strategic learning, and did not consider the potential influence of other 

metacognitive uses of learning strategies. Because students’ degree of SRL plays an important 

role in their learning achievement (Graham & Harris, 2000; Pintrich & De Groot, 1990; Schunk 

 

3 

& Zimmerman, 2007; Zimmerman, 2008; Zimmerman & Bandura, 1994), it is necessary to 

incorporate various types of SRL strategies into any model that aims at a comprehensive 

understanding of the mechanisms through which effective online learning experiences are 

created (Barnard, Paton, & Lan, 2008; Kim, Park, Cozart, & Lee, 2015; Ley & Young, 2001; 

Lin, Zhang, & Zheng, 2017; McMahon & Oliver, 2001; Rockinson-Szapkiw, Wendt, Whighting, 

& Nisbet, 2016; Wang, Shannon, & Ross, 2013). Another limitation of Shea and Bidjerano’s 

revised CoI is its lack of predictive power regarding students’ learning outcomes (academic 

performance, learning satisfaction, or perceived learning). If we are to arrive at valid theoretical 

models that are useful for promoting effective online education, such models must be linked to 

measurable student-achievement outcomes (Akyol & Garrison, 2011a; Garrison, Cleveland-

Innes, & Fung, 2010; Rockinson-Szapkiw et al., 2016). 

Perhaps more importantly, most of the previous studies that looked at how students 

learned online using the CoI and/or SRL frameworks were conducted in postsecondary online-

learning settings (Akyol & Garrison, 2011b; Shea & Bidjerano, 2010, 2012), leaving the rapidly 

increasing population of K-12 online students under-researched (Hawkins, Barbour, & Graham, 

2011). Rather than taking it for granted that there are no important differences between 

postsecondary and K-12 online learning, it is crucial to establish a conceptual framework that is 

sensitive to how students learn online in the K-12 context specifically (Barbour, 2013; Barbour 

& Reeves, 2009; Corry & Stella, 2012; Rice, 2006; Smith et al., 2005). 

In particular, when applying Shea and Bidjerano’s (2010) revised CoI model, two unique 

characteristics of online K-12 learning should be considered: the presence of on-site mentors 

(Borup & Drysdale, 2014), and the use of in-school time for students to learn online courses 

(Roblyer & Marshall, 2002). It is reasonably clear that physical separation between learners and 

 

4 

teachers makes it difficult for online teachers to provide personalized and targeted support to 

their K-12 students (Hawkins et al., 2011). To mitigate this problem, many K-12 online-learning 

institutions assign learners to mentors who are based in the students’ own brick-and-mortar 

schools – usually, existing staff members who have been specially trained for the purpose 

(Barbour & Mulcahy, 2004, 2009; Drysdale, Graham, & Borup, 2014; Harms, Niederhauser, 

Davis, Roblyer, & Gilbert, 2006; Wicks, 2010). Some states, such as Michigan, even require that 

their school districts provide each student who learns online with a mentor who is a state-

certified teacher (Michigan Virtual Learning Research Institute, 2014). As distinct from online 

instructors who cannot be physically present, these mentors can show up in the classroom or 

learning labs during times when students are encouraged or required to be studying (Hannum, 

Irvin, Lei, & Farmer, 2008). 

The second unique feature of the existing K-12 online-learning environment is its use of 

in-school lab learning time. There are two reasons to explain for the use of in-school lab for 

online learning: First, this is to avoid the potential digital divide by a lack of access to technology 

at students’ homes (Roblyer & Marshall, 2002). For example, the 2014 Keeping Pace with K-12 

Online Learning Report showed that students in five charter schools in New Orleans did not 

have stable and reliable Internet access at home. As a solution, they turned to use desktops at 

schools or rely on libraries to access online learning content (Watson, Pape, Murin, Gemin, & 

Vashaw, 2014). Second, it is believed that requiring some in-school learning time for students 

could be helpful in terms of providing additional monitoring and support to high school students, 

who may show less experience in online courses (Roblyer & Marshall, 2002). A removal of seat 

time requirements may lead to low learning performance when learners spend as little time as 

possible on content learning (Rice, 2014). 

 

5 

In this study, I propose a further revision of the CoI framework that incorporates aspects 

of CoI and SRL as well as actual learning outcomes, and which is specially tailored to K-12 

online settings. The next section comprises a literature review structured around CoI, SRL, and 

the two key distinctive features of such settings (i.e., mentors and credit recovery). It is argued 

that, to better understand the mechanism(s) that can predict students’ cognitive presence in 

online learning, several components – including teaching presence, social presence, mentor 

presence, motivation, and online SRL – need to be taken into account. A second aim of the study 

is to understand how learning location may or may not affect students’ perception of teaching 

presence, social presence, mentor presence, cognitive presence, and their learning outcomes. 

At a theoretical level, findings from studies such as this one can help us gain a more 

comprehensive understanding of the mechanisms of students’ online learning. It is hoped that 

this, in turn, will provide future K-12 online-learning studies with a more robust foundation, and 

answer previous scholars’ calls for further investigation of the teaching, social, and cognitive 

aspects of online learning; its impacts on knowledge building; and which of its components are 

critical to the promotion of achievement (Barbour & Reeves, 2009; Rice, 2006; Smith et al., 

2005). It will also shed light on if and how one’s primary learning location may affect students’ 

online learning perceptions and learning outcomes. At a practical level, Cavanaugh, Barbour, and 

Clark (2009) have argued that K-12 online-learning studies should provide best-practice 

guidelines for online teachers. For K-12 online-learning instructors, mentors, and administrators, 

findings from the current study could elucidate not only the complex relationships between the 

studied underlying constructs that exist within online-learning settings, but the importance of the 

quality of teaching presence, mentor presence, social presence, and course characteristics on 

students’ cognitive presence. Findings from the differences in students’ choice of primary 

 

6 

learning location can also inform online educators, researchers, and policy-makers in what ways 

difference may come up, in order to rethink how we can reshape students’ online learning 

experience both at-school and at-home.

 

7 

CHAPTER 2 

LITERATURE REVIEW 

Community of Inquiry 

The CoI framework explains the complexities of online learning by incorporating the 

conceptual components that are believed to be fundamental to successful knowledge-

construction in online-learning environments, focusing on active collaboration and construction 

among active participants in such learning communities (Garrison, 2007; Garrison et al., 2001). 

At its root, CoI reflects deep and meaningful learning through the complex dynamics of design, 

facilitation, and interaction within the course (Akyol & Garrison, 2011a). As well as its three 

aforementioned key components (teaching presence, social presence, and cognitive presence), 

the CoI framework considers how these three components interact with each other, and how such 

interaction results in individuals’ constructive online-learning experiences. 

Teaching Presence. According to the CoI framework, teaching presence has three key 

responsibilities. The first is the design, inception, and organization of the learning content, 

learning activities, and learning schedules. The second is the facilitation of learning discourse 

and students’ online collaborative learning; and the third is direct content-related instruction. It 

plays the most fundamental role in online learning, as it sets the tone for the whole learning 

experience (Anderson, Liam, Garrison, & Archer, 2001; Garrison, Anderson, & Archer, 2010; 

Garrison et al., 2001). It can also facilitate and direct social presence and cognitive presence, in a 

manner that enables learners to generate meaningful learning outcomes (Anderson et al., 2001). 

 

Teaching presence was found to be a significant predictor of self-efficacy in Shea and 

Bidjerano’s (2010) structural equation modeling (SEM) analysis of 3,000 undergraduates taking 

online courses. In addition to its relationship to self-efficacy, teaching presence has a direct and 

 

8 

positive correlation with self-regulation. Two components of teaching presence – facilitating 

discourse and direct instruction – have been found to be directly correlated with metacognition, 

as these responsibilities help to develop one’s metacognitive awareness in monitoring and 

regulating learning (Akyol & Garrison, 2011a). Ebner and Ehri (2016) demonstrated that 

teachers could develop students’ self-regulation abilities through deliberate and structured 

teaching of thinking procedures for online learning, and when Crippen and Earl (2007) provided 

worked examples to online learners, it resulted in better problem-solving skills. 

The generally accepted account of the relationship between teaching presence and 

cognitive presence is that well-structured, carefully facilitated, direct instruction by online 

instructors has a positive influence on the establishment of higher-order cognitive learning 

(Garrison, 2007). Meyer (2004), for example, found that triggering questions raised by teachers 

influenced the quality of students’ responses. Similarly, Celentin (2007) pointed out that the role 

of the teacher was strongly related to whether class discussions reached the highest level of 

cognitive inquiry. Garrison et al.’s (2001) SEM model of the relationships among CoI’s three 

presences found that teaching presence was a significant, direct predictor of cognitive presence. 

Shea and Bidjerano’s (2010) structural model of the relationships among the components of CoI 

found that teaching presence directly and positively predicted cognitive presence. In another 

study of more than 2,000 college students, Shea and Bidjerano (2012) again found that teaching 

presence predicted cognitive presence, even after delivery mode and students’ prior online 

learning experience were controlled for. And more recently, Rockinson-Szapkiw et al. (2016) 

highlighted the singular power of teaching presence, over and above social presence, to 

positively predict the effectiveness of online learning environments. 

 

9 

Previous studies have consistently demonstrated that teaching presence was positively 

associated with satisfaction and/or perceived progress (Akyol & Garrison, 2008; Akyol & 

Garrison, 2014; Arbaugh, 2013; Garrison et al., 2010; Oliver, Osborne, & Brandy, 2009; 

Rockinson-Szapkiw et al., 2016). Garrison et al. pointed out that teaching presence had a major 

and positive impact on learning, which set the tone for the overall learning process. In a study of 

students in an online course, based on transcript analysis and a survey, all of Akyol and 

Garrison’s (2014) respondents reported that teaching presence was the most critical factor in 

their learning. The same study also found that teaching presence was strongly and positively 

correlated with satisfaction and perceived learning. Rockinson-Szapkiw et al., meanwhile, 

affirmed the importance of teaching presence to the quality of course design and implementation. 

Based on the relevant literature, this study proposes to test the following four hypotheses 

related to teaching presence: 

H1: Teaching presence positively predicts students’ online SRL strategies. 

H2: Teaching presence positively predicts students’ self-efficacy. 

H3: Teaching presence positively predicts cognitive presence. 

H4: Teaching presence positively predicts learning outcomes. 

Social Presence. Social presence refers to the degree to which the learner creates 

personal but purposeful relationships, and develops social bonds, in his/her learning context 

(Garrison, 2007; Garrison et al., 2001, 2010). Its three main aspects are effective communication, 

open communication, and group cohesion. Ideally, social presence operates to create open and 

comfortable conditions for students’ inquiry and their high-quality interaction with other parties, 

in the service of a wider educational goal (Garrison et al., 2010). Garrison (2007) sharply 

criticized some previous studies that artificially isolated social presence from CoI’s other two 

 

10 

constructs, and therefore failed to take account of the complex dynamics that took place between 

social presence and the other two in real-world settings. The present research endorses the idea 

that social presence cannot effectively be studied independently of cognitive presence and 

teaching presence. 

It has been hypothesized that a positive social presence should predict students’ self-

efficacy, by serving as a source of “social persuasion” (Shea & Bidjerano, 2010, p. 1724). 

Wighting, Nisbet, and Rockinson-Szapkiw (2013) found that a higher sense of social community 

was associated with a higher confidence in learning; and Shea and Bidjerano (2010) found that 

social presence directly and positively predicted students’ self-efficacy. 

Arguing that CoI should be transformed from a theoretical and descriptive tool into one 

that can be used predictively in quantitative studies, Garrison et al. (2010) created an SEM model 

that included its three presences and their interactive relationships. This model’s results 

suggested that social presence mediated the relationship between teaching presence and 

cognitive presence. Shea and Bidjerano (2008) used cognitive presence as dependent variable in 

hierarchical regression analysis, and found that social presence significantly predicted cognitive 

presence. In a second study, Shea and Bidjerano (2010) structured an SEM model that 

established the direct and indirect relationships among a number of variables in online learning, 

and confirmed a direct and positive relationship between social presence and cognitive presence. 

Subsequently, Shea and Bidjerano (2012) confirmed that social presence was a significant and 

positive predictor of cognitive presence. 

Social presence has also been found to be significantly and positively correlated with 

learning-outcome variables including satisfaction and perceived progress (Akyol & Garrison, 

2008; Arbaugh, 2013; Cobb, 2011; Hawkins, Graham, Sudweeks, & Barbour, 2013; Kang, Liw, 

 

11 

Kim, & Park, 2014; Nisbet, Wighting, & Rockinson-Szapkiw, 2013; Rockinson-Szapkiw et al., 

2016; Zhan & Mei, 2013). For example, echoing Wighting et al., 2013, Nisbet et al. (2013) 

reported that the higher online students’ sense of social community was, the higher they rated all 

three perceived-learning areas, i.e., cognitive, affective, and psychomotor. And a recent meta-

analysis conducted by Richardson, Maeda, Lv, and Caskurlu (2017) showed large positive 

correlations between social presence and satisfaction, and between social presence and perceived 

learning. 

Based on the relevant literature, this study proposes to test three additional hypotheses 

related to social presence, as follows: 

H5: Social presence positively predicts students’ self-efficacy. 

H6: Social presence positively predicts cognitive presence. 

H7: Social presence positively predicts learning outcomes. 

Cognitive Presence. Cognitive presence is the process whereby learners explore 

information, connect ideas, and apply new ideas to other settings through active construction and 

reflection as part of a community of inquiry (Garrison et al., 2001). Cognitive presence in an 

online-learning context can be represented as four phases in the progressive development of 

inquiry: triggering event, exploration, integration, and resolution. According to the CoI 

framework, cognitive presence is a cycle consisting of all four of these phases. 

Rockinson-Szapkiw et al. (2016) examined the relationship between CoI and perceived 

learning outcomes, and found that students with high levels of perceived cognitive presence 

tended to have higher course grades. Using hierarchical regression analysis, Shea and Bidjerano 

(2008) found that cognitive presence was a significant predictor of learners’ overall online 

learning satisfaction; similarly, in three consecutive studies conducted over a two-year period, 

 

12 

Kang et al. (2014) found that cognitive presence significantly predicted students’ learning 

satisfaction. Akyol and Garrison (2011a) found that students’ perceived levels of cognitive 

presence were significantly and positively associated with both their perceived learning and their 

learning satisfaction. Qualitative interviews conducted as part of the same study revealed that 

students considered their class projects as opportunities to synthesize and evaluate course content 

from throughout the semester, and attributed their satisfaction to their perceived cognitive 

presence. 

In light of the findings of the reviewed literature, this study proposes to test the following 

hypothesis related to cognitive presence: 

H8: Cognitive presence positively predicts learning outcomes. 

Self-regulated Learning 

In contrast to the CoI framework, which mainly describes the processes of online learning 

at a macro level, SRL theory helps to explain them at the micro level of individual characteristics 

(Shea & Bidjerano, 2010). As Zimmerman (1986, p. 308) summarized it, SRL is a 

multidimensional construct that regards students as “metacognitively, motivationally, and 

behaviorally active participants in their own learning process”. More specifically, it describes 

students’ proactive self-direction of their own learning processes, whereby their beliefs are 

transformed into learning goals and thus into academic performance (Zimmerman, 2008). Self-

regulation is not a quality that one either possesses or does not possess; rather, it refers to the use 

of one or more specific processes that vary according to the specific learning task one is facing 

(Zimmerman, 2008). 

Early studies of SRL mainly focused on typologies of learning strategies (e.g., 

monitoring, planning, rehearsal, and time management) and/or on how to train students to adopt 

 

13 

such strategies (Zimmerman, 2008). While these strategies were demonstrably effective, 

however, their ability to foster real improvements in learning was found to be limited unless 

students adopted them spontaneously (Zimmerman, 2008). Noting that there is more to SRL than 

merely choosing and using appropriate learning strategies (Schunk & Zimmerman, 2012), 

researchers have greatly expanded the scope of this construct to include additional interrelated 

motivational processes, such as self-efficacy (Zimmerman, 2002; Zimmerman & Schunk, 2012). 

Grounded in social cognitive theory, self-efficacy emphasizes the close interaction between 

behavior, person, and environment and has four types of influences: mastery experience, 

vicarious learning, verbal persuasion, and psychological states (Bandura, 1997). Self-efficacy in 

online learning has been found an effective predictor of students’ use of metacognitive learning 

strategies (Cho & Shen, 2013; Shea & Bidjerano, 2010), and both self-efficacy and the use of 

SRL strategies have been confirmed as necessary to becoming a highly self-regulated learner 

(Zimmerman, 2002). 

It has been convincingly argued that in online-learning contexts, becoming a self-

regulated learner is even more important than it is in face-to-face ones (Tsai, Shen, & Fan, 2013). 

As Zimmerman (1986, 2002) has pointed out, self-regulation strongly emphasizes students’ 

personal choices, goal-setting and learning autonomy. Online learning, because it lacks face-to-

face interaction between teacher and students, requires the learner to be highly proactive and 

organized, planning his or her own learning and evaluating progress regularly (Puzziferro, 2008). 

Lynch and Dembo (2004) found that the self-regulatory attributes that were critical to online 

learning success included motivation, Internet self-efficacy, time management, study-

environment management, and learning-assistance management. Building upon SRL theory and 

the unique characteristics of online learning environments, Barnard, Lan, To, Paton, and Lai 

 

14 

(2009) developed the Online Self-regulated Learning Questionnaire (OSLQ) to examine the 

psychometric properties of SRL in such environments. Their results identified several key 

aspects of online self-regulation, including environment-structuring, goal-setting, time 

management, help-seeking, task strategies, and self-evaluation. And Barnard-Brak, Lan, and 

Paton (2010) distinguished five profiles of online learners ranging from the most competent self-

regulators to the least competent self-regulators, and found that these profile memberships 

differed significantly from one another in terms of academic achievement with the most 

competent self-regulators predicting the highest academic achievement. 

The advantages to students of being highly autonomous and self-regulated have also been 

noted in the case of K-12 online learning. Consistent with findings derived from postsecondary 

online-learning settings, constructs such as self-efficacy, intrinsic motivation, goal-setting, time 

management, and metacognitive skills can be key to achieving K-12 online learning success 

(Cavanaugh et al., 2004; DiPietro et al., 2008; Kim et al., 2015; Lin et al., 2017; Weiner, 2003). 

Moreover, younger learners are likely to possess lower levels of self-regulation ability than adult 

ones, leading the former to need more frequent teacher-student communication, more guidance 

from teachers as online-learning-behavior scaffolders, and more explicit instruction in cognitive 

and metacognitive skills (Cavanaugh et al., 2004). 

Some studies have found that online SRL skills are an important predictor of cognitive 

presence. For example, Garrison (2007) noted that to promote cognitive inquiry, individuals need 

to enhance their own metacognitive awareness of their tasks, and especially the ability to identify 

the level of their own cognitive contributions during class activities. In a similar vein, Pawan, 

Paulus, Yalcin, and Chang (2003) recommended that, to promote higher levels of cognitive 

presence, students should apply a strategy of self-coding to their responses in class. The SEM 

 

15 

results for Shea and Bidjerano’s (2010) model establishing the relationship between CoI, self-

efficacy, and effort regulation indicated that self-efficacy directly predicted the respondents’ 

online cognitive presence, and that online effort regulation also partially mediated the association 

between these learners’ self-efficacy and their cognitive presence. Cho et al. (2017) used cluster 

analysis to classify students into four groups based on different self-regulation levels, and found 

that the high-self-regulation group had a stronger sense of CoI than the low-self-regulation one. 

Wang, Peng, Huang, Hou, and Wang (2008) found that online students’ self-efficacy predicted 

their learning outcomes via the mediating effect of SRL strategies. 

A learner’s degree of self-regulation is strongly and positively correlated with the 

likelihood of his or her academic success (Zimmerman, 2008). As Schunk (2005) put it, “self-

regulated learning is seen as a mechanism to help explain achievement differences among 

students and as a means to improve achievement” (p. 85). In a study of students enrolled in an 

online liberal-arts course, Puzziferro (2008) established a clear relationship between online SRL 

subscales (i.e., time and study environment, effort regulation) and learning performance. The 

same study’s findings also indicated that high learning satisfaction was positively correlated with 

a variety of self-regulation subscales, including both cognitive and metacognitive strategies. 

Barnard et al. (2008) used the OSLQ to measure students’ online SRL skills, and found that 

approximately one-third of the relationship between perceptions of online-course communication 

and academic performance was mediated by online SRL strategies, further highlighting the 

importance of self-regulation in online learning. In the K-12 context, Roblyer and Marshall 

(2002) devised an instrument for detecting the likely success or failure of online students, and 

reported that intrinsic motivation, self-efficacy, time management and goal-setting were critical 

determinants of middle- and high-school students’ online learning success. Roblyer, Davis, 

 

16 

Mills, Marshall, and Pape (2008) found that self-efficacy and SRL skills including goal-setting, 

time management, and self-reflection were significantly capable of predicting whether online 

students would be successful or unsuccessful. 

Based on the literature reviewed above, this study proposes to test the following three 

hypotheses related to SRL: 

H9:  Self-efficacy positively predicts online SRL strategies. 

H10: Self-efficacy positively predicts cognitive presence. 

H11: Online SRL strategies positively predicts cognitive presence. 

Mentors in K-12 Online Courses 

Most U.S. K-12 students who are enrolled in online courses also regularly attend a 

physical school locally, which typically provides each of them with a mentor (Borup & Drysdale, 

2014). Mentors can help maintain a supportive on-site learning environment while also tracking 

students’ learning progress (Harms et al., 2006), and ultimately their role is to ensure that 

students achieve success in their online courses (Wicks, 2010). Barbour and Mulcahy (2004, p. 

11) considered the establishment of the on-site mentor role as “one of the most dramatic 

changes” in distance learning since its extension to the K-12 sphere. It is important that such 

mentors’ support should be both continuous and comprehensive (Barbour, 2012; Freidhoff, 

Borup, Stimson, & DeBruler, 2015; Harms et al., 2006). 

Mentor Presence. For the purposes of the present research, mentors’ roles can be said to 

fall largely into four categories: relationship builders, monitors, content-learning facilitators, and 

problem solvers (e.g., Aronson & Timms, 2003; Barbour & Mulcahy, 2004; Borup & Drysdale, 

2014; de la Varre, Keane, & Irvin, 2011; Drysdale, 2013; Hannum et al., 2008; Kennedy & 

Cavanaugh, 2010; Taylor et al., 2016; Wortmann et al., 2008). The relationship-builder role 

 

17 

refers to the mentor’s endeavors to establish positive relationships with learners to make them 

feel less disconnected and isolated (Barbour & Mulcahy, 2004, 2009; Drysdale, 2013; Hannum 

et al., 2008; Pettyjohn, 2012). This is of critical importance, insofar as students’ connections with 

their online learning environments and teachers have always been of primary concern in online 

learning in both theory and practice (Gunawardena et al., 2009; Swan, 2002). 

A monitor’s responsibility includes reminding students to submit assignments on a 

regular basis, contacting them immediately if they are found to be falling behind, tracking their 

study progress, detecting possible failures, and being aware of absences (Aronson & Timms, 

2003; Barbour & Mulcahy, 2004; de la Varre et al., 2011; Drysdale, 2013). 

