 

 

 

 

 

 

 

 

 

 

 

 

 

P .
J"

 

“1
”i2.
"7“ 33..
r:- 3?
'71; "2"

A ~33.-

¢

-' 3::
tr“
9"}?

2.1.4 -
1"};
'Ywéfém: ‘

' w. v
...

 

1113315

 

Z, UBRARY
2006‘ Michigan State
University

 

 

 

This is to certify that the
thesis entitled

UNDERSTANDING THE INDIVIDUAL-LEVEL ADAPTATION
PROCESS: A NEW CONCEPTUALIZATION AND MODEL

presented by
TARA A. RENCH

has been accepted towards fulﬁllment .
of the requirements for the .

M. A. degree in Psycholcgy

;//ﬂ,: g4,

Major Professor’s Signature
ﬂy / 912M

Date

 

 

MSU is an Afﬁnnative Action/Equal Opportunity Employer

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 KzlProy/Achres/CIRCIDateDueVindd

 

UNDERSTANDING THE INDIVIDUAL-LEVEL ADAPTATION PROCESS: A NEW
CONCEPTUALIZATION AND MODEL

By

Tara A. Rench

A THESIS
Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of
MASTER OF ARTS
Psychology

2009

 

 

 

UNDERSTANDING THE INDIVIDUAL-LEVEL ADAPTATION PROCESS: A NEW
CONCEPTUALIZATION AND MODEL

key process behaviors. Unexpectedly, hypotheses testing individual difference predictors

were not supported.

 

 

 

ACKNOWLEDGEMENTS

Thanks to Rick DeShon, my thesis advisor, for providing support and guidance
throughout the research process for the current paper.

Thanks to the members of my thesis committee, Steve Kozlowski and Kevin
Ford, for providing helpful feedback targeted at improving the current paper.

Thanks to my amazing family and friends for providing emotional support and

encouragement as I strive to reach my goals.

iii

 

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................. vi
LIST OF FIGURES ......................................................................................................... viii
INTRODUCTION .............................................................................................................. 1
Research Perspectives on Adaptation ............................................................................. 6
Adaptive Performance ................................................................................................ 6
Predictors of Adaptation ............................................................................................. 8
Processes (e.g., Training) of Adaptation ................................................................... 13
Summary and Purpose of Current Research ................................................................. 19
Deﬁning Adaptation ...................................................................................................... 20
Adaptation Model ......................................................................................................... 22
Dynamic Adaptation Model .......................................................................................... 41
Speciﬁc Research Focus ............................................................................................... 44
Study Hypotheses .......................................................................................................... 45
METHOD ......................................................................................................................... 52
Participants .................................................................................................................... 52
Computerized Bandit Task ........................................................................................... 52
Rationale for Bandit Task ............................................................................................. 53
Experimental Design ..................................................................................................... 54
Experimental Conditions .............................................................................................. 54
Procedure ...................................................................................................................... 60
Measures ....................................................................................................................... 65
Online Trait Measures ............................................................................................... 65
Process Measures ...................................................................................................... 68
Dependent Variables ................................................................................................. 70
Manipulation Check .................................................................................................. 71
Pilot Test Results .......................................................................................................... 71
Primary Study ............................................................................................................... 73
Analysis Plan ............................................................................................................ 73
RESULTS ......................................................................................................................... 75
Zoomed Analyses .......................................................................................................... 77
Condition Check ....................................................................................................... 77
Manipulation Check .................................................................................................. 83
Unconditional Means Models ................................................................................... 95
Hypothesis Testing .................................................................................................. 100
Summary of Results ................................................................................................ I38
DISCUSSION ................................................................................................................. 140
Supported Predictions and Implications ..................................................................... 141

'iv

Unexpected Findings and Implications ....................................................................... 144

Additional Implications .............................................................................................. 148
Limitations and Future Research ................................................................................ 149
Summary of Limitations and Future Research Directions .......................................... 155
Conclusion .................................................................................................................. 1 58
APPENDICES ................................................................................................................ 160
APPENDIX A ............................................................................................................. 161
APPENDIX B ............................................................................................................. 163
APPENDIX C ............................................................................................................. 170
APPENDIX D ............................................................................................................. 171
APPENDIX E ............................................................................................................. 173
APPENDIX F .............................................................................................................. 174
REFERENCES ............................................................................................................... 175

LIST OF TABLES

Table 1: Description of Change and Warning Manipulations by Experimental Condition

................................................................................................................................... 56
Table 2: Speciﬁc Instructions and Task Details by Experimental Condition ................... 57
Table 3: Procedural Steps ................................................................................................. 60
Table 4: Change Manipulation Results for Zoomed Analyses ......................................... 79
Table 5: Between-Person Descriptive Statistics and Correlations among Key Study

Variables ................................................................................................................... 88
Table 6: Within-Person Correlations for Performance and Process Variables ................. 93

Table 7: Unconditional Means Model Parameter Estimates and Variance Components 100

Table 8: Early Performance Predicting Switching Behavior .......................................... 101
Table 9: Dependability Predicting Performance ............................................................. 102
Table 10: Dutifulness (C3) Predicting Performance ....................................................... 102
Table 11: Dependability Predicting Post-Change Switching Behavior .......................... 106
Table 12: Openness Predicting Performance .................................................................. 109
Table 13: Openness Predicting Post-Change Switching Behavior ................................. 110
Table 14: LGO Predicting Post-Change Switching Behavior ........................................ 111
Table 15: PPGO Predicting Post-Change Switching Behavior ...................................... 111
Table 16: APGO Predicting Post-Change Switching Behavior ...................................... 111
Table 17: APGO Predicting Post-Change Anxiety ......................................................... 112
Table 18: APGO Predicting Post-Change Threat Appraisals ......................................... 113
Table 19: Contrast Comparison Predicting Process Variables ....................................... 1 14
Table 20: Change X Warning Predicting Performance after a Change .......................... 122

vi

Table 21: Change and Time Predicting Boredom ........................................................... 124

Table 22: Change and Time Predicting Satisfaction ....................................................... 127
Table 23: Change and Time Predicting Self-Efﬁcacy .................................................... 127
Table 24: Change and Time Predicting Off-Task and Negative Thoughts .................... 133

vii

LIST OF FIGURES

Figure 1: Linear Adaptation Model .................................................................................. 23
Figure 2: Dynamic Adaptation Model .............................................................................. 43
Figure 3: Full Data Set Examination of Change Manipulation on Performance .............. 76
Figure 4: Condition 1 Performance Graph ........................................................................ 81
Figure 5: Condition 2 Switching Behavior Graph ............................................................ 82
Figure 6: Change Manipulation Check ........................................................... 84
Figure 7: Warning Manipulation Check ........................................................................... 86
Figure 8: Performance Averaged Across Conditions over Time ...................................... 96
Figure 9: Number of Switches Averaged Across Conditions over Time ......................... 97
Figure 10: Response Time Averaged Across Conditions over Time ............................... 98
Figure 11: Process Variables Averaged Across Conditions over Time ............................ 99

Figure 12: Dutifulness X Pre-Post X Change Interaction Predicting Performance (Pre-

Change) ................................................................................................................... 104
Figure 13: Dutifulness X Pre-Post X Change Interaction Predicting Performance (Post-
Change) ................................................................................................................... 105
Figure 14: Dependability X Time Interaction Predicting Switching in Post-Change
Blocks ..................................................................................................................... 108
Figure 15: Contrast Comparison of Change/No Warning Condition vs. All Others in
Predicting Frustration after a Change (Pre-Change) ............................................... 1 15
Figure 16: Contrast Comparison of a Change/No Warning Condition vs. All Others in
Predicting Frustration after a Change (Post-Change) ............................................. 1 16
Figure 17: Contrast Comparison of Change/No Warning Condition vs. All Others in
Predicting SE after a Change .................................................................................. 1 19
Figure 18: Condition*Pre-Post with Satisfaction with Performance .............................. 120

viii

Figure 19:
Figure 20:
Figure 21:
Figure 22:
Figure 23:
Figure 24:
Figure 25:
Figure 26:
Figure 27:
Figure 28:

Figure 29:

Condition*Time with Satisfaction with Performance ................................... 121
Change and Time Predicting Boredom (Pre-Change) ................................... 125
Change and Time Predicting Boredom (Post-Change) ................................. 126
Change and Time Predicting Satisfaction with Performance (Pre-Change). 129
Change and Time Predicting Satisfaction with Performance (Post-Change) 130
Change and Time Predicting Self-Efﬁcacy ................................................... 131
Change and Time Predicting Off-Task Thoughts (Pre-Change) ................... 134
Change and Time Predicting Off-Task Thoughts (Post-Change) ................. 135
Change and Time Predicting Negative Thoughts (Pre-Change) ................... 136
Change and Time Predicting Negative Thoughts (Post-Change) ................. 137
Hypothesized Model Results ......................................................................... 139

ix

INTRODUCTION

The changing nature of today’s organizations — manifested by rapid and
continuous changes in structure, technology and economics among others — has shifted
workers into ambiguous work roles requiring quick and adaptive decision-making to
perform effectively (Burke, Stagl, Salas, Pierce & Kendall, 2006; Ilgen & Pulakos, 1999;
Kozlowski, 2008; Ployhart & Bliese, 2006). Jobs which were once characterized as
stable and routinized are now being transformed into jobs characterized by shifting
demands and roles as the work environment becomes more dynamic (Pulakos, Arad,
Donovan & Plamondon, 2000). This shift implies that “wOrkers need to be increasingly
adaptable, versatile, and tolerant of uncertainty to operate effectively in changing and
varied environments (p. 612; Pulakos et al., 2000). This is increasingly important in
selection and training contexts as the skills and abilities needed to perform a job in the
present may not be the same skills and abilities needed to perform the same job in the
future (Ilgen & Pulakos, 1999). Rather than identify speciﬁc job-related skills and
abilities, organizations should pinpoint the set of characteristics and behaviors that can
effectively predict adaptability to a changing work environment.

For the reasons stated above, understanding adaptation has become a focal
concern for both researchers and practitioners as they try to grasp how to make
organizations more effective (e.g., Burke et al., 2006; Kozlowski & Bell, 2008;
Kozlowski, Gully, Brown, Salas, Smith & Nason, 2001; LePine, Colquitt & Erez, 2000;
Ployhart & Bliese, 2006; Pulakos et al., 2000; 2002). One quick search in Psyclnfo for

“adaptation” results in over 50,000 hits leaving little room for debate about the amount of

research that has examined adaptation. Narrowing the search to speciﬁc [/0 journals
(e.g., Journal of Applied Psychology, Personnel Psychology) still results in hundreds of
publications. With this volume of research, it is no surprise that adaptation researchers
have made numerous conceptual and empirical advances in both the areas of selection
and training. For example, Pulakos and colleagues (2000; 2002) proposed a taxonomy of
situations that may require adaptation of different types, such as crisis situations and
cultural interactions. Other selection-focused researchers have identiﬁed individual
differences (e.g., personality, cognitive ability) that predict performance in adaptive
contexts at both the individual and team level (e.g., LePine et al., 2000). Lastly, in the
training area, researchers (e.g., Bell & Kozlowski, 2008; Burke etal., 2006) have made
advances by developing and testing conceptual models of adaptive transfer, identifying
the key processes and components (e. g., adaptive expertise) that can be used in training
interventions designed to improve adaptation. The brief examples provided above make
it apparent that the work by these researchers has considerably advanced the adaptation
literature. A more detailed review of these developments will be presented shortly.
While advances have been made in our understanding of adaptation, the literature
is still developing (Bell & Kozlowski, in press), as is to be expected. Considering two
fundamental questions about the adaptation process may help structure thinking about
adaptation and highlight aspects of our current understanding that need development.
First, how does an individual determine that adaptation is needed? Second, once it has
been determined that there is a need to adapt, what does the individual do to adapt? The
latter question has received much attention in the literature, speciﬁcally through the work

of Kozlowski, Burke and others exploring the training and learning contexts for

adaptation (e.g., Bell & Kozlowski, 2008; Burke et al., 2006). This work has provided
solid conceptual and empirical understanding of how people adapt, as well as how to
train people to better adapt (e.g., active learning approaches).

The issue highlighted in the ﬁrst question — determining the need to adapt - has
received far less research attention. This is surprising since recognizing the need to adapt
is a precondition for engaging in adaptive actions. Put simply, an individual may possess
the knowledge, skills and abilities needed to adapt, but if he or she does not recognize
that the current conditions require adaptation, the knowledge, skills, and abilities will not
be put to use and the individual will likely fail to adapt. Recognizing or determining the
need to adapt is comprised of two key sub-components. First, the individual must detect
a change in the relationships between environmental cues in behavior-outcome
contingencies. Second, the individual must decide that the detected changes are
meaningful and that actions must be altered in response to the detected changes.

While relatively unexplored in the adaptation literature, change detection has been
studied in both the cognitive psychology and computer science and statistics literatures
(e.g., Beck, Levin, & Angelone, 2007; Levin & Simons, 1997; Rosin, 1997). Computer
scientists employ computer and statistical modeling methods to explore thresholds for
change detection. In these paradigms, thresholds are manipulated to allow for various
levels of noise or discrepancy in the system before a real change is detected. Rosin (1997)
highlights the importance of setting a proper threshold: “. . .the threshold value is critical,
since too low a value will swamp the difference map with spurious changes, while too
high a value will suppress signiﬁcant changes”. Human adaptation research can beneﬁt

from the information gleaned from the computer models if they are viewed as a

representation of the human mind. Analogous to computer systems, it is likely that
humans have internal threshold systems that are used to detect change. Individual and
situational factors are proposed to impact the effectiveness of an individual’s internal
threshold, and thus the likelihood of detecting when a change requires adaptation.

Exploring human change detection, cognitive psychologists have used a variety of
methods and tasks (e. g., intentional vs. incidental detection, ﬂicker paradigm, complex
visual stimuli) to study change detection, with results suggesting that individuals are not
as good at detecting even large changes in their environment as they believe they are
(change blindness; Beck, Levin & Angelone, 2007). Change blindness can lead to
disastrous consequences, as it did in the French Airbus AT320-111 crash costing many
lives, highlighting the importance of understanding how individuals detect and interpret
changes (see Beck et al., 2007). Cognitive researchers have suggested that limited
cognitive resources, such as attention and memory, play a role in detection failure.
Relatedly, the intention to detect change is perceived by individuals to play an important
role in the successful detection of a change (Smilek, Eastwood, Reynolds, & Kingstone,
2007). Intention to detect change can be manipulated by providing cues or warnings
about changes that will occur, thus serving to focus attention and resources toward
monitoring for changes. Adaptation researchers would be served well to integrate the
conceptual and methodological advances these areas have made into the existing
adaptation process literature. If an individual does not appropriately detect a change in
his or her environment, the likelihood of successful adaptation is diminished.

The second sub-component — deciding that detected changes are meaningful and

that actions must be altered in response — is also critical to determining the need to adapt.

If a change is detected, the individual attributes meaning to the change by determining its
relevance (e.g., is this change relevant to what I am doing?), stability (e.g., is this a
temporary or stable change?) and controllability (Weiner, 1985; e.g., can my actions
inﬂuence this change?) Relevant, stable, and controllable attributions should lead to
perceptions that the change is meaningful and that behavioral changes can be impactful,
thus guiding individuals to determine the need to adapt. Faulty attributions will lead to
misguided decisions about the need to adapt. Together, the detection and attribution sub-
components guide our determination of the need to adapt. Failure at either of these stages
will likely result in the failure to adapt.

As highlighted above, understanding how the detection and decision processes
unfold and identifying the factors that impact these processes is critical to understanding
adaptation. However, relatively little attention has been spent modeling these components
of the adaptation process, although organizational research has hit on some tangents of
these components (e.g., cue detection, Burke et al., 2006; quality control literature on
Sigma Six; Linderrnan, Schroeder, Zaheer & Choo, 2003). The purpose of this research
is to gain a better understanding of the detection and decision processes. To do this, the
ﬁrst goal of the current research is to review and summarize a representative set of the
existing adaptation literature with a detection lens to identify how the literature has
addressed (or failed to address) this critical aspect of the adaptation process. The second
objective of this research is to contribute to the conceptual development of the adaptation
literature by proposing a new model of the adaptation process, including the detection
and decision pieces. The third objective of this research is to empirically test a subset of

the proposed adaptation model by incorporating a detection manipulation within an

experimental design allowing for exploration of both within and between subject
differences over time.

A brief outline of the remaining sections is presented here. The ﬁrst section
reviews the existing adaptation research, highlighting conceptual and empirical advances
as well as noting gaps speciﬁcally in the domain of detection. The next section
summarizes where the current research is headed, followed by the conceptualization of
adaptation used for the current study. The stages of the proposed adaptation model are
then outlined, highlighting the person, task and environmental factors believed to impact
each stage. Finally, the entire dynamic model is reviewed and the speciﬁc research focus

and hypotheses of the current study are presented.

Research Perspectives on Adaptation

Although not intended to be an exhaustive review of the adaptation literature, the
following literature review is expected to be representative of the extant research. To
provide structure, the review is organized by the focus of the researchers in their
exploration of adaptation. This framework includes sections on: taxonomies of adaptive
performance, predictors of adaptation (e. g., individual differences — selection focus), and
processes of adaptation (e. g., conceptual models; training focus). Within each section,

both individual and team-level examinations are included if available.

Adaptive Performance

One conceptual advance in the adaptation literature has been made by Pulakos
and colleagues (2000; 2002). As a literature is developing and growing, one common

strategy to put structure on the research is to develop a taxonomic framework for

organization. In response to the expanding adaptation area, Pulakos et a1. (2000)
developed taxonomy of adaptive performance, contributing an “adaptive performance”
dimension to the performance model Campbell and colleagues (1993) had developed
nearly a decade before. The purpose of this research endeavor was to organize the
literature by reviewing and integrating the existing literature on adaptive performance
and related concepts to form a multidimensional construct of adaptive performance. In
her work, Pulakos deﬁned adaptive performance as “altering behavior to meet demands
of a new situation, event, or set of circumstances” (p. 615). A diverse set of jobs and
employees were identiﬁed and a set of critical incidents were provided and later
categorized into one of the eight dimensions, including: handling emergencies or crisis
situations; handling work stress; solving problems creatively; dealing with uncertain and
unpredictable work situations; learning work tasks, technologies, and procedures;
demonstrating interpersonal adaptability; demonstrating cultural adaptability; and
demonstrating physically oriented adaptability.

While this taxonomy can aid our thinking about the features or situational
characteristics that may require adaptation, thus serving as a conceptual thinking tool, it
does not (and was not intended to) address how adaptation plays out (the process of
adaptation). The adaptive performance deﬁnition provided by Pulakos ignores detection,
suggesting adaptive performance is solely based on adjustment of behaviors to new
situations. However, before individuals can adjust their behaviors effectively, they must
ﬁrst detect that a change has occurred that requires that shift. This missing link is critical
to understanding adaptation. Thus, in terms of detection, the work by Pulakos and

colleagues falls short in addressing this issue.

Predictors of A daptation

As highlighted in the opening to this paper, understanding adaptation has
implications for selection procedures used in organizations. Several researchers have
made conceptual and empirical advances targeted at the selection domain (e.g., LePine et
al., 2000; Ployhart & Bliese, 2006). These individuals have theorized about and tested
several individual difference characteristics that are posited to impact the adaptation
process. The goal of this research is to inform organizations about what qualities
individuals coming into an organization may need (or may be ideal) when facing
dynamic, changing environments. LePine and colleagues (2000) listed four ways of
handling adaptation in organizations: I) constantly hire and ﬁre individuals with the skills
needed at the current time, 2) train the proper currently-needed skills, 3) train general
adaptability skills, or 4) select individuals who enjoy and are competent at working in
changing environments. Of these, LePine suggested that targeting the selection domain is
the most reasonable approach for the current state of the literature, given limitations of
the other approaches.

From a conceptual standpoint, several individual difference variables have been
linked to adaptive performance, or adaptation. The most commonly explored variables
include cognitive ability (Burke et al., 2006; LePine, 2003; 2005; LePine, Colquitt, &
Erez, 2000; Ployhart & Bliese, 2006), conscientiousness (LePine, 2003; LePine, Colquitt,
& Erez, 2000; Pulakos et al., 2002 (achievement motivation», openness/ﬂexibility
(Burke et al., 2006; Georgsdottir & Getz 2004; LePine, 2003; LePine, Colquitt, & Erez,
2000; Ployhart & Bliese, 2006), and goal orientation (Kozlowski, Gully, Brown, Salas,

Smith & Nason, 2001; Porter & Tansky, 1999). In some cases, these variables are

examined as direct predictors of outcomes (e.g., LePine et al., 2000), whereas other cases
examine the variables in a larger training or process model (e. g., Kozlowski et al., 2001).
Although the former approach is likely too simplistic, it serves as a guide for further
thinking about the role of distal predictors on the adaptation process. The rationale for
each of these variables’ roles in the adaptation process is logical, although perhaps
incomplete. For example, openness to experience is deﬁned by Costa & McCrae (1992)
as the tendency to be open-minded, curious, creative and ﬂexible. In an environment
characterized by changes and novel stimuli that often demand novel or revised
approaches, researchers suggest that individuals high on openness would be a good ﬁt.
Their ﬂexible and creative mindsets will likely allow them to “think outside the box”,
view changes as a challenge as opposed to a threat and allow them to respond adaptively
(LePine et al., 2000). A few of the empirical examinations of these relationships are
provided below.

LePine, Colquitt and Erez (2000) studied adaptation in a computerized multiple
cue probability learning task (TIDE2; Hollenbeck et al., 1995), predicting that individual
differences, including cognitive ability, openness, and conscientiousness, would be
differentially predictive of pre- and post-change performance due to the increased
complexity and information processing demands imposed after the unexpected change
(Chan, 1996; Lewin, 1951). The results partially supported the authors’ predictions,
ﬁnding that cognitive ability was more strongly related to adaptive performance (after
task cues changed unexpectedly; r = .43) than pre-change performance (r = .23),
openness shifted from being unrelated to pre-change performance (r = -.01) to

moderately, positively related to adaptive performance (r = .35), and conscientiousness

went from slightly positively related to stable performance (r = .10), to moderately,
negatively related to adaptive performance (r = -.29). These ﬁndings provided important
conceptual and empirical advancements in the adaptation literature. First, the differential
ﬁndings across pre-change (routine) and post-change (adaptive) performance provides
support for re-examining the selection procedures currently in use and revising them to ﬁt
the speciﬁc criteria they are meant to target. Second, the signiﬁcant correlations between
the key variables and adaptive performance suggest variables that should be examined
further in the adaptation literature. Third, the study design provides one way of studying
adaptation (abrupt, unexpected change).

Despite the interesting ﬁndings, several holes in this research need to be
highlighted. First, LePine and colleagues conceptualized adaptation in terms of learning
and performance in a novel environment, yet they failed to include any process measures
to assess learning, preventing the explanation of why the relationships occurred. While
noting differences in relationships across stable and changed contexts is important,
LePine and colleagues research is limited in that it fails to explain how those changes
played out over time. Second, and critical to the current research, is the lack of concern
over the detection process. The experimental design did not allow for examination of
whether or not (and how) individuals detected the abrupt, unexpected change. All
participants were unaware of the change and without measures to assess whether and
when detection occurred, the adaptive components of detection and behavioral
adjustment cannot be teased apart. It is possible that the correlations between the

individual difference variables and adaptive performance outcomes may be driven by

10

one’s ability to detect the change, one’s ability to adjust behaviors to the change or a
combination of both. LePine’s study did not allow for that examination.

LePine (2003; 2005) extended his adaptation work to the team-level, using the
same task being performed in teams rather than by individuals. In his 2003 study, LePine
replicated his individual level ﬁndings, with the dependability facets of conscientiousness
relating negatively to post-change performance. In this study, however, LePine
incorporated a mediator, role structure adaptation, deﬁned as “reactive and nonscripted
adjustments to a team’s system of member roles that contribute to team effectiveness” (p.
28). The dependability facets were negatively related to role structure adaptation as well
as post-change performance. Although these ﬁndings provide replication of the
individual level ﬁndings, they still fail to address the detection piece of adaptation.

The second study (2005) made alterations to the task, focusing on a gradual
change in the task as opposed to an abrupt change. LePine incorporated team process
assessments guided by Marks et al. (2000) as an attempt to provide insight into how team
processes impact adaptive performance. Although this study provided more information
about team processes inﬂuence on team adaptive performance which is important to the
literature, it too failed to incorporate detection. However, implicit in LePine’s work is
the notion that the individuals and teams performing the decision-making task must ﬁrst
detect that the abrupt or gradual change has occurred before they would begin making
behavioral adjustments. Therefore, although not explicitly discussed, LePine notes that
detection plays a key role in responding to changes in one’s environment.

Ployhart and Bliese (2006) also emphasized the role of individual differences in

predicting adaptation, proposing the Individual ADAPTability model (I-ADAPT). From

11

their perspective, Ployhart and Bliese suggested that individual adaptability can be
viewed as a composite KSAO that can be used to predict self-regulatory processes (e.g.,
situation perception, knowledge acquisition), which then impacts performance outcomes,
Ployhart and Bliese deﬁned individual adaptability as “... an individual’s ability, skill,
disposition, willingness, and/or motivation, to change or ﬁt different task, social, and
environmental features” and “. . .a reasonably stable individual difference construct that
inﬂuences how a person interprets and responds to different situations” (p.16). From a
selection perspective, the adaptability construct can be assessed to predict how
individuals will perform in situations requiring adaptation. The deﬁnition of adaptability
provided does not suggest explicitly how this construct predicts detection of a change;
however, the construct is proposed to predict situation and appraisal processes which
involve recognizing changes in key situational features and cues and recognizing when
cues have not changed but should have (p. 23). That said, Ployhart and Bliese provide
some conceptual grounding for the importance and role of detection-like processes in the
adaptation model. However, in its current state, the model is too vague to aid prediction.
The abstract level at which these processes are deﬁned and modeled does not make it
clear how detection actually occurs and what speciﬁcally impacts the ability to detect a
change. In addition, empirical work needs to be conducted to explore these relationships,
and speciﬁcally to determine whether or not detection can be teased apart from
behavioral adjustment in this model.

The selection approach to adaptation has provided both conceptual and empirical
understanding into the role of individual differences on adaptation as highlighted above.

However, as can be seen by the review of the literature, gaps exist that need to be ﬁlled.

12

u,"
, .

LI

UR

Speciﬁcally, detection is implicitly important to all of the above work, but not explicitly
addressed. Ployhart and Bliese (2006) discussed detection-like processes, but did not

evaluate how the process unfolds and what impacts it speciﬁcally.

Processes (e. g., Training) of Adaptation

In addition to selection, adaptation research can also have implications for
training in organizations. Research in this area focuses on the processes and factors
underlying transfer of training and learning from practice/training/routine environments
to more complex, difﬁcult environments. Bell and Kozlowski (in press) distinguish
between two types of transfer: 1) near transfer represents the transfer of skills to
problems/tasks/environments similar to those faced in training, and 2) far transfer
“involves using one’s existing knowledge base to change a learned procedure, or to
generate a solution to a completely new problem” (cited from Ivancic & Hesketh, 2000;
p. 1968). Although both types of transfer require some degree of adapting, far transfer is
proposed to represent adaptation. The goal of the majority of this research is to identify
how to train individuals on the skills and competencies they need in order to adapt
previously learned knowledge, skills and strategies from one problem to the next.

Chan (2000) provides a conceptual piece arguing that the individual difference
and training literatures on adaptation can be (and should be) integrated. Chan provides a
general, working deﬁnition of adaptation, suggesting it “refers to the process by which an
individual achieves some degree of ﬁt between his or her behaviors and the new work
demands created by the novel and often ill-deﬁned problems resulting from changing and
uncertain work situations” (p. 4). He posits that while the individual difference

perspective focuses on who is more likely to be adaptive and the learning perspective

l3

focuses on how to train individuals on the skills they need to be adaptive, both
perspectives recognize that adaptation is necessary and fundamentally view the
components of adaptation as being similar. Each approach has limitations, which in part
are addressed by the alternative approach. However, even with his integrated approach,
Chan does not explicitly address the detection piece of adaptation. That said, Chan does
highlight a few key constructs that are common in the training adaptation literature,
which hint at the need for detection. Adaptive expertise is characterized by a deep
structural knowledge base (as opposed to superﬁcial/surface level knowledge), which is
proposed to aid recognition or identiﬁcation of situations in which previously effective
procedures or behaviors do not apply. Metacognition is the second construct Chan
highlights. Metacognitive skills are believed to aid recognition of novelty or change, as
well as monitoring, strategy selection and evaluation (Smith, Ford, & Kozlowski, 1997).
Both of these constructs hint that there is a place for detection in the adaptation process;
however, these constructs serve more to suggest what may impact detection, as opposed
to informing how to model the detection process.

One important contribution of Chan’s conceptual piece is the list of conceptual
and empirical issues that need further development. Speciﬁcally, he suggests that we
need to shift to more dynamic models of adaptation and we need to incorporate more
sophisticated design and analytic techniques that allow researchers to empirically
examine the dynamic models over time (within and between individuals). The current
study was designed being mindful of these suggestions.

Training-related adaptation literature has been conducted at both the individual

and team level. Early work by Kozlowski and colleagues (Kozlowski et at., 2001; Smith,

14

Ford & Kozlowski, 1997) focused on training and learning processes important to
individual-level adaptation. Focusing training efforts toward developing adaptive
expertise (in contrast to routine expertise which was the previous focus of training) was
proposed as key to better performance in changed environments (Smith, Ford &
Kozlowski, 1997). In line with the work Chan (2000) described above, Smith and
colleagues suggested that adaptive expertise is associated with learning outcomes and
metacognition, which aids in recognition of changes and selection of appropriate
responses to the changes. This work has conceptual and practical implications for
training, providing a guide for what training goals should be to develop an adaptive group
of employees. Although constructs related to detection are discussed, the process is not
explained and it is not clear how Smith and colleagues would operationalize and test
detection empirically.

In another individual-level study, Kozlowski and colleagues (2001) conducted an
empirical examination of training-related adaptation processes ﬁnding support for several
mediating variables, including declarative and procedural knowledge and self-efﬁcacy, in
the transfer outcomes (adaptive performance) suggesting that these factors partially
underlie effective adaptive performance (see also Ford et al., 1998). Although this line of
research contributed to the development of process-oriented individual adaptation,
detection can not be distinguished from behavioral or strategic adjustment in this work
because participants in transfer of training research are often told that the transfer task is
different, thus not allowing the study of the detection process.

In more recent works, adaptation research has shifted focus from individual-level

adaptation to team-level adaptation (Bell & Kozlowski, 2008; Bell, Kanar & Kozlowski,

15

2008; Bell & Kozlowski, in press; Burke et al., 2006; Chen, Thomas & Wallace, 2005;
Kozlowski, 2008; Kozlowski & Bell, 2008; Kozlowski, Watola, Nowakowski, Kim &
Botero, in press). One of the major changes occurring in organizations today is a shift to
team-based operations, which has instigated the push for understanding adaptation in
teams. While this is an important ﬁxture direction for adaptation research to move, the
added layer of complexity attributed to examining a team context, makes it even more
difficult to study adaptation. The extant team research focuses on process—oriented
adaptation research, primarily identifying the processes and skills relevant to transfer of
knowledge and skills to transfer problems. Several important conceptual advances have
been made in this area as summarized below.

As mentioned above, researchers examining adaptation in the training transfer
context often hone in on several key characteristics or training outcomes which are
believed to lead to effective performance in a transfer training task (e.g., Chen, Thomas
& Wallace, 2005). Kozlowski and colleagues provide theoretical guidance for
understanding the key process variables in this context and using this theoretical basis for
guiding development of training designs (e. g., Active Learning System (ALS)) promoting
adaptive behavior. The active learning approach (leaming-centered training) is becoming
increasingly important as large organizations adopt technology- and simulation-based
training, which allows employees to train independently. The research on learning-
centered approaches is still deveIOping, as suggested by Bell and Kozlowski (in press):
“while our understanding of how to develop adaptability remains limited, emerging
research suggests that training designs that selectively inﬂuence cognitive, motivational

and affective self-regulatory processes to induce an active approach to learning may hold

16

promise...” (p. 4). These self-regulatory components include practice, self-monitoring
and self-evaluation reaction, which then guide performance in near and far transfer.
These processes operate at both the individual and team level, with multilevel, multiple
goal regulation proposed to be at the core of team learning (Kozlowski & Bell, 2008).
Guided by these conceptual advances, Kozlowski (2008) explored how the active
learning approach could aid development of competencies need by soldiers in Iraq and
Afghanistan, who face rapidly changing and possibly life-threatening situations.
Kozlowski suggests that three competencies are important for soldiers in these
environments: 1) assess situations, 2) solve problems intuitively, and 3) adapt to the
situation. From this, Kozlowski posits that “the basic process of adaptation entails (a)
detection. . ., (b) diagnosis. . ., and (c) adaptation...” (p. 9). This is one of the ﬁrst
conceptual pieces to explicitly state detection as one of the key components in the
adaptation process. Detection in this context includes detecting deviations from the
routine situation, which most closely mirrors the argument presented in the current
research for the role of detection. Although most of the transfer literature informs
participants that the transfer task will not be the same (thus making detection irrelevant),
it is apparent in this example that real-world environments sometimes change too rapidly
for advanced warning of change or are unpredictable, thus requiring individuals to detect
the change This is an important conceptual advance in the adaptation literature and also
highlights the importance of detection in some real-world settings. However, the work by
Kozlowski does not provide understanding on how to empirically tease apart the steps
(e.g., detection vs. adaptation), leaving a gap to be ﬁlled by future research. In addition,

although adaptation does often involve shifts from less to more complex tasks, other

17

types of change (e.g., abrupt changes in underlying rules of the same task) also require
adaptation.

Recent conceptual work by Burke and colleagues (2006) further advances the
theoretical underpinnings of team adaptation. Deﬁning team adaptation as “a change in
team performance, in response to a salient cue or cue stream, that leads to a functional
outcome for the entire team. . .manifested in the innovation of new or modiﬁcation of
existing structures, capacities, and/or behavioral or cognitive goal-directed actions” (p.
1190), Burke et al. provide a more functional deﬁnition of adaptation, laying out some
key components that are required for adaptation. For example, to be considered
adaptation, Burke would suggest that a signal (cue) in the environment or task will result
in a subsequent change, either cognitive or behavioral in nature, by the team and that
change must be successful (i.e., either maintenance or improvement of previous
performance levels). Burke and colleagues’ model of team adaptation includes four key
steps required for adaptation: situation assessment, plan formulation, plan execution and
team learning. Failure at any of these steps results in a failure to adapt, and each step
may be affected by different individual characteristics or processes. Burke suggested that
several predictors (e.g., cognitive ability, openness to experience, mental models) may
impact these steps, but the model has not been tested.

In addition, consistent with the research reviewed previously, the gap in Burke’s
work is related to teasing apart the detection process from behavioral adjustment.
Situation assessment may be necessary to detection, but it is not the equivalent to

detection. Furthermore, an empirical examination of the model that Burke proposed is

18

necessary to guide understanding of how the detection process occurs over time and what

factors actually inﬂuence this particular aspect of the adaptation process.

Summary and Purpose of Current Research

As evidenced in the review of the adaptation literature above, a lot of progress has
been made in both our theoretical and empirical understanding of adaptation. Although
the different reviews and examinations of the adaptation process have provided many
insights into the key components of the adaptation process and the factors that may (and
do) inﬂuence it, the one consistent gap across the literature is the lack of attention paid to
the detection process. Implicitly, detection is a part of adaptation in most of the research;
however, the explicit conceptualization of detection, how it ﬁts into the process, how to
tease it apart from the other components and the speciﬁc factors impacting it are
relatively unexplored. More recent research in the team adaptation literature bring
detection into the conceptual picture, but it is argued here that to test this process
empirically, it is necessary to examine adaptation in a less complex environment to tease
apart the basic processes by which people adapt.

The purpose of the current research is to provide conceptual and empirical
advances to the existing literature by proposing a theoretical model of the adaptation
process at the individual-level that incorporates detection, identifying factors that may
impact (possibly uniquely) each of the adaptation stages, and empirically teasing apart
detection and behavioral adaptation by examining how the process unfolds over time in
varying conditions requiring different levels of detection (detection manipulation).
Speciﬁcally, the contributions of the current study: I) explore the detection piece of

adaptation that is often ignored, 2) examine the adaptation process as it occurs both

19

between individuals and within individuals over time, and 3) tie in process variables that
help explain the “why” of adaptation that is missing from much of the literature (e.g.,

LePine et al., 2000).

Deﬁning Adaptation

Using Burke and colleagues’ (2006) deﬁnition of team adaptation as a guide, the
current research deﬁnes adaptation as a process of monitoring, change detection
recognition and interpretation (attribution), behavioral responses and evaluation, which
results in an eﬂective change in one ’s cognitions, emotions and/or behaviors in response
to the performance cues/demands presented in the environment, such that over time the
adjusted cognitions, emotions and/or behaviors align with the new demands (minimizing
performance decrements). The key components within this deﬁnition will be explained
further below.

The ﬁrst component, which is inherent in the above deﬁnition, is that the
environment or task must change. The nature of the change can vary across tasks and
situations, but the demands (e. g., complexity, priority, payoffs) of the environment must
change in order for an individual to have something to adapt to.

Another key component within the deﬁnition above is the presence of and
recognition of cues in the environment or task that indicate a change needs to be made. A
cue can vary in levels of salience, ranging from an overt alarm signaling that something
has gone array to subtle feedback from the environment indicating a decline in
performance. Not only must a cue be present, but the individual must recognize the cue
and interpret meaning from the cue as to what the problem is. To recognize or detect the

cue, the individual must be monitoring their environment (e. g., explorative behaviors;

20

feedback monitoring). Assigning meaning to the cue is another required component of
adaptation, as an individual cannot determine how they need to change if they do not
know what the new environmental or task demands are.

The next component indicates that the way an individual changes can be
cognitive, aﬂective and/or behavioral in nature. This differs from Burke et al.’s (2006)
deﬁnition of adaptation, which suggested that the change was in team performance. The
focus of the current deﬁnition does not require an individual’s performance to change;
rather maintenance of current performance levels can still be considered adaptation, as
suggested above. What is required is a change in the way individuals pursue the task, by
altering their cognitions (i.e., beliefs or knowledge about the current state of the task),
emotions (e.g., reactions to the task and their performance), or behaviors (e.g., the actions
or choices taken while performing the task). Although cognitions and emotions are
believed to play an important role in the adaptation process, the roles are less understood
in the literature, and therefore the primary focus of the current model is on behavioral
adaptation.

The last component of adaptation is represented by the ﬁrst part of the deﬁnition
above: “successful (e. g, effective) change”, which means that not only must an individual
change, that change must be effective given the new demands of the environment. To be
considered effective, the change an individual makes must be relevant to the new task
and environmental demands. Ineffective changes (e.g., random behaviors, irrelevant
adjustments in behaviors) are not considered to be adaptation. While this

conceptualization is somewhat ambiguous, in that the exact criterion is dependent on the

21

features of the speciﬁc task and the environment, this conceptualization is believed to be

appropriate given what we know about adaptation.

Adaptation Model

A model of adaptation was developed that encompasses the key components in
the deﬁnition above (see Figure 1). The model is composed of ﬁve key steps including:
1) exploiting and monitoring, 2) change detection, 3) attribution and decision process, 4)
behavioral response, and 5) evaluation. The rationale for and the description of these
steps is provided below, followed by the presentation of a dynamic model proposed to
integrate the steps.

Exploiting and Monitoring

Exploiting and monitoring occurs prior to a change, where an individual exploits
current knowledge using strategies that have resulted in high performance in previous
experiences, while monitoring his/her performance and the environment for change.
Person, task, and performance factors present while an individual is performing in a
stable environment will inﬂuence his/her ability to detect a change that occurs in the
environment. In a context where performance feedback provides the signal that a change
has occurred, individual differences or task characteristics leading to higher, less variable
performance levels during periods of stability, will make a change more obvious (as
indicated by a drop in performance). Conversely, greater variability in performance
before a change will make it harder to detect a change, as it will be hard to distinguish the
change from the existing noise in performance levels. These factors are proposed to
inﬂuence both ability to detect and rate of detection in different ways: directly, indirectly

with other factors as mediators and indirectly through moderation. For example, many of

22

 

 

 

 

 

 

 

E3325: .3va
~33?» wagging M
«Saga .3 $83335» no :3?
mouse—‘9 \o 35.8 is: as “:33. ENEES we: MSSuion
3:55:335 \e unoguauuux§ mug—B $43535 SESSEuE 35 Si 3352 «$335
u=te-§i £3... M53396» "SEES: as: Swish .zuﬁaehaeu 8 LEE
@833 uéeﬂﬁ M538 $2: 8 .3335. Sousa? mi 5 3‘..th
QEF—t Mséksﬁv‘m ﬁhwﬂu§$ﬁt “huh 2‘» “SimuuQ L6 =3 nm N§- 83..» NEESNQ 9N n§$=
A war—8:5 33o.— warez—52
. . S noun—.33— omnoamom .5550 m 558:5 a.
Q _a._o_>2_om . . A— Q owns—.0
“Easing . 253:3. ”5.5.5
noun-Ema EMBED
.auaoﬁueagnm

 

 

 

333. «628%ch 33.5

~ 833m

23

the person factors (individual differences) will act indirectly through their relationship
with the task, environmental and performance factors, while performance factors will
have a more direct role in change detection. A description of these factors is provided
below.

Person factors. Person factors are any characteristics of the individual who is
performing the task that may inﬂuence change detection. For example, cognitive ability
will likely inﬂuence the rate of detection through several mediums. First, cognitive
ability is positively related to performance in complex tasks (Hunter & Hunter, 1984), so
individuals high in ability will likely perform at higher levels with less performance
variability in the stable environment, and thus have an easier time detecting changes that
occur. High cognitive ability is also associated with learning and task
knowledge/mastery (e. g., Kozlowski et al., 2001). The more task knowledge an
individual has the more familiar the task and the task environment will be to that
individual. Without task knowledge, the ability to detect whether a relevant ﬂuctuation
has occurred is difﬁcult because the individual has not yet learned all of the intricacies of
the task.

Personality factors will likely play different, but important roles in the exploiting
and monitoring stage. The facets of Costa and McCrae’s (1992) conscientiousness factor
could help explain the behaviors individuals engage in during this stage. Costa and
McCrae propose six facets of conscientiousness, which can be grouped into two larger
subfacets: volition/ achievement (competence, self-discipline, achievement-striving) and
dependability (order, dutifulness, deliberation). In a stable environment, individuals high

in conscientiousness tend to perform better than those low in conscientiousness (Barrick

24

& Mount, 1991; LePine, Colquitt & Erez, 2000). Performing at higher mean levels of
performance during the stable environment will allow individuals to have an easier ability
to detect when a change occurs as they will see the greatest impact on performance (e.g.,
if an individual is not making any errors and suddenly is making a lot, more likely to
realize that something has changed). Related to this, high conscientiousness has been
associated with persistence and rigidity, which represents exploitation. Another
personality factor proposed to impact change detection is openness to experience.
Individuals high on openness are more curious about their surroundings and more likely
to explore alternatives in their environment (Costa & McCrae, 1992). The tendency to
explore is posited to increase monitoring behaviors in a stable environment, thus allowing
for faster change detection rates. Individuals that monitor at a greater rate will receive
the feedback that a change has occurred quicker than those who monitor less frequently
or less intensely. Burke et al. (2006) proposed that openness will “. . .contribute to
noticing and assigning meaning to cues that are often subtle and ﬂeeting...” (p. 1200),
supporting the proposition that openness will aid detection of change.

Attentional factors are likely to impact detection rates as well. For example, if an
individual is focused on the task as opposed to extraneous factors, they will be more in
tune when a change occurs and thus more likely to detect it. Attention to a task may be
impacted by anxiety levels, such that anxiety results in more self-focused attention (e. g.,
rumination, worry) that will detract attention from the task and signal cognitive
withdrawal from the task (Eysenck, Derakshan, Santos & Calvo, 2007). Therefore,
anxiety should indirectly inﬂuence change detection through its impact on task-related

attention due to resource limitations (Kanfer, Ackerman, Murtha, Dugdale & Nelson,

25

1994). In addition, individuals that are higher on attention to detail will notice when
features of their task environment change faster than those individuals that are lower in
attention to detail. Commitment to the task will reduce random guessing while
performing the task, which will also increase the rate at which one will be able to detect
that a change has occurred.

Hyland (1987; 1988) proposed the concept of error sensitivity, which may also
impact change detection. Error sensitivity impacts the salience of errors (or
discrepancies) as an individual is pursuing a goal, such that high error sensitivity will
make errors more salient and thus more detectable than low error sensitivity. Hyland
(1987) proposed that the intensity of the reaction to the same error will vary based on the
level of error sensitivity. Thus, an individual with high error sensitivity will be quicker
and more likely to determine that something has changed than individual low in error
sensitivity.

T ask factors. This section highlights some of the features of the task itself that
can impact the ability for an individual to detect changes in their environment. One task-
related factor that may impact one’s ability to detect a change in the environment is task
complexity (e.g., Wood, 1986). Campbell (1988) reviewed the literature on task
complexity, highlighting the various conceptualizations of task complexity. Using
objective task characteristics to deﬁne task complexity, March and Simon (1958)
proposed that unknown or uncertain alternatives of action as well as uncertain links
between an alternative and an outcome lead to increased complexity. In addition,
Terborg and Miller (I 978) suggest that the number of possible paths that can be taken to

reach a goal is related to complexity, especially when performance on the task is

26

evaluated using “optimal” path data. Based on this research, it is proposed that as a task
becomes more complex, with more interrelationships among the task components and
more alternatives to choose from, it is more difﬁcult for an individual to identify a change
in the environment (i.e., the change will be harder to detect—not as obvious).