As for the role of content-learning facilitator, some studies have maintained that school-

based mentors do not necessarily need to be content experts (Aronson & Timms, 2003; Borup & 

Drysdale, 2014). However, an increasing amount of research has shown that mentors do 

participate in some content-related learning activities, even where direct instruction is not listed 

among their official role responsibilities (Barbour & Mulcahy, 2004; de la Varre et al., 2011; 

O’dwyer, Carey, & Kleiman, 2007; Taylor et al., 2016). 

Last but not least, mentors should work as problem solvers; this role can include 

providing technical assistance to students (Barbour & Mulcahy, 2004; de la Varre et al., 2011; 

Hannum et al., 2008; Keane, de la Varre, Irvin, & Hannum, 2008), and conducting orientations 

to help familiarize students with the online learning environment (Aronson & Timms, 2003; 

Pettyjohn, 2012). 

The Impact of Mentor Presence. The U.S. Department of Education (2007) reported 

that students who sought assistance only from their online instructors were less likely to achieve 

success than those who also took advantage of local on-site support. Roblyer et al. (2008) found 

 

18 

that assigning students on-site mentors, and designating a specific period and location at school 

for online-course participation, were both effective tactics for increasing pass rates, and in some 

cases doubling them. Hannum et al. (2008) found that trained facilitators played a positive role in 

student retention and in encouraging students to spend more time engaging with their course 

content. This in turn echoed Simpson’s (2004) finding that mentors’ constant communication 

with students helped to improve retention rates. Recently, Taylor et al. (2016) reported that 

credit-recovery students who received extensive instructional support from their mentors tended 

to have higher course pass rates than those who had little such support. And Ferdig (2010) found 

that at-risk students who checked in with their mentors on at least two days per week passed at 

least one course, and were likely to express a belief that mentors were a critical factor in helping 

them to achieve success. 

Previous studies also reported mentors’ positive impacts on the promotion of motivation 

and on the enhancement of SRL strategies (de la Varre et al., 2011; Freidhoff et al., 2015; 

Murphy & Rodriguez-Manzanares, 2009; Staker, 2011). For instance, de la Varre et al. found 

that mentors’ presence in the classroom helped to encourage students’ self-efficacy, and that 

mentors’ support was especially beneficial to those who were struggling with their courses or felt 

frustrated by online learning. Freidhoff et al. (2015) interviewed 14 on-site mentors from 

programs with a history of high performance and found that all of them described working with 

students on time-management issues, including educating them about why they needed time-

management skills in online learning, finding solutions for procrastination, and helping arrange 

times for study. 

Based on the relevant literature, this study proposes to test the following four hypotheses 

related to mentor presence: 

 

19 

H12: Mentor presence positively predicts self-efficacy. 

H13: Mentor presence positively predicts online SRL strategies. 

H14: Mentor presence positively predicts cognitive presence. 

H15: Mentor presence positively predicts learning outcomes. 

Learning Locations for K-12 Online Learners 

 

K-12 online learning is not limited to learning behavior that happens at home. The 2015 

Keeping Pace with K-12 Digital Learning Report revealed that a current trend for K-12 online 

learning is to move online learning from being mostly online to frequently combining with onsite 

learning (Watson et al., 2015). The same document reported that many students actually learn 

their online courses from a physical learning institution, such as their local schools or other 

formal learning centers, not from their home. De la Varre et al. (2011) reported that taking an 

online course does not mean that learning happens at a virtual learning environment exclusively, 

and that K-12 online learners actually take their online courses from the school classroom, 

library, or home-school settings.  

However, this does not mean that all K-12 online learners nowadays are using the 

location other than their homes to access online learning contents. In fact, contrary to those who 

chose to learn from school labs or classrooms, there are a considerate number of students choose 

to take the online course from home without seeking support from their physical schools 

(Roblyer & Marshall, 2002; Watson et al., 2015). 

 

Although no studies so far has explicitly compare the learning outcomes between 

students who spend the majority of their learning online contents at-school and at-home, it is 

suggested that the sit-in time at school could be beneficial for online learning. Through 

collecting feedback from virtual high school instructors, Roblyer and Marshall (2002) found that 

 

20 

a common theme from these instructors was about the location of students’ online learning. On 

the one hand, a lack of access to the Internet at some students’ homes may greatly cause Digital 

Divide and affect students’ learning experience and performance. On the other hand, teachers 

believed that by requiring students to actually sit in a room with access to online learning and 

allocating certain time during the school day for online courses help guarantee the quality of 

online learning. This has much to do with the hypothesis that K-12 online learners may show low 

levels of self-regulated learning abilities (both cognitively and metacognitively) when it comes to 

learning independently and autonomously in an online learning environment (Borup & Drysdale, 

2014; Rice, 2006). For example, both Harms et al. (2006) and de la Varre et al. (2011) reported 

that students study in the same physical room with other students spend more time on learning 

contents and less time on off-task behaviors. Studies also found that the physical location may be 

particularly helpful for those who are struggling with learning contents or receive very little 

support from home (Archambault, Diamond, Brown, Cavanaugh, Coffey, Foures-Aalbu, & 

Zygouris-Coe, 2010; Freidhoff et al., 2015; Pettyjohn, 2012). 

 

It is also believed that mentors set the local climate of students’ at-school online learning 

experience (de la Varre, Irvin, Jordan, Hannum & Farmer, 2014), and greatly facilitate those who 

choose to learn their online course at school (Borup & Drysdale, 2014; Freidhoff et al., 2015). As 

those who works as a liaison between the online instructor and the student (de la Varre et al., 

2014), if a student use the school lab as his/her primary online learning location, he/she may 

benefit from mentor’s practice including but not limited to building rapport relationships 

(Barbour & Mulcahy, 2004, 2009; Drysdale, 2013; Hannum et al., 2008; Pettyjohn, 2012), 

progress monitoring (Aronson & Timms, 2003; Barbour & Mulcahy, 2004; de la Varre et al., 

2011; Drysdale, 2013), content facilitation and direct instruction (Aronson & Timms, 2003; 

 

21 

Barbour & Mulcahy, 2004; Borup & Drysdale, 2014; de la Varre et al., 2011; O’dwyer et al., 

2007; Taylor et al., 2016) and trouble-shooting (Aronson & Timms, 2003; Barbour & Mulcahy, 

2004; de la Varre et al., 2011; Hannum et al., 2008; Keane et al., 2008; Pettyjohn, 2012) , as 

discussed in previous sections. 

Accordingly, based on the relevant literature, this study proposes to test the following 

hypothesis related to the primary online learning location.  

H16: Students who use school as their primary online learning locations demonstrate 

higher level of social presence, mentor presence, cognitive presence, self-efficacy, online SRL 

strategies, and learning outcomes, compared with those who choose to spend their majority time 

of online learning at home. 

Research Questions 

The present study aims to extend Shea and Bidjerano’s (2010) revised CoI framework by 

incorporating additional components drawn from SRL theory as well as the effects of mentor 

presence, it will then examine the resultant new framework in terms of its power to predict 

learning outcomes in K-12 online settings. This study also aims to test whether or not student’s 

primary online learning location matters.  

It will be guided by the following four research questions: 

1.  Looking at Shea and Bidjerano’s (2010) model, does adding a path from teaching 

presence to SRL make the model a better fit? 

Shea and Bidjerano’s (2010) model is shown in Fig. 1, and the hypothesized alternative 

model building on Shea’s model that added the path from teaching presence to self-

regulated learning is shown in Fig. 2.  

 

 

22 

 

Figure 2. The Hypothesized Alternative Model 

H1: Teaching presence positively predicts students’ online SRL strategies. 

 

H2: Teaching presence positively predicts students’ self-efficacy. 

H3: Teaching presence positively predicts cognitive presence. 

H5: Social presence positively predicts students’ self-efficacy. 

H6: Social presence positively predicts cognitive presence. 

H9: Self-efficacy positively predicts online SRL strategies. 

H10: Self-efficacy positively predicts cognitive presence. 

H11: Online SRL strategies positively predicts cognitive presence. 

2.  How well does the model explain cognitive presence if mentor presence is added? 

 

23 

The hypothesized alternative model with mentor presence is shown on Fig. 3.  

Figure 3. The Hypothesized Alternative Model with Mentor Presence 

 

H12: Mentor presence positively predicts self-efficacy. 

H13: Mentor presence positively predicts online SRL strategies. 

H14: Mentor presence positively predicts cognitive presence. 

3.  How well does the model explain learning outcomes (i.e., satisfaction, perceived 

progress, final grade)? 

The hypothesized model with mentor presence and learning outcomes is shown in Fig. 4.  

 

 

 

 

24 

 

Figure 4. The Hypothesized Alternative Model with Mentor Presence and Learning 
Outcomes 

 

The three types of learning outcome were tested separately.  

H4: Teaching presence positively predicts learning outcome. 

H7: Social presence positively predicts learning outcome. 

H15: Mentor presence positively predicts learning outcome. 

H8: Cognitive presence positively predicts learning outcome. 

4.  How do the associations between social presence, cognitive presence, mentor presence, 

self-efficacy, online SRL strategies and learning outcomes vary according to whether the 

student’s primary online learning location is home or school? 

The hypothesis relevant to this question is H16 is as follows: 

 

25 

H16: Students who use school as their primary online learning locations demonstrate 

higher level of social presence, mentor presence, cognitive presence, self-efficacy, online SRL 

strategies, and learning outcomes, compared with those who choose to spend their majority time 

of online learning at home. 

 

 

26 

CHAPTER 3 

METHODS 

Context 

This study was conducted at a nonprofit virtual school in the Midwestern United States 

during the spring semester of 2017. Although termed a school, this institution is a non-profit, 

state-wide supplemental program, which was authorized by the state and overseen by state 

agencies. It is one of the largest online-learning schools in the country. Although this 

institution’s online courses were the main focus of inquiry, it should be noted that all the students 

who participated in this study also attended their brick-and-mortar schools on a daily basis. 

Typically, each student would take one or two of their courses from the virtual school, but the 

majority at their own school. 

Each virtual-school course was self-paced and asynchronous. Most of the communication 

between students and their online teachers took place online through discussion forums and 

messages. Each student had an adult offline mentor, in addition to their online instructor(s) 

whose main responsibility was to deliver learning content. For the most part, students were 

encouraged to complete their online-learning activities at seat times, in labs located in their local 

schools, under the supervision of their mentors. However, some seat-time waivers were issued, 

entitling certain students to be mentored entirely online (see Rice, Huerta, Shafer, Barbour, 

Miron, Gulosino, and Horvitz, 2014). Assessment methods for online learners included, but were 

not limited to, essays, discussions, projects, and computer-graded quizzes. 

Participants 

All 696 individuals who participated in this study, via responding to an online survey, 

were taking online high-school level courses at the virtual school described above at the time of 

 

27 

their participation. As shown in Table 1, 72% of the participants were female (N = 502) and 28% 

were male (N = 193). An additional student did not indicate the gender. Most of the participants 

were 12th graders (N = 313, 45%), followed by 11th graders (N = 163, 23%), 10th graders (N = 

132, 19%), 9th graders (N = 59, 8%), 8th graders (N = 25, 4%) and 7th graders (N = 4, 1%).  

Table 1. Descriptive Statistics 

Mean 

Min 

Max 

Alpha 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

SD 
.45 
.37 
.15 
.17 
.25 
.19 
.08 
.19 
.28 
.39 
.42 
.50 
.28 
.46 
.30 
.37 
.35 
.41 
.23 
1.05 
.82 
.96 
.91 
.80 
.86 
.81 
1.12 
.84 

.28 
.84 
.02 
.03 
.07 
.04 
.01 
.04 
.08 
.19 
.23 
.45 
.08 
.30 
.10 
.16 
.14 
.22 
.06 
2.95 
3.93 
3.74 
3.72 
3.87 
4.13 
3.59 
3.93 
4.20 
80.14 

 
Male 
Caucasian 
African American 
Hispanic 
Asian 
Other ethnicity 
7th grade 
8th grade 
9th grade 
10th grade 
11th grade 
12th grade 
English 
Foreign languages 
Math 
Science 
Social science 
Other subjects 
Credit-recovery 
Prior knowledge 
Teaching presence 
Mentor presence 
Social presence 
Cognitive presence 
Self-efficacy 
Online SRL strategies 
Satisfaction 
Perceived progress 
Final grade 
Teaching presence, Mentor presence, Social presence, Cognitive presence, Self-efficacy, Online 
SRL strategies, Satisfaction, and Perceived progress are the composite scores of the measured 
items. 
 
 

N 
695 
696 
696 
696 
696 
696 
696 
696 
696 
696 
696 
696 
694 
694 
694 
694 
694 
694 
690 
696 
694 
696 
696 
694 
696 
696 
695 
696 
696 

0 
0 
0 
0 
0 
0 
0 
0 
0 
0 
0 
0 
0 
0 
0 
0 
0 
0 
0 
1 
1 
1 
1 
1 
1 
1 
1 
1 
0 

1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
5 
5 
5 
5 
5 
5 
5 
5 
5 

.93 
.96 
.88 
.94 
.92 
.87 
.90 
.90 

 

24.50 

100 

 

28 

Regarding the ethnicity of the participants, the majority were Caucasians (84%), while 

2% were African Americans, 3% Hispanics, 7% Asians, and 4% other ethnicities.  

The virtual-school subjects that the participants were enrolled in included Foreign 

Languages (30%), Science (16%), Social Science (14%), Math (10%), English (8%), and others 

(22%). Two students did not report the subjects of the online courses they were taking. Most 

participants were enrolled on a non-credit-recovery basis (N = 650; 94%) and around 6% of them 

(N = 40) for credit-recovery purposes, while 6 respondents did not answer this question. When 

the students were asked about their prior knowledge on the subject in this course, around 10% of 

them rated it as very poor, 21% as poor, 37% as fair, 26% as good, and 6% as very good.  

Other background questions about each learner covered their previous online-learning 

experience and primary learning location. With regard to the former, 38% reported that they had 

never taken an online course before, and 33% that they had taken only one; an additional 13% 

had taken two, and 16% had taken three or more. Two students did not answer this question. 

When asked about their primary location for completing most of their online coursework, around 

58% students (N = 399) reported that they completed it at their own school and around 42% 

students (N = 295) reported that they completed it at home. One student did not answer this 

question. 

Procedure 

Participants were recruited primarily with the assistance of two gatekeepers at the 

research site, who offered help in contacting IT staff there to resolve technical issues with the 

survey settings, announced the study on the virtual school’s learning-management systems, and 

ensured that the Qualtrics survey link was displayed when each student logged in to such 

systems. The gatekeepers set a window of five days during which students could complete the 

 

29 

study. To help ensure a high response rate, the researcher provided 20 gift cards worth $20 each 

that were randomly awarded to 20 students who completed the survey. 

The first page of the online survey was a parental consent form (see Appendix A). Before 

answering any questions, each students first had to give this form to their parents and obtain their 

approval for study participation. The second page of the survey was a student assent form (see 

Appendix B), completion of which was also required before anyone could start the survey 

proper. On both these forms, the researcher provided informed about the nature of the study; 

made it clear that participation would be completely voluntary, and could be terminated by the 

student at any point; and that all responses would remain confidential and not have any effect on 

course grades. The time it would take for a given student to complete the survey was typically 

20-25 minutes. 

Survey 

Survey approaches are typically used to measure the opinions, attitudes, or characteristics 

of a sample of the population, and have been very extensively utilized in educational research 

(Creswell, 2005). The use of a survey method is appropriate for the present study, given the large 

number of students involved and the need to conduct statistical analyses. The survey questions 

are provided in Appendix C. 

Demographic Information. This part of the survey instrument asked about students’ 

gender, grade level, ethnicity, and subject of the online course currently being taken. If they were 

taking more than one course, they were asked to choose the one that they thought best 

represented their online-learning experience. The multiple-choice response categories for course 

subject included English, Foreign Languages, Math, Science, Social Science, and “Other”. This 

section also asked the students their reasons for enrollment (i.e., credit-recovery or non-credit-

 

30 

recovery), the major location in which they completed most of their online coursework (i.e., 

home, school, or other places), and their prior knowledge of the subject they were taking (i.e., 

very poor, poor, fair, good, or very good). 

Community of Inquiry. The participants’ perceptions of teaching, social, and cognitive 

presences were measured using a modified version of an instrument originally developed by 

Arbaugh, Cleveland-Innes, Diaz, Garrison, Ice, Richardson, & Swan (2008) based on the CoI 

framework. Specifically, some of Arbaugh et al.’s items were slightly modified to better fit the 

current study’s participants and context. In the modified instrument used here, each item was 

measured on a five-point Likert scale, ranging from 1 = strongly disagree to 5 = strongly agree. 

Of its 34 items, 12 were designed to assess teaching presence, six to assess social presence, and 

12, cognitive presence. The reliability and validity of the factor structure of the three constructs 

in the original instrument have been widely confirmed (Arbaugh et al., 2008; Garrison et al., 

2010; Rockinson-Szapkiw et al., 2016; Shea & Bidjerano, 2010, 2012). 

Teaching Presence. Teaching presence comprises three dimensions: design and 

organization, facilitation, and direct instruction. One sample item for design and organization is 

The instructor clearly communicated important course topics. One sample item for facilitation is 

The instructor was helpful in guiding the class towards understanding course topics in a way 

that helped me clarify my thinking. And for direct instruction, a sample item is My instructor 

provided useful illustrations that helped make the course content more understandable to me. 

The Cronbach’s alpha for teaching presence was computed as .93. 

Social Presence. Social presence consisted of two dimensions: affective expression and 

open communication. One sample item for affective expression is I have a sense of belonging in 

 

31 

the course. One sample item for open communication is I felt comfortable conversing through 

the online medium. The Cronbach’s alpha score for social presence was .88. 

Cognitive Presence. Cognitive presence included triggering event, exploration, 

integration, and resolution. One sample item for triggering event is Problems posed increased my 

interest in course issues. A sample item for exploration is I utilized a variety of sources to 

explore problems. A sample item for integration is Learning activities helped me construct 

solutions; and a sample item for resolution is I can describe ways to apply the knowledge created 

in this course. The Cronbach’s alpha computed for cognitive presence was .94. 

Self-efficacy. The four-item instrument used in the present research to measure students’ 

self-efficacy was adopted from the Motivated Strategies for Learning Questionnaire (MSLQ; 

Pintrich, Smith, Garcia, & McKeachie, 1993). It was answered using a five-point Likert scale, 

again ranging from 1= strongly disagree to 5 = strongly agree. This instrument has been deemed 

to have good construct validity by previous online-learning studies (Shea & Bidjerano, 2010, 

2012). The Cronbach’s alpha for self-efficacy was found to be .92. One sample item for self-

efficacy is I believe I will receive an excellent grade in this class. 

Online Self-regulated Learning Strategies. Ultimately, the instrument used to examine 

SRL strategies was based on a 24-item questionnaire developed by Barnard et al. (2009) to 

measure students’ online self-regulation skills in terms of five different aspects: goal-setting, 

help-seeking, task strategies, self-evaluation, and time management. Shea and Bidjerano (2012) 

further refined the original questionnaire by excluding items that were conceptually ambiguous, 

and confirmed the factor structure of the new version using exploratory factor analysis (EFA). 

The results showed that it contained three latent factors: goal-setting, strategic learning, and 

help-seeking. The current study adopted Shea and Bidjerano’s version of the survey, and re-

 

32 

phrased some items to better fit the specific situation of the targeted online-learning institution. 

One sample item for goal-setting is I set goals to help me manage studying time for my online 

courses. A sample item for strategic learning is I work on extra problems or do additional 

readings in my online course beyond the assigned ones to master the course content. And a 

sample of a help-seeking item is I find someone who is knowledgeable in course content so that I 

can consult with him/her when I need help. The Cronbach’s alpha for online self-regulated 

learning strategies was .92. 

Mentor Presence. This part of the present study’s survey was adapted from previous 

literature regarding mentors’ roles (Borup & Drysdale, 2014), and includes 16 items that 

describe a variety of such roles, including problem-solver, social-relationship builder, progress 

tracker, and content instructor. The participants responded using the same five-point Likert scale 

described above. A sample of problem-solver is The mentor helped me become familiar with the 

course platform; a sample of social-relationship builder is The mentor expressed appreciation for 

my contribution; a sample of progress tracker is; and a sample of content instructor is The mentor 

helped me with content learning. The Cronbach’s alpha for online self-regulated learning 

strategies was .96. 

Satisfaction. Kuo, Walker, Schroder, and Belland (2014) developed five items to 

measure students’ satisfaction with online learning. The present study adopted four of these 

items, as the other one was not appropriate to K-12 online-learning settings. Students responded 

via the same five-point Likert-scale discussed above. The Cronbach’s alpha score for satisfaction 

was .90, and a sample item is Overall I am satisfied with this class. 

Perceived Progress. The current study’s instrument for measuring students’ perceived 

progress was adopted from Lin, Zheng, and Zhang (2016), and uses the same five-point Likert 

 

33 

scale as above. The Cronbach’s alpha for perceived progress was computed as .90, and a sample 

item is I understand most of the learning content in my class. 

Data Analysis 

 

SEM was the main form of statistical analysis used in the current study. A multivariate 

technique, it examines the complex relationships among variables in a hypothesized model to 

establish the extent to which the model fits the data (Schumacker & Lomax, 2010). In contrast to 

other statistical methods, SEM allows researchers to consider all the variables in the model 

simultaneously and to make decisions on whether to keep, reject, or modify the model (Kline, 

2011). Conducting SEM always involves the creation of a measurement model and a structural 

model. The first tests the relationships of various factor loadings with the latent factor, while 

second captures the direct and indirect structural relationships among multiple variables, with the 

inclusion of measurement errors (Kline, 2011). 

Fit statistics in SEM provide information about the degree to which a given hypothesized 

model is supported by the data. The chi-square test, a traditional approach to measuring the 

fitness of a proposed model, compares the difference between the sample and fitted covariance 

matrices (Hu & Bentler, 1999). A non-significant chi-square value indicates a good model fit and 

little difference between the sample covariance matrix and the reproduced matrix (Schumacker 

& Lomax, 2010). However, since the chi-square test is very sensitive to sample size, a large 

sample can easily result in a statistically significant chi-square value. Multiple additional 

goodness-of-fit indices are therefore usually computed (Schumacker & Lomax, 2010). Two 

useful alternative fit statistics are comparative fit index (CFI) and root mean square error of 

approximation (RMSEA) (Hu & Bentler, 1999). CFI compares the sample’s covariance matrix 

with the baseline model, taking sample size into account, while RMSEA examines how well the 

 

34 

model fits the baseline covariance matrix with optimal parameter estimates (Schumacker & 

Lomax, 2010). CFI values larger than .9 are considered to reflect an adequate fit, and larger 

than .95, a good fit. RMSEA values smaller than .08 suggest an acceptable fit, and smaller 

than .05, a good fit (Hu & Bentler, 1999). If the initial model proves to be unacceptable, it needs 

to be modified through the addition or deletion of paths until all remaining paths achieve 

statistical significance (Kline, 2011). All SEM analyses in the current study were conducted 

using Mplus7 software. 

Instrument Validation. A double-split cross-validation technique was employed to 

examine the validity of the latent factor structure in the present study’s data (Hair, Black, Babin, 

Anderson, & Tatham, 2006). This requires splitting the entire dataset into two randomly selected 

sub-datasets that are then used to validate one another (Cudek & Browne, 1983). Specifically, 

EFA is first employed to identify the factor structure of one half of the sample, and then 

confirmatory factor analysis (CFA) used to confirm such structure using the other half. 