The rate of feedback from the environment concerning performance and
environmental variability should also impact the ability of an individual to detect a
change during the exploiting and monitoring stage. As an individual is performing in a
task environment, the slower the feedback (e.g., sporadic or completely absent) is
presented to the individual, the slower the participant will be able to detect a change has
occurred. On the other hand, if the individual is receiving feedback continuously (i.e.,
after every action taken), detecting a change in the environment or one’s performance
will be quicker.

The clarity of the feedback presented in a task environment can also impact one’s
ability and speed at which change detection occurs. For example, if an individual cannot
decipher what the feedback is relaying about the current situation, it will be harder to
ﬁgure out when the feedback is giving them cues that something has changed.
Ambiguity of feedback will slow down the ability of an individual to detect a change in
his/her environment.

Performance factors. Performance factors are inﬂuenced by both the person and
task factors presented above. Performance variability will play a key role in detection of
cues in the environment. Lower performance variability will provide a more stable
picture against which a change occurring will stand out, whereas highly variable

performance will make it harder to detect a change, as the impact of the change will not

27

be able to be teased apart from the existing noise in performance levels. Similarly, mean
performance will impact detection rates, such that those individuals performing at a
higher level of performance will be more impacted by a change in the environment (i.e.,
will see the greatest performance decrement), than those performing at a lower level of
performance (making a lot of errors prior to the change).

Change Detection Process

Change Detection is the step at which individuals perceive a cue in their
environment suggesting that change is needed. For example, one cue could be increased
levels of negative emotions, including frustration or anxiety, which signals that one needs
to change. Aside from individual differences, task and performance factors, normal
environmental variability and the nature of the change itself will also impact the ability of
an individual to detect when a change has occurred in the environment.

Environmental factors. Tasks are performed in an environmental context that has
features that impact the ability to detect changes occurring within that environment.
Environments that tend to have high variability, represented by frequent or constant
shifting in the demands imposed on the tasks that are operating in that environment, will
likely feel more unpredictable to individuals, thus reducing their chances at detecting
when an actual (relevant) change has occurred. Environments that are highly volatile
may produce many changes that are irrelevant to the task that is being performed by a
certain individual, but that volatility will increase the uncertainty about when a change is
relevant to the individual. Examples of this can be found in the quality control literature
on Sigma Six (Linderrnan, Schroeder, Zaheer & Choo, 2003). Companies seeking

quality control typically set bounds within which they will accept variability in their

28

product, but when a product (or a certain percentage of products) exceeds those bounds,
cues will be sent out to the company that something is out of whack. If an environment
allows for a lot of variability (and experiences a lot of variability), it will take longer for
an individual to detect a change. However, if there are tight “quality control” boundary
conditions placed on the environment reducing the amount of variability allowed before
determining something has changed (e. g., variability that is “out of bounds”), the quicker
a change will be detected.

The nature of the change itself will also impact the ability to detect the change.
For example, if there is a warning or an obvious cue that calls attention to the fact that the
environment or the task is about to change, monitoring will increase in preparation,
making it easier to detect the change when it happens. That is, the change will become
more salient. On the other hand, if an individual is not provided with any warning or
cues that the system is about to change, monitoring will likely remain at lower intensity
and thus result in slower detection of a change. Another feature of the change event itself
is whether it occurs gradually over time or abruptly. An abrupt change may be more
obvious to individuals performing in an otherwise stable environment, whereas a gradual
change will have subtler cues that something has changed in the beginning, thus
increasing the time it takes for an individual to detect the change. Similarly, the intensity
of the change will inﬂuence whether the change is more subtle (thus harder to detect) or
more obvious (easier to detect). Stronger changes with large impacts on performance and
the environment should be detected faster than changes that have more minor impact on
the current state.

Attribution and Decision Process

29

Once a possible change has been detected, the next step is Attribution and
Decision Making. During this stage, individuals will make attributions about the change
they detected (e. g., Is it stable? Is it under my control?), and then decide whether to
continue with their current strategy (i.e., ride it out) or make a change. This process
could also result in emotional reactions, with certain attributions guiding decisions in the
next step.

Causal attribution dimensions. Weiner (1985) proposed that individuals attribute
outcomes and events based on three causal dimensions: locus, stability and
controllability. Locus refers to whether an individual attributes an outcome or event to
internal causes (e. g., cognitive ability, effort) or external causes (e.g., luck, environment).
Stability refers to whether an individual attributes an outcome or event to a temporary
ﬂuctuation (unstable) or a relatively stable cause. Lastly, controllability refers to whether
an individual attributes an outcome or event as under their volitional control or out of
their control. Weiner (1985) argues that these attributions will guide behavior and
“. . .that adaptation is not possible without causal analysis” (p. 549). Therefore, it can be
expected that based on the attributions an individual makes, differences may appear in
whether or not individuals decide that a change in their environment is relevant and
requires problem solving. It follows that if a change is attributed to temporary (unstable)
causes, one is less likely to decide that a change is relevant. However, if one attributes a
change to stable causes, the individual will expect that past outcomes will reoccur, such
that performance decrements following a change will likely reoccur unless the individual
reacts thus increasing the likelihood that the individual will decide the change is relevant

(Ford, 1985; 1987; Weiner, 1985). Changes that are attributed to factors outside the

30

individual’s control could possibly lead to one of two decisions: 1) the individual will
decide that they cannot do anything about the change and its effects, and therefore either
withdrawal or persist with the previous strategy, or 2) the individual will acknowledge
that the change is relevant, but instead of changing courses of action will instead impose
control on the environment by increasing effort (Ford, 1985; 1987). Additionally,
changes that are attributed to controllable and/or internal forces are likely to be
determined relevant and individuals will then proceed with problem solving in the next
stage (Ford, 1985; 1987).

Person threshold variables. In addition to the attribution process described
above, several individual difference variables are proposed to inﬂuence the decision
process about whether to ride a change out (i.e., determine a change doesn’t require
action) or pursue other problem solving strategies. For example, while the dependability
facets are expected to be beneﬁcial to performance during the exploitation stage, they are
likely to be detrimental to decision making in the face of change (LePine, Colquitt &
Erez, 2000). These facets have been associated with persistence and rigidity, such that
individuals high on the dependability facets will likely impose previously effective
strategies on a changing environment to maintain control over their environment (LePine,
Colquitt & Erez, 2000; Stewart & Nandkeolyar, 2006). Similar to the threat-rigidity
hypothesis proposed later, rigidity associated with the dependability facets should lead to
more “ride it out” decisions about changes. In addition, the deliberation facet of
conscientiousness is related to cautiousness and error avoidance (Costa & McCrae, 1992;
Gully, Payne, Koles & Whiteman, 2002). Following this same logic, the need for control

is another individual difference that is believed to result in decision strategies similar to

31

those determined by the dependability facets of conscientiousness. Although previous
research has not examined speciﬁc facet predictions with deliberation that this author has
found, it is reasonable to believe that high deliberation would be associated with the risk-
averse tenets of prospect theory, such that individuals high in deliberation would be less
likely to seek out alternatives or explore, deciding instead to ride the change out.

The openness factor of the Big 5 on the other hand has been associated with more
curiosity and exploratory behaviors, suggesting that in the face of a change, an individual
high on openness will be more likely to decide the change is relevant and then move onto
different strategies to ﬁgure out how to counteract the effects of that change. In fact the
conceptualization of openness is in contradiction to the rigidity effects found in facets of
conscientiousness, suggesting that while openness may not be as beneﬁcial to
performance in stable environments (where exploitation will be more rewarding), it will
likely be positively related to performance in changing environments by increasing an
individual’s decision to make a behavioral change based on the change detected in the
environment (LePine, Colquitt & Erez, 2000). In support of this, Burke et al. (2006)
posit that “. . .open individuals are less likely to become entrenched in routines and are
more accepting of novel solutions to problems” (p. 1200; see also LePine et al., 2000).

The organizational strategy literature provides insight into another potential
individual difference variable that may impact the decision process at this stage. Miles
and Snow’s (1978) typology of strategic types include prospectors, defenders, analyzers
and reactors. Prospectors are proposed to succeed in unpredictable or changing
environments because they are risk-taking, always searching for new alternatives to excel

beyond the competition. Bringing this logic down to the individual level, it is proposed

32

that individuals that are high risk-seekers or high in novelty or sensation-seeking will be
more likely to take similar approaches in the face of a change. That is, these individuals
will be more likely to decide to explore and make behavioral changes (i.e., pursue other
alternatives) in the face of a change.

Another individual difference proposed to be related to the decision process is
fear of failure. Individuals with a high fear of failure will likely desire more conﬁrmation
that a change is real before deciding to change courses of action. One explanation for
delaying this decision can be found in Anderson’s (2003) article on the psychology of
doing nothing. Anderson proposes that the outcomes resulting from doing nothing are
associated with less anticipated regret than the same outcomes resulting from taking a
course of action. Those with a fear of failure may choose to ride out a change because
they fear less regret from failure if they do nothing than if they make a change that results
in failure. Similarly, performance avoid goal orientation (Vandewalle, 1997) is posited to
be related to fear of failure.

Gain versus loss ﬁame. Experiencing threat and/or anxiety during the attribution
and decision-making stage will impact the behavioral change choice stage; however, two
perspectives have been proposed that differentially predict the direction of this impact.
Prospect theory (Kahneman & Tversky, 1979) and the threat-rigidity hypothesis (Staw,
Sandelands & Dutton, 1981) provide differential behavioral predictions based on the
same attribution of threat, but both predictions are posited to lead to a decision that a

change needs to be reacted to, and therefore will lead to the behavioral change choice

stage.

33

Prospect theory (Kahneman & Tversky, 1979) proposes that when a change is
framed as a gain (opportunity), individuals are more likely to engage in risk-averse
behaviors, such as less exploration and more exploitation, because they do not want to
change what is working. Conversely, when a change is framed as a loss (threat),
Kahneman and Tversky propose that individuals are more likely to engage in risk-seeking
behaviors, such as more exploration and less exploitation, to attempt to counteract the
threat. Anderson and Nichols (2007) reported that the more diverse information collected
through exploration, the greater the reduction in the perceived threat.

Contradictory to the tenets of prospect theory, the threat-rigidity hypothesis
(Staw, Sandelands, & Dutton, 1981) posits that threats actually lead to a narrowing of
search, reducing exploration and leading to increased persistence (exploitation). In their
1981 article, Staw and colleagues suggest that a perceived threat (deﬁned broadly as
impending loss or cost to the entity) will increase anxiety and stress, which then act to
restrict information-processing and constrict control leading to rigidity in response.
Response rigidity is deﬁned as “tendency toward well-leamed or dominant responses” (p.
503), such that in the face of a threat, an individual is less ﬂexible, employing instead a
previously effective strategy (Staw et al., 1981). When a threat stems from a real change
requiring strategy or structural adjustment, this rigid response is proposed to be
maladaptive. Although Staw et al. suggest that increased search may occur directly after
a threat has been perceived, they posit that the information individuals receive during this
search will be similar to previous searching as they will use standard operating and
information procedures to guide their search.

Behavioral Response

34

If individuals decide to make a change in the previous step, they will move to the
Behavioral Response stage, where they will make a decision about what type of change
to make to meet the new demands. Newell and Simon (1972) suggest that the problem
space representation formed by individuals will guide their choice of problem solving
strategies. The previous stages of the proposed model have framed the problem space
from which the individuals will be guiding their behavioral response. Behavioral
adaptation is proposed to either lead to a qualitative change or a quantitative change in
behavior. A qualitative change is characterized by making an actual change in the
strategy or the structure of behavior (e. g., choosing another path/option to pursue). Once
an individual has decided to qualitatively adapt, there are two branches that he/ she could
pursue: 1) “one-shot” problem solving or 2) exploration (Fowler, 1965; Kaehlbling,
Littman & Moore, 1996). One-shot problem solving involves quickly deciding on and
exploiting an alternative behavior that currently appears to have the highest payoff. The
potential costs of this strategy are high if exploration would have revealed a better option.
However, if it results in the best course of action, the payoffs are maximized.
Exploration on the other hand involves trial and error learning, where an individual
searches his/her environment for possible alternatives until settling on what seems like
the best alternative. The effectiveness of this decision is largely based on the features of
the situation which will be described below.

A quantitative change involves persisting with the current strategy or behavior,
but increasing the effort put into the behavior (i.e., effort allocation change). Again, the
effectiveness of this choice will depend on the nature of the change. If the change is only

temporary, persisting through the change without adjusting strategies will be effective.

35

However, if the change requires a qualitative change, this choice will be ineffective (Staw
etaL,l981)

Person factors. In addition to the attributions made about the environment,
several individual difference variables are proposed to inﬂuence the behavioral response.
Some researchers (e.g., Dweck, 1986; Vandewalle, 1997) propose that there are three
dimensions of goal orientation: learning goal orientation (LGO), performance prove goal
orientation (PPGO) and performance avoid goal orientation (APGO). LGO has been
conceptualized and empirically related to more explorative learning strategies, such that
individuals high in LGO are less concerned with performance outcomes than with
learning the task and the environment (Payne, Youngcourt & Beaubien, 2007;
Vandewalle, 1997). Logically it follows then that individuals high in LGO will be more
likely to choose an exploratory approach to behaviorally adapt than either exploitation or
effort allocation. Conversely, PPGO and APGO are characterized as being focused on
performance outcomes and avoiding failure as opposed to learning the environment
(Vandewalle, 1997). These individuals would tend to avoid facing risk of failure in their
environment, and so will likely avoid the trial and error learning inherent in exploration,
where errors are likely to occur. Instead, individuals high in PPGO should be more likely
to exploit the option that currently appears to have the biggest payoffs (one-shot problem
solving), whereas individuals high in APGO will likely engage in either one-shot
problem solving (viewed as less risky; Garcia, Calantone & Levine, 2003) or effort
allocation (possibly reduced effort leading to withdraw).

Similar to LGO, openness is expected to be associated with choosing exploration

as a strategy approach. The conceptualization of openness in the Big 5 (see Costa &

36

McCrae, 1992) is congruent with exploration, as it refers to the tendency to be more open
and curious about new alternatives and new surroundings. On the other hand, both the
dependability facets of conscientiousness and the need for control are expected to be
more in line with effort allocation, or persistence (with increased effort) of the current
strategy. As reviewed before, both of these constructs are associated with the need to
impose order on one’s environment and tend to lead to methodical and rigid thinking and
behaving (LePine, Colquitt & Erez, 2000). Previous research examining the relationship
of the dependability facets with performance after a change in the environment have
found negative correlations, suggesting that the problem solving approach used is
ineffective (LePine, Colquitt & Erez, 2000). Although the actual strategies or processes
used by these individuals have not been empirically explored, it is expected that the
rigidness in cognitions and behaviors experienced by high dependability and need for
control individuals results in persisting with previously effective strategies when the
current conditions require a change (Staw etal., 1981). Similarly, another individual
difference expected to play a role in the behavioral response process is anxiety (or stress).
In a paper by Staw and colleagues (1981), anxiety and stress lead to restricted
information processing and increased constriction of control, which then leads to rigidity.
The rigidity experienced by these individuals is likely to reduce the likelihood of
exploration and increase effort allocation/persistence with previously learned (once
effective) information and strategies.

Process factors. In addition to relatively stable individual differences, process
variables are expected to impact decision making at this stage of the adaptation process as

well. Process variables refer to state-like reactions that are triggered by the task or

37

stimulus an individual is facing. Task-speciﬁc self-efﬁcacy has been well documented
(e.g., Kozlowski et al., 2001) in the learning and adaptation literature with higher self-
efﬁcacy predicting better adaptive performance. Although the role of affect on judgment
and decision making is not as well understood, there are a few paths by which affect is
proposed to impact behavioral choice (Maner & Gerand, 2007). Along with goals, Maner
and Gerand report that emotions (affect) inﬂuence judgment formation, risk-taking and
decision making under uncertain conditions (see also Weiss & Cropanzano, 1996).

Affect has been posited to serve as a signal of change (threat or opportunity) in one’s
environment which then leads to either approach or avoidance of the situation (see Maner
& Gerend, 2007). Although a clear relationship is lacking, positive affect is proposed to
relate to approach behaviors, while negative affect (e. g., anxiety) is suggested to relate to
avoidance behaviors (Manet & Gerend, 2007).

Taking these relationships a step further, negative affect should lead to more off-
task-focused attention and behaviors (avoidance) resulting in the behavioral response of
cognitive withdrawal, whereas more positive affect should result in non-withdrawal
choices (e. g., invest more effort either through exploration, exploitation or increased
effort allocation).

Situational factors. Environmental and performance factors will impact the
behavioral response stage as well, with certain situational features being more in line with
certain problem solving strategies than others. For example, the time constraints imposed
on individuals by the environment will likely inﬂuence the problem solving approach
they choose. If there is time pressure to make a decision and/or on receiving the

payback/reward, an individual will be less likely to choose exploration as a strategy and

38

will instead be more likely to engage in one-shot problem solving (exploit the currently
perceived best strategy; Garcia, Calantone & Levine, 2003). Similarly, feedback rates are
likely to inﬂuence problem solving choice. Exploration will not be very efﬁcient or
effective if feedback about the effectiveness of a decision is slow or absent. Exploration
is more applicable when the environment provides continuous (or at least rapid) feedback
that will allow individuals to move through the trial and error process. When feedback
about a choice is slow or absent, exploitation (or quick problem solving) will be more
likely. In addition, the costs associated with a wrong decision will likely impact decision
making. If choosing a wrong course of action results in high costs, individuals will take
more time to explore and consider all possible alternatives (exploration to gain support
for a decision), whereas, if the cost of a wrong choice will not weigh heavily on the
individual, one-shot problem solving may be used more frequently. Past performance (or
previous choice effectiveness) should also be related to decision making at this stage. If
previous performance using a strategy has met with high success, the individual is less
likely to switch from that strategy, but rather invest more effort into persisting with that
strategy (Siegler & Lemaire, 1997).
Evaluation

After a course of action has been determined, individuals will engage in the
behavior and then evaluate if the behavior was effective in the current environment
(Evaluation). A few individual differences will play a role in this stage. For example,
individuals may have different standards for what is “effective” while they are engaging
in this process, leading some to move out of the stage fairly quickly while others cycle

through to gain more conﬁdence in the effectiveness of their chosen course. For

39

example, in a paper on consumer decision-making, Mitchell and Walsh (2006) suggest
that perfectionists will take more time and explore more alternatives before buying an
item because they want it to be superior. So while evaluating a new course of action,
perfectionists may strive to ﬁnd the alternative that perfectly matches the current
conditions (even though this may not exist) resulting in long periods of evaluation before
deeming their choice a success.

Affective reactions may also play a role in evaluating the effectiveness of a
behavioral choice. For example, Carver’s (2004) self-regulatory process of action and
affect proposes that as individuals pursue goals they compare their perception of their
performance with their performance goal, resulting in affective reactions. If a behavioral
response results in a continued discrepancy between perceived state and goal state, the
individual will experience negative affective reactions, indicating that the choice of
action was not effective. However if a behavioral response reduces the discrepancy to
accepted levels, the individual will experience more neutral or positive affective
reactions, indicating that the choice of action was effective. That said, negative affective
cues (e.g., frustration, anxiety, dissatisfaction) signaling an ineffective behavioral choice
are proposed to cycle the individual back to the behavioral response stage, where another
behavioral choice will be made. If the individual cycles repeatedly between the response
and evaluation stages without success, negative affect may lead to cognitive withdrawal
as evidenced by off-task thoughts. From a resource allocation perspective (e. g., Kanfer et
al., 1994), as more resources are devoted to off-task thoughts, less are available for on-
task attention, suggesting a form of cognitive withdrawal. Successfully adapting affect in

a changing environment entails regulating the negative affect experienced such that it

40

does not lead to cognitive withdrawal, but rather acts as a signal that guides an individual

to try another course of action.

Dynamic Adaptation Model

Based on the description of the different stages above, the dynamics of this
process are outlined here and presented in Figure 2. The top part of the model (above the
dotted line) represents the exploitation process that occurs before an individual makes a
decision to adapt or change. Once an effective strategy has been learned for coping with
the demands of the environment, individuals will exploit the effective strategy, while
remaining vigilant of their environment and performance levels. As they are monitoring
their environment, they will receive feedback that will form both a perception of the
environment as well as their performance. Environmental input outside of the current task
environment may also impact environmental perceptions at this point. For example, the
global economy or new organizational competitors will inﬂuence an individual’s
environmental perceptions and these environmental perceptions will feed into the
performance perceptions. The perceptions formed will be compared to a standard (e. g.,
goal, norm, previous performance) and if the perceptions are within an acceptable range
of the standard, the individual will determine there is no change and continue exploiting
and monitoring. However, if a change is detected (i.e., perceptions are not within an
acceptable range of the standard), the individual will move onto the attribution and
decision process. This process will result in a decision to either ride the change out or to
make a behavioral change. If he decides to ride it out, he will cycle back to the
monitoring and evaluation stage and continue exploiting his current strategy. On the

other hand, if he decides to make a behavioral change he will proceed to the behavioral

41

response stage, where he will decide on one of three courses of action: one-shot problem
solving, exploration or effort allocation. After engaging in the behavioral change chosen,
the individual will evaluate the effectiveness of this choice. If deemed acceptable, the
individual will cycle back to the monitoring and exploiting stage, where he will exploit
the new strategy he has chosen while monitoring the environment for further disruptions.
However, if the choice is not effective, the individual will cycle through the behavioral
change choice and evaluation stages until a choice is found to be acceptable, at which
time he will cycle back to the monitoring and exploiting phase.

As reviewed in the previous sections, several individual differences and
situational factors will impact these stages along the way. Decisions can only be
determined “adaptive” at each stage if they match the current demands of the
environment. For example, if the disturbance in the environment is only a temporary
glitch, then the most adaptive decision process would be persisting with the current
strategy (if effective) and monitoring the environment. However this same course of
action will not be effective if the disturbance is stable and requires a behavioral change.
Therefore the effectiveness of any one decision at each stage is dependant on the
conditions of the environment. In addition, individual differences that guide people to
make different decisions may be beneﬁcial in some environments, while detrimental in
others. For example, individual differences leading to behavioral rigidity will lead to
better performance in stable environments (persisting through temporary ﬂuctuations),

while it could be maladaptive in changing situations (trying to persist when change is

needed).

42

 

Behavioral Adaptation
Process

 

9.589% 8: 098:0

 

 

 

8.38.? team

 

 

 

 

38233.
383206

 

 

 

32228.2 «SEA—Emacs E829: 8:: 3:95 .262

 

 

 

 

 

 

 

 

 

Exploitation Process

 

 

 

 

 

\./

330E eaten—gm

 

 

 

 

 

 

382m

 

 

.53qu
a. 8:332

 

38309 oweano

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

coups—nam—
mszom 528: .v.
m
C
h r r rrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrr :6 rrrrrr
3
m
h
C
vﬂwwﬁm 55320“. , cosh—020m
sew—39:00 gauge—om 2.258.55—

 

 

 

5o 3 02x

 

/—0@Dﬁoo.m\.
36890 0320 e Z Engagem—

=o_§_o_axm

wean
wctozeoz

 

 

 

the: SBSQSGV £5323
m 055

43

Specific Research Focus

The adaptation model presented above incorporates more variables and processes
than a single investigation could reasonably incorporate. Therefore, the current
investigation will focus on a subset of variables and processes that are thought to be
particularly impactful in the adaptation process. To narrow the scope of the project,
“effective” or “successful” adaptation is operationalized as (proximally) engaging in the
appropriate behaviors given the current state of the environment (i.e., exploitation during
pre-change blocks, exploration during initial post-change blocks), which (distally) will
lead to post-change performance that reaches pre-change levels. In addition, this study
incorporates two manipulations of the task environment which are proposed to lead to
differential relationships between individual difference variables and adaptation
behaviors and performance. The rationale for the two manipulations (change and
warning) chosen for this study is outlined below.

Manipulation Explanation and Rationale. The decision to manipulate change and
warning levels between individuals serves to address the gaps in the adaptation literature
discussed above. Speciﬁcally, the manipulations will allow this study to contribute to the
research in the following ways: 1) teasing apart detection of a change and adjustment to
that change, 2) examining between person adaptation behaviors over time, and 3)
incorporating process related measures to understand why outcomes vary by individuals.

To test these aspects of the model and contribute to the literature, the change and
warning manipulations will be used to create different environments that tap different
parts of the model. First, an environment that allows for a normal, uninterrupted
assessment of learning and performance needs to be created to serve as a control group.

Change and warning manipulations will also be used to create environments needed to

44

test the attribution process, speciﬁcally by providing warnings of a change to individuals
when a change has not actually occurred, requiring individuals to make the decision to
ride it out. Lastly, varying warning levels within changed environments will help
evaluate the distinction between detection and behavioral response. Exploring the impact
of these manipulations over time contributes to the existing literature by allowing for the
examination of differences in behaviors and performance across different environmental
conditions over time.

Individual Differences. Several individual differences are proposed to be relevant
to the current study. Conscientiousness (speciﬁcally the dependability facets of
conscientiousness), openness to experience, trait anxiety, need for control and goal
orientation are the trait individual differences focused on in this study. Previous research
has highlighted the importance of these traits in adaptation, and the impact of these traits
is hypothesized to differentially relate to adaptive performance across conditions.

Process Variables. The process variables that are the focus for the current study
include boredom, frustration, anxiety, satisfaction with performance, goal commitment,
self-efﬁcacy, threat and challenge appraisals and off-task thoughts. Each of these

variables is proposed to inﬂuence the adaptation process in the current study.

Study Hypotheses

The speciﬁc hypotheses and the rationale for the hypotheses are listed below. For
the following hypotheses to be explored, the change manipulation must be successful. If
successful, the change manipulation should result in the following three outcomes. First,
the change manipulation must result in performance differences pre- and post-change.

Second, the change must result in different patterns of behaviors (pre- vs. post-change),

45

highlighted by exploitation before a change and exploration (e. g., switching behaviors)
after a change. Third, the change must result in different response times, with response
time decreasing before a change (indicating that a behavioral response pattern has been
learned) and then increase after a change (indicating that a behavioral change needs to
occur). These three conditions will indicate that the change manipulation was successful.
For all of the hypotheses, the “change” is an abrupt change presented which alters the
underlying structural rules for governing performance on a task.

The study hypotheses proposed below are each targeted to address at least one of
the intended contributions of the current study. Speciﬁcally, all of the study hypotheses
help fulﬁll the contribution of examining the adaptation process over time, by
incorporating a time component in each of the relationships proposed. In addition,
hypotheses 2 through 8 target key individual differences proposed in the literature to test
the impact of these trait variables on different adaptation outcomes (e. g., serve to both
replicate and extend previous research). Hypotheses 8 through 14 contribute to the
literature by examining a breadth of theoretically relevant process variables, to
understand how individuals affectively react to a change over time. The intended
contribution of exploring the role of detection in the adaptation process is targeted by
hypotheses 9 and 10, by examining the impact of different combinations of change and
warning (i.e., detection aids). A summary of the hypotheses will be presented at the end
of the results section.

General Performance Hypothesis

46

H1: Higher performance levels during periods of stability will lead to increased switching
behaviors after a change, representing a switch to exploration behaviors due to change
detection.
Rationale. Individuals who are performing well experience few errors, so when a
change occurs that forces errors to occur (initially), well performing individuals will
notice the performance decrement faster than individuals who are lower performers
and begin exploring their environment.
Individual Diﬂerence Hypotheses
H2: The dependability facets of conscientiousness will be positively related to
performance in a period of stability, but negatively related to performance after an abrupt
change.
Rationale. It is well documented that during periods of stability, conscientiousness
facets are positively related to performance (e. g., Barrick & Mount, 1 991). Recently
however, research has found that the dependability facets are negatively related to
performance in changed environments (e. g., LePine et al., 2000).
H3: The dependability facets of conscientiousness will be positively related to
convergence (exploitation behaviors) after a change and negatively related to divergence
(exploratory behaviors) after a change.
Rationale. To impose order on their environment as quickly as possible, individuals
high on the dependability facets will likely narrow in on a strategy quickly and reduce
exploration (which would lead to more errors).
H4: Openness to experience will be positively related to performance, with this

relationship being stronger in after a change versus in periods of stability.

47

Rationale. Previous research on openness and performance has found small positive
relationships with performance in a stable environment, with larger positive
relationships being found in changed environments (e. g., LePine et al., 2000).
H5: Openness to experience will be positively related to divergence (exploratory
activities) after a change and negatively (or un-) related to convergence (exploitation
activities) after a change occurs in a task environment.
Rationale. Openness by its very deﬁnition is associated with greater exploration and
curiosity (Costa & McCrae, 1992).
H6: Learning goal orientation will be positively related to divergence (exploratory
behaviors) after a change occurs in a task environment.
Rationale. Learning goal orientation is associated with the desire to learn one 's
environment. Exploratory behaviors are one way by which individuals can learn
their new environment after things have changed (Vandewalle, I 99 7).
H7: Prove and avoid performance goal orientations will be negatively related to
divergence (exploratory behaviors) after a change occurs in a task environment.
Rationale. Prove and avoid performance goal orientation are associated with the
desire to prove ability or avoid failure in achievement situations as opposed to
learning or mastery (Vandewalle, 1997). Exploration (learning through trial and
error) will likely lead to errors initially, so would be avoided by performance
oriented individuals.
H8: Avoid performance goal orientation will be positively related to anxiety and threat

appraisals after a change occurs in a task environment.

48

Rationale. Avoid performance goal orientation is associated with fear of failure
(Vandewalle, I 99 7). An unexpected change in the environment will lead inevitably to
initial performance decrements (failure), which should increase these individuals’
anxiety levels and perceived threat levels.
Situation-specific Hypotheses
H9: Relative to any other possible combination of warning and change, when a change
occurs in a task environment and a warning is not presented beforehand, individuals will
experience greater levels of frustration and anxiety, and lower levels of self-efﬁcacy and
satisfaction with performance.
Rationale. When change occurs unexpectedly (without warning) individuals will
likely experience more negative reactions as a result of unpredicted performance
decrements, whereas, uncertainty or unpredictability of the environment will
decrease if an individual is warned of the change or if no change occurs. Warnings
provide detection aids that remove the demand on the individual to detect the change,
instead allowing them to focus resources on adjusting to the change.
H10: Level of change (change versus no change) will moderate the impact of warning
level on performance after the warning, such that when a change actually occurs,
performance will be highest when a speciﬁc warning is given immediately before the
change occurs and lowest when no warning is given about the change, whereas when a
change does not occur, performance will be highest when a warning is not given about a
change and lowest when a speciﬁc warning is given.
Rationale. Increased specificity of warnings are argued to be helpful when a situation

is actually changing, allowing for increased detection speed and additional resources

49

to devote to adjusting to the change. However, if a false warning is presented, in that
there is actually not a change, individuals will likely invest resources in trying to
ﬁgure out what changed as opposed to investing resources into exploiting the strategy
that is working.

Process Hypotheses

H11: Boredom levels will increase over time when performing a particular task, with

relatively higher levels of boredom occurring when the task remains stable than when a

change occurs in the task environment.
Rationale. After performing the same task repeatedly, novelty is expected to wear off
and boredom to set in. However when a change is introduced into the environment.
novelty will once again increase, decreasing boredom levels. This is suggested by
Kanfer and Ackerman ’s (1996) motivational control component of regulatory skills,
which proposes that motivational control is a skill that can keep attention on task
despite boredom, which is especially important in situations of low attentional
demand When tasks remain stable over long periods of time and thus become
relatively automatic, the attentional demand will drop and boredom will increase.
Similarly, Csikszentmihalyi (1990) proposed that boredom occurs when an individual
has the skills to perform a more challenging task than he is actually involved in.
Automaticity reduces challenge, while change (novelty) can increase challenge.

H12: Satisfaction with performance will decrease after a change occurs in a particular

task environment, while satisfaction with performance will increase over time during

periods of stability.

50

Rationale. As performance decrements will likely occur following a change,
satisfaction with performance should decrease. However, as behavior becomes well-
learned during stable periods, performance will likely improve, thus leading to
greater satisfaction with performance.
H13: Self-efﬁcacy levels will decrease after a change occurs in the task environment
relative to both within person previous levels and between person levels during periods of
stability.
Rationale. State self-efficacy is proposed to decline aﬁer failure or poor performance,
as the belief that one is capable of performing the task successfully will falter when
receiving negative feedback (Judge, Jackson, Shaw, Scott & Rich, 200 7).
H14: Off-task thoughts and affective (negative) thoughts will increase after a change
occurs in the task environment relative to both within person previous levels and between
person levels during periods of stability.
Rationale. In the face of unpredictable discrepancies in performance, oﬂ-task
thoughts can be used to cognitively withdraw from the task to reduce negative affect

(Kanfer et al., I 994).

51

METHOD

Participants

A total of 262 undergraduate students from a large Midwestern university
participated in this study. Individuals received course credit for participation. Data from
six individuals was discarded due to random responding during the lab session. The ﬁnal
sample used for analyses consisted of 256 individuals. The ﬁnal numbers by condition
are: Condition 1 (n = 32), Condition 2 (n = 35), Condition 3 (n = 35), Condition 4 (n =
35), Condition 5 (n = 32), Condition 6 (n = 27), Condition 7 (n = 31), and Condition 8 (n
= 29). Ninety-eight percent of the sample was between the ages of 18 and 22, with the
full range spanning from 18 to 27. Twenty-eight percent of the sample was male (n =
72), and 83% of the sample was Caucasian. Freshman comprised almost half of the
sample (49%), with 26% of the sample comprised of sophomores, 17% juniors and 8%

seniors.

Computerized Bandit Task

The computerized bandit task used in this study is a probability decision-making
task, similar to multiple cue probability leaming (MCPL) tasks that have been used in the
adaptation literature in the past (e.g., LePine etal., 2000). For the present study, the task
is framed as a human resources decision-making task, where an individual has four
potential agencies from which to attempt to hire a new employee. Each agency
(represented by a button labeled: Agency 1, Agency 2, Agency 3 and Agency 4) has an

underlying probability of hiring success, which is unknown to the individual (i.e., the

52

participant) making the hiring decision. A hiring attempt is made by clicking on one of
the agency buttons. If the decision leads to a successful hire, the participant’s score
increases by 1, whereas if it leads to an unsuccessﬁil hire, the participant’s score does not
change. The task provides immediate feedback in terms of the number of hiring attempts
made during the current block, the number of successful hires made, and the overall goal
the individual is trying to reach for the current block (30 successful hires out of 50 hiring
attempts). The probabilities for the task were manipulated such that all agencies did not

have equal probabilities for hiring success.

Rationale for Bandit Task

The Bandit task, which is a probability task, is appropriate for the current study as
it allows for detection of behavioral change over time to manipulated conditions. Output
from the task includes the decision made for each trial (i.e., which agency did they
choose), success/failure for each trial, and response time between decisions. Having
these behavioral indicators of adaptation is important to the present study and to
understanding the behavioral adaptation process. The simplicity of the task allows for
initial learning to occur after relatively few blocks of trials. This is important for the
present study because if an individual is still learning their environment when a change is
introduced into the environment, the individuals will likely not be able to detect it (too
much random noise). The simplicity of this task also allows for increased experimental
control over the features of the task, so interpreting what is causing changes in an
individual’s behavior is easier. This task has been piloted in other studies in various
forms and has been found to be an adequate representation of a simplistic adaptation

environment.

53

Experimental Design

A 4 (warning: not told vs. told-general beginning vs. told-general after block 5 vs.
told-speciﬁc after block 9) X 2 (change: change vs. no change) between subjects design

with repeated measures over 18 blocks was used in this study (see Table 1).

Experimental Conditions

The change manipulation is operationalized as the presence or absence of an
abrupt change in the structural rules (e. g., probabilities of success associated with each
agency choice) occurring midway through the task. The warning manipulation varies the
level of warning about a change presented to individuals. Speciﬁcally, the warning
ranges in both timing of warning and speciﬁcity of warning, from no warning to told-
general beginning to told-general post-block 5 to told-speciﬁc post-block 9 (immediate
before the change).

The manipulations result in eight conditions, with each condition serving a
purpose linked to the theory proposed at the beginning of this paper. These conditions
are explained in greater detail below as well as in Table 2.

Condition 1: Control group. Condition 1 is the control group for the current
study. Participants in this condition performed the Bandit task without any
environmental manipulations. All 18 blocks of trials were performed with the same
underlying probability distribution for success. In addition, the participants were only
given general task instructions, with no mention of a potential change. See additional

details in Table 2.

54

Condition 2: No change/told-general beginning group. The participants in
condition 2 also performed all 18 blocks of trials with the same underlying probability

distribution for success. However, in this condition, in addition to the general task

55

 

 

 

 

«.83 £8 «.83 28
«.83 58 8228 «882 2.2 828 «882 2.2 842820 828 «882 2.2 8829 «.83 58
8.820 ..8 «83 828 um. «.83 828 88828 .. 2 «.83 8882 83828 .228 «882 2.2 8.829 89829
8.828 8 M22282 28.28282 8 32282 «8.28288. 8 M22283 ..8M2828 8 M22283
822.82% “as 8.2.889» 888% “2 8.2.2889» «888% .8 82.2889» 82 Km. 8.2.2889
«882 2.. 88828 82 «82 2.2 83828 82
«.82 2.2 832828 .6 «.822 828 882828 .. 2 «.83 8882 8888 «9.82 2.2 8.88 898 8
82 8888 8 M22282 «8.28282 8 $2283 «8.28282 8 32283 82 89828 8 M22288. .5 2
8,2282% .8 8.2.889» 888% 5. 8.2.2880» «888% Km 8.2.2880» 82 .3 8.2.2880»
s «28 m «.88 88.2.8888
.228 emboumursﬁ 828 288286.385 8.82.8885 382 282

 

 

 

 

 

 

8.2.2880 «828888.2m >8 882882.282 M22282 28 8882 .O 8 8.22.2935

_ 838,—.

56

 

3:53.: :5... on .o .8... a.
.82: ow::..8 8.2.32.8... 3.2....
.95.... m: 82.89.... 2.3m.

39:2: 2:... a .o .8... a.
.82: ow::..o $2.338... 3.2....
.95.... a: 2.2.3.... 2.5m.

3.52.: 25.... 8 .c .8... a.
.82: 8w::..8 8.2.32.8... 3.2...
.95.... a: 82.89.... 2:29

3.52.. :5... o. .o .83 a.
.82: ow::..o 8.2.32.8... 3.2....
.95.... 2. 2.2.3.... 2:89

82.82...

 

m288... ..: .8. 28:...
2.. .3... 3.2.32.8... 3.2....
:...ow::..o
38: o>:.. 22.. 2.. .8 282...:
2:22... 5. 288... 2.. 9.8.2.
2.»... ”5.52:... 2.2.88.2:
...:8.mm3.
32.2.2.5 2.28. w:...:.. 28.
2... .8 9.88... .885. 22.8.9288
2. ...3 :8> .28. 2... .8 2883
. 2:88.22 22:82: w:...... on
.8 .8... 28m. 282.82: 2...... cm
.8 .:8 29:82: w:...3 5.88:...
on 82:... 8. 2 .:8w 58> .88..
2... 8. 2... 2:83 :8» 8:83.
2833 .:8..: 8.288.. .58..
2:88.22 .:... :82:.. 28:89..
2.. :8 22.8 2.282: 2......

: 22:. 8 .. .68.. .8282: w:.:3 :
2.:2. 8. 8... 2:83 :8» 38:89..
3833 .:8..: 38.28.. 9.2:...
3 =5 :9. :3. :5 a...
3...... 92......“ 9.82. 82.89....

8.883 ..: .8. 2:8
2.. 2:... 8.2.338... 3.2....
...ow::..8 .3... v.22
2.. .8 2888...: 2:288 .23. 8...
288.... 88.: :8» m: 22.8223.
8 .8... 2: 2&2. 2....
u:...:2...: 2.2.89.5 28.3.8
382.2898 2.3.8. w:...:.. 28:.
2... .8 3.883 .885. 828.8888
2. ...3 :8> .23. 2... .8 288...
. 2:88.22 22:52: w:.:3 cm
.8 .8 28:”. 2922.: w:...... on
.8 .:8 28.282: w:...... .3238...
cm 2.2: 8. 2 .:8w 28> .58...
8.... 8. 2... 2:83 :9. 38:8»...
2833 .:8..: 8.288.. .58..
2:88.82 .22 82:3 28:09..
2.. :8 28.... 8:32: w:......

: 3.2: 8 ... 2.8.. .9282: w:...... :
22:. 8. 8... 2:83 :8» 28:83.
2833 .:8..: 88.282. w:...::.
2. =3 :9. .28. :5 s...
3.8:. 22...... 9.8.2. 82.89.:—

283 =: .8. 8.2::
2.. 2:... 8.2.32.8... 3.2....

...ow::..8 2:... v.8. o... .8 282...:
2:28 .22.. 8... 288.... 38.:
:8» m: 22.2.2282 28.38...
382.289... 2.3.8. w:...:.. 28.
2... .8 8.883 .238. w:..o...:.88
2. =.3 :8> .22.. 2....8 288...
. 2:88.52 222.82: 2...... on
.8 .8... 28.... 22:52: 2...... cm
.8 .:8 28:32: 9...... .3233.
an 8.2: 8. 2 .:8w 58> .88...
2... 8. 8... 2:83 :8» 28:83.
2833 .:8..: 8.282. .58..
2:823. .:... :82:.. zu:ow<
2.. :8 22.8 2.2.2.: 2......

: 8.2: 8 .. .88.. 22.2.: w:...... :
2...... 8. 8... 2:83 :8» 38:83.
2833 .:8..: 2.8.28... 2.2:...
2. =3 .5. 8.8. :5 s...
3.2.. ”2...... 9.8.2. 28288.5

8.88... ..: .8. 2:3
2.. b... 8.2.32.8... 3.2...

.3288. 38:25.28
£39 2.5.. .8. 2...... 9.83
.238 w:..o_...:88 8.. ...3 :8 >
8.2.. 2... .8 2883 . 2:88.82
2:52.: w:...... on .8 .8... 32......
292.82: 2...... cm .8 .8 885
.8 .:8 29:82: 2...... .3288...
cm 82:... 8. 2 .:8w 58>
.58.. 2... 8. 2... 2:83 :83
mo:ow< ..o...3 .:8..: 8.282.
58.. 2:88.22 2:... :82:..
28:83. 8... :8 28.... 29:82:
w:...... : 2.2: 8 .. .88.. .8882:
w:..... : 22:. 8. 2... 2:83 :83
28:83. 3833 .:8..: 2.8.282.
2%: 3 =5 .5. .85
2... .8... 3.2.. 3......“ 9.8.2.
28:89.5 28.2.88: .8280.

92829
82

 

 

. «8.8 8.2. .2834...

 

u. «.83 8.8 3.82.5835

 

2.22.2282 8.828%.38h

 

8.8. .82

 

 

 

8.2.889 «8.285.822... A3 2.22%. «8.. 88 228282222. 8.28%.

N 83:...

57

instructions given to control participants, the no change/told-general participants were
told at the beginning of the task that a change may occur at some point during the task
(see Table 2 for more details). The presence of a general waming in an environment
without change increases the ambiguity in the environment, as is consistent with a real
world environment.

Condition 3: No change/told-general after block 5 group. This condition mirrors
condition 2, with the exception that the general warning was presented aﬁer block 5 as
opposed to prior to the task starting, which allowed for initial learning to take place. The
underlying success probabilities did not change as in conditions 1 and 2 (see Table 2 for
more details). Similar to condition 2, the warning provided will increase the ambiguity in
the environment.

Condition 4: No change/told-specific after block 9 group. Similar to conditions 1,
2, and 3, participants did not incur a change in the underlying success probabilities across
blocks of trials. The general task instructions were also identical to those presented to
participants in the previous conditions, with the exception that the no change/told-speciﬁc
participants were told that a change had occurred after the 9th block of trials (see Table 2).
This condition presents a false warning, representing environments where false alarms or
warnings of irrelevant changes need to be evaluated and the decision to continue with the
current strategy needs to be made.

Condition 5: Change/not told group. In condition 5, participants were only given
the general task instructions given to the control participants in condition 1. However,
unlike the control condition, the participants actually experienced a change in the

underlying task probabilities after the 9"1 block of trials, such that the previously most

58

effective agency (i.e., highest probability of success) was no longer the optimal choice
(see Table 2). Participants in this condition had the greatest detection burden because
they are given no warning that a change may occur.

Condition 6: Change/told-general beginning group. In condition 6, participants
were given the general task instructions given to all other conditions, but were also told at
the beginning of the task that a change may occur at some point during the task (like
condition 2). However, unlike condition 2, the participant actually experienced a change
in the underlying probabilities (like condition 5). The general warning in this condition
should help individuals detect changes by making the possibility of a change salient at the
beginning of the task.

Condition 7: Change/told-general after block 5 group. This condition mirrors
condition 6, with the exception that the general warning was presented after block 5 (like
condition 3) as opposed to prior to the task starting, which allowed for initial learning to
take place. However, unlike condition 3, the participant actually experienced a change in
the underlying probabilities (like condition 5 and 6). The general warning in this
condition should help individuals detect changes by making the possibility of a change
salient after initial learning has taken place.