As the current study involves CoI, online SRL strategies, and mentor presence, it was 

decided to conduct both EFA and CFA on the measured items to validate those belonging to the 

theorized latent constructs. All participants (N = 696) were randomly split into two datasets, one 

drawn from 348 participants and the other from 348. The first was used for EFA and the second 

for CFA. The 67 items included in this testing contained measurements on teaching presence (12 

items), social presence (6 items), cognitive presence (12 items), mentor presence (16 items), self-

regulated learning strategies (10 items), self-efficacy (4 items), satisfaction (3 items), and 

perceived progress (4 items). 

Exploratory Factor Analysis. EFA is used to find the latent variables that can account for 

observed variation and covariation among the observed variables (Costello & Osborne, 2005). 

 

35 

An exploratory method, the small number of latent constructs it identifies can account for a 

larger set of variables. 

To decide whether the data is suitable for factor analysis, two types of tests – KMO and 

Bartlett’s Test of Sphericity are first conducted to evaluate sampling adequacy (Bartlett, 1950; 

Kaiser, 1974). Kaiser suggested that when judging the result of KMO, values less than .49 were 

unacceptable; .50 to .59, miserable; .60 to .69, mediocre; .70 to .79, middling; .80 to .89, 

meritorious; and .90 to 1.00, marvelous. The null hypothesis for the Bartlett’s Test of Sphericity 

is that the sample intercorrelation matrix is from a population in which the variables are not 

collinear. The results consist of chi-square statistics, the significance of which can be used to 

decide whether to accept or reject the null hypothesis. If the null hypothesis (i.e., that the test 

result is significant) is rejected, it is suitable to conduct factor analysis for the measured 

variables. 

Following Kaiser’s rule, eigenvalues larger than 1 and the scree test examining the graph 

of the eigenvalues and looking for the break point in the data were used to decide the number of 

factors retained (Kaiser & Rice, 1974). Both the unrotated results and the rotated results were 

examined. Since the original loadings from the unrotated results could not provide a direct 

interpretation, rotating the factors was chosen to present the result. The assumption of the data 

was that the resulting factors were correlated with each other, so oblique rotation (Promax 

rotation) was preferred over orthogonal rotation. The factor loading of the item should be greater 

than 0.4 under the relevant component and less than 0.4 under all the other components (Stevens, 

1996). 

Confirmatory Factor Analysis. CFA answers questions about the extent to which the 

items under the latent constructs in a model indeed measure those latent constructs (Wang & 

 

36 

Wang, 2012). CFA’s primary concern is how well the model fits the data. In the current study, 

chi-square statistics, together with multiple goodness-of-fit indices including RMSEA and CFI, 

were applied to determine the fitness of the models used (Hu & Bentler, 1999). If a hypothesized 

CFA model is deemed a good fit for the data, it means that the tested factorial structure in the 

measurement model is valid. 

As well as using model-fit indices, researchers should refer to the convergent and 

discriminant validity of their measurements. Convergent validity is how well a factor is 

measured by its associated survey items (Campbell & Fiske, 1959). There are two ways to decide 

the convergent validity of the measured items: 1) testing the factor loading of each of them, and 

2) comparing the average variance extracted (AVE) and the composite reliability (Fornell & 

Larcker, 1981; Hair et al., 2006). A factor loading greater than .70 is thought to indicate a well-

defined construct (Gefen, Straub, & Boudreau, 2000; Hair et al., 2006). In this study, all factor 

loadings of the items under each latent construct were computed and compared to see whether 

they are greater than .70. Another way of testing convergent validity is to compare AVE against 

composite reliability. To compute AVE, one needs to sum up each squared factor loading and 

divide it by the number of indicators in the latent construct (Hair et al., 2006). If 1) the composite 

reliability is greater than the AVE, and 2) the AVEs are greater than .50, convergent validity is 

deemed satisfactory. 

Discriminant validity is evaluated to test whether a latent construct can explain more 

variance in the observed variables than either measurement error or other latent constructs under 

the measurement model (Fornell & Larcker, 1981). Specifically, the AVEs of any two latent 

constructs should both exceed the shared variance (i.e., square of the correlation) between these 

 

37 

two constructs. In other words, the square root of the AVE of each latent construct should exceed 

the inter-construct correlation between this latent construct and another latent construct. 

Structural Model. The second step of SEM is to examine the structural model using 

statistical-significance tests of the path coefficients, to test the researchers’ hypotheses (Kline, 

2011). A structural model captures the structural relations among multiple variables, both latent 

and observed, while simultaneously taking measurement errors into account (Kline, 2011). 

According to Cohen (1988), a standardized path coefficient between .1 and .3 reflects a small 

effect size, .3 to .5 a medium effect size, and >.5, a large effect size. 

Testing Mediation. SEM allows researchers to test direct, indirect, and total effects 

simultaneously (Kline, 2011). A direct effect is the direct path from one construct to another; an 

indirect one is the path from one construct to another through one or more mediators in the 

model; and the total effect is the sum of the direct and indirect effects. 

Bootstrapping is a method used to determine the confidence intervals (CIs) for mediation 

effects. It generates an empirical approximation of the sampling distribution through repeated 

resampling, and establishes CIs and calculates p-values through such resampling (MacKinnon, 

Lockwood, & Williams, 2004). Bootstrapping with 5,000 resamples was employed to build CIs 

(Preacher & Hayes, 2006). 

RQ1. Shea’s Model and the Alternative Model. Two separate SEM models, Shea’s 

Model and the Alternative Model (adding one more direct path from TP to SRL) were 

established and compared. Chi-square significance testing and model-fit indices were used as 

criteria for deciding whether a better model for the data in the present study could be developed. 

Any insignificant direct effects were removed. 

 

38 

RQ2. Model with Mentor Presence. Based on the model identified from RQ1, to 

answer RQ2, one more variable (i.e., Mentor Presence) was added to the SEM model. Again, 

model-fit indices and standardized path coefficients were examined to determine whether the 

model could be accepted. Then, paths with insignificant direct effects were removed, and the 

final model tested for mediation effects. 

RQ3. Model with Learning Outcomes. Since previous studies found that the two 

outcome variables used in this study, satisfaction and perceived progress, differed in terms of 

their relationships with learning variables (Eom, Wen, & Ashill, 2006), it was decided that they 

would be used as endogenous variables, separately in two different models. Final grade would 

also be used as a spate learning outcome variable. As with the SEM analysis conducting in 

connection with RQ2, model-fit indices and standardized path coefficients were first analyzed in 

each model. Then, if a direct effect between two variables was identified as insignificant, its path 

was removed from the SEM model. Then, the bootstrap method with 5,000 resamples was used 

to determine the significance of mediation effects. 

RQ4. Learning at school vs. Learning at Home. To answer RQ4, regarding the 

difference between students who completed the majority of their whether at school or at home, 

descriptive statistics and ANOVA were used. These two groups of students were compared 

based on the differences in the items of different presences, including social presence, cognitive 

presence, mentor presence, self-efficacy, self-regulated learning strategies, satisfaction, 

perceived progress, and final grade.

 

39 

CHAPTER 4 

RESULTS 

This chapter includes statistical analyses of this study’s data, including tests of the 

validity of the latent constructs, examination of descriptive statistics, correlations, the 

measurement model, and analyses leading to the answers for each research question. 

Exploratory Factor Analysis 

As previously mentioned, the data subset for possible EFA was drawn from 348 

participants randomly selected from the entire sample (N = 696). The KMO test and the 

Bartlett’s Test of Sphericity were then conducted to decide the appropriateness of conducting 

EFA on this data. The KMO result was .95, and the Bartlett’s Test of Sphericity result was 

20127.40 (df = 2211, p < .001), which showed that the intercorrelation matrix in the sample was 

suitable for factor analysis. 

Next, PCA with Promax rotation was conducted to arrive at the factor solution of the 

measured variables. Following Kaiser’s rule and scree-plot results (Kaiser & Rice, 1974), a 

seven-factor solution with eigenvalues >1 was extracted, and explained 85.04% of the variance 

among the 67 tested items (see Table 2). The first factor (eigenvalue = 26.99) explained 55.79% 

of the variance, and contained all the 16 items from the mentor presence and one item from self-

regulated learning strategies (i.e., If needed, I try to ask my online teacher/mentors about the 

question that I don’t understand) which also reflected the help that a student received from 

mentors. The factor loadings were all larger than .40, ranging from .40 to .80 (see Table 3). 

Table 2.  Test of the Validity of the Latent Constructs – Eigenvalue of the Exploratory 
Factor Analysis (N = 348) 

 
Factor 1 

Eigenvalue 

Difference 

Proportion 

Cumulative 

26.99 

22.11 

.5579 

.5579 

 
 

 

40 

Table 2 (cont’d) 

Factor 2 
Factor 3 
Factor 4 
Factor 5 
Factor 6 
Factor 7 

 

4.88 
2.84 
2.24 
1.77 
1.33 
1.10 

2.04 
.60 
.48 
.43 
.23 
.07 

.1008 
.0587 
.0463 
.0365 
.0275 
.0227 

.6587 
.7174 
.7637 
.8002 
.8277 
.8504 

Table 3.  Test of the Validity of the Latent Constructs - Exploratory Factor Analysis (N = 
348) 

 

TP_1 
TP_2 
TP_3 
TP_4 
TP_5 
TP_6 
TP_7 
TP_8 
TP_9 
TP_10 
TP_11 
TP_12 
SP_1 
SP_2 
SP_3 
SP_4 
SP_5 
SP_6 
CP_1 
CP_2 
CP_3 
CP_4 
CP_5 
CP_6 
CP_7 
CP_8 
CP_9 
CP_10 
CP_11 
CP_12 
EFFI_1 
EFFI_2 

 

Factor

Factor 

Factor  

Factor

Factor 

Factor

Factor 

1 
.23 
.24 
.21 
.21 
.29 
.29 
.28 
.23 
.36 
.23 
.22 
.31 
.30 
.26 
.21 
.25 
.29 
.28 
.15 
.20 
.21 
.16 
.22 
.14 
.17 
.24 
.28 
.08 
.17 
.07 
.18 
.18 

2 
.21 
.21 
.14 
.19 
.26 
.28 
.34 
.27 
.23 
.24 
.25 
.25 
.39 
.35 
.33 
.31 
.37 
.40 
.50 
.60 
.62 
.55 
.56 
.63 
.66 
.67 
.62 
.68 
.71 
.66 
.09 
.20 

4 
.25 
.20 
.17 
.18 
.24 
.20 
.06 
.11 
.04 
.16 
.15 
.11 
.23 
.11 
.07 
.23 
.33 
.21 
.16 
.23 
.27 
.11 
.07 
.17 
.19 
.22 
.28 
.34 
.33 
.30 
.78 
.83 

3 
.69 
.64 
.57 
.40 
.67 
.67 
.58 
.63 
.57 
.73 
.68 
.61 
.42 
.40 
.23 
.28 
.26 
.19 
.31 
.30 
.25 
.20 
.26 
.28 
.37 
.32 
.27 
.21 
.22 
.09 
.13 
.16 

41 

5 
.03 
.08 
.10 
.05 
.10 
.18 
.21 
.16 
.22 
.06 
.08 
.07 
.25 
.24 
.13 
.18 
.18 
.19 
.13 
.22 
.27 
.29 
.31 
.13 
.14 
.17 
.13 
.07 
.07 
.02 
.14 
.14 

6 
.11 
.10 
.03 
.07 
.13 
.14 
.18 
.10 
.07 
.03 
.06 
-.02 
.12 
.10 
.01 
.01 
.00 
.06 
.01 
.00 
.06 
.09 
.09 
.09 
.09 
.06 
.11 
.08 
.03 
.03 
.03 
.03 

7 
.11 
.08 
.19 
.03 
.10 
.10 
.02 
.03 
.05 
.05 
.04 
.09 
.25 
.30 
.30 
.24 
.24 
.17 
.23 
.19 
.10 
-.08 
.04 
.14 
.11 
.07 
.04 
.04 
.00 
.09 
.01 
.02 

Table 3 (cont’d) 

.20 
.14 
.21 
.10 
.18 
.17 
.16 
.20 
.22 
.23 
.40 
.36 
.79 
.79. 
.76 
.77 
.72 
.80 
.64 
.68 
.63 
.49 
.47 
.48 
.61 
.68 
.70 
.63 
.22 
.19 
.12 
.18 
.13 
.16 
.22 

EFFI_3 
EFFI_4 
SRL_1 
SRL_2 
SRL_3 
SRL_4 
SRL_5 
SRL_6 
SRL_7 
SRL_8 
SRL_9 
SRL_10 
MP_1 
MP_2 
MP_3 
MP_4 
MP_5 
MP_6 
MP_7 
MP_8 
MP_9 
MP_10 
MP_11 
MP_12 
MP_13 
MP_14 
MP_15 
MP_16 
SATIS_1 
SATIS_2 
SATIS_3 
PROG_1 
PROG_2 
PROG_3 
PROG_4 
TP = Teaching Presence, SP = Social Presence, CP = Cognitive Presence, MP = Mentor 
Presence, SRL = Online SRL Strategies, EFFI = Self-efficacy, SATIS = Satisfaction, PROG = 
Perceived Progress 
 

-.02 
.14 
-.03 
-.07 
.02 
.11 
.14 
.18 
.22 
.17 
.07 
.08 
.16 
.15 
.13 
.12 
.04 
.02 
-.08 
.18 
-.03 
-.02 
.11 
.05 
.06 
.05 
.04 
.00 
.45 
.58 
.52 
.09 
.22 
.16 
.22 

.14 
.27 
.23 
.17 
.22 
.16 
.10 
.18 
.23 
.33 
.07 
.12 
.06 
.07 
.15 
.24 
.15 
.15 
.11 
.20 
.16 
.16 
.10 
.04 
.16 
.17 
.14 
.13 
.47 
.39 
.38 
.40 
.44 
.45 
.45 

.15 
.22 
.16 
.15 
.15 
.13 
.09 
.04 
.11 
.16 
.28 
.31 
.19 
.18 
.11 
.16 
.19 
.23 
.18 
.22 
.15 
.15 
.15 
.16 
.23 
.20 
.23 
.22 
.39 
.36 
.27 
.12 
.12 
.12 
.13 

.81 
.68 
.37 
.19 
.40 
.25 
.04 
.05 
.10 
.19 
.13 
.13 
.13 
.17 
.12 
.10 
.22 
.15 
.20 
.21 
.17 
.13 
.07 
.10 
.10 
.08 
.12 
.19 
.37 
.35 
.38 
.63 
.67 
.59 
.59 

.17 
.13 
.52 
.69 
.61 
.69 
.39 
.45 
.55 
.51 
.39 
.39 
.16 
.13 
.01 
.07 
.16 
.19 
-.02 
.08 
.04 
.04 
.08 
.10 
.16 
.11 
.11 
.05 
.12 
.10 
.12 
.14 
.03 
.04 
.08 

-.01 
.03 
.06 
.06 
.12 
.02 
.14 
.21 
.02 
.09 
-.01 
.02 
.04 
-.03 
-.05 
.02 
.11 
.14 
.17 
.27 
.36 
.61 
.65 
.56 
.56 
.18 
.17 
.09 
.13 
.06 
.07 
.10 
.11 
.15 
.11 

The second factor (eigenvalue = 4.88) explained 10.08% of the variance and contained all 

the 12 items from cognitive presence (factor loadings ranging from .50 to .71), one item from 

 

42 

satisfaction (i.e., Overall I am satisfied with this class. Factor loading was .47), and all the four 

items from perceived progress (factor loadings ranged from .40 to .45). 

The third factor (eigenvalue = 2.84) explained 5.87% of the variance and contained all 

the 12 items from teaching presence and two items from social presence. The factor loadings for 

teaching presence ranged from .40 to .73, and the factor loadings for social presence was .42 

and .40.   

The fourth factor (eigenvalue = 2.24) explained 4.63% of the variance and contained four 

items, all from the self-efficacy items. Their factor loadings were .78, .83, .81 and .68 

respectively. 

The fifth factor (eigenvalue = 1.77) explained 3.65% of the variance. The factor loadings 

larger than .40 were all from the self-regulated learning items. Of the 10 items, 7 items had factor 

loadings ranged from .45 to .69. The remaining three items showed factor loadings of .39, which 

was close to the .40 threshold.  

The sixth factor (eigenvalue = 1.33) explained 2.75% of the variance and contained four 

items from the mentor presence, ranging from .56 to .65. These four items were also shown in 

the first factor. 

The last factor (eigenvalue = 1.10) explained 2.27% of the variance and contained three 

items, all from the satisfaction items. Their factor loadings were .45, .58, and .52 respectively. 

The results of PCA showed that the 67 tested items reflected seven components, which 

largely reflected the eight latent variables measured in this study. Of the seven components, 

cognitive presence and perceived progress were found under the same factor. A potential 

explanation was that both latent variables measured students’ conceptions of their cognitive 

 

43 

achievement in the course. Thus, the classification from the EFA using the half dataset from the 

study is consistent with the proposed measurement model in this study.  

Confirmatory Factor Analysis 

The other data subset (N = 348), representing another half of the total data, was used for 

CFA of the construct validity of the latent factors. As shown in Table 4, there were eight latent 

variables in the model: teaching presence (12 items), social presence (6 items), cognitive 

presence (12 items), mentor presence (16 items), online SRL strategies (10 items), self-efficacy 

(4 items), satisfaction (3 items), perceived progress (4 items), The measurement model yielded χ2 

(2116) = 8115.876, p < .001, RMSEA = .065 (90% CI from .064 to .067), CFI = .831, and 

SRMR = .047. According to the modification suggestions, correlations were added within each 

latent variable, including two correlations in teaching presence, four correlations in social 

presence, six correlations in cognitive presence, four correlations in online SRL strategies, and 

11 correlations in mentor presence. The revised model yielded χ2 (2089) = 4632.357, p < .001, 

RMSEA = .043 (90% CI from .041 to .044), CFI = .929, and SRMR = .040, indicating a 

sufficient fit for the data.  

Table 4.  Test of the Validity of the Latent Constructs – Confirmatory Factor Analysis (N = 
348) 

 

Item 
TP_1 
TP_2 
TP_3 
TP_4 
TP_5 
TP_6 
TP_7 
TP_8 
TP_9 
TP_10 
TP_11 
TP_12 

 
 
TP 
 
 
 
 
 
 
 
 
 
 
 

 

Standardized 

β 

.74*** 
.68*** 
.69*** 
.51*** 
.79*** 
.80*** 
.76*** 
.74*** 
.76*** 
.77*** 
.67*** 
.68*** 

44 

S.E. 
.03 
.03 
.03 
.04 
.02 
.02 
.03 
.03 
.03 
.02 
.03 
.03 

SP_1 
SP_2 
SP_3 
SP_4 
SP_5 
SP_6 

 

CP_1 
CP_2 
CP_3 
CP_4 
CP_5 
CP_6 
CP_7 
CP_8 
CP_9 
CP_10 
CP_11 
CP_12 

 

SRL_1 
SRL_2 
SRL_3 
SRL_4 
SRL_5 
SRL_6 
SRL_7 
SRL_8 
SRL_9 
SRL_10 

EFFI_1 
EFFI_2 
EFFI_3 
EFFI_4 

 

MP_1 
MP_2 
MP_3 
MP_4 
MP_5 
MP_6 
MP_7 
MP_8 

Table 4 (cont’d) 

SP 
 
 
 
 
 
 
CP 
 
 
 
 
 
 
 
 
 
 
 
 
SRL 
 
 
 
 
 
 
 
 
 

EFFI 
 
 
 
 
MP 
 
 
 
 
 
 
 

 

 

.84*** 
.77*** 
.73*** 
.70*** 
.70*** 
.62*** 

 

.56*** 
.70*** 
.74*** 
.64*** 
.72*** 
.73*** 
.81*** 
.78*** 
.79*** 
.70*** 
.74*** 
.72*** 

 

.73*** 
.71*** 
.76*** 
.77*** 
.40*** 
.53*** 
.66*** 
.69*** 
.54*** 
.50*** 

.91*** 
.93*** 
.84*** 
.77*** 

 

.79*** 
.77*** 
.74*** 
.76*** 
.85*** 
.84*** 
.76*** 
.77*** 

45 

.02 
.03 
.03 
.03 
.03 
.04 

 

.04 
.03 
.03 
.04 
.03 
.03 
.02 
.02 
.02 
.03 
.03 
.03 

 

.03 
.03 
.03 
.03 
.05 
.04 
.04 
.03 
.04 
.05 

.01 
.01 
.02 
.02 

 

.02 
.02 
.03 
.03 
.02 
.02 
.03 
.02 

Table 4 (cont’d) 

.02 
.03 
.04 
.03 
.03 
.03 
.02 
.03 

.78*** 
.72*** 
.62*** 
.67*** 
.73*** 
.77*** 
.77*** 
.71*** 

MP_9 
MP_10 
MP_11 
MP_12 
MP_13 
MP_14 
MP_15 
MP_16 

 
 
 
 
 
 
 
 
 
SATIS 
 
 
 
PROG 
 
 
 
***p < .001. TP = Teaching Presence, SP = Social Presence, CP = Cognitive Presence, MP = 
Mentor Presence, SRL = Online SRL Strategies, EFFI = Self-efficacy, SATIS = Satisfaction, 
PROG = Perceived Progress 
 

PROG_1 
PROG_2 
PROG_3 
PROG_4 

.73*** 
.87*** 
.85*** 
.84*** 

SATIS_1 
SATIS_2 
SATIS_3 

 

 

.01 
.01 
.02 

.03 
.02 
.02 
.02 

 

 

.92*** 
.91*** 
.80*** 

 

 

Composite Reliability. Table 5 presents the composite-reliability scores of each latent 

construct. Four steps were used to calculate them: 1) all factor loadings were summed; 2) this 

sum was squared to obtain the Square of the Sum of All Factor Loadings under the Same Latent 

Construct (SSI); 3) all error variances of each indicator were summed to obtain the Sum of Error 

Variances (SEV) of each indicator; and 4) composite reliability was computed using the formula 

SSI/(SSI+SEV). According to Nunnally and Bernstein (1994), a composite reliability larger than 

.70 is recommended. The computed composite-reliability scores for each of the eight latent 

constructs implied that all were satisfactory in this regard, with TP = .99, SP = .99, CP = .99, MP 

= .98, SRL = .97, EFFI = .99, SATIS = .99 and PROG = .99. 