Condition 8: Change/told-speci/ic after block 9 group. In the last condition,
participants experienced a change in their environment after the 9th block of trials (like
conditions 5-7), and additionally, were given a speciﬁc warning informing them of a
change after the 9‘h block. Participants in this condition did not have to detect the change

because they were told that the task had changed. Therefore, participants in this

59

condition were expected to differ based on their ability to problem solve about how to
cope with the change.

Pilot Testing. To test whether the manipulations proposed above are effective
(i.e., result in differences across conditions), a pilot test using the change and warning
manipulations was conducted before data collection began for the current study. The
same manipulation check questions were administered to the pilot study participants that
were administered to the actual study participants. Results from this pilot study are

presented at the beginning of the study results section.

Procedure
The general details of the procedure for the current study are outlined in Table 3.

A more detailed description of the procedure is provided below.

 

 

Table 3
Procedural Steps

Order of . .
Procedural Steps Description of Steps
1. Online 0 The subjects gave their initial consent online and completed
Consent & Trait the background and trait measures (demographics,
Survey conscientiousness facets, openness facets, trait anxiety, goal

orientation, and need for control) approximately one week
prior to the lab session.

 

2. Randomization 0 Prior to the in-person lab experiment, participants who signed
up for the experiment and completed the online trait survey
were randomly assigned to a condition group. Because of the
nature of the instruction manipulation, each experimental
session consisted of only one condition.

 

 

3. In-Person Lab 9 All participants were seated at a computer in the lab at the

Experiment start of the session.

0 All the subjects were given the same brief description about
the Bandit task at the beginning of the session (not condition
speciﬁc).

0 For Conditions 1 and 5, participants were given general
instructions on how to begin the Bandit task.

 

 

 

6O

Table 3 (cont’d).

 

 

o For Conditions 2 and 6, participants were given the general
task instructions as well as the general warning informing
them of the potential of a change.

0 For Conditions 3, 4, 7, and 8, participants were given only the

Jeneral task instructions at this point.

 

Bandit Task and Process Measures

 

 

 

 

 

 

 

4. Bandit Task 0 Participants alternated between completing a block of 50
Blocks [-9 and trials (i.e., 50 hiring attempts) and completing a set of process
Process Measures measures (current boredom, current frustration, current
anxiety, satisfaction with performance, current self-efﬁcacy,
goal commitment, personal goal, threat/challenge appraisals
and off-task thoughts).
0 Underlying success probabilities for all conditions: Agency 1
= .3; Agency 2 = .3; Agency 3 = .3; Agency 4 = .6 (choosing
Agency 4 is optimal choice).
0 Both the task and the process measures are completed on the
computer.
5. General 0 For Conditions 3 and 7: after the 5th block of trials and
Warning after process measures, participants were given a general warning
Block 5 that a change may occur in their environment.
6. Speciﬁc 0 For Conditions 4 and 8: after the 9th block of trials and
Warning after process measures, participants were given a speciﬁc warning
Block 9 that a change has occurred in their environment.
7. Bandit Task 0 For Conditions 5, 6, 7 and 8: beginning with Block 10, the
Blocks 10—18 and underlying success probabilities will be changed. Underlying
Process Measures success probabilities: Agency 11 = .3; Agency 2 = .6; Agency
3 = .3; Agency 4 = .3 (choosing Agency 2 is now the optimal
choice).
0 For Conditions 1, 2 3 and 4: probabilities stay the same as for
Blocks 1-9. Underlying success probabilities: Agency 1 = .3;
Agency 2 = .3; Agency 3 = .3; Agency 4 = .6 (choosing
Agency 4 is still the optimal choice).
0 Participants continued alternating between completing a
block of trials and a set ofjrocess measures.
8. Manipulation o All participants were given a short manipulation check after
Check the completion of all blocks and process measures which
assessed their knowledge of the task hypotheses as well as the
presence/absence of a change in the task.
9. Debrieﬁng 0 Subjects were debriefed.

 

 

61

 

Initial Recruitment. Participants were recruited voluntarily through the
Department of Psychology subject pool at Michigan State University. Upon signing up
for the experiment, participants were administered an online consent form (see Appendix
A) that outlined the details of the experiment and their rights as participants. Only upon
acceptance of the terms of the informed consent were participants linked to the ﬁrst part
of the experiment.

Online Trait Survey. Immediately upon accepting the terms of the informed
consent, participants were linked to the online trait survey battery (see Appendix B). The
online battery included a background questionnaire which asked participants for
information about their gender, ethnicity, age, year in school, and GPA/SAT/ACT scores.
After completing the background questionnaire, participants responded to a series of
personality and individual difference measures online, including trait anxiety,
conscientiousness facets, openness facets, need for control and goal orientation. The
traits measured at this point are proposed to have theoretical relationships with
performance on the Bandit task. This survey was completed by participants
approximately one week prior to their scheduled lab session. The time lag between the
trait survey administration and the lab session was important to reduce the likelihood that
trait responses inﬂuence how the individual perceived or reacted to the task.

Randomization. Upon completion of the online trait survey, the lab sessions that
individuals signed up for were randomly assigned to different conditions. Sessions rather
than participants were randomly assigned because the verbal instructions presented in the

conditions differ, thus making it necessary to keep all participants within a session in one

62

condition. The randomization method used was an online random number generator,
with the condition set that an equal number of groups per condition were to be generated.

Bandit Task Instructions. As participants arrived to the lab session, they were
seated at individual computers around the lab. Verbal instructions were given to
participants informing them about the task, as well as introducing the speciﬁc context that
the task will be set in (see Appendix C). This presentation was consistent across
conditions. Once the general instructions were given, individuals in conditions 2 and 6
were presented with the general warning that the task may change during the session. All
individuals were then told to begin the Bandit task by clicking on the appropriate icon on
the computer screen.

Alternation of Bandit Task Blocks and Process Measures. Each block of the
Bandit task consisted of 50 trials (or hiring attempts). The participant moved through
each block of trials by making decisions about which agency to make a hiring attempt
from. Each click (hiring attempt) represented one trial. At the end of each block of 50
trials, participants received feedback on the computer screen about their performance
(e. g., “You scored 30 out of 50.”). After receiving the feedback, participants moved to a
series of process questions, which were also administered on the computer. The process
measure included items tapping boredom, frustration, anxiety, satisfaction with
performance, personal goal, goal commitment, self-efﬁcacy threat/challenge appraisals
and off-task thoughts (see Appendix D). When all individuals in the group were ﬁnished
with the set of questions, the participants were moved on to the next block of trials. This
alternating process continued for 18 iterations, for a total of 18 Bandit task blocks of 50

trials and 18 administrations of the process measures. Process measures were assessed

63

20'

aﬁer every block to assess change over time within individuals, both as indicators of
change detection (e.g., does frustration increase after an unexpected change occurs?) and
change in reactions to the task.

Told-General after block 5 Warning Instructions. For conditions 3 and 7,
participants were interrupted after the 5th block of the Bandit task and completion of the
5‘h set of process measures. The experimenter verbally informed the participants in these
conditions that certain aspects of the task may change over the course of the task. After
this verbal interruption, participants were moved onto the next block of the Bandit task
(refer back to Table 2 for speciﬁc wording).

Told-Specific Warning Instructions. For conditions 4 and 8, participants were
interrupted after the 9th block of the Bandit task and completion of the 9th set of process
measures. The experimenter verbally informed the participants in these conditions that
certain aspects of the task have now changed. After this verbal interruption, participants
were moved onto the next block of the Bandit task (refer back to Table 2 for speciﬁc
wording).

Manipulation Check. After participants completed all Bandit task blocks and
process measures, a manipulation check was administered on paper to assess individual’s
knowledge of the hypotheses as well as their insights about what, if anything, changed in
the task (see Appendix E).

Debrieﬁng. At the end of the experiment, participants were given a paper
debrieﬁng form that provided information regarding the purpose of the study and contact

information (see Appendix F )-

64

Measures
Online Trait Measures

Background Information and Cognitive Ability. The ﬁrst survey that individuals
completed was a background information survey. The background information survey
consisted of eight self-report questions assessing various demographic characteristics,
including gender, age, ethnicity, year in school, GPA, SAT and ACT scores. The SAT
and ACT scores serve as a measure of general cognitive ability, as cognitive ability is a
strong predictor of task performance (Hunter & Hunter, 1984).

Trait Anxiety. Trait anxiety was assessed with the 10-item IPIP Neuroticism:
Anxiety facet (http://ipip.ori.org/). The items on this facet are expected to tap general,
trait levels of anxiety. Participants rated these items on a 5-point Likert-type scale
ranging from 1 = Strongly Disagree to 5 = Strongly Agree, indicating how well each item
represents how they generally are. Half of these items were worded in the positive
direction (i.e., higher agreement equals more anxiety; e.g., “Worry about things”), and
half of the items were worded in the negative direction (i.e., higher agreement equals less
anxiety; e. g., “Am not easily bothered by things”). The latter items were reverse-scored,
such that higher scores represent higher levels of trait anxiety. The scale reliability was
good for this sample (a = .87).

Conscientiousness Facets. Conscientiousness was assessed with the six IPIP
facets of conscientiousness (http://ipip.ori.org/). The facets are: C1: Self-efﬁcacy (belief
in competence to perform or handle tasks; e.g., “Come up with good solutions”; a = .78),
C2: Orderliness (tendency to be organized and structured; e.g., “Love order and

regularity”; a = .86), C3: Dutifulness (tendency to be rule-abiding; e. g., “Try to follow

65

the rules”; a = .78), C4: Achievement-striving (tendency to be hard-working and set high
standards; e.g., “Turn plans into actions”; a = .82), C5: Self-discipline (tendency to self-
driven and reliable; e. g., “Am always prepared”; = .88); and C6:
Cautiousness/Deliberation (tendency to avoid impulsivity and act in a rigid, error-
avoidant way; e. g., “Stick to my chosen path”; a = .82). Each facet was assessed by 10
items, which participants rated on a 5-point Likert-type scale ranging from 1 = Strongly
Disagree to 5 = Strongly Agree, indicating how well each item represents how they
generally are. Higher scores represent higher levels of each facet. Three facets
(orderliness, dutifulness, and cautiousness) have been posited to represent a higher order
facet of dependability (e. g., LePine et al., 2000). Scores for these three facets were
aggregated to form a composite score for dependability (a = .90). Similarly, the other
three facets (self-efﬁcacy, achievement striving and self-discipline) have been proposed
to comprise a higher order achievement facet. The scores from these facets were
combined to form a composite achievement score (a = .92). The overall
conscientiousness scale comprised of all six facets also had a good scale reliability ((1 =
.94).

Openness Facets. Openness to experience was assessed with the six IPIP facets of
openness (http://ipip.ori.org/). The facets are: 01: Imagination (e.g., “Have a vivid
imagination”; o. = .85), O2: Artistic Interests (e. g., “See beauty in things that others might
not notice”; a = .85), O3: Emotionality (e.g., “Experience my emotions intensely”; a =
.85), O4: Adventurousness (e.g., “Prefer variety to routine”; or = .78), OS: Intellect (e.g.,
“Can handle a lot of information”; a = .84); and O6: Liberalism (e.g., “Believe that there

is no absolute right or wrong”; a = .77). Each facet was assessed by 10 items, which

66

participants rated on a 5-point Likert-type scale ranging from 1 = Strongly Disagree to 5
= Strongly Agree, indicating how well each item represents how they generally are.
Higher scores represent higher levels of each facet. The overall openness scale
comprised of all six facets also had a good scale reliability ((1 = .93).

Need for control. Need for control was assessed using a self-constructed, 4-item
scale. The literature does not contain a measure for need for control that I felt
represented the construct that I was trying to assess. An example item from the
constructed scale is “When I face situations that I do not have control over, I try my best
to regain control.” The four items were rated on a 5-point Likert-type scale ranging from
1 = Strongly Disagree to S = Strongly Agree, indicating how well each item represents
how they generally are. Higher scores represent higher need for control. The rationale
for including a need for control scale is to validate whether or not the aggregated
dependability facet of conscientiousness is positively correlated with need for control as
suggested in previous literature (e. g., LePine et al., 2000). The overall need for control
scale had an acceptable scale reliability (or = .73).

Goal orientation. Trait goal orientation was assessed using Vandewalle’s (1997)
measure of goal orientation, which is comprised of 13 items measuring three dimensions:
learning goal orientation (LGO), prove performance goal orientation (PPGO) and avoid
performance goal orientation (APGO). LGO refers to the desire to develop oneself by
gaining mastery over the environment (5 items; a = .82; e.g., “I often look for
opportunities to develop new skills and knowledge”). PPGO refers to the desire to prove
one’s competence in performance situations (4 items; a = .75; e.g., “I’m concerned with

showing that I can perform better than others”). APGO is the desire to avoid failure in a

67

performance context (4 items; a = .83; e.g., “I prefer to avoid situations where I might
perform poorly”). Individuals were asked to rate each item on a S-point Likert-type scale
ranging from 1 = Strongly Disagree to 5 = Strongly Agree. Higher scores on each

dimension represent higher levels of that dimension.

Process Measures

The process measure consists of items that were mainly self-constructed, single-
item measures of various state variables. The rationale for using single-item ratings for
each variable was to keep the process measure short enough that it could be assessed after
all 18 blocks of trials. The single-item scales included one item for boredom, one item
for frustration, one item for anxiety and one item for satisfaction with performance (see
Appendix D for speciﬁc items). All of these items were given referents of “how the
individuals felt during the previous block of 50 trials”. All of these items were rated on a
5-point Likert-type scale ranging from 1 = Strongly Disagree to 5 = Strongly Agree.

Although individuals were provided with an assigned goal of 30 hiring successes
out of 50 hiring attempts for each block, an item assessing individuals’ personal goals
was included in the process measure to control for or at least evaluate what goal the
individuals were actually working toward. In addition, goal commitment (to the assigned
goal) was assessed using a single item adapted from Kanfer et al. (1994), rated on a 5-
point scale from 1 = Strongly Disagree to 5 = Strongly Agree.

Self-eﬂicacy was measured with two items assessing both the level of conﬁdence
in and perceived likelihood of reaching the assigned goal during the next set of 50 trials.
Ratings were made on S-point scales ranging from 1 = not at all conﬁdent (not at all

likely) to 5 = extremely conﬁdent (extremely likely). The correlation between these two

68

items was high (r = .89), justifying the aggregation of these items into a single state self-
efﬁcacy measure.

Threat and challenge appraisals were assessed using two questions, tapping an
individual’s perception of whether the task is threatening or challenging. The questions
were based on Lazarus and F olkman’s (1984) model of primary and secondary stress
appraisals. The ﬁrst question assessed primary appraisal by identifying the potential
demands or stress expected from the upcoming task (e.g., “How demanding do you
expect the next set of hiring attempts to be?”). Secondary appraisal, or perceived coping
ability, was assessed with the second question (e.g., “How able are you to cope with the
demands of the next set of hiring attempts?”). The items were rated on a 5-point Likert-
type scale ranging from 1 = Not at all demanding (not at all able to cope) to 5 =
Extremely demanding (Extremely able to cope). Threat and challenge appraisals were
derived from the ratio of perceived demands to perceived coping. The lower the ratio,
the more challenged an individual is, whereas the higher the ratio, the more threatened an
individual is.

Lastly, off-task thoughts and negative (affective) thoughts were assessed via an
adapted version of Kanfer et al.’s (1994) measure of these constructs. Two items tapped
off-task thoughts and one tapped the negative (affective) thoughts that were experienced
during the prior 50 trials of the Bandit task (see Appendix D for speciﬁc items).
Individuals rated each item on a 5-point Likert-type scale ranging from 1 = Never to 5 =
Constantly. The correlation between the two items tapping off task thoughts (r = .25)
was lower than expected, so the composite off-task thought variable should be interpreted

cautiously.

69

Dependent Variables

Several dependent variables were explored in this study. The operationalizations
for each of the dependent variables are described below.

Performance is deﬁned as the number of successful hires (i.e., decisions) an
individual makes per block. Each block has 50 decision choices, so maximum
performance for each block is 50 successful hires. However, because the highest
probability associated with a given agency was .6, the typical maximum performance
level is 30. The performance for each block was computed by aggregating the success
for each trial within a given block. Thus, performance is a block level variable.

Exploitation and exploration behaviors are operationalized in terms of the number
of switches between the four agencies within a given block. A switch was deﬁned as a
change in agency choice from the previous choice. Higher number of switches within a
block represents higher exploration and lower exploitation. The degree of exploration
and exploitation is also represented as a block level variable.

Response time is deﬁned as the number of milliseconds that lapse between two
decision choices, with time starting immediately following a button click representing a
decision and stopping when the next click is made. The response times for each trial
within a block were aggregated to form an average block level response time variable.

In addition, several of the process variables described in the previous section were
used as dependent variables for some of the analyses. For example, boredom, anxiety,
threat appraisals, frustration, self-efﬁcacy, satisfaction with performance and off-task and

negative-affective thoughts were used as DVs for at least one hypothesis. These

70

variables are operationalized as they were described in the previous section, with higher

scores on each item or scale representing higher levels of the variable of interest.

Manipulation Check

Several questions assessing the effectiveness and perception of the manipulations
were administered to individuals at the end of the session. In addition to providing
validation that the manipulations are effective, responses were also examined to provide
insight into what individuals experienced as they performed the task in the varying
conditions. Questions focused on both the warning manipulation (sample item: “I was
informed that a change may or would occur in the task environment”) and the change
manipulation (sample item: “Thinking back over the 18 blocks of trials, did any aspects
of the task change?) Based on responses to these questions, individuals were prompted
to provide more detailed accounts of what they perceived as occurring in their
environments. Open-response formats were provided in addition to the multiple choice

questions to allow for further possible explication (see Appendix E for speciﬁc items).

Pilot Test Results

A pilot study testing 192 undergraduates was conducted prior to the current study
to evaluate the effectiveness of the change and warning manipulations. Pilot study
participants (and their data) are not included as a part of the actual study. Manipulation
check data were collected following the pilot study. The results from this examination
are provided below.

The change manipulation was assessed by asking individuals whether or not

anything changed during the task. Out of the 102 individuals in the no change

71

conditions, 44 (43%) individuals reported that there was a change in the environment,
compared to 70 (78%) of the 90 individuals reporting a change in the change conditions.
Although the number of individuals reporting that a change occurred in the no change
condition was higher than we expected, examining the qualitative follow-up questions to
this question indicated that other variations were being reported as changes. For
example, some individuals reported a change if their performance changed from block to
block, suggesting that there is noise in our measurement. However, the general wording
of this question is intended to prevent the use of “leading questions”. That said, the
substantially higher percentage of individuals reporting a change in the change condition
suggests that the change manipulation was effective. Noise due to measurement is
proposed to account for the less than perfect effect.

At the time of the pilot study, there were only three levels of warning as opposed
to four (did not have the told-general after block 5 warning). However, I believe that
examining the general pattern of results for these three conditions will provide the
information needed to speak to the effectiveness of the waming manipulation. The
warning manipulation check was composed of two quantitative questions. First,
participants were asked if they were informed of a change occurring during the
experiment. For the not told conditions, 9 out of 63 (14%) reported being warned of a
change, 29 out of 65 (45%) of those in the told-general beginning condition reported
being warned, and 61 out of 61 (100%) of those in the told-speciﬁc after block 9
condition reported being warned of the change. Other than the told-general beginning
condition, there is strong evidence that the warning manipulation worked. For those in

the told-beginning condition, it is possible that they did not pull out the warning from the

72

rest of the task instructions presented before the task, thus reducing the salience of the
warning. However, although the told-general manipulation did not produce as high of
numbers as I would expect, there was a substantially higher percentage of individuals in
the told-general condition reporting a warning than in the no change condition,
suggesting that the warning was somewhat effective (14% vs. 45%, respectively).

The second question assessing the warning manipulation tapped when individuals
reported being informed of the change (never, before task started, or middle of the task).
The patterns were generally as expected, with the majority of the not told individuals
reporting never being told. Of the 9 individuals in the not told condition who reported a
warning in the ﬁrst question, eight of these individuals reported being warned before the
task began, suggesting they may have interpreted the general task instructions as a
warning. The told-general beginning condition showed a split between never being told
(n =23) and being told before the task began (n = 24), indicating again that the salience of
this warning was weaker. The majority of individuals in the told-speciﬁc after block 9
condition reported receiving the warning in the middle of the task (80%), as expected.

Overall, results from this pilot study suggest that the manipulations were effective

enough to justify using them in the current study.

Primary Study

Analysis Plan

To test the research hypotheses, I used linear mixed modeling analyses (SPSS v.
17) with random intercepts and slopes over individuals to account for the non-

independence of the data. The models estimated the within- and between-person variation

73

in the primary dependent variables (performance and switching behavior) accounted for
by the manipulations (e.g., level of change or warning), trait variables (e.g.,
conscientiousness), process variables (e.g., frustration) or some interaction of these
variables in my study. In addition, the within- and between-person variation in the
process variables was estimated by other models. Time was included in all of the models.
Effect size estimates were computed for the hypothesized relationships using the

procedure outlined by Rosenthal and Rubin (2003). The effect size, represented as

requivalenta is calculated using the F-value and denominator degrees of freedom for each

source test (\/(F/(F + [N-2]). The effect size is interpreted as a correlation, with

2 . . .
requivalent , representing the percentage of vanance 1n the DV accounted for by each

source.

Prior to testing the study hypotheses, I estimated an unconditional means model
for each of the DVs used in the study hypotheses. These models tested whether there was
systematic within- and between-person variation in these variables, which would justify
incorporating predictors of variation into the models. An ICC is computed for each of the
models, which represents the percentage of total variance in the DV accounted for
between-person differences. If the ICC is greater than .05, conditional growth models that

incorporate predictors can be tested (Barcikowski, 1981).

74

RESULTS

Before testing all of the study hypotheses, the success of the change manipulation
was explored. As noted previously, a successful change manipulation should result in
changes in performance, switching behaviors (exploration/exploitation), and response
time after a change for those in the change condition, but not for those in the no change
condition. Failure to meet these conditions precludes examination of the remaining study
hypotheses, as the focus of these hypotheses is on predicting the effects of the change. If
the change manipulation does not result in a change in outcomes, there is no variability to
predict.

The ﬁrst step I took to examine the success of the change manipulation was
graphing the change by time interaction for each outcome. Examining the interaction
patterns in the graphs made it apparent that using the full range of data may mask change
manipulation effects on outcomes that are actually present. Speciﬁcally, the graphs
illustrated that the majority of activity occurred within a small window surrounding the
change indicated by the vertical lines in the ﬁgure (see Figure 3).

To be comprehensive, aﬁer interpreting the graphs, I also analyzed these
relationships using growth modeling. The growth model analyses failed to ﬁnd a
signiﬁcant effect of the change manipulation on these outcomes. The results from these
analyses were in line with my interpretation of the graphs, suggesting that further
analyses with the full data set should not take place.