Table 5.  Test of the Validity of the Latent Constructs - The Standardized Factor Loading, 
SSI, SEV, AVE, and Composite Reliability of the Latent Constructs (N = 348) 

 

SSI 

SEV 

AVE 

Composite 
Reliability 

TP 

73.79 

.34 

.52 

.99 

 

46 

Table 5 (cont’d) 

.18 
.35 
.37 
.06 
.31 
.04 
.09 

.53 
.52 
.41 
.75 
.57 
.77 
.68 

6.92 
10.82 

.99 
.99 
.97 
.99 
.98 
.99 
.99 

19.01 
74.48 
39.56 
11.90 
145.20 

SP 
CP 
SRL 
EFFI 
MP 
SATIS 
PROG 
SSI = the square of the sum of all factor loadings under the same latent construct; SEV = the sum 
of all error variances of each indicator; AVE = the sum of each squared factor loading; 
Composite Reliability = SSI/(SSI + SEV). 
TP = Teaching Presence, SP = Social Presence, CP = Cognitive Presence, MP = Mentor 
Presence, SRL = Online SRL Strategies, EFFI = Self-efficacy, SATIS = Satisfaction, PROG = 
Perceived Progress 
 

Convergent Validity. Two criteria were used to test the convergent validity of the 

measured items: 1) the factor loading of each, and 2) comparing the Average Variance Extracted 

(AVE) and the composite reliability (Fornell & Larcker, 1981; Hair et al., 2006). 

Factor Loading. The standardized factor loadings of each item on the latent constructs 

are shown on Table 4. Of the 67 factor loadings, 51 items showed factor loadings larger than .70, 

the threshold recommended by Hair et al. (2006) and Gefen et al. (2000). For the remaining 16 

items, 10 items had factor loadings larger than .60, 5 items had factor loadings larger than .50, 

and 1 item showed factor loading larger than .40. Therefore, the construct validity of the 

measurement model can be deemed satisfactory.  

Average Variance Extracted. According to Hair et al. (2006), convergent validity is 

adequate when 1) composite reliability is greater than AVE, and 2) all AVEs are larger than .50. 

As shown in Table 5, all the latent constructs’ composite reliabilities were greater than the 

corresponding AVEs, all of which in turn were greater than .50 (i.e., TP = .52, SP = .53, CP = 

.52, EFFI = .75, MP = .57, SATIS = .77, and PROG = .68) except in the case of SRL (.41). 

 

47 

Nevertheless, as most AVEs exceeded the required level and all the composite reliabilities were 

greater than the AVEs, it is reasonable to suggest that convergent validity was satisfactory. 

Discriminant Validity 

As mentioned above, the AVEs of any two latent constructs should be greater than the 

shared variance (i.e., square of the correlation) between those constructs (Fornell & Larcker, 

1981). Table 6 shows the inter-construct correlation between each latent construct (off the 

diagonal) and the square roots of their AVEs (on the diagonal). All the AVE values were greater 

than all the inter-construct correlations, and the measurement model therefore met the 

discriminant-validity criterion. 

Table 6.  Test of the Validity of the Latent Constructs - Discriminant Validity for the 
Measurement Model (N = 348) 

 
 
 
 
 

 
 
 
 
 
 

 
 

(.72 a) 

 
 
 

(.64 a) 

 
 
 
 

 (.73 a) 

(.72 a) 

 

TP 

SP 

CP 

SRL 

EFFI  MP 

SATIS  PROG 

.44 
.38 
.43 
.51 
.67 
.41 

.56 
.32 
.23 
.29 
.44 
.47 
.29 

Latent 
construct 
TP 
SP 
CP 
SRL 
EFFI 
MP 
SATIS 
PROG 
a Diagonals in parentheses are square roots of the AVE from the latent construct. Off-diagonals 
are the inter-construct correlations between each of the latent construct. 
TP = Teaching Presence, SP = Social Presence, CP = Cognitive Presence, MP = Mentor 
Presence, SRL = Online SRL Strategies, EFFI = Self-efficacy, SATIS = Satisfaction, PROG = 
Perceived Progress 
 
 
Descriptive Statistics 

.27 
.31 
.32 
.43 
.29 

.31 
.30 
.37 
.28 

(.88 a) 

.51 

(.76 a) 

.44 
.27 

(.86 a) 

.29 
.55 
.41 

(.82 a) 

 
 
 
 
 
 
 

As mentioned earlier, Table 1 shows the descriptive statistics of the items measured in 

the study. The items included measurements of teaching presence (12 items), social presence (6 

items), cognitive presence (12 items), mentor presence (16 items), online SRL strategies (10 

items), self-efficacy (4 items), satisfaction (3 items), perceived progress (4 items), and final 

 

48 

grade. The reliabilities of these variables were: teaching presence (α = .93), social presence (α = 

.88), cognitive presence (α = .94), mentor presence (α = .96), online SRL strategies (α = .87), 

self-efficacy (α = .92), satisfaction (α = .90), perceived progress (α = .90). Table 1 also shows the 

demographic information of this study, including gender, grade, ethnicity, subject, and whether 

or not they take the course for credit recovery purpose. Table 7 presents the correlations between 

each of the measured variables, and indicates that all variables were significantly correlated with 

each other at the p = .001 level. 

Table 7.  Correlation Table  

 
1. TP 
2. MP 
3. SP 
4. CP 
5. EFFI 
6. SRL 
7. SATIS 
8. PROG 
9. Final grade 

2 
 

1 
1 
.61***  1 
.68*** 
.68*** 
.45*** 
.45*** 
.62*** 
.52*** 
.27*** 

3 
 
 

4 
 
 
 

5 
 
 
 
 

6 
 
 
 
 
 

.51***  1 
.51*** 
.38*** 
.49*** 
.45*** 
.43*** 
.19*** 

.69***  1 
.49*** 
.51*** 
.68*** 
.54*** 
.26*** 

.56***  1 
.56*** 
.70*** 
.65*** 
.32*** 

.44***  1 
.59*** 
.70*** 
.25*** 

7 
 
 
 
 
 
 

8 
 
 
 
 
 
 
 

9 
 
 
 
 
 
 
 
 

.47***  1 
.45*** 
.24*** 

.70***  1 
.40*** 

.26***  1 

***p < .001. TP = Teaching Presence, SP = Social Presence, CP = Cognitive Presence, MP = 
Mentor Presence, SRL = Online SRL Strategies, EFFI = Self-efficacy, SATIS = Satisfaction, 
PROG = Perceived Progress 
 

Measurement Model 

CFA was performed to test the fitness of the latent variables for the data. As shown in 

Table 8, there were eight latent variables in the model: teaching presence (TP, 12 items), social 

presence (SP, 6 items), cognitive presence (CP, 12 items), mentor presence (MP, 16 items), 

online SRL strategies (SRL, 10 items), self-efficacy (EFFI, 4 items), satisfaction (SATIS, 3 

items), perceived progress (PROG, 4 items). The measurement model yielded χ2 (2116) = 

8115.876, p < .001, RMSEA = .065 (90% CI from .064 to .067), CFI = .831, and SRMR = .047. 

According to the modification suggestions, correlations were added within each latent variable, 

 

49 

including two correlations in teaching presence, four correlations in social presence, six 

correlations in cognitive presence, four correlations in online SRL strategies, and 11 correlations 

in mentor presence. The revised model yielded χ2 (2089) = 4632.357, p < .001, RMSEA = .043 

(90% CI from .041 to .044), CFI = .929, and SRMR = .040, indicating a sufficient fit for the 

data.  

Table 8.  Measurement Model 

 
 
TP_1 
TP_2 
TP_3 
TP_4 
TP_5 
TP_6 
TP_7 
TP_8 
TP_9 
TP_10 
TP_11 
TP_12 
SP_1 
SP_2 
SP_3 
SP_4 
SP_5 
SP_6 
CP_1 
CP_2 
CP_3 
CP_4 
CP_5 
CP_6 
CP_7 
CP_8 
CP_9 
CP_10 
CP_11 
CP_12 
SRL_1 
SRL_2 
SRL_3 

 
 
TP by 
 
 
 
 
 
 
 
 
 
 
 
SP by 
 
 
 
 
 
CP by 
 
 
 
 
 
 
 
 
 
 
 
SRL by 
 
 

 

Standardized 

S.E. 
.02 
.02 
.02 
.03 
.02 
.01 
.02 
.02 
.02 
.02 
.02 
.02 
.02 
.02 
.03 
.03 
.02 
.03 
.01 
.01 
.01 
.03 
.02 
.02 
.02 
.02 
.02 
.02 
.02 
.02 
.02 
.02 
.02 

β 

.76*** 
.71*** 
.67*** 
.52*** 
.81*** 
.83*** 
.77*** 
.75*** 
.75*** 
.78*** 
.71*** 
.70*** 
.81*** 
.75*** 
.64*** 
.67*** 
.71*** 
.63*** 
.62*** 
.75*** 
.77*** 
.64*** 
.71*** 
.75*** 
.81*** 
.81*** 
.80*** 
.71*** 
.75*** 
.68*** 
.70*** 
.69*** 
.76*** 

50 

 
 
Table 8 (cont’d) 

SRL_4 
SRL_5 

.73*** 
.43*** 

.02 
.04 

 

 

.03 
.03 
.02 
.03 
.03 

.01 
.01 
.01 
.02 

 

 

.87*** 
.92*** 
.87*** 
.78*** 

.54*** 
.67*** 
.73*** 
.55*** 
.54*** 

SRL_6 
SRL_7 
SRL_8 
SRL_9 
SRL_10 
 
EFFI_1 
EFFI_2 
EFFI_3 
EFFI_4 
 
MP_1 
MP_2 
MP_3 
MP_4 
MP_5 
MP_6 
MP_7 
MP_8 
MP_9 
MP_10 
MP_11 
MP_12 
MP_13 
MP_14 
MP_15 
MP_16 
 
SATIS_1 
SATIS_2 
SATIS_3 
 
PROG_1 
PROG_2 
PROG_3 
PROG_4 

 
 
 
 
 
 
EFFI by 
 
 
 
 
MP by 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
SATIS 
 
 
 
PROG 
 
 
 
***p < .001. TP = Teaching Presence, SP = Social Presence, CP = Cognitive Presence, MP = 
Mentor Presence, SRL = Online SRL Strategies, EFFI = Self-efficacy, SATIS = Satisfaction, 
PROG = Perceived Progress 

.78*** 
.77*** 
.71*** 
.77*** 
.82*** 
.86*** 
.73*** 
.80*** 
.78*** 
.70*** 
.64*** 
.67*** 
.78*** 
.78*** 
.79*** 
.72*** 

.02 
.02 
.02 
.02 
.01 
.01 
.02 
.02 
.02 
.02 
.02 
.02 
.02 
.02 
.02 
.02 

.74*** 
.88*** 
.87*** 
.86*** 

.91*** 
.90*** 
.80*** 

 

 

 

 

.01 
.01 
.02 

.02 
.01 
.01 
.01 

 

 

51 

RQ1. Shea’s Model and the Alternative Model 

To answer RQ1, regarding two models of the structural relationships among teaching 

presence, social presence, self-efficacy, online SRL strategies, and cognitive presence, two SEM 

analyses were conducted using Mplus7. The first model (see Fig. 1) was developed by Shea and 

Bidjerano (2012). Building on it, the alternative model (see Fig. 2) proposed by the current study 

added one more path in which teaching presence predicts online SRL strategies.  

Both models had the same five latent variables. The five latent variables were TP, SP, 

SRL, EFFI, CP. The latent construct of TP includes 12 indicators, measuring teacher’s direct 

instruction, facilitation, and design/organization of the course. CP was measured by 12 

indicators, measuring triggering event, exploration, integration, and resolution. SP was measured 

by 6 items, including affective expression and open communication. SRL was measured by 10 

items, including goal-setting, strategic learning, and help seeking. EFFI was measured by 4 

indicators.   

Table 9 presents a comparison between Shea’s model and the alternative model in terms 

of model fit. For Shea’s model, the results were χ2 (878) = 2368.510 (p < .001), RMSEA = .050 

with a 90% CI from .048 to .053, CFI = .926, and SRMR = .074. For the alternative model, they 

were χ2 (877) = 2295.049 (p < .001), RMSEA = .049 with a 90% CI from .047 to .052, CFI = 

.930, and SRMR = .045. 

Table 9.  Fit Indices of Shea's Model and the Alternative Model 

 
χ2 
df 
p 
RMSEA 
90% CI of RMSEA 
CFI 
SRMR 
 

Shea’s Model 

Alternative Model 

2368.510 

2295.049 

878 

< .001 
.050 

877 

< .001 
.049 

.048, .053 

.047, .052 

.926 
.074 

.930 
.045 

 

52 

Since the first model was nested in the second model, chi-square difference testing could 

be used to compare the two models. The chi-square difference was 73.461 (df = 1, p < .001). 

Therefore, the alternative model proposed by the present study, with one extra path from TP to 

SRL, can be deemed more appropriate than Shea’s model to explaining the structural 

relationships between the components of CoI and SRL among this sample of K-12 online 

learners. 

Table 10 shows the standardized coefficients of the paths between CoI, EFFI, and SRL. 

The path coefficients of the hypothetical alternative model are also shown in Fig. 5, with all 

straight lines representing significant predictions, and all dotted lines, non-significant 

predictions. As the figure indicates, all the path estimates were statistically significant except for 

the path from TP to EFFI (β = .08, p > .05). Therefore, it was decided to remove this 

insignificant path and run a new model named the Modified Alternative Model.  

Table 10.  The Structural Model for the Alternative Model 

 
 
TP  
SP 
EFFI 
TP  
SRL 
EFFI 
TP 
SP 

 
 
EFFI on 
 
SRL on 
 
CP on 
 
 
 
***p < .001. TP = Teaching Presence, SP = Social Presence, CP = Cognitive Presence, SRL = 
Online SRL Strategies, EFFI = Self-efficacy 
 
 

S.E. 
.08 
.08 
.04 
.04 
.04 
.04 
.06 
.06 

Standardized 

β 
.08 

.50*** 
.37*** 
.37*** 
.24*** 
.13*** 
.24*** 
.40*** 

 

 

 

 

53 

Figure 5. The Path Coefficients of the Alternative Model 

 

The standardized coefficients of the structural model for the Modified Alternative Model 

are shown in Table 11 and Fig. 6. The path from SP to EFFI had a large effect size (β = .57). The 

path from SP to CP had a medium effect size, with a standardized factor loading of .41. All the 

rest of the paths had small effect sizes (EFFI to SRL: .38, TP to SRL: .37, SRL to CP: .24, TP to 

CP: .23, and EFFI to CP: .12). Around 33% of the total variance in EFFI was accounted for by 

SP, around 40.5% of the total variance in SRL was accounted for by TP and EFFI, and around 

72.4% of the total variance in CP was accounted for by TP, SP, SRL, and EFFI. 

 

 

54 

Table 11.  The Structural Model for the Modified Alternative Model 

 
 
SP 
EFFI 
TP  
SRL 
EFFI 
TP 
SP 

 
 
EFFI on 
SRL on 
 
CP on 
 
 
 
***p < .001. TP = Teaching Presence, SP = Social Presence, CP = Cognitive Presence, SRL = 
Online SRL Strategies, EFFI = Self-efficacy 
 

S.E. 
.03 
.04 
.04 
.04 
.04 
.06 
.07 

Standardized 

β 

.57*** 
.38*** 
.37*** 
.24*** 
.12*** 
.23*** 
.41*** 

Figure 6. The Path Coefficients of the Modified Alternative Model 

 

 

55 

 

RQ2. Model with Mentor Presence 

To establish the structural relationships between CoI constructs, SRL, self-efficacy and 

mentor presence, the Modified Alternative Model presented above was modified further via the 

addition of MP. Again, all the SEM analyses were conducted using Mplus7. To differentiate this 

new model from previous ones, it will be referred to as the “Hypothesized Alternative Model 

with Mentor Presence”. It consisted of six latent variables – TP, SP, SRL, CP, EFFI, and MP. 

The indicators for the latent constructs TP, SP, SRL, CP, and EFFI were the same as those in the 

Modified Alternative Model. For the newly added latent variable MP, 16 measured indicators 

were included. Fig. 3 shows this new model.  

The new model was used to explore the hypotheses that MP predicts EFFI, SRL, and CP. 

Tests of model fit yielded results of χ2 (1670) = 3846.492, (p < .001), RMSEA = .044 with a 90% 

CI from .042 to .046, CFI = .929, and SRMR = .043. 

Table 12 presents the standardized coefficients for the Hypothesized Alternative Model 

with Mentor Presence, and its standardized coefficients are shown in Fig. 7. The straight lines 

represent significant predictions, and the dotted lines non-significant ones. As Fig. 7 shows, only 

two path estimates were statistically non-significant: from MP to CP (β = -.03, p > .05), and from 

MP to EFFI (β = .09, p > .05). Again, it was decided to remove the two insignificant paths and 

run a new model named the “Modified Alternative Model with Mentor Presence”. 

Table 12.  The Structural Model for the Model with Mentor Presence 

 
 
SP 
MP 
EFFI 
TP  
MP 

 
 
EFFI on 
 
SRL on 
 
 

 

Standardized 

β 

.51*** 

.09 

.35*** 
.22*** 
.24*** 

S.E. 
.05 
.05 
.04 
.05 
.05 

56 

Table 12. (cont’d) 

.04 
.04 
.06 
.07 
.04 

-.03 

SRL 
EFFI 
TP 
SP 
MP 

.24*** 
.12*** 
.25*** 
.41*** 

CP on 
 
 
 
 
***p < .001. TP = Teaching Presence, SP = Social Presence, CP = Cognitive Presence, MP = 
Mentor Presence, SRL = Online SRL Strategies, EFFI = Self-efficacy 
 
 
Figure 7. The Path Coefficients of the Alternative Model with Mentor Presence 

 

 

The new model without the two insignificant paths is illustrated in Fig. 8. As shown in 

Table 13, the path from SP to EFFI (β = .58, p < .001) had a large effect size, while the paths 

from EFFI to SRL (β = .35, p < .001) and from SP to CP (β = .41, p < .001) both had medium 

 

57 

effect sizes. All the other paths’ loadings were of small effect size, with TP to SRL = .22 (p 

< .001), MP to SRL = .24 (p < .001), SRL to CP = .24 (p < .001), EFFI to CP = .12 (p < .01), and 

TP to CP = .24 (p < .001). Around 33.4% of the variance in EFFI was accounted for by SP; 

around 43.2% of the total variance in SRL was accounted for by TP, EFFI, and MP; and around 

72.3% of the total variance in CP was accounted for by TP, SP, SRL, and EFFI. 

Table 13.  The Structural Model for the Modified Model with Mentor Presence 

 
 
SP 
EFFI 
TP  
MP 
SRL 
EFFI 
TP 
SP 

 
 
EFFI on 
SRL on 
 
 
CP on 
 
 
 
**p < .01, ***p < .001. TP = Teaching Presence, SP = Social Presence, CP = Cognitive 
Presence, MP = Mentor Presence, SRL = Online SRL Strategies, EFFI = Self-efficacy 
 
 

.58*** 
.35*** 
.22*** 
.24*** 
.24*** 
.12** 
.24*** 
.41*** 

Standardized 

β 

S.E. 
.03 
.04 
.05 
.05 
.04 
.04 
.06 
.07 

 

 

 

 

 

 

 

 

 

 

 

 

 

58 

Figure 8. The Path Coefficients of the Modified Alternative Model with Mentor Presence 

RQ3. Model with Learning Outcomes – Satisfaction 

To explore the structural relationships among CoI, SRL, self-efficacy, mentor presence, 

and satisfaction, Satisfaction (SATIS) was added to the model described in the previous section 

(see Fig. 4), with the expectation that TP, SP, MP, and CP would all have direct effects on 

SATIS. The overall model results were χ2 (1848) = 4213.685 (p < .001), RMSEA = .044 with a 

90% CI from .042 to .045, CFI = .928, and SRMR = .043. 

 

Table 14 shows the standardized coefficients for the model after adding SATIS. Fig. 9 

gives the standardized coefficients of the factor loadings. Again, the straight lines represent 

significant predictions, and the dotted lines non-significant ones. As Fig. 9 shows, all the path 

 

59 

estimates were statistically significant except for the path from TP to SATIS (β = .01, p > .05), 

and from MP to SATIS (β = -.06, p > .05). 

Table 14.  The Structural Model for the Hypothesized Model with Satisfaction 

Standardized 

β 

.61*** 
.35*** 
.22*** 
.24*** 
.24*** 
.12** 
.25*** 
.39*** 
.36*** 

 
 
SP 
EFFI 
TP  
MP 
SRL 
EFFI 
TP 
SP 
CP 
TP 
SP 
MP 

 
 
EFFI on 
SRL on 
 
 
CP on 
 
 
 
SATIS on 
 
 
 
**p < .01, ***p < .001. TP = Teaching Presence, SP = Social Presence, CP = Cognitive 
Presence, MP = Mentor Presence, SRL = Online SRL Strategies, EFFI = Self-efficacy, SATIS = 
Satisfaction. 
 

S.E. 
.04 
.05 
.05 
.04 
.04 
.04 
.06 
.07 
.06 
.06 
.08 
.04 

.01 

.57*** 

-.06 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

60 

Figure 9. The Path Coefficients of the Alternative Model with Mentor Presence and 
Satisfaction 

 

 

After the insignificant path from TP to SATIS and MP to SATIS were removed, the 

resulting new model had results of χ2 (1850) = 4216.289 (p < .001), RMSEA = .044 with a 90% 

CI from .042 to .045, CFI = .928, and SRMR = .043. Fig. 10 presents this modified model with 

standardized path coefficients on each path, and Table 15 shows its standardized coefficients. In 

the modified model, two paths showed large effect size: from SP to EFFI (β = .61, p < .001), and 

from SP to SATIS (β = .53, p < .001). Medium effect sizes were found for three paths, i.e., from 

SP to CP (β = .39, p < .001), CP to SATIS (β = .36, p < .001), and TP to SRL (β = .35, p < .001). 

Small effect sizes were found for the six remaining paths, i.e., from TP to SRL (β = .22, p 

 

61 

< .001), MP to SRL (β = .24, p < .001), SRL to CP (β = .24, p < .001), EFFI to CP (β = .12, p 

< .01), and TP to CP (β = .25, p < .001). 

Table 15.  The Structural Model for the Modified Model with Satisfaction 

 
 
SP 
EFFI 
TP  
MP 
SRL 
EFFI 
TP 
SP 
CP 
SP 

 
 
EFFI on 
SRL on 
 
 
CP on 
 
 
 
SATIS on 
 
**p < .01, ***p < .001. TP = Teaching Presence, SP = Social Presence, CP = Cognitive 
Presence, MP = Mentor Presence, SRL = Online SRL Strategies, EFFI = Self-efficacy, SATIS = 
Satisfaction. 
 

.61*** 
.35*** 
.22*** 
.24*** 
.24*** 
.12** 
.25*** 
.39*** 
.36*** 
.53*** 

Standardized 

β 

S.E. 
.03 
.04 
.05 
.05 
.04 
.04 
.06 
.07 
.06 
.06 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

62 

Figure 10. The Path Coefficients of the Modified Alternative Model with Mentor Presence 
and Satisfaction 

 

Mediation was then tested using the bootstrapping procedures previously described. The 

mediation between SP and SATIS, had an unstandardized total effect of .77 (p < .001), a total 

indirect effect of .20 (p < .001), and a direct effect of .57 (p < .001). The 95% CI of the 

unstandardized indirect effect ranged from .12 to .31. The total indirect effect from SP to SATIS 

was significant. The relationship between SP and SATIS was mediated by: 1) CP (B = .15, 95% 

CI from .09 to .24); and 2) EFFI, SRL, and CP (B = .02, 95% CI from .01 to .04). Both 

mediation effects were significant. 

 

63 

RQ3. Model with Learning Outcomes – Perceived Progress 

Another hypothesized model, using perceived progress (PROG) as the learning-outcome 

variable in place of SATIS, is presented in Fig. 4. This model aims to elucidate the structural 

relationships among CoI, SRL, EFFI, MP, and PROG. It was hypothesized that TP, SP, MP, and 

CP would all have direct effects on PROG. The overall model results were χ2 (1910) = 4493.232 

(p < .001), RMSEA = .045 with a 90% CI from .043 to .047, CFI = .923, and SRMR = .046. 

Table 16 shows the standardized coefficients for the model using PROG as its learning-

outcome variable, and Fig. 11 its standardized coefficients. All paths’ estimates were statistically 

significant, except for the path from MP to PROG, and from TP to PROG. After removing these 

insignificant paths (see Table 17 and Figure 12), the new model yielded χ2 (1912) = 4499.775 (p 

< .001), RMSEA = .045 with a 90% CI from .043 to .047, CFI = .922, and SRMR = .047. In this 

model, the path from SP to EFFI (β = .60, p < .001), and from CP to PROG (β = .60, p < .001) 

showed large effect sizes. A medium effect size was found for two paths, i.e., from EFFI to SRL 

(β = .35, p < .001), and from SP to CP (β = .37, p < .001). All the other paths showed small effect 

sizes, with TP to SRL (β = .22, p < .001), MP to SRL (β = .24, p < .001), SRL to CP (β = .24, p 

< .001), EFFI to CP (β = .16, p < .001), TP to CP (β = .24, p < .001), and SP to PROG (β = .21, p 

< .01). 

Table 16.  The Structural Model for the Hypothesized Model with Perceived Progress 

 
 
SP 
EFFI 
TP  
MP 
SRL 
EFFI 
TP 
SP 
CP 

 
 
EFFI on 
SRL on 
 
 
CP on 
 
 
 
PROG on 

 

Standardized 

S.E. 
.03 
.04 
.05 
.05 
.04 
.04 
.06 
.07 
.06 

β 

.61*** 
.35*** 
.22*** 
.24*** 
.24*** 
.16*** 
.27*** 
.34*** 
.61*** 

64 

Table 16. (cont’d) 

SP 
MP 
TP 

 
 
 
***p < .001. TP = Teaching Presence, SP = Social Presence, CP = Cognitive Presence, MP = 
Mentor Presence, SRL = Online SRL Strategies, EFFI = Self-efficacy, PROG = Perceived 
Progress. 

.09 
.04 
.07 

.33*** 

.04 
-.01 

 

Figure 11. The Path Coefficients of the Alternative Model with Mentor Presence and 
Perceived Progress 

Table 17.  The Structural Model for the Modified Model with Perceived Progress 

 

 
 
SP 
EFFI 
TP  
MP 

 
 
EFFI on 
SRL on 
 
 

 

Standardized 

β 

.60*** 
.35*** 
.22*** 
.24*** 

S.E. 
.03 
.04 
.05 
.05 

65 

Table 17. (cont’d) 

SRL 
EFFI 
TP 
SP 
CP 
SP 

CP on 
 
 
 
PROG on 
 
**p < .01, ***p < .001. TP = Teaching Presence, SP = Social Presence, CP = Cognitive 
Presence, MP = Mentor Presence, SRL = Online SRL Strategies, EFFI = Self-efficacy, PROG = 
Perceived Progress. 
 

.24*** 
.16*** 
.24*** 
.37*** 
.60*** 
.21** 

.04 
.04 
.06 
.07 
.06 
.07 

Figure 12. The Path Coefficients of the Modified Alternative Model with Mentor Presence 
and Perceived Progress 

 
As with the SATIS-outcome model described in the previous section, mediation was then 

tested using bootstrapping procedures. The mediation was between SP and PROG. Here, the 

bootstrapped unstandardized total effect was .36 (p < .001), the total indirect effect was .21 (p 

 

66 

< .01), and the direct effect, .15 (p > .05). The 95% CI of the unstandardized indirect effect 

ranged from .12 to .42. The relationship between SP and PROG was mediated by: 1) CP (β 

=  .15, p < .01, 95% CI from .09 to .32), 2) EFFI and CP (β = .04, p < .05, 95% CI from .01 

to .08), and 3) EFFI, SRL, and CP (β = .02, p < .01, 95% CI from .01 to .04).  

RQ3. Model with Learning Outcomes – Final Grade 

The last hypothesized model used students’ final grade as the learning-outcome variable. 

As shown in Fig. 5, the model explores the structural relationships among CoI, SRL, EFFI, MP, 

and final grade. The hypothesis was that TP, SP, MP, and CP would all relate with final grade. 

The model yielded χ2 (1728) = 3934.584 (p < .001), RMSEA = .044 with a 90% CI from .042 to 

.045, CFI = .928, and SRMR = .043. 

Table 18 shows the standardized coefficients for the model using final grade as its 

learning-outcome variable, and Fig. 13 its standardized coefficients. All paths’ estimates were 

statistically significant, except for the path from TP to final grade (β = .05, p > .05), from MP to 

final grade (β = -.05, p > .05), and from SP to final grade (β = .03, p > .05). After removing these 

insignificant paths (see Table 19 and Figure 14), the new model yielded χ2 (1731) = 3935.637 (p 

< .001), RMSEA = .044 with a 90% CI from .042 to .045, CFI = .928, and SRMR = .043. In this 

model (see Table 18), the path from SP to EFFI (β = .58, p < .001) showed large effect size. A 

medium effect size was found for two paths, i.e., from EFFI to SRL (β = .35, p < .001), and from 

SP to CP (β = .41, p < .001). All the other paths showed small effect sizes, with TP to SRL (β 

= .22, p < .001), MP to SRL (β = .24, p < .001), SRL to CP (β = .24, p < .001), EFFI to CP (β 

= .12, p < .001), TP to CP (β = .24, p < .001), and CP to final grade (β = .29, p < .01). 

 

67 

 

 

Table 18.  The Structural Model for the Hypothesized Model with Final Grade 

 
 
 
 
SP 
EFFI on 
EFFI 
SRL on 
TP  
 
MP 
 
SRL 
CP on 
EFFI 
 
TP 
 
SP 
 
Final Grade on  CP 
TP 
 
SP 
 
 
MP 
**p < .01, ***p < .001. TP = Teaching Presence, SP = Social Presence, CP = Cognitive 
Presence, MP = Mentor Presence, SRL = Online SRL Strategies, EFFI = Self-efficacy 

.58*** 
.35*** 
.22*** 
.24*** 
.24*** 
.12** 
.24*** 
.41*** 
.25*** 

S.E. 
.03 
.04 
.05 
.05 
.04 
.04 
.06 
.07 
.08 
.09 
.11 
.05 

Standardized 

β 

.05 
.03 
-.05 

 

68 

 

Figure 13. The Path Coefficients of the Alternative Model with Mentor Presence and Final 
Grade  

Table 19. The Structural Model for the Modified Model with Final Grade 

 

Standardized 

β 

 
 
 
 
SP 
EFFI on 
EFFI 
SRL on 
TP  
 
MP 
 
SRL 
CP on 
EFFI 
 
TP 
 
SP 
 
Final Grade on  CP 
**p < .01, ***p < .001. TP = Teaching Presence, SP = Social Presence, CP = Cognitive 
Presence, MP = Mentor Presence, SRL = Online SRL Strategies, EFFI = Self-efficacy 
 

.58*** 
.35*** 
.22*** 
.24*** 
.24*** 
.12** 
.24*** 
.41*** 
.29*** 

S.E. 
.03 
.04 
.05 
.05 
.04 
.04 
.06 
.07 
.04 

 

 

69 

Figure 14. The Path Coefficients of the Modified Alternative Model with Mentor Presence 
and Final Grade 

RQ4. Studying at Home or at School 

 

 

First, descriptive tables were used to get an understanding of the difference of the 

location choice for students to complete the majority of their work (see Table 20). Again, of the 

696 participants, about 58% reported that the primary location for them to complete most of their 

online coursework was at school and the rest 42% reported that they completed most of their 

work at home. 

 

 

 

70 

Table 20.  The Demographic Differences between “At-school” Students and “At-home” 
Students 

 
Gender 
 
Grade 
 
 
 
 
 
Subject 
 
 
 
 
 
Credit-
recovery 
 

Numbers of 
online courses 
before 

 

 
Male 
Female 
7th grade 
8th grade 
9th grade 
10th grade 
11th grade 
12th grade 
English 
Foreign languages 
Math 
Science 
Social science 
Other subjects 
Credit-recovery 
students 
Non-credit-recovery 
students 
1 course 
2 courses 
3 courses 
4 or more courses 

At-school 
102 
297 
2 
16 
33 
80 
80 
189 
27 
110 
40 
62 
56 
103 
22 

374 

165 
137 
47 
50 

At-home 

90 
205 

2 
9 
26 
52 
82 
124 
31 
97 
28 
49 
41 
49 
18 

275 

101 
89 
40 
64 

Total 
192 
502 

4 
25 
59 
132 
162 
313 
58 
207 
68 
111 
97 
152 
40 

649 

266 
226 
87 
114 

As shown from Table 20, a slightly more students chose to complete the majority of their 

work at school regardless of gender, grade, subject, and whether or not the student took the 

online course for credit-recovery purposes. Another interesting thing to point out from Table 20 

was, when it came to check the difference of the choice of the primary learning location in terms 

of the number of online courses that they have taken before, those who have taken 1, 2, or 3 

online course before all preferred more to learning at school, while for those who has already 

taken 4 or more than 4 online courses, these group of students showed a preference of learning at 

home (N = 64) as compared with learning at school (N = 50). 

Next, this study examined whether students’ primary choice of online learning location 

related to students online learning perceptions and outcomes including their perceptions of social 

 

71 

presence, cognitive presence, mentor presence, online self-regulated learning strategies, and 

learning outcomes. One-way ANOVAs were conducted to evaluate the impact of two different 

primary online learning locations on various online learning presences and outcomes. As shown 

in Table 21, the ANOVAs showed significant group effects on the following items:  

Table 21.  Group Comparisons with Means and ANOVAs 

 
 
SP 
 
 
 
 
 
 
CP 
 
 
 
 
 
 
 
 
 
 
 
 
SRL 
 
 
 
 
 
 
 
 
 
 
EFFI 
 
 
 

 

 
 
SP_1 
SP_2 
SP_3 
SP_4 
SP_5 
SP_6 
 
CP_1 
CP_2 
CP_3 
CP_4 
CP_5 
CP_6 
CP_7 
CP_8 
CP_9 
CP_10 
CP_11 
CP_12 
 
SRL_1 
SRL_2 
SRL_3 
SRL_4 
SRL_5 
SRL_6 
SRL_7 
SRL_8 
SRL_9 
SRL_10 
 
EFFI_1 
EFFI_2 
EFFI_3 
EFFI_4 

F 
 
3.16 
4.29* 
2.01 
.68 
.12 
.02 
 
6.77** 
6.55* 
4.95* 
.55 
.00 
1.64 
.62 
2.47 
.94 
.39 
.56 
.10 
 
1.14 
6.26* 
4.16* 
.11 
.03 
.82 
.37 
1.93 
1.50 
4.06* 
 
.32 
.01 
.05 
1.46 

M (SD) 

At-home 
3.58 
3.39 
3.33 
3.89 
4.15 
3.75 
 
3.31 
3.56 
3.52 
3.96 
3.95 
3.88 
3.82 
3.83 
3.86 
4.04 
3.97 
4.07 
 
3.99 
3.73 
3.86 
3.72 
3.47 
2.90 
3.35 
3.41 
3.69 
3.46 
 
4.11 
4.14 
4.29 
3.93 

At-school 
3.73 
3.59 
3..47 
3.83 
4.12 
3.76 
 
3.51 
3.79 
3.71 
4.02 
3.95 
3.99 
3.88 
3.94 
3.94 
4.09 
4.02 
4.10 
 
4.07 
3.95 
4.02 
3.74 
3.50 
2.81 
3.41 
3.53 
3.80 
3.65 
 
4.07 
4.13 
4.31 
4.02 

72 

Table 21. (cont’d) 

MP_1 
MP_2 
MP_3 
MP_4 
MP_5 
MP_6 
MP_7 
MP_8 
MP_9 
MP_10 
MP_11 
MP_12 
MP_13 
MP_14 
MP_15 
MP_16 
 
SATIS_1 
SATIS_2 
SATIS_3 
 
PROG_1 
PROG_2 
PROG_3 
PROG_4 
 
Final grade 

MP 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
SATIS 
 
 
 
PROG 
 
 
 
 
Final grade 
*p < .05, **p < .001. SP = Social Presence, CP = Cognitive Presence, MP = Mentor Presence, 
SRL = Online SRL Strategies, EFFI = Self-efficacy, SATIS = Satisfaction, PROG = Perceived 
Progress. 

3.68 
3.74 
3.87 
3.77 
3.86 
3.79 
4.29 
4.02 
3.87 
4.04 
3.42 
3.45 
3.49 
3.58 
3.60 
3.76 
 
4.03 
3.94 
3.89 
 
4.29 
4.24 
4.14 
4.12 
 
81.41 

3.41 
3.53 
3.64 
3.61 
3.72 
3.57 
4.20 
3.99 
3.85 
3.99 
3.54 
3.59 
3.47 
3.49 
3.44 
3.65 
 
4.00 
3.87 
3.83 
 
4.24 
4.19 
4.09 
4.08 
 
78.39 

8.24** 
5.40* 
6.56* 
3.09 
2.43 
5.39* 
1.32 
.12 
.08 
.28 
1.51 
1.67 
.02 
.87 
2.54 
1.33 
 
.15 
.60 
.38 
 
.50 
.42 
.37 
.20 
 
2.59 

 

In terms of social presence, only one measured item showed significant difference: “I 

never feel isolated in this online course”. The mean for at-school students was 3.59 and the mean 

for at-home students was 3.39 (p < .05). 

In terms of cognitive presence, three items showed significant differences. These items 

were: “Problems posed increased my interest in course issues” (M = 3.51 for at-school students 

and M = 3.31 for at-home students, p < .01), “Course activities raised my curiosity” (M = 3.79 

 

73 

for at-school students and M = 3.56 for at-home students, p < .05), and “I felt motivated to 

explore content related questions”(M = 3.71 for at-school students and M = 3.52 for at-home 

students, p < .05). 

In terms of mentor presence, at-school students reported significantly higher mentor 

presences on the following four items: “The mentor helped me get used to the online learning 

environment”(M = 3.68 for at-school students and M = 3.41 for at-home students, p < .01), “The 

mentor helped me become familiar with the course platform”(M = 3.74 for at-school students and 

M = 3.53 for at-home students, p < .05), “The mentor helped me when I met technical 

problems”(M = 3.87 for at-school students and M = 3.64 for at-home students, p < .05), and “The 

mentor fostered a sense of learning community”(M = 3.79 for at-school students and M = 3.57 

for at-home students, p < .05). 

Lastly, in terms of self-regulated learning strategies, three items demonstrated significant 

differences: “I set short-term (daily or weekly) goals as well as long-term goals (monthly or for 

the semester)” (M = 3.95 for at-school students and M = 3.73 for at-home students, p < .05), “I 

keep a high standard for my learning in my online courses” (M = 4.02 for at-school students and 

M = 3.86 for at-home students, p < .05), and “If needed, I try to ask my online teacher/mentors 

about  questions that I don’t understand” (M = 3.65 for at-school students and M = 3.46 for at-

home students, p < .05).

 

74 

CHAPTER 5 

DISCUSSION 

Both the CoI framework and SRL theory have been widely accepted and used to provide 

explanations for students’ online learning process and outcomes. The former focuses on active 

collaboration and construction among learners and teachers aimed at building deep and 

meaningful learning through teaching-, social-, and cognitive presences (Garrison, 2007; 

Garrison et al., 2001); while the latter considers the metacognitively, motivationally, and 

behaviorally active learning processes initiated by learners (Zimmerman, 1986). The primary 

goals of the present study have been to extend Shea and Bidjerano’s (2010) model to include CoI 

presence and SRL as potential explanations of students’ online-learning process, and to evaluate 

the predictive power of this model using a large sample of data from an actual K-12 online-

learning setting. 

In spite of – or perhaps because of – the profusion of online-learning research in higher-

education settings, very few frameworks have explicitly focused on the unique features of 

younger online learners or on differences in online-learning environments between the school 

and college levels (Barbour, 2012, 2013; Barbour & Reeves, 2009; Borup et al., 2013; 

Cavanaugh et al., 2009; Rice, 2006). The present research is a response to these and other 

scholars’ calls for a study of existing theoretical frameworks’ applicability in K-12 online 

learning settings. 

This study expands on previous research in four ways: 1) by testing whether the most 

prominent previously established model can usefully be applied to K-12 online learners; 2) by 

testing the addition of mentor presence to that model; 3) by linking the previously established 

model to learning outcomes (i.e., satisfaction, perceived learning, and final grade); and 4) by 

 

75 

testing whether students who mostly learn at-school or at-home show similar or different level of 

online learning presences and learning outcomes. Overall, the results provide strong evidence in 

support of the hypotheses presented in Chapter 2, above, and help to explain the complicated 

relationships among CoI, self-efficacy, SRL strategies, mentor presence, and learning-outcome 

variables in K-12 online learning. Table 22 summarizes the results of the tested hypotheses, and 

the following sections discuss the main findings relating to each research question. 

Table 22.  Summary of the Tested Hypotheses 

Not Supported 

Supported 

Not Supported 

Supported 

Supported 

Result 
Supported 

Hypothesis 
H1: Teaching presence positively predicts students’ 
online SRL strategies. 
H2: Teaching presence positively predicts students’ self-
efficacy. 
H3: Teaching presence positively predicts cognitive 
presence. 
H4: Teaching presence positively predicts learning 
outcome. 
H5: Social presence positively predicts students’ self-
efficacy. 
H6: Social presence positively predicts cognitive 
presence. 
H7: Social presence positively predicts learning outcome.  Supported1 
H8: Cognitive presence positively predicts learning 
Supported 
outcome. 
H9: Self-efficacy positively predicts online SRL 
strategies. 
H10: Self-efficacy positively predicts cognitive presence.  Supported 
H11: Online SRL strategies positively predict cognitive 
Supported 
presence. 
H12: Mentor presence positively predicts self-efficacy. 
H13: Mentor presence positively predicts online SRL. 
H14: Mentor presence positively predicts cognitive 
presence. 
H15: Mentor presence positively predicts learning 
outcome. 
H16: Students who use school as their primary online 
learning locations demonstrate higher level of social 
presence, mentor presence, cognitive presence, self-
efficacy, online SRL strategies, and learning outcomes, 
compared with those who choose to spend their majority 
time of online learning at home. 

Supported 

Not Supported 
Supported 
Not Supported 

Not Supported 

Partially Supported2 

 

76 

Table 22. (cont’d) 

1 Supported in the satisfaction and perceived progress model but not supported in the final grade 
model. 
2 Students who use school as their primary online learning locations did not demonstrate 
significantly higher level of self-efficacy and learning outcomes, compared with those who 
choose to spend their majority time of online learning at home. 

 

Comparing Shea’s Model against the Alternative Model 

As previously discussed, Shea and Bidjerano’s (2010) SEM model confirmed the 

relationship between the CoI framework, self-efficacy, and self-regulation (see Fig. 1). Guided 

by prior literature on students’ self-regulation under teachers’ guidance, the present study added 

one more direct path to Shea’s model – from teaching presence to self-regulation – with the aim 

of improving its fit with a sample of data from K-12 online learners. SEM analyses showed that 

the hypothesized model was a better fit than Shea’s model. The new model also found 1) a 

significant relationship between teaching presence and SRL that was not specified in Shea’s 

model, and 2) that teaching presence did not predict self-efficacy significantly, even though 

Shea’s model had shown a strong correlation between these two constructs. These two 

differences and why they were found are discussed in detail below. 

Teaching Presence and Self-regulated Learning Strategies. The new model proposed 

and tested in the current study confirmed a direct path between teaching presence and self-

regulation, with a medium effect size (.45). Such a finding echoes SRL theory on the importance 

of teachers empowering students to become self-aware and self-directed during the instruction 

process (Zimmerman, 2002). Zimmerman, Bonner, and Kovach (1996) highlighted that teachers 

need to prepare students to learn on their own, through techniques such as setting goals and self-

evaluating learning: i.e., to become self-regulated. The present findings also echo the idea that it 

is through online teaching presence that students become metacognitively self-regulated (Akyol 

 

77 

& Garrison, 2011b). The medium effect size of the relationship between teaching presence and 

self-regulation are consistent with previous quasi-experimental studies’ findings that students’ 

SRL skills could be improved through intentional and structured teaching practice (Crippen & 

Earl, 2007; Ebner & Ehri, 2016). 

When examining the correlation between teaching presence and self-regulation, Shea and 

Bidjerano (2010) reported a .19 correlation. In the current study, in contrast, the correlations 

between the latent factor of teaching presence and the latent factor of self-regulation was .33. 

There are two possible explanations for this discrepancy. First, since most of the courses in the 

current study were purchased from a third party, it is possible that they were uniformly well-

made and thoughtfully designed to facilitate students’ self-regulation development. As a general 

matter, it is possible that these and other course-design considerations may relate to whether and 

how well teachers can simultaneously deliver content knowledge and metacognitive-thinking 

guidance. 

The second potential explanation resides in scholars’ differing choices of self-regulation 

measurements. Although the current study and Shea and Bidjerano’s (2010) were both clearly 

situated within the concept of self-regulation, each used a different scale to measure it. Notably, 

Shea and Bidjerano used four items from the “effort regulation” section of the MSLQ (Pintrich et 

al., 1993) as indicators for self-regulation; while the current study incorporated concepts such as 

goal-setting and help-seeking in addition to effort regulation. The additional goal-setting and 

help-seeking measurements may account for the stronger observed correlation between teaching 

presence and SRL. 

Teaching Presence and Self-efficacy. Shea and Bidjerano (2010) found that teaching 

presence significantly predicted self-efficacy, whereas the present study did not find that the 

 

78 

same relationship was significant. The reason for this may reside in the differences between the 

participants selected in the two studies. Shea and Bidjerano’s respondents comprised both 

blended learning-learning and fully online students, while in the present study all students took 

their relevant courses online only; and as Shea and Bidjerano conceded, they found much 

stronger correlations between teaching presence and self-efficacy among blended-learning 

students (who engaged in some forms of face-to-face interaction) than among students in fully 

online courses. 

The question, then, is why teaching presence is less relevant (or in our case, not relevant 

at all) to self-efficacy in online-learning settings. Research on traditional face-to-face settings has 

found that teachers can foster students’ self-efficacy through a range of practices, such as 

providing specific and accurate feedback, setting challenging tasks, individualizing learning, 

showing care, and demonstrating effective modeling practices (Linnenbrink & Pintrich, 2003; 

Pajares, 2012). It is possible that online-learning environments are deficient in vicarious learning 

experiences and/or verbal persuasion from teachers, due to the limits such environments place on 

teacher-student interaction. If so, it would be especially challenging for online learners to form 

strong self-efficacy. Previous studies have also found that it is difficult for teachers to adjust their 

lesson plans and tailor learning to each individual in K-12 online language courses (Lin & 

Zheng, 2015; Oliver et al., 2009), which were being taken by 29% of our sample. Therefore, 

given that the context of the current study involved an asynchronous learning format and less 

active roles on the part of the instructors than in Shea and Bidjerano’s sample, it is possible that 

there really was no significant relationship between teaching presence and self-efficacy to be 

found. 