As a result of these analyses, I made the decision to cut down the data set to

maximize the chances of ﬁnding an effect if one is present.

75

cow—8:0 ll
communes: II

baggage D

83V 85.

w~§c_m_v_m~N_:c_a
_____—_ __

w

h
_

 

 

 

 

 

Im—

ION

INN

IVN

aaumuoy ad rgpucg

[0N

le

 

 

ooggotom so scum—sag: amass—U mo couaﬁﬁaxm How «am ==m

m 2:3

76

The original data set included 18 performance blocks, each composed of 50 trials,
resulting in a total of 900 trials of data for each participant. The ﬁrst 9 blocks were pre-
change blocks, with learning expected to plateau around block 5 or 6 (from pilot data).
Based on the patterns of the graphed data, I created a restricted data set, which zoomed in
on the blocks where the effects seemed to occur. Speciﬁcally, the zoomed in data set was
restricted to the two blocks before a change (blocks 8 and 9), representing the pre-change
blocks and the two blocks immediately after a change (blocks 10 and 11), representing
the post-change blocks. As a result of these decisions, the data set was reduced to 200
trials, representing these four blocks. Process variables assessed after each of these
blocks (8, 9, 10 and 11) were kept to measure state changes over time. The 200 trials
were then re-divided into performance blocks consisting of 10 trials per block to allow
for a more ﬁne-grained examination of behavior variability. The outcome variables were
recomputed based on these new blocks using the same procedure as presented in the
methods section. The ﬁnal data set used for the rest of the analyses was thus comprised of
20 performance blocks, 10 of these pre-change and 10 of these post-change.

It should be noted here that due to the reduced data set, the immediate impact of
the told-general beginning and told-general after block 5 manipulations were not able to
be explored. However, it is proposed that the beneﬁt of reducing the data set to be able to

explore what is happening surrounding the change is greater than the costs of shrinking

the data set.

Zoomed Analyses

Condition Check

77

As with the full data set, the ﬁrst analyses I conducted were aimed at testing the
success of the change manipulation on performance, switching behaviors, and response
time. The predictors entered into the model included time (0-19), change condition
(collapsed across warning) and pre-post block (dichotomous split of blocks before and
after change), along with all two- and three-way interactions. The DVs for each of the
three models were performance, switching behaviors, and response time, respectively
(see Table 4).

Condition 1: Performance. The ﬁrst condition predicted that performance would
be similar in slope and intercept in the pre-change blocks (blocks 0-9) for both change
and no change conditions, whereas in the post-change blocks (blocks 10-19),
performance in the change condition would decrease relative to pre-change levels and
compared to the no change condition (who should see a steady increase or maintenance
of performance across both pre- and post-change conditions). The
condition*time*prepost 3-way interaction was signiﬁcant (F (1 , 4603) = 4.44, p = .04),
accounting for less than 1% of the variance. The effect size of this interaction is very
small. However, looking at the graph in Figure 4, it is evident that both the change and
no change conditions maintained similar performance during the ﬁrst 10 blocks assessed,
which captured pre-change time. Between blocks 9 and 10, a change occurred in the
change condition, but not in the no change condition. The performance level of the no
change group stays fairly consistent in level and slope across the remaining performance
blocks. However, the performance level for the change group diverges from the other

condition immediately after the change occurred (block 10 performance dropped

78

signiﬁcantly). However, over the course of the remaining time (time 11-19), the

performance levels climbed more steeply as they approached pre-change levels.

Table 4
Change Manipulation Results for Zoomed Analyses

 

 

 

 

Num df Den df F-Value p r-equiv
DV: Performa_ng§
Time 1 1227 10.24 .00 .09
Change 1 1038 13.34 .00 .1 1
Pre-Post 1 4603 24.87 .00 .07
Time*Change l 1227 1.39 .24 .03
Change*Pre-Post 1 4603 17.39 .00 .06
Time*Pre-Post 1 4603 5.56 .02 .03
Change*Pre-Post*Time l 4603 4.44 .04 .03
DV: Switching
Time 1 549 12.05 .00 .15
Change 1 336 .01 .98 .01
Pre-Post 1 4604 .67 .41 .01
Time*Change 1 549 1 .74 .19 .06
Change*Pre-Post 1 4604 1 .33 .25 .02
Time*Pre-Post 1 4604 .33 .57 .01
Change*Pre-Post*Time l 4604 3.83 .05 .03
DV: Response Time (med)
Time I 698 16.35 .00 .15
Change 1 367 1.12 .29 .06
Pre-Post 1 4604 .43 .5 l .01
Time*Change 1 698 .63 .43 .03
Change*Pre-Post 1 4604 .04 .84 .00
Time*Pre—Post 1 4604 .91 .34 .01
Change*Pre-Post*Time l 4604 .00 .99 .00

 

Notes: med = median; Num df = numerator degrees of freedom; Den df =
denominator degrees of freedom; r-equiv is the effect size for each source
(sqroot(F/(F + Den df))
Condition 2: Switching Behavior. The second model tested whether the number
of switches at each point in time (0-19) differed across change and no change conditions

as a function of the presence of a change. The three-way interaction of

changemanipulation*time*prepost borders on signiﬁcance (F (1, 4604) = 3.83, p = .05).

79

Again, the effect size for this relationship was very small, accounting for less than 1% of
the variance. The graph depicting this relationship (see Figure 5) shows that across time,
prior to a change (times 0-9), both conditions show similar patterns in terms of the
number and pattern of switching behaviors. Speciﬁcally, these patterns suggest a slight
downward trend over time, such that the number of switches decreases over time
(suggesting less exploration and more exploitation). However, after a change occurs in
the change condition, the patterns diverge. The change condition individuals show a
spike in switches immediately following a change followed by a steady level of switches.
The no change condition individuals however show a continuing decline in the number of
switches across all 20 time points. This pattern provides support for this condition, and
justiﬁes testing the remainder of the hypotheses testing predictors of switching behavior
over time.

Condition 3: Response Time. The third condition explored whether response
times varied over time as a function of change. It was hypothesized that response times
would increase in the post-change blocks (10-19) for the change conditions (as a result of
having to re-evaluate the task to identify how to change response patterns), but not for the
no change conditions. The pattern of response times across both conditions was similar,
and therefore this condition was not supported (F (1, 4604) = .00, p = .99). There was a
signiﬁcant main effect for time (F (1 , 698) = 16.35, p = .00), such that response times

showed a general decline over time across all conditions.

80

Figure 4

Condition 1 Performance Graph

“19.00
“18.00
‘17.00

 

 

 

"U
i: ED :3
‘3 £1 a
3 a a
‘3 D U
l
o I I
\\
~\‘
\\‘
~\'."
~\.
.\
.\
.\
V) _ V!”
V) V
aouuuuopad

81

‘l6.00
15.00
14.00
13.00
12.00
11.00
10.00
'.00

7.00
.00

4.00
3.00
2.00
1.00

Figure 5

Condition 2 Switching Behavior Graph

ConditionType
-- Unchanged

—- Changed

 

 

 

-19.00
-18.00
~17.00
-16.00
—15.00
—14.00
—13.00
—12.00
—11.00
—10.00
—9.00
—8.00
—7.00
—6.00
—5.00
—4.00
—3.00
-2.00
+1.00

 

 

3.50“

é é. é.

«i N N

(moms #) mums 3WD)!“

82

1.50‘

Although this condition was expected to be supported, the nature of the task may
explain why it was not. When performance feedback suggested that the task had
changed, the simplicity of the task allowed for individuals to react to the change fairly
quickly, with minimal disruption. Response time was not a key DV in any of the study
hypotheses, thus even though this ﬁnding was unexpected, the rest of the models were

tested.

Manipulation Check

The quantitative and qualitative results of the manipulation check were also
explored to see whether the change and warning manipulations were effective. The
change manipulation was assessed by asking individuals whether or not anything
changed during the task (see Figure 6). Out of the 137 individuals in the no change
conditions, 73 (53%) individuals reported that there was a change in the environment.
This was not expected, as the task did not change for these individuals. Reviewing the
qualitative follow-ups to this question suggested that individuals felt that several different
things were changing in their environments —— from their particular behaviors to the
likelihood of reaching the goal. The general wording of this question may have led to
erroneous “yes” answers that do not mirror the actual change I was expecting them to
report. However, the question was worded as it was to prevent leading individuals in
specifying the change I was looking for. For those in the change condition, out of 119
participants, 88 (74%) reported that a change occurred in their environment. While this is
higher than the no change condition, which is what I would expect, there are still a
sizeable number of individuals who did not report a change. This could again be a result

of the wording of the question, or could be due to low pre-change performance.

83

Emaﬁu
_

258%..5

33325
_

 

 

 

a

g

 

 

 

 

 

g?

paﬁunqg use; Buruodax slunprxrpul jo 33311133134

5

 

 

«8&9 29:33:..333 sagas
S can

84

That is, if individuals performed poorly before a change, they may not have noticed that a
change had occurred. Even though the clarity of the change manipulation check was
weak, it is still expected that the “real” change was detected by the majority of
individuals.

The warning manipulation check was composed of two quantitative questions.
First, participants were asked if they were informed of a change occurring during the
experiment (see Figure 7). For the not told conditions, 6 out of 64 (9%) reported being
warned of a change, 31 out of 62 (50%) of those in the told-general beginning condition
reported being warned, 61 out of 66 (94%) of those in the told-general after block 5
condition reported being warned, and 63 out of 64 (98%) of those in the told-speciﬁc
after block 9 condition reported being warned of the change. Other than the told-general
beginning condition, the numbers indicate that the warning manipulation worked. For
those in the told-beginning condition, it is possible that they did not pull out the warning
from the rest of the task instructions presented before the task, thus reducing the salience
of the warning.

The second question assessing the warning manipulation tapped when individuals
reported being informed of the change (never, before task started, or middle of the task).
The patterns were generally as expected, with the majority of the not told individuals
reporting never being told. The told-general beginning condition showed a split between
never being told (n = 16) and being told before the task began (n = 27), indicating again
the poor warning salience of this condition. Both the told-general after block 5 and the
told-speciﬁc after block 9 conditions showed that the majority of individuals reported

receiving the warning in the middle of the task, as expected.

85

3:32.55
swim nos m 0.85 ”can :8. gwm :8. was“? oz

 

 

é

_ 4

 

.15

g

 

 

 

g

.3

 

 

 

 

aﬁunqa n 10 pauuom Bumodar slcnprArpur 10 aﬁrnuaauad

 

 

g

 

 

«8&9 =o2£z$§2 wﬁES:
a 2%

86

Overall, results from the manipulation check suggest that the manipulations worked as
expected.

Table 5 presents the means, standard deviations, and correlations for all of the key
study variables across conditions. The general patterns of correlations across variables are
as I expected. As noted in the table, the correlations in this table are between-person
correlations for the study variables. Considering the trait variables, one speciﬁc area of
interest for this study is the pattern of correlations among the facets of conscientiousness,
as several of my hypotheses are based on a subset of these facets. The correlational
patterns lend support for the breakdown of conscientiousness into dependability and
achievement factors, as predicted and supported in past research. Also noteworthy here
is that the process variables used in this correlation table are aggregated across the four
time points (aﬁer blocks 4, 9, 14, and 19) during which they were collected. Likewise,
the outcome variables (switching, response time and performance) are aggregated across
all 20 performance blocks as well. Correlations between process and outcome variables
were stronger than those between trait and outcome variables, as would be expected.

In Table 6, within-person correlations are presented for the process and
performance variables. The process variable correlations represent averaged within-
person correlations for the four time points these variables were assessed. Two sets of

performance, switching and response time variables are also provided.

87

 

:8. :3- 8. :8. 8.- 8.- 8. 8. 8. 8. a... 83 83135808

 

No. *N_.u 0O. co. *m_.. no. 00. m_. No. *2. mm. moN Emzﬂon: 00 .VN
ELV- uuNmr co. 3.0m. we; unONf 3,2. ON. _ _. co. hm. 00m 30:3:— mO .MN
aim-N. uuhmr _ _ .. *ch. 3.6—6 _.:.,Nm.- _o..- _N. N_. nor 0v. hwy-M Baco>c< v0 .NN
auMN. E..MN.- No. unNN. :. aimN. _ _.- nor no. no. mm. vwd ham—acozoEm no ._N
neON. {*ONB _o. nun-N. _o.- _o.u —o. «*mv. .or me. hm. mag” otﬂg NO .ON
*3 . a: .u _ _. *nm—. VG. _0. 0o. N_. we. _o.- vm- gum GOSQEme— _O .o—
uemh. aimN... _o. 1.0.». SEN. a: .u no... NN. wo- n: . wm. med A533

3282850 .m—

«nmm. *2... co: _.:._Nm. ..:.__N. _o.- nor 3. o_. o_. :4. Norm A533
8:58:88 .:

nemw. a... _ m... mo. unNn. *m _ . ucVN... Nor _N. WC. aim _ . mv. m©.m A533
EoEo>oEu< .o—

ueov. 2.- _o. 3.6—. N_. nor vo. mN. he. _ _. mm. _M.m 30:553sz 00 .m—
uncm. SEN-u co. aim-M. _ _. aim—f wor 2. cor wo- mo. WNM. oEEEﬂDnzom WU .3
n15. unoNv :. «awn. ©_. *m—r No. _N. :. caa. :0. on...” EoEo>oEo< #0 .m—
cnam. SEN-u _o.- S'Nv. *2 . 2.- _o.- NM. *3. :- 2... 09? 3.053339 MU .N-
anom- mo.- *N_... .24—N. **MN. _ _. nub—r #0.. QC. #0. G. “um-M 30512qu N0 .:
_ 4.31m... .31 . «raw. :- _.:..NM.- no. a . :. taw— . mm. vwﬁ ~Aowe—hmm._n=om _U .o—

_ ..:..mN. ..:..nm.- aiNN. unmm. no. 5N. *2 .u Nov 3.. NoN OOA< :8... .0

_ no. _.:..NN. co. aim-N. 2 . mo... 2 . we. mwd 00mm :qu .w

_ ac. SEN-u :. 0°. *2. :. mm. :5 001:3;- .h

_ **WN. nor 2 . Ma. No. on. Emm- _ob:oU he @002 .0

_ *©_ .- No. mo. mcr V6. nod bomxc< “3.:- .m

_ .434. 2.- {now- _N.m NYVN FU< .v

_ *NVK 0_. vW—ov voémm— F<m .m

_ 8.- 3... 8.2 owe. .m

_ Om. ”—.m (£0 .—

8~¢§m§§k §§ﬁh

2 o m N. o m v m N _ Om 2

 

$33.8; SEM- AoM Maegan 85:33.89 ~88 85-23m- mﬁitome zcmxmkémmacwm

m 03m..-

88

 

 

No. 33:6 co. umc. nor ch vo- No. 8. "5.— mné ooEﬁEO-ﬁom .5”
co. numo. numc. Ibo. 3.2. avo- _o. tax-o. 35o. :lug madVM 08:. Omcoawod .hm
..:..mO.- 26¢. a. .3..— _.- uan the.- cor _o.- _o. 2 .m Ave-N win—033m 6m
835.3; 2:385
fro—.- _.:..M_. unvo- cnwcr 3.":- nuhor .30.. No. _o. On; _m.N 3:95.:-
woz 35m .8
fume.- ..imc. mo. 3..th aim—- _o.- aiNN. oo- 5. ﬁx. MK.— tO 33m .VM
No.- 1.2. _o. 3.3.- **o_. «mo. .2..— —. oc- euvo. mh. ba- Qa<uwo§ 39m .mm
name. .3..— _.- 3.,ch E..— _. .3..— _.- gin—.- oo.- .3..— _. .3..— _.- no; m.N mm 83m .Nm
oo- No. CC. name. No. S. :NN. tamo- smoc wa- th $5800 300 38m Am
too. .306.- ..imor uumo. EEO.- ..iwor «*ON. oo. _o.- mmd wm.wN 300 when 38m .OM
.Ibo. :N_.- enmcr aim-c. nine.- 5. L: .- «*mo. 5. 3; EN .ton— 3m 8.3m 6N
sine.- ..im— . No. 23¢.- ..z..MN. mo.- .53. sine. caho- vo; _N.N 3253‘ 33m .wN
niecr **N_. ean. {nor .33. fume. *uNN. CO. 55—. RNA WBN norm-awe.“ 83m .hN
Nor nmc. Elb- **v_.- 2.3. _.:.._ _.- :2. *mo.- Sumo.- m_-_ N9". Save-5m 85m 6N
835.3; 8.82%

o_ o w h w v m N _ Om $—

 

.anoov m 033-

89

 

 

..55.- 3..m5.- 3.55. ...m5. 355. N5. 3.55. ...m5. 3.55. —5. 55— 3.5 oocmctoton— .wm
3..N—. 3..5— . 3.55. 3.55. N5. and. —5. m5. 3..— —. m5. —5.5w— mod5m 05:. omconmoy— Rm
3.55. 3.2. 3.55.- 3..m.5.- ..:..—.5.- 3R5.- ......55.- 3.55.- ..:..w5.- —5. 5—.m 55.N wad—355m .5m
8~QQ~L§§ Nsﬁvbag
3.55. 3.35. m5.- —5. ..:..55.- N5.- 3.55.- N5.- —5.- 3.55. 5m.— —M.N ﬂaw—EFF woZ 89m .mm
N5. *m5.- 3.55. 3.55. N5. 3.55. ...M5. 3.55. ...m5. 3..— —. 5w. mh— .—.-—.O 33m .5M
:2..- S. 8.- 8.- 8. :2 .- 3.- .8. 8. :3. ﬂ. 8. assé 23m .3.
.25—.- 3..:.- 3..— —. 3.55. _.:..m—. 3.55. 3.3—. 3.5—. m5. 3.5—. m5.— m.N mm 33m .Nm
...m5. 355. m5. 355. N5.- 3.55. —5. 3.55.- 3..— —. —5. 55. NBA 35800 :30 33m . — m
3.55.- _.55.- 3..m—. 3.5—. 3.5—. 3.55. 3..5—. ..:..m—. 3.2. 3.55. mm.5 mm.wN FOO Eon— 0—5m .5m
—5. 3.55.- 3.55. 3.55. :55. 3..m5. ...m5. N5. 3.55. 55. 5— .— —o.N tun— umm 38m .5N
3.55. —5. m5.- 3..m5.- 55. 3.55.- 55. 3.35. ...m5.- —5.- 55.— —N.N .00—5 33m .wN
3.55. ...m5. 55. 55. —5. 3.55.- N5.- 3.55. 55. 3.55. hN— WEN catabman— 88m .5N
3R.5.- _5. 3..55.- 315.- ..:..m5.- 3.35.- 3.55.- 3.55.- —5.- 3.35.- m—.— N5.m EO—uouom 85w .5N
8~Q§Nh§§ H§U§t
3cm”. 3..NN.. .5— . 55. ..:..w—. w5.- —5.- 3.5N. 3.5m. —5.- 5m. 55.»... A533 mmocﬁx—O .mN
3.5m. 3.5N. N— .- — —.- 5—.- 55.- .15—.- n5.- —5.- *5— .- mm. moN Eﬂ—Eon—x— 5O .5N
3..—m. 3..m5. 3..—m. 3..—N. 3.5»: 5—. 3.5N. 3.5m. 3.5m. 5—. NW. 556 50:85 no .MN
3..Nm. ......N5. 55. 55.- 1.5—. 3..5—.- m5. «*hN. 3.5N. 55K 55. 556. o§:o>—u< 50 .NN
:55. :hm. 3..NN. 1.5—. 3.. — N. —5.- >5. 3.5M. .35”. 3..5—. mm. 5w.m b—Eco—uoEm— MO .—N
— 33.5. N— . 55. *m— . m5.- m5.- 3..mN. 3.5m. N5. hm. m5.m o—am—E NO .5N
— m5.- >56 —5. .15—.- _..m—.- 5—. ‘25—. 5—.- 5m. 5nd cots—:92:— —O .5—
— 3.. —0. ......N5. 3.3.. 3.55. 3.2.. 3.55. 3.3.. mm. m5.m €53 mac—Eo—omcoo .w—
. :8. :S. :8. :8. :E. 2...... :.. 8.... ES... assauioo .:
— 3.5m. 3.55. 3.55. 3..N5. 3.5m. m5. m5.m A533 acoEo>oEo< .5—
— 3..Nm. 3.3M. 3.55. 3.55. mm. —M.m 30:33:60 50 .m—
— 3.55. 3.35. 3..Nm. m5. mNd o=——Q—0m—O-.——om WU .5—
— 355. 3...m5. 55. 55d EoEo>o—~—o< 50 .m—
— 3..m5. 5m. 55.5 anus—33:0 no .N—
— —5. 55.»... 305—555 NU .— —

5N o— w— h— 5— m— 5— m— N_ —— Om $—

 

........8. m 0.3

9O

 

 

3.5N. 3.55. 3.5—.- 3.5N.- 3.55.- 55.- 55. N5.- N5. 55.- 55— 55.5 occur—totem .55
3.55.- 3.2.- 3.55. 3..— —. 3.5—.- 3.2. 3.N—. 3.5—. 3.55. 3.5—. —5.55— 55.555 05—5 3:058.”— .55
3.5N.- 3.55.- 3.5—. .. — N. 3.55. 3.55. 3.55. —5. N5. 3.55. 5— .5 55.N chS—Bm .55
$353: 2.38.5
....NN.- 3.55.- 3.55. 3. — 5. 3.55. 3.55. —5. —5. 3.55.- 3.5—. 55.— —5.N map—wave;
32 23m .2
3..NN.- 3.5N.- 355. 3..N5. 3.55. .55. 3.55. 55. N5.- ....5—. 55. 55.— :0 385 .55
3.N5.- 3.5N.- 3.55. 3.55. 3.— —. N5.- —5. ....55.- 3.55.- 3.55. 55. 55. 55.5—«ouch. 885 .55
3.55. 3.5. .....w—.- 355.- 3.5N.- 3.55.- 3.55.- —5. ..55. 3.55.- 55.— 5N mm 385 .N5
3.5N. 3.5—. —5.- 3.55.- 3.5N.- 3.55. 55. 3.55. 355. 55. 55. N55 “—5800 :30 885 .—5
— 3.5N. 3.N—.- 3.5—.- 3.N—.- ..55.- 3.5— ... ..55. N5. —5.- 55.5 55.5N EGO Eun— 385 .55
— 35—.- 3.55.- ...5—6 —5. .....55. —5. 3.55. ..55.- 5—.— —5.N tun— «mm 835 .5N
— .....N5. ....55.- 55. —5. 3.N—.- 3.55.- 3.5—. 55.— —N.N .00—52. 395 .5N
— 355. —5. ..55.- 3.55.- 3.55.- 3.55. 5N.— 55.N cotabmam 83m .5N
— 3.55.- 3.55.- 3.55.- 355.- —5.- 5—.— N5.5 Soup—om Baum .5N
8~Q~N~k§k 0.8.995“
— 3.55. 3. —5. 3.55. 3.55. 55. 55.5 A833 305—050 .5N
— 3.5N. 3.5N. 3.N5. 55. 55.N sum—3.55: 5O .5N
— 3.55. 3.55. 55. 55.5 30:3:— nO .5N
— 3.55. 55. 55.5 83:03:... 50 .NN
_ mm. ...... £38.85... 8 ..N
833.3; 5955

55 5N 5N 5N 5N 5N 5N 5N NN — N Om $—

 

32.8. m 2.3

91

6:... 8:38.. W. .5— .m...w=o£ 03852. ﬂ 3:53...
52. $2530... gnaw—mo ﬂ :0 Jam—8%... 2.0.... m_ ...—33...: $8050.23 m. mm 35.5—88.5 Bow m. #888 Row
:85 12.8.2. m. Row 32. ”gauge-ton 5.3 cotoﬂmuam m. .tom Sm .358 0:... 5 ... 583mm“ 83.3.? E55035
895 maﬁa—.9 $.80..— .oE: .55 EB 0302. $9.8 gown—2.50 05802. 0.... $5538.00 ..o.vn—3. Kev... .8qu

 

— 3.55.- 3.55.- 3.5N.- 3.5—.- 3.5—.- 3.55. .....N—. 55.— 55.5 Dosage-tun— .55
. ...... :2. :2. :8. ::.- .8. ...... 3.... 2.... 3.5%.... .2

. ......N. :3. :8. :m... .....- 2a .3 3.2.3 .8

«5383\— 2.38.5

. :om. ......M. :on- :2..- on. .3 3.5.5.?