 

79 

Mentor Presence in K-12 Online Learning 

Mentor presence was included in the current study’s modeling to reflect a key feature of 

K-12 online learning, one that has been called a critical change to such learning (Barbour & 

Mulcahy, 2004) and an important factor in helping K-12 students succeed online (Ferdig, 2010; 

Roblyer et al., 2008; Taylor et al., 2016). De la Varre (2011) recommended that mentors’ work 

be considered as a component of the CoI framework to reflect their facilitation and engagement 

activities, as in real-life online learning settings, teaching and facilitating practices were actually 

distributed through the contribution from the instructor and the mentor separately and 

individually. 

Building on Shea and Bidjerano’s (2010) model, this study added one more factor – MP – 

to reflect this trait of K-12 online learning. The SEM results showed a satisfactory fit for the 

resulting model, confirming that adding mentor presence to the existing combined SRL/CoI 

framework could provide a more comprehensive picture of the mechanisms that predict K-12 

online learning. 

Mentor Presence. Previous studies have subdivided K-12 mentors into four roles: 

relationship builders, monitors, content-learning facilitators, and technical-problem solvers (e.g., 

Aronson & Timms, 2003; Barbour & Mulcahy, 2004; Borup & Drysdale, 2014; de la Varre et 

al., 2011; Drysdale, 2013; Hannum et al., 2008; Kennedy & Cavanaugh, 2010; Taylor et al., 

2016). The current study examined the mentoring practices used by K-12 online-learning 

mentors, and found that their practices include all roles that previous studies have mentioned. 

This is especially reflected through the fact that all 16 items of the measured mentor presence 

loaded on to the latent factor of mentor presence with factor loadings greater than .60. Taken as a 

whole, this study confirmed the relevance of these four pre-existing categories. 

 

80 

 

Descriptive statistics showed that the composite score of mentor presence was perceived 

at moderate levels by K-12 online learners (M = 3.74), and the means for the 16 sub-scales of 

mentor presence that reflected mentor’s different roles in facilitating online learning ranged from 

3.47 to 4.25. This reflected the fact that mentors can play a variety of roles in facilitating 

students’ learning, rather than just limiting themselves to the original purpose of tracking 

students’ progress (Borup & Drysdale, 2014; Drysdale, 2013; Kennedy & Cavanaugh, 2010; 

Taylor et al., 2016). Specifically, this finding confirmed previous studies’ findings about the 

commonplace, albeit unofficial, extension of mentors’ roles into direct content instruction 

(Barbour & Mulcahy, 2004; de la Varre, 2011; O’Dwyer et al., 2007). 

Mentor Presence, Self-efficacy, and Self-regulated Learning Strategies. After the 

addition of MP, the model illustrated in Fig. 8 indicated that such presence did not significantly 

predict students’ self-efficacy. The insignificant direct effect of mentor presence on self-efficacy 

did not support previous studies’ findings about the positive impact of mentoring on students’ 

confidence in online learning (de la Varre et al., 2011; Freidhoff et al., 2015; Murphy & 

Rodriguez-Manzanares, 2009). A potential explanation is that those studies explored how 

mentor’s explicit encouragement affected students beliefs on whether they would achieve 

success in online learning; while the mentor presence measured in the current study were mainly 

interested in areas including relationship building, monitoring, content-learning instruction or 

facilitation, and technical-problem solving, with only one item asking encouragement-related 

question (i.e., “The mentor encouraged students when needed”) out of the 16 items. Since this 

study is the first and the only study that use quantitative approach to describe mentor presence, a 

replication of such method is strongly needed. This finding also stated the need that perhaps in 

order to promote one’s self-efficacy, mentors need to use explicit self-efficacy promoting 

 

81 

strategies, which echoed with some professional-development programs that have specifically 

trained mentors in strategies for improving students’ self-efficacy and motivation (Staker, 2011).  

K-12 online learners have been found to have relatively low metacognitive skills, which 

makes their online learning a challenging experience (Borup & Drysdale, 2014; Cavanaugh et 

al., 2009). It is therefore crucial for each child to have an on-site mentor who can help them with 

the development of such metacognitive skills (Borup & Drysdale, 2014). The present study’s 

modeling also identified a direct relationship between mentor presence and online SRL 

strategies, implying that students with higher levels of perception of their mentors’ presence 

were more likely to demonstrate higher levels of online SRL strategies. The finding of such 

relationship directly supported previous findings regarding mentors’ positive role in developing 

students’ SRL skills (Freidhoff et al., 2015; Harms et al., 2006), further confirmed the 

importance of mentors to teach students with learning skills rather than merely supervising their 

content learning (Harms et al., 2006; Wicks, 2010). 

In the context of the current study, students enrolled in only one or two online courses 

offered by the research site as a supplement to the curricula of their own brick-and-mortar 

schools (Watson et al., 2011). In addition, it was required by law that each student be equipped 

with one mentor employed by his or her school district, who would be responsible for 

“determining appropriate instructional methods for each pupil, diagnosing learning needs, 

assessing pupil learning, prescribing intervention strategies, reporting outcomes, and evaluating 

the effects of instruction and support strategies” (Freidhoff et al., 2015, p. 112). More than half 

of the participants in this study (57.5%) used school as the primary learning location through 

attending lab sessions overseen by their mentors on school days to learn virtual-school course 

content and complete virtual-school assignments. It is possible that during the above-mentioned 

 

82 

lab sessions, mentors’ practice in keeping track of the progress, reminding students of the 

assignment due dates, keeping students stay focused on learning help to develop students self-

regulated learning skills such as time-management, goal-setting, or self-monitoring and self-

reflection.  

Connecting the CoI Models with Learning Outcomes 

In this study, the three learning-outcome variables SATIS, PROG, and final grade were 

included in separate hypothesized models, all of which also contained CoI, SRL, EFFI, and MP. 

Overall, the results from both learning-outcome models provided strong support for the 

hypothesized relationships between all these variables, as explained more fully below. 

Incorporating Learning Presence into the CoI Model. The models demonstrated 

positive relationships between CoI-framework and SRL-theory variables, on the one hand, and 

learning-outcome variables on the other. Social- and cognitive presences were both found to 

positively predict students’ perceptions of their learning progress and learning satisfaction, while 

only cognitive presence was found to predict their final grade. These findings are consistent with 

prior findings about the important roles played by social presence and cognitive presence in 

directly correlating with students’ satisfaction with learning and their perceptions that learning 

has occurred (Akyol & Garrison, 2008; Arbaugh, 2013; Kang et al., 2014; Rockinson-Szapkiw et 

al., 2016). Since most of the previous studies of online learning using the CoI framework focused 

on post-secondary environments, the present study has also extended the relationship between 

the CoI framework and learning outcomes to K-12 online learning contexts, and confirmed that it 

is useful in that context, following certain modifications. 

This dissertation’s models have also directly answered scholars’ calls for a natural 

consideration and integration of learning presence into the CoI framework (Barnard et al., 2008; 

 

83 

Cho et al., 2017; Lin et al., 2017; Shea & Bidjerano, 2010). None of the empirical evidence 

presented in Chapter 4, above, suggests that learning presence – including self-efficacy and 

metacognitive learning strategies – played any major role in explaining how learners transferred 

their perceptions of teaching-, mentor-, or social presence into learning outcomes. In general, the 

present study’s finding that students with high learning presence are likely to perceive high 

cognitive presence and in turn show high learning outcomes is consistent with previous studies’ 

findings about the intercorrelations between CoI and SRL predictors (Cho et al., 2017; Garrison, 

2007; Lin et al., 2017; Lynch & Dembo, 2004; Pawan et al., 2003; Puzziferro, 2008; Roblyer & 

Marshall, 2002; Shea & Bidjerano, 2010, 2012; Weiner, 2003). 

This study’s identification of learning presence’s mediation effect on the relations 

between social presence and learning outcomes also adds empirical support to the mediating role 

of SRL in online learning, and answers calls for increased attention to learning presence within 

the CoI framework (Barnard et al., 2008; Cho et al., 2017; Shea & Bidjerano, 2010; Wang et al., 

2008). For example, Wang et al. concluded that self-efficacy affected online-learning outcomes 

via learning strategies, which highlighted the importance of self-directed learning in online 

contexts. Barnard et al. found that approximately one-third of the relationship between students’ 

perceived social presence and their learning achievement was explained by their online SRL 

behaviors. Similarly, the current study found that about half of the relationship between students’ 

perceived social presence and their satisfaction was accounted for by learning presence (i.e., self-

efficacy and SRL strategies). The mediations tested above have also shown in detail how social 

presence predicted learning outcomes under the mediation of learning presence, thereby 

revealing the operation of a metacognitive, motivational, and behavioral learning process in the 

transformation of perceptions of online learning into learning outcomes. 

 

84 

Distinctive Learning Outcomes. Although the proposed three learning outcome models 

are similar to each other in terms of their prediction paths, one path highlighted the difference 

between them and is therefore worthy of further discussion. The finding that social presence 

directly predicted satisfaction and perceived progress, but not final grade, lead us to reconsider 

two things: the first is the role of social presence in students’ self-perceived learning progress 

(both emotionally and cognitively), and the second is the need to differentiate perceived learning 

progress and actual final grade as two distinct variables in evaluating online learning. 

Both satisfaction and perceived progress were believed to be important components of 

successful online learning (Kuo et al., 2014; Lin et al., 2016), and were highly correlated with 

each other (Eom, Wen, & Ashill, 2006; Lin et al., 2016). The former refers to “the favorability of 

a student’s subjective evaluation of the various outcomes and experiences associated with 

education” (Elliott & Shin, 2002), and the latter focuses on measuring one’s subjective feeling 

that “that learning is taking place” (Rockinson-Szapkiw et al., 2016, p. 21). The fact that social 

presence directly predicted these two outcomes showed that when asking about students self-

perceived outcomes in learning, whether about psychological emotion (Bolliger & Martindale, 

2004), or about cognitive aspect of learning (Eom et al., 2006), students’ social experiences in 

the entire learning process really matter. That is to say, a strong social presence can greatly 

decide how students “perceive” their entire learning journey (Elliott & Shin, 2002).  

Second, the distinct findings between the perceived progress model and the final grade 

model confirmed the existence of the difference between students’ self-perceived learning 

outcomes and their actual learning outcomes although both measures reflect students’ cognitive 

learning progress outcomes (Barbour, 2010; 2013; Ferdig et al., 2009). Lin et al. (2016) used 

perceived progress in measuring students’ cognitive learning outcomes due to the lack of access 

 

85 

to grade-related data, and especially pointed out that “We could not verify the degree to which 

these responses were accurate reflections of students’ online learning, as opposed to optimistic or 

pessimistic projections. Thus, where possible, future research should incorporate additional data 

stored in online learning management systems” (p. 70). This slightly divergent findings from the 

two models supported this hypothesis and therefore concluded that we need to use multiple 

sources of data to measure students’ cognitive learning outcomes in a more accurate and 

comprehensive way. 

Learning At-home vs. Learning At-school 

This dissertation’s final research question examined whether students’ primary learning 

location (i.e., learning at-home or learning at-school) made a significant difference on students’ 

perceived social presence, mentor presence, self-efficacy, online SRL strategies, and learning 

outcomes. These two groups of students did not differed significantly in most of the measured 

items. However, they still differed in some items as described in chapter 4, and these significant 

differences actually suggested some interesting patterns. To sum, the significant differences 

concentrate on the following aspects: feeling of isolation in social presence, triggering event in 

cognitive presence, solving technical problem in mentor presence, and goal-setting and help-

seeking in online SRL strategies. Albeit with these differences, students did not demonstrate 

significant differences in three measured learning outcomes across groups. 

Feeling of Isolation. Social interaction correlates with K-12 students’ academic 

performance (Hawkins et al., 2013). The item “I never feel isolated in this online course” is the 

only item in the latent factor of social presence that showed significant difference between 

learning-at-home students and learning-at-school students. Not surprisingly, those whose 

primary online learning location is school showed significant less sense of isolation than those 

 

86 

students whose primary online learning location is at home. This finding is consistent with 

previous findings about how the difference in physical space may affect students online learning 

(Borup & Drysdale, 2014; Drysdale, 2013; Freidhoff et al., 2015), and how the physical show-up 

of students at a same classroom help to eliminate the sense of isolation that generate from online 

learning (e.g., Borup & Drysdale, 2014; Charania, 2010; Pettyjohn, 2012). As Murphy and 

Rodriguez-Manzanares (2008) descbribed in their study, as an addition to the “organized yet 

formal” online learning communication between students and teachers, online students believed 

that the face-to-face interaction that take place at labs make students feel at ease and 

comfortable. The “non-course-specific or social interaction” (Drysdale et al., 2014, p. 18) gave 

students extra psychological support and reassurance in addition to interactions regarding 

academic contents.  

It should be noted that although one item listed above showed significant difference 

between two types of learners, in general, of the six measured social presence items, five out of 

six showed insignificant difference between at-home learners and at-school learners. On the one 

hand, it showed the diverse nature of the online learners: different learners perceive their social 

presence in different ways and in general, this were not affected by where the learner decided to 

complete the majority of their learning activities. On the other hand, this finding reflected that a 

high level of social presence is related with complex factors that do not only limit to learning 

location. The well-design of a supportive and rapport learning environment, a healthy and active 

relationship with peers and teachers in the virtual environment seem more important why 

determining one’s perception of social presence online (de la Varre et al., 2011), so that students 

know that “there is a person behind the computer and not a robot” (Murphy & Rodriguez-

Manzanares, 2009, p.5). In specific, the role that online instructors take in building social rapport 

 

87 

(DiPietro et al., 2008; Murphy & Rodriguez-Manzanares, 2009) through their teaching practices 

may play a more important role in the current study.  

Triggering Event in Cognitive Presence. Cognitive presence measured in this study 

contained four subsets that show four different phases of the practical inquiry cycle: triggering 

event, exploration, integration, and resolution (Garrison et al., 2001). There are three items in 

each four subset. All three items in triggering event were found to show significant difference 

between at-home and at-school students. There were no significant differences detected in 

exploration, integration, and resolution. Triggering event measured whether students’ motivation 

and curiosity were raised by the problems posed in the course. Students who completed the 

majority of their online learning at school demonstrated significant higher scores on triggering 

event than those at-home students.  

Although no studies in K-12 online learning context has revealed such difference among 

learning locations, studies conducted in higher education settings showed consistent findings. 

For example, Akyol and Garrison (2011a) found that much of the triggering events occurred 

during the face-to-face sessions than the online sessions. In another study, Vaughan and Garrison 

(2005) found that there were more triggering events coded in face-to-face settings (13%) than in 

online settings (8%). Perhaps more importantly, some quotes from the participants in this study 

explained why curiosity is more likely to emerge in a physical setting, “(the face-to-face setting) 

develops a sense of community”, “face-to-face --- the discussions are in the moment and I often 

forget afterwards what was said” (p. 6). In this study, it is then possible that the actual presence 

of people in a same physical location helped to develop a strong sense of community – it even 

does not matter that whether the student showing up in a same room actually belongs to a same 

community under a same subject. The physical presence of the students, regardless of learning 

 

88 

contents, have built up a similar sense of community, which then helped to promote students’ 

cognitive presence in triggering event. The physical presence offered some additional cues for 

learners to initiate their learning curiosity and become motivated in learning (Vaughan & 

Garrison, 2005). 

Perceived Mentor’s Help. The mentor’s practice in this study fall largely into four 

categories: relationship builders, monitors, content-learning facilitators, and problem solvers. 

The results suggested that mentor showed a stronger presence in being a problem solver for those 

who chose to use school as their primary online learning location. In addition, mentor’s practice 

in fostering a sense of learning community was scored higher for those at-school students.  

Other than meeting students’ academic needs, online learning programs should also 

consider students’ technical needs during online learning (Watson, 2007; Pettyjohn, 2012). A 

considerate number of students did show relatively low levels of technical skills before starting 

online learning (Oliver et al., 2009), which may then limit their access and performance 

(Cavanaugh et al., 2009). In many K-12 online learning programs, mentors were assigned by 

schools to provide technical supports such as conducting orientations to familiarize students with 

the online learning environment, help to troubleshoot technical problems throughout the entire 

online learning journey (Aronson & Timms, 2003; Barbour & Mulcahy, 2004; de la Varre et al., 

2011; Hannum et al., 2008; Keane et al., 2008; Pettyjohn, 2012; Borup & Drysdale, 2014). The 

findings of this study not only confirmed the existence of such role, but also extended our 

understanding of the mentor’s role as a problem-solver. Those who chose to complete the 

majority of their online learning at school labs perceived a better presence of their mentors when 

it comes to helping them solve technique problems.  

 

89 

Goal-setting & Help-seeking Learning Strategies. When asking about student’s online 

SRL strategies, study-at-school students showed significantly higher score on goal-setting and 

help-seeking strategies, with all the other strategies insignificant between groups.  

Goal-setting belongs to the forethought phase of self-regulated learning (Zimmerman, 

1986). Learners use this strategy prior to  or during their online learning. Help-seeking is in the 

second phase – performance control of learning (Zimmerman, 1986), and showed how learners 

use strategies to seek help to adjust their learning. Findings from this study provided empirical 

support to previous hypotheses about K-12 online learners’ low levels of self-regulated learning 

abilities (Borup & Drysdale, 2014; Cavanaugh et al., 2009; Rice, 2006), possible benefits for 

students to have sit-in time at school (Roblyer & Marshall, 2002), and the differences between 

learning-at-school and learning-at-home students, especially in terms of self-regulated learning 

ability differences (Borup & Drysdale, 2014; Rice, 2006). In particular, students who complete 

the majority of their online learning at school demonstrated significantly higher goal-setting 

strategy and help-seeking strategy can be thought as an addition to the already finding that 

students learning online contents in the same physical room spend less time on off-task behaviors 

(de la Varre et al., 2011; Harms et al., 2006). Such difference may be explained by the practice 

of mentors during students’ physical sit-in times at school, as mentors were found to act as a 

problem-solver and promote self-regulated learning strategies (Aronson & Timms, 2003; 

Barbour & Mulcahy, 2004; de la Varre et al., 2011; Hannum et al., 2008; Keane et al., 2008; 

Pettyjohn, 2012). 

Limitations and Recommendations for Future Research 

Several limitations of this study need to be noted. First, learning outcomes in this study 

comprised learner satisfaction, perceived learning, and final grade. Learning outcomes need not 

 

90 

be limited to these three aspects alone. For example, Beldarrain (2008) collected students’ grades 

attained in each exam, in addition to their overall final course grades; while Borup et al. (2013) 

included students’ changes in disposition as a dimension of course outcomes. It has also been 

recommended that standardized test results be considered when examining the effectiveness of 

learning (U.S. Department of Education, 2008). Future studies should consider a wider variety of 

learning-outcome metrics when examining CoI, SRL, and related factors in online learning. 

A second limitation concerns the sources of the data collected in this study. Due to 

limited access to the target school’s learning-management system and records, TP, MP, and SP 

could only be measured based on students’ self-reported perceptions. Future studies should 

include multiple sources of data (e.g., learning-management system, learning records, teacher 

perceptions, teacher and student interviews, open-ended questions, and/or mentor journals) to 

verify students’ self-reported data (Barbour, 2012; Borup et al., 2013; Cavanaugh et al., 2009). 

Such data will also provide in-depth insights into how perceived CoI and SRL elements may 

correlate with each other. 

A third limitation regards the random-sampling method used in this study. This approach 

may run the risk of not being inclusive or representative enough for a specific population of 

students (Creswell, 2005). The sample for this research consisted primarily of females (69.9%), 

high-school students (95%), and Caucasians (82.3%). Moreover, the participants cannot be 

considered “fully” online learners, as they still attended their brick-and-mortar schools on a daily 

basis. Researchers should therefore be cautious about trying to generalize from the findings of 

this study to the experience of K-12 students with other demographic, cultural, and learning 

backgrounds. More studies should be conducted to test whether the revised CoI models proposed 

in this study can be generalized. 

 

91 

Last but not least, SRL is a complex construct that entails students’ active, goal-oriented 

regulation of behavior, motivation, and cognitive and metacognitive learning tasks (Pintrich et 

al., 1993; Zimmerman, 2008); yet, learning presence as measured in the present study only 

included self-efficacy and SRL. Future studies of similar topics should consider also including 

intrinsic goal orientation and task value within learning presence. 

Implications for Practitioners 

Based on the findings of the present study, it is clear that online learning should 

emphasize students’ roles as strategic learners (see also McMahon & Oliver, 2001; Shea & 

Bidjerano, 2010; Wang et al., 2013). Online learning is a complicated process that involves self-

efficacy and SRL beyond what is represented in the existing CoI framework. The present study’s 

more inclusive framework was effective in predicting students’ learning outcomes. This has 

several implications for online educators. 

Learning Presence. First, the learning presence issues identified in this dissertation’s 

revised CoI model could inform course design, online instruction, and off-line facilitation. 

Online instructors and mentors should structure an online-learning environment that activates 

students’ motivational and SRL processes in order to help them reach better learning outcomes 

(Cho et al., 2017; Kim et al., 2015; McMahon & Oliver, 2001; Wang et al., 2013). Although 

online learning encompasses a variety of learning resources and activities, it is often assumed 

that these resources and activities can be mastered without the guidance of teachers (McMahon 

& Oliver, 2001). Ley and Young (2001) proposed four principles to embed SRL into teaching: 1) 

structuring an effective learning environment, 2) integrating cognitive and metacognitive 

learning strategies into instructional activities, 3) modeling each student’s self-monitoring 

process through goal-setting and feedback-giving, and 4) allowing students opportunities for 

 

92 

self-monitoring. This dissertation’s findings tend to support those principles. Its other practical 

implications include the importance of teachers sharing their thinking load with students, to 

make the latter more aware of their own understandings. Teachers should also design activities 

such as writing reflective journals, using problem-based learning or inquiry-based learning, 

designing authentic learning activities or using concept mapping to promote SRL (Artino, 2008; 

McMahon & Oliver, 2001). 

Online educators also need to develop teaching strategies to enhance students’ self-

efficacy. It is important to identify students with low self-efficacy and develop action plans to 

support them (Murphy & Rodriguez-Manzanares, 2009). Artino (2008) recommended that online 

teachers seek to improve students’ self-efficacy in two specific ways. The first is helping 

students identify and set goals that are challenging yet realistic, thus increasing their willingness 

to exert effort to achieve them, and the likelihood that they will commit to the goal and sustain 

their learning effort. The second is providing students with timely and honest feedback that 

focuses exclusively on the task itself. This will help learners understand their progress and where 

there is room for improvement, while developing their self-efficacy as they master the set goals. 

Therefore, teachers should explicitly specify the requirements for each assignment or learning 

module on the syllabus; set deadlines to facilitate goal completion; communicate regularly on 

goals and performance; and use online gradebooks to allow self-monitoring. 

Mentor Presence. Within the sphere of U.S. K-12 education, the facilitation of learning 

by mentors is unique to online learning. The physical separation between learners and online 

teachers makes it necessary for mentors to provide targeted, timely, and individualized support 

for K-12 online learners (Hawkins et al., 2011). The present study identified a range of mentors’ 

responsibilities, and found that mentor presence was closely related to students’ SRL strategies. 