32 38m .2

. :om. :ch :3- .... m... to 29m .3.

. :2..- :2 .- m». 3. 338.... 3% .8

. :N... 8.. m... mm 3% .mm

. .... NE :58 .80 3% ..m

«6353\— 333%

 

55 55 55 mm 55 55 N5 .5 Gm _>_

 

......8... m 2......

92

 

 

.8. . .. 2. 2. 8. 8. 8. 8.- 2. 8. 8.- 8. 2.8.5,. .8
2. 8. .... ... ... 2. 2. 8.. 2. 8. 8.- 8. 2.9.3.2 .8
... 8. 8. 8. 8. 8. 8. 2.- 2. 2. 8.- 8. ESE-8.8
2...- 2.- 2.- o..- 8.- 8.- 8.- .m. 2.- 8.- 8. ....- 85»: .8
2.- ....- 2 .- . ..- 8.- 8.- 2 .- 8. 8.- 8.- 8. ...- .89. .2
.8.- 2.- 2.- o..- 8.- 8.- 8.- 8. 8.- 8.- E. 2.- .388 .2
2.- 8.- 8.- 8. 8. .o. o..- 2. 8.- ....- 8. 8.- 88.80 .2
2. 8. 2. 8. 8. .... 2. 2.- 2. 8. 8.- .... 8.8... .2
8. .... 2. 2. .... .... 8. 8.- .N. 8. 8.- 8. 8853.. .2
2. 8. 8. 8.- 8.- 8.- 8. 2.- o..- ... 8.- 2. 8828 ....
8.- 8.- 8.- m . .- 8.- 2.- ...- 8. 8.- 8.- 8. 2.- .532 .2
8.. 8. 8.- 2. . ... 2. 8. .m- 8. 8. 8.- 8. San? .2
8.. . .. 8. 8. 8. 8. m . .- 8. 2. ... .- 2. 53-2. .. .
8.. 8. 8. 8. 8. 8.- 2. 8. 8.- 8.- 53-2 8.
8.. 8. 8. 8. 2.- 8. 2. 2.- 8.- ..mub. .8
8.. 8. 8. 8.- 8. 8. 8.- 2.- 85:5. ...
8.. 8. 8.- 8. 8. 8.- 2 .- 81.5. s
8.. 8.- 8. 8. 8.- 8.- 53-3 .8
8.. 8.- 8.- 8. 8.- tom-ms. .m
8.. 8. ...- 2.- 8:5. ...
8.. 8.- 8.- 818.88% .8
8.. ...- onloosﬁotom .N
8.. .88... ..

2 . . o. 8 w h o n v m 8 .

 

.3353; 889$ .85 823229-me 3.58.2923»..er 238.5235:

o 033-

93

.30... on ..0 .00 ..000 5.8. 00:... 0 005.000 0.03 .0...

0030...? 80005 0.0 8255.... 02.0505 255.2 .50 A0235... 0.00.5.0. :0 :002 500.0500 .002... 005.....-
.3000E0-t0m 50.5 5.550.). .2005 .0580... .005; 50:055-055. .....3 5.50.0305. 5.00.05 6:05.550“.
.000. 500.000 8.0.0.5.. £0.8me... 50.00.05 5.0.5050 05:0..0 0.. 0... :. 8.003 :0 5.. 0:0 5......50

05:0..0 5.. 8.00... 05:05-0... 5.. 00.9.0 .0550. 0 500050 5.. 0080... 500. ..0 50:5: u v< .5 500050 .0.
00805 500. ..0 .0950: u 5< x:0....0:00 05:0:0 5.. 8.003 05:05-82. 5.. 00.0.5 .05....0. N 50550 5.. 0080...
500. ..0 .0050: u 2 .. 500050 5.. 0080... 500. .0 .0050: u .< 50.6. 0.003 500.00 50:500. .05. .0... o. 0:.

0. 00.052550 0.0 EsmI0<-_< 0:0 6.1.8.5. 6.4.0:...»— .o.l:00:.._.m .c.l5:...0..3m .c.l00:0:..0...0m .205.
0.003 500.00 :0... _0>0_ .0... cm 0... 0. 3.050.550 m0:_0> 0.0 amt-5. 5:0 owl5:.:0..>»m 65550555595005.
0:... 5.500.. n .5. x0555 .050 n . 6500.5 05.0.. n 8 25:55:05 0. .0050... 80.02

 

 

8.. 8. 8. 8.- 8.- 8.- 8.- 8. 8. .8. 8.- 28:28
8.. 8. :1 8.- 8.- 8. 2.. 2.. 8.- ....- 50.00.28
8.. 8.- 8.- 8.- 8.- .N. 8. .... o..- 2.3.58
8.. 8. 8. 8. 8.- 8.- 8.- 8. 0882 .8
8.. 8. 8. 8.- 2..- .8..- 8. .80. .2
8.. 8. 8.- 8.- 8.- 8.. .028 .2
8.. 8.- 2.- 2..- 2. 0.00.80 .2
8.. 8. 8. . ..- 00.52 .0.
8.. .... 2.- 8.5.8.... .2
8.. 2.- 050200....
8.. 53-372
8 8 .8 8 2 2 2 2 2 ... 2

 

.AU..:00V o 030..-

94

The ﬁrst set, as noted in the table, represent the within-person correlations across
50 trials (representing the 50 trials prior to each process assessment). The second set
represents the within-person correlations across 10 trials, which are based on the
groupings for the zoomed data set. The patterns in the data set are in line with what we
hypothesized. Speciﬁcally, the within-person process variables are correlated in the
expected directions. For example, frustration and anxiety are moderately positively
correlated, while frustration and boredom are only weakly positively correlated.
Frustration is also correlated negatively with self-efﬁcacy, goal commitment, satisfaction
with performance and personal goals, while it is positively correlated with negative and
off-task thoughts. Performance and switching are negatively correlated, suggesting that
individuals who fail to ﬁnd and stick with the most effective response perform worse than
those who do. Process and performance variables are also correlated in the expected
direction.

Graphs depicting the general patterns over time, averaged across conditions, for
performance, switching, response time, and process variables are displayed in Figures 8-

ll, respectively.

Unconditional Means Models

Before testing the study hypotheses, I ran unconditional means models for each of
the DVs. The parameter estimates, variance components (within and between), and ICCs
for the unconditional means models are presented in Table 7. The percentage of total
variance in the DVs accounted for by between-person differences ranged from 13%
(performance) to 84% (boredom). These results indicate that there is sufﬁcient between-

person variation in the DVs, supporting the exploration of predictors.

95

Figure 8

Performance Averaged Across Conditions over Time

 

 

—19.00
”18.00
’17.00
'16.00
”15.00
”14.00
“13.00
“12.00
“-11.00
—10.00 g

"-8.00
—7.00
-6.00
-5.00
—4.00
"-3.00
‘2.00
”1.00

 

 

5.2‘

49 1:.
v v
aousuuoyad unaw

96

4.4"

4.2“

Figure 9

Number of Switches Averaged Across Conditions over Time

 

 

”19.00
-18.00
”17.00
“16.00
-15.00
“14.00
'13.00
“12.00
*1 1.00
"-10.00
'9.00
'8.00
Lmo
“6.00
-5.00
"4.00
‘3.00
‘2.00
—l.00
H00

 

 

J3
N.
M

g "F é
«s a bi

(saqoms #) 10!“an Moms mm

97

2.25“

 

 

—19.00
’18.00
'17.00
"l6.00
'15.00
"-14.00
—13.00
"-12.00
“-11.00
"10.00 a

‘8.00
-7.00
*‘6.00
"-5.00
—4.00
—3.00
”2.00
“-1.00

 

 

420.00-

Response Time Averaged Across Conditions over Time

Figure 10

is é. é. é. é.
§ § § § §
(Human) emu asuodsau uuaw

300.00*

98

mmcaoz u u
oocgotom.-ll
Donna. I
cougmam ..l
Soup—oml

8&3

as...— EF 232..

 

 

 

oo.m 8N co; 8.
_ _ p h
TON
..... iiilliuuuuuunu Fm.~
III IJ.\IIVAI I\| I‘ll III II
\\\..\\\ /:/:/
l ....... I no m m
a
n
lm.m
\
\
led

 

 

35$ .53 .325ch 363V hmMEmaV 833ch 36.8.5
: 2:5

99

Table 7

Unconditional Means Model Parameter Estimates and Variance Components

 

 

Ave. SD SD

Intercept (within) (between) ICC Hyp.
%
Performance 4.73 1.81 .69 .13 2, 4, IO
Switching 2.60 1.86 2.56 .65 I, 3, 5, 6, 7
Process DVs
Frustration 2.75 .61 1.1 I .77 9a
Anxiety 2.21 .44 .94 .82 8a, 9b
Boredom 3.62 .46 1.03 .84 l 1
Self-eﬁicacy 2.49 .48 .91 .78 9c, 13
Sat. with Performance 2.91 .84 .77 .45 9d, 12
Threat Appraisals .97 .33 .68 .77 8b
Off-Task Thoughts 1.73 .31 .79 .78 14a
Negative Thoughts 2.31 .62 1.14 .77 14b

 

Notes: All estimates are signiﬁcant at p < .05. Models are averaged across all
conditions. Ave. Intercept is the average intercept across individuals for the DV.
SD (within) represents the variation within persons in the DV. SD (between)
represents the variation between persons in the DV. ICC represents the percentage
of variation in the DV accounted for by between person diﬁerences. Hyp. is the
hypothesis number that the null model serves as the basic model for.

Hypothesis Testing

Hypothesis 1. I predicted that the change manipulation (change vs. no change)
would interact with pre-change performance (blocks 0-9) in predicting post-change
(blocks 10-19) switching behaviors, such that good performance in the pre-change blocks
would differentially relate to the number of switches in the post-change blocks (see Table
8). Speciﬁcally, I expected that in blocks 10-19, those with higher performance early on
in the change condition would show an increase in switching behavior aﬁer a change
(signaling that the change was detected and exploration has begun), whereas those in the
no change condition would show a steady decline in switching in the later blocks.
However, the three-way interaction between changemanipulation, prepost blocks and
performance was not signiﬁcant (F (1, 4824) = .77, p = .38). The effect size for this

relationship was only .01, indicating that the three-way interaction accounted for

100

basically no variance in switching behavior. There was however a signiﬁcant main effect

of performance on switching behavior (F (1, 4774) = 26.32, p = .00), suggesting that

across conditions and time, better performance predicted fewer number of switches.

However, this effect was also very small (Tequiv = .07).

Table 8

Early Performance Predicting Switching Behavior

 

 

Num df Den df F-Value p r-equiv

DV: Switching

Time I 3735 .01 .94 .00
Change 1 372 .00 .96 .00
Pre-Post 1 4628 5.53 .02 .03
Performance 1 4774 26.32 .00 .07
Change*Pre-Post l 4758 .48 .49 .01
Change*Time 1 580 2.15 .14 .06
Pre-Post*Time 1 4597 .88 .3 5 .01
Pre-Post* Performance 1 4645 6.76 .0 1 .04
Change*Performance 1 4802 .1 I .74 .00
Time*Performance I 47 75 2.51 .1 1 .02
Change*Pre-Post*Performance 1 4824 .77 .38 .01

 

Notes: Num df = numerator degrees of freedom; Den df = denominator degrees of
freedom; r-equiv is the effect size for each source (sqroot(F/(F + Den df))

Personality-related Hypotheses

Hypothesis 2. It was hypothesized that the dependability aspect of

conscientiousness would differentially interact over time across the change conditions in

predicting performance (see Table 9). Speciﬁcally, it was expected that dependability

would be beneﬁcial to performance in the early blocks for both conditions, but be

detrimental post-change for the change condition only. However, the results of my

analyses failed to ﬁnd support for this hypothesis (F (1 , 770) = .34, p = .56), which was

not surprising given the extremely small effect (requiv = .02).

101

Table 9
Dependability Predicting Performance

 

Num df Den df F-Value p r-equiv

 

DV: Performance

Time 1 1215 .05 .82 .01
Change 1 276 .01 .92 .01
Pre-Post I 4603 .30 .59 .0 1
Dependability 1 782 .00 1 .00 .00
Change*Pre-Post 1 813 1.84 .18 .05
Change‘Time I 1215 1.41 .24 .03
Pre-Post‘Time 1 4603 4.97 .03 .03
Pre-Post* Dependability 1 4603 .26 .61 .0 I
Change*Dependability 1 252 .75 .39 .05
Time*DependabiIity 1 1215 .35 .55 .02
ChggeWre-PosﬂDependability 1 770 .34 .56 .02

 

Notes: Num df = numerator degrees of freedom; Den df = denominator degrees of
freedom; r-equiv is the effect size for each source (sqroot(F/(F + Den df))
I reran these analyses with the individual facets of dependability- order (C2),
dutifulness (C3) and cautiousness (C6)- being entered in place of dependability. The

models with order and cautiousness failed to reach signiﬁcance.

Table 10
Dutifulness (C3 ) Predicting Performance

 

Num df Den df F-Value p r-equiv

 

DV: Performance

Time 1 1222 .82 .36 .03
Change 1 270 1.04 .3 I .06
Pre-Post 1 461 1 .02 .88 .00
Dutifulness l 782 .10 .75 .01
Change*Pre-Post 1 797 5.86 .02 .09
Change’Time 1 1222 1.51 .22 .04
Pre-Post*Time 1 4603 4.98 .03 .03
Pre-Post*Dutifulness 1 4610 I . 1 3 .29 .02
Change‘Dutifulness l 252 .15 .70 .02
Time*Dutifulness I 1222 1.50 .22 .04
CEge‘Pre-Post‘Dutifulness l 766 3.13 .08 .06

 

Notes: Num df = numerator degrees of freedom; Den df = denominator degrees of
freedom; r-equiv is the effect size for each source (sqroot(F/(F + Den df))

102

However, when dutifulness was entered into the model, a trend emerged (F (1, 766) =
3.13, p = .08), although the three-way interaction accounted for less than 1% of the
variance (see Table 10). Examining the interaction graphs mapping this relationship (see
Figures 12 and 13), shows that in the early blocks, as expected, higher dutifulness was
related to higher performance across both conditions.

However, in the later blocks, contrary to the hypothesis, higher levels of
dutifulness were still associated with higher levels of performance for those in the change
condition, although the overall level of performance dropped across all individuals in that
group. Dependability was virtually unrelated to performance in the later blocks of the no
change condition, either due to ceiling effects in performance (although the maximum
performance was not reached) or other factors. The nature of this three-way interaction
between dutifulness, change condition and prepost performance blocks actually shows
some evidence that the interaction between these variables may operate differently than I
expected. However, as this relationship is only a trend, and only emerged for one of the

three facets comprising dependability, the results should be interpreted with caution.

103

GB 2%;

 

 

 

 

8 ...wE 8 ”s3
Imp...»
lwé
lmwé M
m.
.2. m
[36
cowcﬁo -II
cows—£25 I
oghcﬂquoo In

 

8.8 3.8:. $55.9...

Emﬁaxbdxxﬁ moaaztoxsm MEBNBEK actuaxgé 63:30 N Bonﬁmxm N mam=~§3m~
2 85w:

104

So. 38.3.So

 

 

8 “E: 8 “.84
1v
.\
.\.
.\. 1.. .v
.\.
\\\
lvé mm
W.
.3. m
[was
8826 -l
woman—~83 Ill
oghcﬁauaou In

 

 

 

are: 38:. «"5588.—

«mwaagbemenc muzuztcxwmm MSBSEK 26.2.6333 63:35 N 3.973% N umm=~:\.~3Q
2 28E

105

Hypothesis 3. Hypothesis 3 focused on only a subset of performance blocks, and
speciﬁcally on the change condition individuals. It was proposed that dependability
would be negatively related to the number of switches in key presses (exploration

behaviors) after a change occurs (see Table 11).

Table 1 l
Dependability Predicting Post-Change Switching Behavior

 

Num df Den df F -Va1ue p r-CCMV

 

DV: Switching

Time 1 1 18 4.00 .05 .18
Dependability 1 1 18 1.97 .16 .13
Time*Dependability 1 1 18 4.15 .04 .18

 

Notes: Num df = numerator degrees of freedom; Den df = denominator degrees of
freedom; r-equiv is the effect size for each source (sqroot(F/(F + Den df))

As this hypothesis does not propose a contrast between the conditions or pre-post

change performance, this model was limited to only post-change blocks (blocks 10-19)

for the change condition individuals. Dependability had a small effect (requiv = .13) on

post-change switching behavior, although it was not signiﬁcant (F (1, 118) = 1.97, p =
.16), failing to support the hypothesis that dependability would be negatively related to
switching. The model was run again with the independent facets of dependability, and
the model still failed to reach signiﬁcance. Examining a graph of the main effect shows
that although not signiﬁcant, the general pattern is opposite of the expected effect, with
higher levels of dependability resulting in more switching. However, as the main effects
were not signiﬁcant, interpreting the trend in the graph must be done with caution.

However, in the original model, there was a signiﬁcant time*dependability

interaction (F(1, 118) = 4.15,p = .04; requiv = .18). The interpretation of the graph

106

displaying this interaction in Figure 14, suggests that right after the change (block 10),
higher dependability was associated with more switches than low dependability. Over
the rest of the post-change blocks, however, the switching behaviors of low and high
dependability individuals converged, and eventually ﬂipped. This pattern can explain the
lack of a main effect for dependability, as well as provide insight into the general trend
found between dependability and switching. It appears that conversely to what was
expected, high dependability actually results in higher levels of switching immediately
following a change. Although dependability was expected to result in rigidity, and
therefore less exploration, it may be that a third factor played a role. Dependability may
be associated with better performance in pre-change blocks, allowing for the change to be
immediately evident to these individuals, whereas, lower dependability may result in
poor pre-change performance, thus making change detection, and thus the need for
exploration, more difﬁcult. Testing this explanation, I ran a model examining
dependability and performance in pre-change blocks. Dependability was not
signiﬁcantly related to performance in pre-change blocks (F(1, 118) = .12, p = .73),
suggesting that the dependability-performance relationship does not explain the

dependability-switching relationship.

107

«EFF
8L2 oomwﬁ 8L2 8mg 8L2 ORE coma ooW— ooh: coho—

 

 

(saqouMs #)-|0!A31138 moms

 

 

. no?
,.
,_
. 18a
£82.83 in -l ,
£385 33 II
néogoa

 

3.68m owngbéem S Mamauaim Maturemnm zetusmEN 255 N bmweentmva
3 23mm

108

Table 12

Openness Predicting Performance

 

Num df Den df F-Value p r-equiv

 

DV: Performance

Time I 1214 .10 .75 .01
Change 1 275 1.57 .21 .08
Pre-Post 1 4619 1.81 .18 .02
Openness 1 785 1.31 .25 .04
Change*Pre-Post l 803 .00 1 .00 .00
Change*Time 1 1213 1.45 .23 .03
Pre-Post‘Time l 4601 4.97 .03 .03
Pre-Post*0penness 1 4621 . 1 1 .74 .00
Change*0penness 1 252 .29 .59 .03
Time‘Openness 1 1214 .45 .50 .02
Change‘Pre-Post‘Openness 1 764 .58 .45 .03

 

Notes: Num df = numerator degrees of freedom; Den df = denominator degrees of
freedom; r-equiv is the effect size for each source (sqroot(F/(F + Den df))

Hypothesis 4. I hypothesized that openness would interact with change condition
and prepost performance blocks in predicting performance, such that in the pre-change
blocks openness will impact performance similarly across conditions, whereas in the
post-change blocks, openness will have an increasingly beneﬁcial effect on performance
for those in the change condition only (see Table 12). However, the results indicated a
very small effect that failed to ﬁnd support for this hypothesis (F (1, 764) = .56, p = .45).
I reran these analyses with the two individual facets of openness that are theoretically
most relevant- Adventurousness (O4) and Intellect (05)- being entered in place of overall
openness. These models failed to reach signiﬁcance as well. Therefore, hypothesis 4
was not supported.

Hypothesis 5. As with hypothesis 3, this hypothesis focused on only a subset of
performance blocks, and speciﬁcally on the change condition individuals. It was
proposed that openness would be positively related to the number of switches in key

presses (exploration behaviors) aﬁer a change occurs. As this hypothesis does not

109

propose a contrast between the conditions or pre-post change performance, this model
was limited to only post-change blocks (blocks 10-19) for the change condition
individuals. As displayed in Table 13, the main effect for openness was not signiﬁcant
(F (1 , 118) = .74, p = .39), failing to support the hypothesis that openness would be

positively related to switching.

Table 13
Openness Predicting Post-Change Switching Behavior

 

 

Num df Den df F-Value J r—equiv
DV: Switching
Time I 1 18 .07 .80 .02
Openness 1 I 18 .74 .39 .08
Time‘Openness 1 1 1 8 .06 .81 .02

 

Notes: Num df = numerator degrees of freedom; Den df = denominator degrees of
freedom; r-equiv is the effect size for each source (sqroot(F/(F + Den df))

The model was run again with the two theoretically-relevant facets of openness,
and the model still failed to reach signiﬁcance. The time*openness interaction also did
not reach signiﬁcance (F (1, 118) = .06, p = .81). Hypothesis 5 was not supported.
Effects for both the main effect and interaction were very small, accounting for less than
1% of the variance in switching behaviors.