 

93 

The findings also revealed a clear classification of mentor presence into just two main 

types, i.e., academic and non-academic. By the same token, this dissertation has confirmed that 

mentors play multiple roles in supporting students’ learning, and are not limited to the original 

purpose of “monitoring learning” (de la Varre et al., 2011). 

In the proposed revision to the CoI framework, MP was found to correlate significantly 

with SRL. The results call further attention to the responsibility of each mentor to provide aid in 

the development of self-regulation skills (Borup & Drysdale, 2014; Harms et al., 2006), 

especially as pre-college-aged students may lack sufficient SRL abilities to thrive in online 

environments, which are typically more autonomous and student-centered than face-to-face 

schooling (Weiner, 2003; Wicks, 2010). Previous sections have discussed how teachers and 

mentors can enhance learning presence through a range of different instructional approaches. 

Besides all those mentioned above, it is particularly important for a mentor to work as a “role 

model” and a “cheerleader” for students (Wortmann et al., 2008, p. 12). From a mentor-as-role 

model, students can learn how to develop necessary cognitive and metacognitive learning 

strategies that can be beneficial for online learning, through observing the mentor’s behavior, as 

an adjunct to receiving direct support for the development of SRL strategies. The mentor-as-

cheerleader, meanwhile, should establish a relationship of trust with students and celebrate their 

success, with the aim of keeping them motivated to learn. 

Online Learning at School. Completing K-12 online learning does not necessarily mean 

that online learning only takes place in virtual environments. Many K-12 online learners either 

are required or have the option to complete their learning in a fixed location at school, in most 

cases, in school labs with computers and mentors (de la Varre et al., 2011; Watson et al., 2015). 

The findings in this study do not indicate that studying at-school is necessarily better than 

 

94 

studying at-home. In fact, the insignificance between two groups of students on three types of 

learning outcomes indicates that learning location does not matter that much to the sample in this 

study. However, the findings about some significant differences in terms of the measured items 

suggested that studying online learning at school labs may have some certain benefits, which 

further leads us to think about this question: how can we make sure the same quality of online 

learning can be delivered to those who choose to spend the majority of their online learning time 

at home? In particular, online learning programs need to make sure students possess a strong 

social presence, get guidance in developing self-regulated learning strategies, and receive help 

once needed regardless of their primary online learning locations. Another possible solution can 

be found from Borup and Drysdale’s (2014) study about the use of online-facilitator as a monitor 

and liaison between students and their online courses. Although these online facilitators do not 

share the same physical space with students, they can still function as on-site facilitators to make 

impact on student learning.

 

95 

CHAPTER 6 

CONCLUSION 

The number of K-12 online learners has been increasing dramatically over the past few 

years. It is therefore critically important to understand K-12 students’ online-learning processes 

and the factors that predict their online-learning outcomes. A theoretical model that captures the 

students situated in online learning, and reflects the interaction among each party, and examines 

the influence of those interconnections on students’ online learning experiences is thus highly 

expected (de la Varre et al., 2011). 

Based on the CoI framework and SRL theory, this study is among the first to shed light 

on the relationships among existing CoI variables, self-efficacy, SRL, and learning outcomes in 

the K-12 online-learning context. To help understand the influence of K-12 mentoring – which is 

unique to online learning in the U.S. – mentor presence was also included as a factor in this 

dissertation’s modeling. In addition to examining the relationships among this expanded range of 

factors, this study also investigated whether a student’s primary online learning location mattered 

in online learning.  

This study’s four main findings can be summed up as follows. First, in contrast to Shea 

and Bidjerano’s (2010) widely used model, it found a significant relationship between teaching 

presence and the use of self-regulated learning strategies, and a non-significant relationship 

between teaching presence and self-efficacy. Second, its models confirmed mentor’s practices in 

four domains and found that they significantly predicted student online SRL strategies. Third, it 

found that satisfaction, perceived progress and final grade provide strong support for the 

hypothesized relationship among these learning outcomes, CoI, and SRL, supporting previous 

calls for a natural consideration and integration of learning presence into the CoI framework – 

 

96 

though it should be noted that minor differences exist between final grade model and other two 

models. And fourth, this study’s comparison of learning-at-home and learning-at-school students 

suggested that in general, there were no significant differences on most of the items in TP, SP, 

MP, CP, online SRL strategies, and learning outcomes between these two groups of students. 

However, those whose primary online learning location is at school showed significantly less 

feelings of isolation, higher ability in generating curiosity once the online learning starts, higher 

perceptions on mentor’s practice as a problem solver, and higher goal-setting and help-seeking 

strategies. 

In short, this dissertation has established a theoretical framework for K-12 online learning 

that enables us to identify important components of such learning while striking a fine balance 

between extensiveness and parsimony. The findings indicate that SRL serves important functions 

in connecting CoI presences and learning outcomes. It is hoped that the processes identified in 

this study will be useful and relevant to K-12 online-learning institutions and educators seeking 

to improve online learning via a wide range of approaches. Given the important role of SRL in 

online learning, the results of the modeled relationships also warrant further academic 

investigation in other K-12 online-learning contexts. 

 

97 

APPENDICES

 

98 

APPENDIX A: Parental Consent Form 

We are inviting your child to participate in a research study designed to help us 

understand how to support students in online courses. If you agree to allow your child to 

participate in this study we will ask your child to complete a questionnaire designed to better 

understand their online learning.  

There are no correct or incorrect answers, and your child's responses here will in no way 

impact their standing in their online course. The researchers are interested only in their online 

learning process.  Their participation is voluntary and they may decline to answer the 

questionnaire or may skip any items that they feel uncomfortable answering. Your child can 

withdraw from the study any time without penalty. Your child's final course grade will also be 

collected. All data, including grades and responses are confidential and their privacy will be 

protected to the maximum extent allowable by law. They will be given a unique identifier and 

following the completion of the online course, all documents will contain only this unique 

identifier. There are no direct benefits to participating in this study, although we hope that they 

will gain more insight into their learning through participation.  

Upon completing the survey, your child will be given the opportunity to be entered in a 

drawing to receive a $20 Amazon gift card for your child’s time while he/she is in this study 

(there will be twenty gift cards  in total).  

If you have any concerns of questions about this research study, such as scientific issues 

or how to do any part of it, or to report an injury, please contact the following investigators:  

 • Dr. Chin-Hsi Lin, responsible project investigator, 513A Erickson Hall, Michigan State 

University, East Lansing, MI, 48824. chinhsi@msu.edu (517) 353-5400.  

 

99 

 • Yining Zhang, secondary investigator, Erickson Hall, Michigan State University, East 

Lansing, MI, 48824. zhangy58@msu.edu (517) 580-9736.  

If you have any questions or concerns about your role and rights as a research participant, 

would like to obtain information or offer input, or would like to register a complaint about this 

research study, you may contact, anonymously if you wish, the Michigan State University’s 

Human Research Protection Program at 517-355-2180, FAX 517-432-4503, or e-mail 

irb@msu.edu, or regular mail at: 408 W. Circle Dr. Rm. 207 Olds, East Lansing, MI 48824. 

By entering your initials and clicking the "yes" button you mark your consent to have 

your child participate.  

 

100 

APPENDIX B: Student Assent Form 

We are inviting you to participate in a research study designed to help us understand how 

to support students in online courses. If you agree to participate in this study we will ask you to 

complete a questionnaire designed to better understand your online learning.  

There are no correct or incorrect answers, and your responses here will in no way impact 

your standing in your online course. Your participation is voluntary and you may decline to 

answer the questionnaire or may skip any items that you feel uncomfortable answering. You can 

withdraw from the study any time without penalty. Your final course grade will also be 

collected. All responses, including grades and responses are confidential and your privacy will 

be protected to the maximum extent allowable by law. You will be given a unique identifier and 

following the completion of your online course, all documents will contain only this unique 

identifier. There are no direct benefits to participating in this study, although we hope that you 

will gain more insight into your learning through participation.   

Upon completing the survey, you will be given the opportunity to be entered in a drawing 

to receive a $20 Amazon gift card for your time in this study (there will be twenty gift cards 

in total).  

If you have any concerns of questions about this research study, such as scientific issues 

or how to do any part of it, or to report an injury, please contact the following investigators:  

 • Dr. Chin-Hsi Lin, responsible project investigator, 513A Erickson Hall, Michigan State 

University, East Lansing, MI, 48824. chinhsi@msu.edu (517) 353-5400.  

 • Yining Zhang, secondary investigator, Erickson Hall, Michigan State University, East 

Lansing, MI, 48824. zhangy58@msu.edu (517) 580-9736.  

 

101 

If you have any questions or concerns about your role and rights as a research participant, 

would like to obtain information or offer input, or would like to register a complaint about this 

research study, you may contact, anonymously if you wish, the Michigan State University’s 

Human Research Protection Program at 517-355-2180, FAX 517-432-4503, or e-mail 

irb@msu.edu, or regular mail at: 408 W. Circle Dr. Rm. 207 Olds, East Lansing, MI 48824. 

By entering your initials you mark your willingness to participate.

 

102 

APPENDIX C: Survey 

Demographic 

1.  What is your gender 

a.  Female 
b.  male 

2.  What is your grade level? 

a.  5th grade 
b.  6th grade 
c.  7th grade 
d.  8th grade 
e.  9th grade 
f.  10th grade 
g.  11th grade 
h.  12th grade 

3.  What is your ethnicity? 

a.  White 
b.  African American 
c.  Hispanic 
d.  Asian 
e.  Other 

4.  What is the subject of the online course you are taking with MVU? (If you take multiple 

courses this semester, choose one that best represents your experience in this school) 

a.  English 
b.  Foreign languages 
c.  Math 
d.  Science 
e.  Social science 
f.  Other subjects 

5.  What is the name of this online course? 
6.  How would you rate your prior knowledge on this subject? 

a.  Very poor 
b.  Poor 
c.  Fair 
d.  Good 
e.  Very good 

7.  Are you taking this course for credit recovery? 
8.  How many online courses have you taken before? 

a.  1 
b.  2 
c.  3 or more 

9.  Where did you complete most of your online coursework? 

a.  School 
b.  Home 
c.  other 
Teaching presence 

 

103 

Design and organization 

1.  The instructor clearly communicated important course topics. 
2.  The instructor clearly communicated important course goals. 
3.  The instructor provided clear instructions on how to participate in course learning 

activities. 

4.  The instructor clearly communicated important due dates/time frames for learning 

activities. 

Facilitation 

5.  The instructor was helpful in guiding the class towards understanding course topics in a 

way that helped me clarify my thinking. 

6.  The instructor helped to keep course participants engaged and participating in productive 

dialogues. 

7.  The instructor helped keep the course participants on task. 
8.  The instructor encouraged course participants to explore new concepts in this course. 
9.  Instructor actions reinforced the development of a sense of community among course 

participants. 

Direct instruction 

10. My instructor provided useful illustrations that helped make the course content more 

understandable to me. 

11. My instructor presented helpful examples that allowed me to better understand the 

content of the course. 

12. My instructor provided clarifying explanations or other feedback that allowed me to 

better understand the content of the course. 

Social presence 
Affective expression 

1.  I have a sense of belonging in the course. 
2.  I never feel isolated in this online course.  
3.  Online or web-based communication is an excellent medium for social interaction. 

Open communication 

4.  I felt comfortable conversing through the online medium. 
5.  I felt comfortable participating in the course activities. 
6.  I felt comfortable interacting with others in this course. 

Cognitive presence 
Triggering event 

1.  Problems posed increased my interest in course issues. 
2.  Course activities raised my curiosity. 
3.  I felt motivated to explore content related questions. 

Exploration 

4.  I utilized a variety of sources to explore problems. 
5.  Brainstorming and finding relevant information helped me resolve questions. 
6.  The course was valuable in helping me appreciate different perspectives. 

Integration 

7.  Learning activities helped me construct solutions. 
8.  Learning activities helped me construct explanations. 
9.  Reflection on course content helped me understand fundamental concepts in this class. 

Resolution 

 

104 

10. I can describe ways to apply the knowledge created in this course. 
11. I have developed solutions to course problems that can be applied in practice. 
12. I can apply the knowledge in this course to other non-class related activities. 

Self-efficacy   

1.  I believe I will receive an excellent grade in this class. 
2.  I'm confident I can do an excellent job in this course. 
3.  I expect to do well in this class.  
4.  I'm certain I can master the skills being taught in this class.  

Self-regulated learning strategies 
Goal setting 

1.  I set standards for my assignments in online courses. 
2.  I set short-term (daily or weekly) goals as well as long-term goals (monthly or for the 

semester). 

3.  I keep a high standard for my learning in my online courses. 
4.  I set goals to help me manage studying time for my online courses. 

Strategic learning 

1.  I take thorough notes for my online courses 
2.  I work extra problems or do additional readings in my online courses beyond the assigned 

ones to master the course content. 

3.  I try to schedule time every day or every week to study for my online courses. 
4.  I reflect on my learning in online courses to examine. 

Help seeking 

5.  I find someone who is knowledgeable in course content so that I can consult with him/her 

when I need help. 

6.  If needed, I try to ask my online teacher/mentors about the question that I don’t 

understand. 

7.  I interact with my online teacher/mentors to help me understand how I am doing in my 

online classes. 

Mentor presence 

1.  The mentor helped me get used to the online learning environment. 
2.  The mentor helped me become familiar with the course platform.  
3.  The mentor helped me when I met technical problems. 
4.  The mentor gave me suggestions on solving technical problems. 
5.  The mentor expressed appreciation for my contribution. 
6.  The mentor fostered a sense of learning community . 
7.  The mentor was friendly when interacting with me or other students. 
8.  The mentor encouraged students when needed.  
9.  The mentor was interested in students as individuals. 
10. The mentor kept tracking of my progress. 
11. The mentor kept constant check-in with me. 
12. The mentor reminded me of the assignment due dates. 
13. The mentor kept me stay focused on learning. 
14. The mentor addressed my misunderstanding about content learning. 
15. The mentor helped me with content learning. 
16. The mentor answered my questions about content.  

Satisfaction  

 

105 

1.  Overall I am satisfied with this class. 
2.  I would recommend this online course to other students. 
3.  I would take an online course like this again in the future. 

Perceived progress 

1.  I have met most of the requirements teachers made in the class. 
2.  I understand most of the learning content in my class. 
3.  I learned to identify the central issues of the course. 
4.  I developed the ability to communicate clearly about the subject. 

 

106 

REFERENCES

 

107 

REFERENCES 

 

Akyol, Z., & Garrison, D. R. (2008). The development of a community of inquiry over time in an 

online course: Understanding the progression and integration of social, cognitive and 
teaching presence. Journal of Asynchronous Learning Networks, 12, 3-22. 
doi:10.24059/olj.v12i3.66  

Akyol, Z., & Garrison, D. R. (2011a). Understanding cognitive presence in an online and 

blended community of inquiry: Assessing outcomes and processes for deep approaches to 
learning: Cognitive presence in an online and blended community of inquiry. British 
Journal of Educational Technology, 42, 233–250. doi:10.1111/j.1467-8535.2009.01029.x 

Akyol, Z., & Garrison, D. R. (2011b). Assessing metacognition in an online community of 

inquiry. The Internet and Higher Education, 14(3), 183–190. 
doi:10.1016/j.iheduc.2011.01.005 

Akyol, Z., & Garrison, D. R. (2014). The development of a community of inquiry over time in an 

online course: Understanding the progression and integration of social, cognitive and 
teaching presence. Journal of Asynchronous Learning Networks, 12, 3-22. 

Anderson, T., Liam, R., Garrison, D. R., & Archer, W. (2001). Assessing teacher presence in a 

computer conferencing context. Journal of the Asynchronous Learning Network, 5(2). 
Retrieved from http://auspace.athabascau.ca/handle/2149/725 

Arbaugh, J. B. (2013). Does academic discipline moderate CoI-course outcomes relationships in 

online MBA courses? The Internet and Higher Education, 17, 16-28. 
doi:10.1016/j.iheduc.2012.10.002  

Arbaugh, J. B., Cleveland-Innes, M., Diaz, S. R., Garrison, D. R., Ice, P., Richardson, J. C., & 

Swan, K. P. (2008). Developing a community of inquiry instrument: Testing a measure of 
the Community of Inquiry framework using a multi-institutional sample. The Internet and 
Higher Education, 11(3–4), 133–136. doi:10.1016/j.iheduc.2008.06.003 

Archambault, L., Diamond, D., Brown, R., Cavanaugh, C., Coffey, M., Foures-Aalbu, D., & 
Zygouris-Coe, V. (2010). Research committee issues brief: An exploration of at-risk 
learners and online education. International Association for K-12 Online Learning. 
Retrieved from http://eric.ed.gov/?id=ED509620 

Aronson, J. Z., & Timms, M. J. (2003). Net choices, net gains: Supplementing high school 

curriculum with. San Francisco: WestEd. Retrieved from 
https://works.bepress.com/michael_timms/9/download/  

Artino, A. (2008). Practical guidelines for online instructors. TechTrends, 52, 37-45. doi: 

10.1007/s11528-008-0153-x 

Bandura, A. (1997). Self-efficacy: The exercise of control. New York: Greeman 

 

108 

Barbour, M. K. (2012). Models and resources for online teacher preparation and mentoring. In K. 
M. Kennedy & L. Archambault (Eds.), Lessons learned in teacher mentoring: Supporting 
educators in K-12 online learning environments (pp. 83–102). Vienna, VA: International 
Association for K-12 Online Learning. 

Barbour, M. K. (2013). The landscape of K-12 online learning: Examining what is known. 

Handbook of Distance Education, 3, 574–593. doi:10.4324/9780203803738.ch36 

Barbour, M. K., & Mulcahy, D. (2004). The role of mediating teachers in Newfoundland’s new 

model of distance education. The Morning Watch, 32 (1-2), 1–14. Retrieved from 
http://www.mun.ca/educ/faculty/mwatch/fall4/barbourmulcahy.htm 

Barbour, M. K., & Mulcahy, D. (2009). Beyond volunteerism and good will: Examining the 

commitment of schoolbased teachers to distance education. Retrieved from 
http://digitalcommons.sacredheart.edu/ced_fac/177/ 

Barbour, M. K., & Reeves, T. C. (2009). The reality of virtual schools: A review of the literature. 

Computers & Education, 52, 402–416. doi:10.1016/j.compedu.2008.09.009 

Bartlett, M. S. (1950). Tests of significance in factor analysis. British Journal of Mathematical 

and Statistical Psychology, 3(2), 77-85. doi: 10.1111/j.2044-8317.1950.tb00285.x  

Barnard, L., Lan, W. Y., To, Y. M., Paton, V. O., & Lai, S.-L. (2009). Measuring self-regulation 
in online and blended learning environments. The Internet and Higher Education, 12, 1–
6. doi:10.1016/j.iheduc.2008.10.005 

Barnard, L., Paton, V., & Lan, W. (2008). Online self-regulatory learning behaviors as a 

mediator in the relationship between online course perceptions with achievement. The 
International Review of Research in Open and Distance Learning, 9. 
doi:10.19173/irrodl.v9i2.516  

Barnard-Brak, L., Lan, W. Y., & Paton, V. O. (2010). Profiles in self-regulated learning in the 

online learning environment. The International Review of Research in Open and Distance 
Learning, 11, 61–78. doi:10.4018/978-1-61692-901-5.ch002  

Beldarrain, Y. (2008). Engaging the 21st century learner: An exploratory study of the 

relationship between interaction and achievement in the virtual high school. Capella 
University, Minnesota. 

Bernard, R. M., Abrami, P. C., Borokhovski, E., Wade, C. A., Tamim, R. M., Surkes, M. A., & 
Bethel, E. C. (2009). A meta-analysis of three types of interaction treatments in distance 
education. Review of Educational Research, 79, 1243–1289. 
doi:10.3102/0034654309333844 

Black, E. W., Ferdig, R. E., & DiPietro, M. (2008). An overview of evaluative instrumentation 

for virtual high schools. American Journal of Distance Education, 22, 24–45. 
doi:10.1080/08923640701713422 

 

109 

Bolliger, D. U., & Martindale, T. (2004). Key factors for determining student satisfaction in 

online courses. International Journal on E-learning, 3(1), 61-68. Retrieved from 
https://www.learntechlib.org/p/2226 

Borup, J., & Drysdale, J. S. (2014). On-site and online facilitators: Current and future direction 
for research. In R. Ferdig & K. Kennedy (Eds.), Handbook of research on K-12 online 
and blended learning (pp. 325–346). ETC Press. Retrieved from 
http://press.etc.cmu.edu/files/Handbook-Blended-Learning_Ferdig-Kennedy-
etal_web.pdf   

Borup, J., Graham, C. R., & Davies, R. S. (2013). The nature of adolescent learner interaction in 

a virtual high school setting. Journal of Computer Assisted Learning, 29, 153–167. 
doi:10.1111/j.1365-2729.2012.00479.x   

Campbell, D. T., & Fiske, D. W. (1959). Convergent and discriminant validation by the 

multitrait-multimethod matrix. Psychological Bulletin, 56, 81-105. 
doi:10.1037/h0046016  

Cavanaugh, C., Gillan, K. J., Kromrey, J., Hess, M., & Blomeyer, R. (2004). The effects of 

distance education on K-12 student outcomes: A meta-analysis. Naperville, IL: Learning 
Point Associations. Retrieved from http://eric.ed.gov/?id=ED489533 

Cavanaugh, C. S., Barbour, M. K., & Clark, T. (2009). Research and practice in K-12 online 
learning: A review of open access literature. The International Review of Research in 
Open and Distributed Learning, 10(1). doi: 10.19173/irrodl.v10i1.607  

Celentin, P. (2007). Online training: Analysis of interaction and knowledge-building patterns 

among foreign language teachers. Journal of Distance Education, 21(3), 39–58. 
Retrieved from http://ijede.ca/index.php/jde/article/view/29/35 

Charania, A. K. (2010). Preparing future teachers for virtual schooling: Assessing their 
preconceptions and competence. (Doctorial dissertation). Iowa State University. 
Retrieved from http://lib.dr.iastate.edu/etd/11447 

Cho, M. H., & Shen, D. (2013). Self-regulation in online learning. Distance Education, 34, 290–

301. doi:10.1080/01587919.2013.835770 

Cho, M. H., Kim, Y., & Choi, D. (2017). The effect of self-regulated learning on college 

students' perceptions of community of inquiry and affective outcomes in online learning. 
The Internet and Higher Education, 34, 10-17. doi:10.1016/j.iheduc.2017.04.001  

Cobb, S. C. (2011). Social presence, satisfaction, and perceived learning of RN-to-BSN students 

in web-based nursing courses. Nursing Education Perspectives, 32, 115-119. 
doi:10.5480/1536-5026-32.2.115  

Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: 

Lawrence Earlbaum Associates. 

 

110 

Corry, M., & Stella, J. (2012). Developing a framework for research in online K-12 distance 

education. The Quarterly Review of Distance Education, 13(3), 133–151. Retrieved from 
http://go.galegroup.com.proxy1.cl.msu.edu.proxy2.cl.msu.edu/ps/i.do?p=AONE&u=msu
_main&id=GALE%7CA327816151&v=2.1&it=r&sid=summon&authCount=1 

Costello, A. B., & Osborne, J. W. (2005). Exploratory factor analysis: Four recommendations for 

getting the most from your analysis. Practical Assessment, Research, and Evaluation, 
10(7), 1-9.  

Creswell, J. W. (2005). Educational research: Planning, conducting, and evaluating quantitative 

and qualitative research. Upper Saddle River, N. J.: Merrill. 

Crippen K. J. & Earl, B. L., (2007). The impact of Web-based worked examples and self-

explanation on performance, problem solving, and self-efficacy. Computers & Education, 
49, 809-821. doi: 10.1016/j.compedu.2005.11.018  

Cudek, R. & Browne, M. W. (1983). Cross-validation of covariance structures. Multivariate 

Behavioral Research, 18, 147-167. doi: 10.1207/s15327906mbr1802_2 

de la Varre, C., Irvin, M., Jordan, A. Hannum W., & Farmer, T. (2014). Reasons for student 

dropout in an online course in a rural K-12 setting. Distance Education, 35, 324-
344.doi:10.1080/01587919.2015.955259 

de la Varre, C., Keane, J., & Irvin, M. J. (2011). Dual perspectives on the contribution of on-site 
facilitators to teaching presence in a blended learning environment. Journal of Distance 
Education (Online), 25(3). Retrieved from 
http://www.ijede.ca/index.php/jde/article/view/751/1285  

De Laat, M., Lally, V., Lipponen, L., & Simons, R.-J. (2007). Online teaching in networked 
learning communities: A multi-method approach to studying the role of the teacher. 
Instructional Science, 35(3), 257–286. doi:10.1007/s11251-006-9007-0 

DiPietro, M., Ferdig, R. E., Black, E. W., & Preston, M. (2008). Best practices in teaching K-12 

online: Lessons learned from Michigan Virtual School teachers. Journal of Interactive 
Online Learning, 7(1), 10–35. Retrieved from 
http://www.ncolr.org/jiol/issues/pdf/7.1.2.pdf  

Drysdale, J. S. (2013). Online facilitators and sense of community in K-12 online learning 

(Unpublished doctoral dissertation). Bringham Young University - Provo. 

Drysdale, J. S., Graham, C. R., & Borup, J. (2014). An online high school “shepherding” 

program: Teacher roles and experiences mentoring online students. Journal of 
Technology and Teacher Education, 22, 9-32. 

Ebner R. J. & Ehri, L. C. (2016). Teaching students how to self-regulate their online vocabulary 

learning by using a structured think-to-yourself procedure. Journal of College Reading 
and Learning, 46, 62-73. doi:10.1080/10790195.2015.1075448 

 

111 

Elliott, K. M., & Shin, D. (2002). Student satisfaction: An alternative approach to assessing this 
important concept. Journal of Higher Education Policy and Management, 24, 197-209. 
doi: 10.1080/1360080022000013518  

Eom, S. B., Wen, H. J., & Ashill, N. (2006). The determinants of students’ perceived learning 

outcomes and satisfaction in university online education: an empirical investigation. 
Decision Sciences Journal of Innovative Education, 4, 215–235. doi:10.1111/j.1540-
4609.2006.00114.x 

Ferdig, R. E. (2010). Understanding the role and applicability of K-12 online learning to support 

student dropout recovery efforts. Lansing, MI: Michigan Virtual University. Retrieved 
from http://www.mivu.org/Portals/0/RPT_RetentionFinal.pdf 

Fornell, C., & Larcker, D.F. (1981). Evaluating structural equation models with unobservable 

variables and measurement error. Journal of Marketing Research, 18, 39-50. doi: 
doi.org/10.2307/3151312  

Freidhoff, J., Borup, J., Stimson, R., & DeBruler, K. (2015). Documenting and sharing the work 

of successful on-site mentors. Journal of Online Learning Research, 1(1), 107–128. 
Retrieved from https://www.learntechlib.org/p/149918 

Garrison, D. R. (2007). Online community of inquiry review: Social, cognitive, and teaching 

presence issues. Journal of Asynchronous Learning Networks, 11(1), 61–72. Retrieved 
from https://eric.ed.gov/?id=EJ842688  

Garrison, D. R., Anderson, T., & Archer, W. (2001). Critical thinking, cognitive presence, and 
computer conferencing in distance education. American Journal of Distance Education, 
15, 7–23. doi:10.1080/08923640109527071 

Garrison, D. R., Anderson, T., & Archer, W. (2010). The first decade of the community of 

inquiry framework: A retrospective. The Internet and Higher Education, 13, 5–9. 
doi:10.1016/j.iheduc.2009.10.003 

Garrison, D.R., Cleveland-Innes, M., & Fung, T. S. (2010). Exploring causal relationships 

among teaching, cognitive and social presence: Student perceptions of the community of 
inquiry framework. The Internet and Higher Education, 13, 31-36. doi: 
10.1016/j.iheduc.2009.10.002 

Gefen, D., Straub, D. W., & Boudreau, M. (2000). Structural equation modeling and regression: 

Guidelines for research practice. Communications of the Association for Information 
Systems, 4(7), 1–78.
Retrieved from 
http://cits.tamiu.edu/kock/NedWebArticles/Gefenetal2000.pdf 

Graham, S., & Harris, K. R. (2000). The role of self-regulation and transcription skills in writing 

and writing development. Educational Psychologist, 35, 3–12. 
doi:10.1207/s15326985ep3501_2 

 

112 

Gunawardena, C., Hermans, M. B., Sanchez, D., Richmond, C., Bohley, M., & Tuttle, R. (2009). 

A theoretical framework for building online communities of practice with social 
networking tools. Educational Media International, 46, 3–16. 
doi:10.1080/09523980802588626 

Hair, Jr., J.F., Black, W.C., Babin, B.J., Anderson, R.E., Tatham, R.L., 2006. Multivariate data 

analysis (6th Ed.), Pearson-Prentice Hall, Upper Saddle River, NJ. 

Hannum, W. H., Irvin, M. J., Lei, P., & Farmer, T. W. (2008). Effectiveness of using learner

centered principles on student retention in distance education courses in rural schools. 
Distance Education, 29, 211–229. doi:10.1080/01587910802395763 

Harms, C. M., Niederhauser, D. S., Davis, N. E., Roblyer, M. D., & Gilbert, S. B. (2006). 

Educating educators for virtual schooling: Communicating roles and responsibilities. The 
Electronic Journal of Communication, 16(1 & 2). Retrieved from 
http://public.vrac.iastate.edu/~gilbert/papers/JP2007-harms-niederhauser.pdf 

Hawkins, A., Barbour, M. K., & Graham, C. R. (2011). Strictly business: Teacher perceptions of 

interaction in virtual schooling. International Journal of E-Learning & Distance 
Education, 25(2). Retrieved from http://www.ijede.ca/index.php/jde/article/view/726 

Hawkins, A., Graham, C. Sudweeks R., & Barbour, M. (2013). Academic performance, course 
completion rates, and student perception of the quality and frequency of interaction in a 
virtual high school. Distance Education, 34, 64-83. doi: 10.1080/01587919.2013.770430 

Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: 

Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1-55. doi: 
10.1080/10705519909540118  

Kaiser, H. F. (1974). An index of factorial simplicity. Psychometrika, 39, 31–36. doi: 

10.1007/bf02291575  

Kaiser, H. F., & Rice, J. (1974). Little jiffy, mark IV. Educational and psychological 

measurement, 34, 111-117. doi: 10.1177/001316447403400115  

Kang, M., Liw, B. T., Kim, J., & Park, Y. (2014). Learning presence as a predictor of 

achievement and satisfaction in online learning environments. International Journal on 
E-Learning, 13, 193–208. Retrieved from https://eric.ed.gov/?id=EJ1035981 

Keane, J., de la Varre, C., Irvin, M. J., & Hannum, W. (2008). Learner-centered social support: 

Enhancing online distance education for underserved rural high school students in the 
United States. ALT-C 2008 Research Proceedings, 39-48. Retrieved from 
http://repository.alt.ac.uk/435/  

Kennedy, K., & Cavanaugh, C. (2010). Development and support of online teachers: The roles of 
mentors in virtual schools. Journal of Technology Integration in the Classroom, 2(3), 37–
42. Retrieved from http://connection.ebscohost.com/c/articles/55301570/development-
support-online-teachers-roles-mentors-virtual-schools 

 

113 

Kim, C., Park, S. W., Cozart, J., & Lee, H. (2015). From motivation to engagement: The role of 

effort regulation of virtual high school students in mathematics courses. Educational 
Technology & Society, 18(4), 261–272. Retrieved from 
http://www.ifets.info/journals/18_4/20.pdf 

Kline, R. B. (2011). Principles and practice of structural equation modeling. New York, NY: 

Guilford.  

Kuo, Y.-C., Walker, A. E., Schroder, K. E. E., & Belland, B. R. (2014). Interaction, Internet self-

efficacy, and self-regulated learning as predictors of student satisfaction in online 
education courses. The Internet and Higher Education, 20, 35–50. 
doi:10.1016/j.iheduc.2013.10.001 

Ley, K., and Young, D. B. (2001). Instructional principles for self-regulation. Educational 

Technology Research and Development 49, 93–103. doi: /10.1007/bf02504930  

Lin, C.-H., Zhang, Y., & Zheng, B. (2017). The role of learning strategies in online language 

learning: A structural equation modeling analysis. Computers & Education, 113, 75-85. 
doi:10.1016/j.compedu.2017.05.014 

Lin, C.-H., & Zheng, B. (2015). Teaching Practices and Teacher Perceptions in Online World 

Language Courses. Journal of Online Learning Research, 1(3), 275-303. Retrieved from 
https://www.learntechlib.org/p/171055  

Lin, C.-H., Zheng, B., & Zhang, Y. (2016). Interactions and learning outcomes in online 
language courses. British Journal of Educational Technology, 48, 730-748. doi: 
10.1111/bjet.12457 

Linnenbrink, E. A., & Pintrich, P. R. (2003). The role of self-efficacy beliefs in student 

engagement and learning in the classroom. Reading & Writing Quarterly, 19(2), 119–
137. doi:10.1080/10573560308223 

Liu, F., & Cavanaugh, C. (2012). Factors influencing student academic performance in online 

high school algebra. Open Learning: The Journal of Open, Distance and E-Learning, 27, 
149–167. doi:10.1080/02680513.2012.678613 

Lynch, R., & Dembo, M. (2004). The relationship between self-regulation and online learning in 

a blended learning context. The International Review of Research in Open and 
Distributed Learning, 5(2), 1–16. doi:10.19173/irrodl.v5i2.189 

MacKinnon, D. P., Lockwood, C. M., & Williams, J. (2004). Confidence limits for the indirect 

effect: Distribution of the product and resampling methods. Multivariate Behavioral 
Research, 39, 99-128. doi: 10.1207/s15327906mbr3901_4  

McMahon, M., & Oliver, R. (2001). Promoting self-regulated learning in an online environment. 
Retrieved from http://ro.ecu.edu.au/cgi/viewcontent.cgi?article=5815&context=ecuworks  

 

114 

Meyer, K. A. (2004). Evaluating online discussions: Four different frames of analysis. Journal of 

Asynchronous Learning Networks, 8, 101–114. Retrieved from 
http://onlinelearningconsortium.org/sites/default/files/v8n2_meyer_1.pdf   

Michigan Virtual Learning Research Institute. (2014). Mentoring fundamentals: A guide for 

mentoring online learners, Version 1. Lansing, MI: Michigan Virtual University. 
Retrieved from https://micourses.org/resources/pdf/toolkit/mentor_guide_14.pdf 

Murphy, E., & Rodriguez-Manzanares, M. (2009). Learner-centredness in high-school distance 
learning: Teachers’ perspectives and research-validated principles. Australasian Journal 
of Educational Technology, 25, 597–610. doi: 10.14742/ajet.1110  

Nisbet, D., Wighting, M., & Rockinson-Szapkiw, A. (2013). Measuring sense of community and 

academic learning in graduate education. The International Journal of Interdisciplinary 
Educational Studies, 7(1), 1-8. 

Nunnally, J. C., & Bernstein, I. H. (1994) Psychometric theory (3rd ed.). New York, NY: 

McGraw-Hill, Inc. 

O’dwyer, L. M., Carey, R., & Kleiman, G. (2007). A study of the effectiveness of the Louisiana 
Algebra I online course. Journal of Research on Technology in Education, 39, 289–306. 
doi:10.1080/15391523.2007.10782484 

Oliver, K., Osborne, J., & Brady, K. (2009). What are secondary students’ expectations for 

teachers in virtual school environments? Distance Education, 30, 23–45. doi: 
10.1080/01587910902845923  

Pajares, F. (2012). Motivational role of self-efficacy beliefs in self-regulated learning. In D. H. 

Schunk & B. J. Zimmerman (Eds.), Motivation and self-regulated learning: Theory, 
research, and applications (pp. 111–140). New York, NY: Routledge. 

Pawan, F., Paulus, T. M., Yalcin, S., & Chang, C.-F. (2003). Online learning: Patterns of 

engagement and interaction among in-service teachers. Language Learning & 
Technology, 7, 119–140.  

Pettyjohn, T. J. (2012). Stakeholder’s perceptions of supplemental online learning for credit 

recovery. Retrieved from http://digitalcommons.georgiasouthern.edu/etd/402/ 

Pintrich, P. R., & De Groot, E. V. (1990). Motivational and self-regulated learning components 

of classroom academic performance. Journal of Educational Psychology, 82, 33–40. 
doi:10.1037/0022-0663.82.1.33 

Pintrich, P. R., Smith, D. A., Garcia, T., & McKeachie, W. J. (1993). Reliability and predictive 

validity of the Motivated Strategies for Learning Questionnaire (MSLQ). Educational 
and Psychological Measurement, 53, 801–813. doi:10.1177/0013164493053003024 

 

115 

Preacher, K. J., & Hayes, A. F. (2004). SPSS and SAS procedures for estimating indirect effects 
in simple mediation models. Behavior Research Methods, Instruments, & Computers, 36, 
717-731. doi: 10.3758/bf03206553  

Puzziferro, M. (2008). Online technologies self-efficacy and self-regulated learning as predictors 

of final grade and satisfaction in college-level online courses. American Journal of 
Distance Education, 22, 72–89. doi:10.1080/08923640802039024 

Rice, K. L. (2006). A comprehensive look at distance education in the K–12 context. Journal of 

Research on Technology in Education, 38, 425–448. 
doi:10.1080/15391523.2006.10782468 

Rice, K. (2014). Research and history of policies in K-12 online and blended learning. In R. 

Ferdig & K. Kennedy (eds.), Handbook of research on K-12 online and blended Learning 
(pp. 51–82). Pittsburgh, PA: ETC Press.  

Rice, K., Huerta, L., Shafer, S., R., Barbour, M. K., Miron, G., Gulosino, C., & Horvitz, B. 
(2014). Virtual Schools in the U.S. 2014: Politics, performance, policy, and research 
evidence. Retrieved from http://nepc.colorado.edu/publication/virtual-schools-annual-
2014 

Richardson, J. C., Maeda, Y., Lv, J., & Caskurlu, S. (2017). Social presence in relation to 

students’ satisfaction and learning in the online environment: A meta-analysis. 
Computers in Human Behavior, 71, 402–417. doi:.10.1016/j.chb.2017.02.001 

Roblyer, M. D., Davis, L., Mills, S. C., Marshall, J., & Pape, L. (2008). Toward practical 

procedures for predicting and promoting success in virtual school students. American 
Journal of Distance Education, 22, 90–109. doi:10.1080/08923640802039040 

Roblyer, M. D., & Marshall, J. C. (2002). Predicting success of virtual high school students: 

Preliminary results from an educational success prediction instrument. Journal of 
Research on Technology in Education, 35, 241–255. 
doi:10.1080/15391523.2002.10782384 

Rockinson-Szapkiw, A., Wendt, J., Whighting, M., & Nisbet, D. (2016). The predictive 

relationship among the community of inquiry framework, perceived learning and online, 
and graduate students’ course grades in online synchronous and asynchronous courses. 
The International Review of Research in Open and Distributed Learning, 17(3). doi: 
10.19173/irrodl.v17i3.2203  

Schumacker, E. S., & Lomax, R. G. (2010). Structural equation modeling. New York, NY: 

Palgrave Macmillan. 

Schunk, D. H. (2005). Self-regulated learning: The educational legacy of Paul R. Pintrich. 

Educational Psychologist, 40, 85–94. doi:10.1207/s15326985ep4002_3 

 

116 

Schunk, D. H., & Zimmerman, B. J. (2007). Influencing children’s self-efficacy and self-

regulation of reading and writing through modeling. Reading & Writing Quarterly, 23, 7–
25. doi: 10.1080/10573560600837578 

Schunk, D. H., & Zimmerman, B. J. (2012). Motivation and self-regulated learning: Theory, 

research, and applications. New York, NY: Routledge. 

Shea, P., & Bidjerano, T. (2008). Measures of quality in online education: An investigation of 

the community of inquiry model and the net generation. Journal of Educational 
Computing Research, 39, 339–361. doi:10.2190/EC.39.4.b 

Shea, P., & Bidjerano, T. (2010). Learning presence: Towards a theory of self-efficacy, self-

regulation, and the development of a communities of inquiry in online and blended 
learning environments. Computers & Education, 55, 1721–1731. 
doi:10.1016/j.compedu.2010.07.017 

Shea, P., & Bidjerano, T. (2012). Learning presence as a moderator in the community of inquiry 

model. Computers & Education, 59, 316–326. doi:10.1016/j.compedu.2012.01.011 

Simpson, O. (2004). The impact on retention of interventions to support distance learning 

students. Open Learning, 19(1), 79–95. doi:10.1080/0268051042000177863 

Smith, R., Clark, T., & Blomeyer, B. (2005). A synthesis of new research in K–12 online 

learning. Naperville, IL: Learning Point Associates. 

Staker, H. (2011). The rise of k-12 blended learning: Profiles of emerging models. Innosight 

Institute. Retrieved from http://eric.ed.gov/?id=ED535181 

Swan, K. (2002). Building learning communities in online courses: The importance of 

interaction. Education, Communication & Information, 2(1), 23–49. 
doi:10.1080/1463631022000005016 

Taylor, S., Clements, P., Heppen, J., Rickles, J., Sorensen, N., Walters, K., Michelman, V. 

(2016). Getting back on track the role of in-person instructional support for students 
taking online credit recovery. Washington, D.C: American Institutes for Research. 
Retrieved from http://www.air.org/system/files/downloads/report/In-Person-Support- 
Credit-Recovery.pdf. 

The U.S. Department of Education. (2007). Connecting students to advanced courses online. 

Washington, DC. Retrieved Sep 16, 2016, from 
http://www.ed.gov/admins/lead/academic/advanced/index.html. 

Tsai, C.-W., Shen, P.-D., & Fan, Y.-T. (2013). Research trends in self-regulated learning 

research in online learning environments: A review of studies published in selected 
journals from 2003 to 2012: Colloquium. British Journal of Educational Technology, 44, 
E107–E110. doi:10.1111/bjet.12017 

 

117 

Vaughan, N. & Garrison, D. (2005). Creating cognitive presence in a blended faculty 

development community. Internet and Higher Education, 8, 1-12. doi: 
10.1016/j.iheduc.2004.11.001 

Wang, Y., Peng, H., Huang, R., Hou, Y., & Wang, J. (2008). Characteristics of distance learners: 

Research on relationships of learning motivation, learning strategy, self-efficacy, 
attribution and learning results. Open Learning, 23(1), 17-28. doi: 
10.1080/02680510701815277  

Wang, C.-H., Shannon, D. M., & Ross, M. E. (2013). Students’ characteristics, self-regulated 

learning, technology self-efficacy, and course outcomes in online learning. Distance 
Education, 34, 302–323. doi:10.1080/01587919.2013.835779 

Wang, J. & Wang, X. (2012). Structural equation modeling: Applications using Mplus: Methods 

and applications. West Sussex, UK: Higher Education Press. 

Watson, J., & Ryan, J. (2007). A national primer on K-12 online learning. Washington, DC: 

North American Council on Online Learning. 

Watson, J., Murin, A., Vashaw, L., Gemin, B., & Rapp, C. (2011). Keeping pace with K-12 

online learning: An annual review of policy and practice, 2011. Evergreen Education 
Group. Retrieved from http://eric.ed.gov/?id=ED535912 

Watson, J., Pape, L., Murin, A., Gemin, B., & Vashaw, L. (2014). Keeping pace with K-12 

digital learning: An annual review of policy and practice. Retrieved from 
https://files.eric.ed.gov/fulltext/ED558147.pdf 

Watson, J., Pape, L., Murin, A., Gemin, B., & Vashaw, L. (2015). Keeping pace with K-12 

digital learning: An annual review of policy and practice. Retrieved from 
http://www.kpk12.com/wp-content/uploads/Evergreen_KeepingPace_2015.pdf 

Weiner, C. (2003). Key ingredients to online learning: Adolescent students study in cyberspace - 
the nature of the study. International Journal on E-Learning, 2(3), 44–50. Retrieved from 
https://www.learntechlib.org/p/14497  

Wicks, M. (2010). A National Primer on K-12 Online Learning. Version 2. International 

Association for K-12 Online Learning. Retrieved from http://eric.ed.gov/?id=ED514892  

Wighting, M. J., Nisbet, D., & Rockinson-Szapkiw, A. J. (2013). Measuring sense of community 

and academic learning in graduate education. The International Journal of 
Interdisciplinary Educational Studies, 7(1), 1–8. doi:10.18848/2327-
011x/cgp/v07i01/53294 

Wortmann, K., Cavanaugh, C., Kennedy, K., Beldarrain, Y., Letourneau, T., & Zygouris-Coe, V. 

(2008). Online teacher support programs: Mentoring and coaching models. North 
American Council for Online Learning. 

 

118 

Zhan, Z., & Mei, H. (2013). Academic self-concept and social presence in face-to-face and 

online learning: Perceptions and effects on students’ learning achievement and 
satisfaction across environments. Computers & Education, 69, 131–138. 
doi:10.1016/j.compedu.2013.07.002 

Zimmerman, B. J. (1986). Becoming a self-regulated learner: Which are the key subprocesses? 

Contemporary Educational Psychology, 11, 307–313. doi:10.1016/0361-476X(86)90027-
5 

Zimmerman, B. J. (2002). Become a self-regulated learner: An overview. Theory Into Practice, 

41(2), 64–70. doi:10.1207/s15430421tip4102_2 

Zimmerman, B. J. (2008). Investigating self-regulation and motivation: Historical background, 

methodological developments, and future prospects. American Educational Research 
Journal, 45, 166–183. doi:10.3102/0002831207312909 

Zimmerman, B. J., & Bandura, A. (1994). Impact of self-regulatory influences on writing course 
attainment. American Educational Research Journal, 31, 845–862. doi:10.2307/1163397 

Zimmerman, B. J., Bonner, S. & Kovach, R. (1996). Developing self-regulated learners: Beyond 

achievement to self-efficacy. Washington, DC: American Psychological Association. 

Zimmerman, B. J., & Schunk, D. H. (2012). Motivation: An essential dimension of self-

regulated learning. In D. H. Schunk & B. J. Zimmerman (Eds.), Motivation and self-
regulated learning: Theory, research, and applications (pp. 1–30). New York, NY: 
Routledge. 

 

 

 

119