Hypotheses 6 and 7. Hypotheses 6 and 7 also focused only on switching behavior
in the post-change performance blocks for change condition individuals. Hypothesis 6
predicted a positive relationship between LGO (Table 14) and switching behaviors after a
change, whereas hypothesis 7 predicted a negative relationship between both PPGO

(Table 15) and APGO (Table 16) and switching behaviors.

110

Table 14
LGO Predicting Post-Change Switching Behavior

 

 

Num df Den df F-VaIue p r-equiv
DV: Switching
Time 1 118 .13 .26 .03
LGO 1 118 .27 .61 .05
Time*LGO 1 1 18 1.39 .24 .1 1

 

Notes: LGO = Ieaming goal orientation; Num df = numerator degrees of freedom; Den df
= denominator degrees of freedom; r-equiv is the effect size for each source (sqroot(F/(F
+ Den df))

Table 15
PPGO Predicting Post-Change Switching Behavior

 

 

Num df Den df F-Value p r-equiv
DV: Switching
Time 1 1 18 .23 .64 .04
PPGO 1 1 18 .39 .54 .06
Time*PPGO 1 1 18 .27 .60 .05

 

Notes: PPGO = performance goal orientation; Num df = numerator degrees of freedom;
Den df = denominator degrees of freedom; r-equiv is the effect size for each source
(sqroot(F/(F + Den df))

Table 16
APGO Predicting Post-Change Switching Behavior

 

 

Num df Den df F-Value p r-equiv
DV: Switching
Time 1 1 18 .58 .45 .07
APGO 1 1 18 .08 .78 .03
Time*APGO 1 1 18 .55 .46 .07

 

Notes: APGO = performance avoid goal orientation; Num df = numerator degrees of
freedom; Den df = denominator degrees of freedom; r-equiv is the effect size for each
source (sqroot(F/(F + Den df))

The main effects for LGO (F(1, 118) = .27,p = .61), PPGO (F(1, 118) = .39,p = .54),
and APGO (F (1 , 118) = .08, p = .78) were all very small (< 1% of the variance), failing

to signiﬁcantly predict switching behaviors. Two-way interactions between time and

111

goal orientation were also tested, and none reached signiﬁcance. Therefore, hypotheses 6
and 7 were not supported.

Trait-Process Hypotheses

Hypotheses 8a and 8b. Theoretically, it was expected that APGO would be
related to higher levels of state anxiety and threat appraisals after a change. Focusing on
only post-change reactions for change condition individuals, a model testing the main
effect of APGO on state anxiety and threat appraisals was examined. The main effects of
APGO on state anxiety (F (1, 118) = .74, p = .39; see Table 17) and threat appraisals (F (1,
118) = .74, p = .39; see Table 18) resulted in very small effects that were not signiﬁcant,
failing to support hypotheses 8a and 8b. APGO-time interactions also did not reach
signiﬁcance. These ﬁndings are not consistent with previous research linking APGO
with anxiety. However, most of this research has been cross-sectional and failed to
consider this relationship over time. Thus, this ﬁnding suggests future research should be

done to ﬁrrther understand this relationship.

Table 17
APGO Predicting Post-Change Anxiety

 

 

Num df Den df F-Value p r-equiv
DV: Anxieg
Time 1 l 18 .00 .96 .00
APGO 1 1 18 .98 .33 .09
Time*APGO 1 1 18 .02 .89 .01

 

Notes: APGO = performance avoid goal orientation; Num df = numerator degrees of
freedom; Den df = denominator degrees of freedom; r-equiv is the effect size for each
source (sqroot(F/(F + Den df))

112

Table 18
APGO Predicting Post-Change Threat Appraisals

 

Num df Den df F -Value p r-equiv
DV: Threat Appraisals

Time 1 1 18 .99 .32 .09
APGO 1 1 18 .05 .83 .02
Time‘APGO 1 118 1.51 .22 .1 l

 

Notes: APGO = performance avoid goal orientation; Num df = numerator degrees of
freedom; Den df = denominator degrees of freedom; r-equiv is the effect size for each
source (sqroot(F/(F + Den df))

Contrast Comparisons

Hypotheses 9a — 9d. This set of hypotheses predicted that as compared to any
other condition, individuals in the change condition who were not given any warning of
the change before it occurred would experience: higher frustration (a), higher anxiety (b),
lower self-efﬁcacy (c), and lower satisfaction with performance (d), after a change in the
environment (see Table 19). This contrast comparison was used to test how the burden of
having to detect a change (relative to being warned about a change or not experiencing a

change) impacts adaptation outcomes, and speciﬁcally, affective outcomes. Very small

effects were found for each of these outcomes (Tequiv = .01—.07).

Consistent with hypothesis a, the condition*prepost*time interaction was
signiﬁcant (F(1, 4585) = 21 .92, p = .00). The graphs in Figures 15 and 16 show the
nature of this interaction. Speciﬁcally, the change/no warning condition experienced
lower levels of frustration than the rest of the conditions in the early blocks, although this
difference narrowed over the early blocks. Conversely, after the change occurred (in the
post-change blocks), the change/no warning group experienced greater levels of

frustration than the other groups, with this discrepancy being larger immediately

113

following a change, and becoming less discrepant over time. Individuals who
experienced a change without warning had higher levels of frustration immediately

following the change than any other groups, thus hypothesis 9a was supported.

Table 19

Contrast Comparison Predicting Process Variables

 

Num df Den df F-Value p r-equiv
DV: Frustration (9a)

Time 1 400 7.79 .01 .14

Pre-Post I 4585 72.51 .00 .12

ChangeNoWamS 1 298 5.22 .02 .13

Pre-Post*Time 1 4585 21 .72 .00 .07

Pre-Post*ChangeNoWam5 1 4585 74.52 .00 . 1 3

ChangeNoWamS‘Time 1 400 7.73 .01 .14
1

PrePost’ChangeNoWamS” Time 4585 21.92 .00 .07

DV: Anxiety (9b)

 

Time 1 402 .32 .57 .03
Pre-Post 1 45 85 .00 1 .00 .00
ChangeNoWamS 1 289 .13 .72 .02
Pre-Post*Time 1 4586 1.81 .18 .02
Pre-Post‘ChangeNoWarnS 1 4585 1.58 .21 .02
ChangeNoWam5*Time 1 402 .01 .94 .00
PrePost*ChangeNoWam5* Time 1 4586 .18 .67 .01
DV: Self-Efﬁcacy (9c)

Time 1 344 4.69 .03 . 12
Pre-Post 1 4574 49.16 .00 .10
ChangeNoWamS l 283 .14 .71 .02
Pre-Post*Time 1 45 74 .90 .34 .01
Pre-Post‘ChangeNoWarnS 1 4574 l 1.92 .00 .05
ChangeNoWamS *Time 1 344 2.30 .13 .08
PrePost*ChangeNoWam5* Time 1 4574 .90 .34 .0]
DV: Satisfaction (9d)

Time 1 415 6.90 .01 .13
Pre-Post 1 4607 50.57 .00 . 10
ChangeNoWarnS 1 367 3.16 .08 .09
Pre-Post*Time 1 4618 6.69 .01 .04
Pre-Post*ChangeNoWam5 1 4607 9.28 .00 .04
ChangeNoWam5*Time 1 415 6.10 .01 . 12
PrePost*ChangeNoWam5‘ Time 1 4618 .45 .50 .01

 

Notes: Num df = numerator degrees of freedom; Den df = denominator degrees of
freedom; r-equiv is the effect size for each source (sqroot(F/(F + Den df))

114

95,—.

8d 8.x 8.5 8.0 cad 86 cod 8d 8.. 8.
_ _ _ h L _ _ p _ _

 

 

 

 

 

 

 

 

.Immd
-J. LE
/
-1 W
a
B
u
[ed M
m
o.
u
12
8.. -l
8.1 / .
953828520 L: N
8.... 3.8.: $5.68..

«mmﬁébdxat mMSaB a 5%» =e.:a§§.¢ wﬁtﬁpﬁ E 22:0 3.. ...: Sturgeb $5.53: QZEMSSULE tentemseb 3.5.260
m. and...

115

84 -l
8. l
EBOZOMCEU

 

 

 

©~ 053nm

116

It is important to consider why the discrepancy in frustration levels existed prior
the change, as this wasn’t hypothesized. When considering that the comparison group
consists of all other levels of warning and change however, it makes sense. Three of the
conditions comprising this group were told that a change was going to occur (or may
occur) and it didn’t —- this deception or ambiguity in their environment may have led to
initial frustration with the task. However, over time this frustration likely dissipated as
the time since warning increased. Regardless of the reason for this pre-change difference
in frustration levels, however, it actually suggests that change/no warning participants
were greatly affected, as their frustration levels not only increased after the change but
overcame the greater initial frustration levels of the other conditions.

Although frustration and anxiety are expected to be similar emotional responses,
and thus ebb and ﬂow similarly, the model predicting differences in anxiety between the
change/no warning condition and all other conditions was not signiﬁcant (F (1 , 45 86) =
.18, p = .67), thus hypothesis 9b was not supported. The main effects and two-way
interactions in the model also failed to reach signiﬁcance. The lack of ﬁndings for
anxiety could be due to the relatively low mean levels of anxiety experienced across
individuals — overall, individuals did not ﬁnd the task to be very anxiety-producing.

Hypothesis 9c examined the impact of change and warning on self-efﬁcacy. A
signiﬁcant condition*prepost interaction was found (F (1, 4574) = 11.92, p = .00).
Interpreting the graph of this interaction in Figure 17, suggests that in the pre-change
blocks, those in the change/no warning condition experienced higher levels of self-
efﬁcacy than the individuals in the other conditions. However, in the post-change blocks,

there is an immediate drop in change/no warning individuals’ level of self-efﬁcacy, while

117

individuals in the other conditions show a fairly even level of self-efﬁcacy across pre-
and post-change blocks. The drop reduces the discrepancy between self-efﬁcacy levels
in pre-change blocks, such that they are approximately equal following a change.
Examining this relationship across all blocks captured what otherwise would have been
missed — that is, the change without warning did have a profound inﬂuence on
individuals’ self-efﬁcacy levels. However, if only post-change levels were examined, it
would appear that self-efﬁcacy levels were fairly similar across all conditions. The
longitudinal nature of the data thus allows for discovery of this relationship when
otherwise it may have been missed.

Consistent with hypothesis 9d, a signiﬁcant condition*prepost interaction (F (1 ,
4607) = 9.23, p = .00) and a signiﬁcant condition*time interaction (F(1, 415) = 6.10, p =
.01) were found for satisfaction with performance. The graphs in Figures 18 and 19
depicting the condition*prepost interaction shows that the nature of this relationship is
consistent with the hypothesis. That is, prior to the change in the task, individuals in the
change/no warning condition experienced slightly higher levels of satisfaction with
performance than those in the other conditions, but in the post-change blocks, the
relationship switched, such that individuals in the change/no warning condition saw a
signiﬁcant drop in satisfaction levels that dropped them below the other conditions. The
time interaction graph tells the same story with a little more detail. Speciﬁcally, the
comparison group showed a fairly steady level of satisfaction with performance across all
time points (both pre- and post-change blocks), whereas, the change/no warning group
experienced a large drop in satisfaction levels immediately following the change, with a

subsequent increase over time (eventually closing the gap between conditions).

118

8.. -1
8. ..I
EBOZDmcﬁo

:8»...

 

 

lbw .N

s

Kenya-nag unaw

[OWN

load

 

 

mmteﬁv a new? .mMMEBEEK S 2.35 Ex as 29.53.69 35:23: eZEMSEDKo teatemgeb 3.3369

: 2%.”.

119

8.. -|
8. 1
gaozowaﬁo

282..

 

 

,/ Imam

lwd

lad

 

aoumuoyad my“ nogoejspes unaw

[NM

 

 

mo2222§§~ 22.: 2e..Be\....22m «2.2: 26232.? 26.2.8290
... 2am...

120

 

 

 

mmmmmmmmmmmmmmmmmmmm
ﬁﬂﬁﬁﬂﬂﬂﬂﬂﬂﬁﬂﬂﬁﬁdﬁﬁﬂ.
THEN
13m
m
S
lg"... n
m.
u
m.
lmm
m.
[gm
:3

 

 

mutcsxcxwmk ~33 =e.:oc\u.aam 3.5» 2E5... 202.8ch
2 2am

Thus hypothesis 9d was supported.

Hypothesis 10. The impact of the change*warning interaction on performance in
the post-change blocks was examined in this model. This hypothesis targeted one of the
primary purposes of this study: what is the role of detection vs. behavioral change in the
adaptation process? Speciﬁcally, it was expected that in the change conditions, those
receiving a speciﬁc warning immediately before the change would perform best, while
those receiving no warning would perform worst. Conversely, it was expected that in the
no change conditions, those receiving a (false) speciﬁc warning would perform the worst,
while those receiving no warning would perform best. Neither the two-way interaction
between change and warning (F (l , 183) = .95, p = ..42; see Table 20), nor the three-way
interaction between change, warning and time (F (1, 159) = .74, p = .53) reached

signiﬁcance.

Table 20
Change X Warning Predicting Performance after a Change

 

Num df Den df F-Value p r-eguiv
DV: Performance

Time I 159 7.97 .01 .22
Change 1 183 19.79 .00 .31
Warning 3 I83 .99 .40
Change'Time I 159 8.83 .00 .23
Waming*Time 3 159 1.13 .34
Chan ge*Warning 3 183 .95 .42
Chﬂge*Waminﬂime 3 159 .74 .53

 

Notes: Num df = numerator degrees of freedom; Den df = denominator degrees of
freedom; r-equiv is the effect size for each source (sqroot(F/(F + Den df))

The change manipulation signiﬁcantly interacted with time to predict performance (as
evidenced in condition 1), but the warning manipulation did not appear to be strong

enough to elicit differences in performance. Examining the manipulation check data

122

suggests that the warning manipulation was not as clear cut as the change manipulation,
providing one reason for why this hypothesis was not supported. Since the warning
manipulation was weak, I was not able to gain a clear understanding of how the detection
and behavioral change process interact in the adaptation process.

Process Hypotheses

Hypothesis 11-14. The remaining four hypotheses examined changes in process
variables over time as a function of change and time. It was predicted that the presence
of a change in one’s environment would induce discrepancies in reactions to the task
compared to pre-change or no change levels. The following process variables were
examined in these hypotheses: boredom (H11), satisfaction with performance (H12), self-

efﬁcacy (H13), off-task thoughts (Hl4a), and negative thoughts (H14b). The change

manipulation had a fairly small effect on each of these outcomes (Tequiv = .03-.07).

Consistent with hypothesis 11, a signiﬁcant change*prepost*time interaction was
found in predicting boredom levels (F(1, 4598) = 4.03, p = .045; see Table 21). It was
expected that boredom levels would decrease in the post-change blocks for individuals
experiencing a change, relative to their pre-change levels and the post-change levels of
individuals not experiencing a change. The nature of this interaction is captured by the
graphs in Figures 20 and 21. In pre-change blocks, boredom levels show the same
general pattern for both conditions, with a jump in boredom around block 4. The mean
level of boredom is consistently higher for the change condition than the no change
condition in the pre-change blocks, however, which was not expected. In the post-change
blocks, this same relationship holds immediately following the change, but approximately

halfway through the post-change blocks, boredom levels spike in the no change

123

condition, but do not show the same jump in the change condition. At the end of the task,
those in the no change condition experience greater levels of boredom than those in the
change condition as expected.

Although the interaction was signiﬁcant, the pattern depicted in the graphs
suggests that if the change did impact boredom, it had a delayed effect. That is, the
boredom levels of the no change condition did not surpass the boredom levels of the
change condition until several trials after the change occurred. Thus, although the
relationship makes theoretical and empirical sense, it is possible that the change did not

directly impact boredom levels.

Table 21
Change and Time Predicting Boredom

 

Num df Den df F-Value p r-equiv

 

DV: Boredom

Time I 362 8.24 .00 .15
Change 1 280 .43 .5 l .04
Pre-Post l 4598 .05 .82 .00
Change*Pre-Post 1 4598 3.00 .08 .03
Change'Time l 362 .33 .57 .03
Pre-Post*Time l 4598 .42 .52 .01
Change*Pre-Post*Time 1 4598 4.03 .05 .03

 

Notes: Num df = numerator degrees of freedom; Den df = denominator degrees of
freedom; r-equiv is the effect size for each source (sqroot(F/(F + Den df))

124

 

 

 

 

cowSEU -II
vowcncocD III

oghaoEEoU

3:;
8.3 8.w_ 8.5 8.2 8.2 84; 8.2 8.2 oo.: ood—
_ _ _ _ _ r _ _ _ _

 

 

 

[No.m

:2

 

M
moparog unaw

o'o
‘0.
m

 

Ind

 

LEM

 

 

3.-.: 3.8:: 38:. «“5532...

«owtcabéoxt thmxem 32.5%me 235 38 93:59
a 0.3me

126

Hypotheses 12 and 13 predict that satisfaction with performance (Table 22) and
self-efﬁcacy (Table 23), respectively, will decrease after a change relative to pre-change
levels and compared to those in the no change condition. Consistent with hypothesis 12,
a signiﬁcant change*prepost*time interaction was found (F (l , 4601) = 6.94, p = .01) for

satisfaction with performance.

Table 22
Change and Time Predicting Satisfaction

 

Num df Den df F-Value p r-equiv

 

DV: Satisfaction with Performance

Time 1 418 1.39 .24 .06
Change I 367 4.88 .03 .1 l
Pre-Post l 4600 58.06 .00 .1 1
Change‘Pre-Post 1 4600 24.70 .00 .07
Change‘Time l 418 .02 .89 .01
Pre-Post’Time 1 4601 23.74 .00 .07
ChggﬂPre-Post‘Time l 4601 6.94 .01 .04

 

Notes: Num df = numerator degrees of freedom; Den df = denominator degrees of
freedom; r-equiv is the effect size for each source (sqroot(F/(F + Den df))

Table 23
Change and Time Predicting Self-Eﬂicacy

 

Num df Den df F-Value p r-equiv

 

DV: Self-Efﬁcacy

Time I 350 2.27 .13 .08
Change I 283 .80 .37 .05
Pre-Post l 4579 45.46 .00 . 10
Change*Pre-Post 1 4579 4.73 .03 .03
Change*Time l 350 1.07 .30 .06
Pre-Post*Time l 4581 5.84 .02 .04
Change*Pre-Post*Time l 4581 .97 .33 .01

 

Notes: Num df = numerator degrees of freedom; Den df = denominator degrees of
freedom; r-equiv is the effect size for each source (sqroot(F/(F + Den df))

Examining the interaction graphs in Figures 22 and 23 demonstrates that during the pre—

change blocks, satisfaction levels are fairly similar and stable across change conditions,

127

although there is a slight dip in the change condition levels. However, in the post-change
blocks, the same steady level of satisfaction persists for those in the no change condition,
while a large dip in satisfaction occurs in for those in the change condition immediately
following the change. These levels begin to rebound over time, but do not reach the
previous levels (or the no change levels). These results support the predictions made in
hypothesis 12.

Hypothesis 13 was also supported, with a signiﬁcant change*prepost interaction
predicting self-efﬁcacy levels (F0, 4579) = 4.73, p = .03). Consistent with my
predictions, both change and no change conditions experienced similar levels of self-
efﬁcacy in the pre-change blocks, with those in the change condition experiencing a
signiﬁcant drop in self-efﬁcacy after a change occurred and those in the no change
condition maintaining fairly stable levels (and slightly increasing) in the post-change

blocks (see Figure 24).

128

g:
cod 8.x 8.5 86 cod 86 8M cod 8; 8.
P P _ h F P _ _ _

 

 

 

 

 

 

 

F
tad
J.
y [mad

. w

. a

, lad m
. S
. m.
,. loam m.

a

. m.

, u

. Iwo.~ m.

. m1.

,. a

., .. w.

/ W

’ [8 m «w

Base .| .
caucuses: .II 13 m
oghcﬁaoo
at: 8.923 3.8:— «Wan—Uta...—

 

«mmagboxog muzcﬁxcxsk 5:: 2c.=oc\w.:um mﬁtﬁwﬁ 3:5 35» mM2§U
mm 8&3

129

«a;
9&3 8m: 8m.— oohc. 8.2 8.3 comm. 8W. 8”: echo—

 

lvd

 

 

 

Jo
N
aaumuoyad 1mm uoyanjspes ueaw

$09.30 -i
Emcee—25 i

oghsgoo l~.m
8...: 3.8.... 3.8... given...

 

 

 

QM§.~U-.8% 826.50th 3.3. taughtam @2283ka 2&5 .28 mwtgb
mm 8a....

130

vowzﬁu il
cowcﬁocD i

093303.20 U

.252.

 

 

IONN

./ [.3

i

/I
[Comma-Has usaw

/
é
oi

loud

 

 

AoQu§ﬂ$mm M:...o...§.& 8.5.5 .35 «M585
. ..N 2......

131

Hypothesis 14 examined the role of change and time in predicting off task
thoughts (a) and negative thoughts (b). Results outlined in Table 24 show that a
signiﬁcant change*prepost*time interaction was found for predicting both off task
thoughts (F(1, 4604) = 18.96, p = .00) and negative thoughts (F(1, 4596) = 24.34, p =
.00). See Figures 25 and 26 for the interactions of H14a, and see Figures 27 and 28 for
the interactions of Hl4b. The interaction pattern for off-task thoughts does not
completely line up with the predictions of this hypothesis. Mean levels of off-task
thoughts are consistently higher for the no change condition, across both pre- and post-
change blocks. However, when looking at the pattern of off-task thoughts in the change
condition, although mean levels are lower across time relative to the no change condition,
there is a jump in the experience of off-task thoughts immediately following the change
compared to pre-change levels. Thus, within the change condition, the expected pattern
of off-task thoughts was found, but the no change condition experienced greater overall
levels of off-task thoughts, which was not expected.

The graphs depicting the interaction of change and time for negative thoughts
support my predictions. During pre-change blocks, the level of negative thoughts
experienced across change and no change conditions converge (although those in the no
change condition experience slightly higher levels across these blocks). However,
immediately aﬁer the change, those in the change condition experience increased levels
of negative thoughts relative to their pre-change levels and compared to those in the no
change condition. The increased level of negative thoughts begins to decrease over the

later blocks, but never reaches pre-change levels. Thus, hypothesis 14b was supported.

132

Table 24

Change and Time Predicting Off-T ask and Negative Thoughts

 

 

 

 

Num df Den df F-Value p r-equiv

DV: Off-Task Thoughts (14a)

Time 1 387 6.04 .01 . 12
Change 1 280 .31 .58 .03
Pre-Post 1 4604 1 .90 . 17 .02
Chan ge‘ Pre-Post 1 4604 19.83 .00 .07
Change‘Time l 387 . 12 .73 .02
Pre-Post‘Time l 4604 1 .45 .23 .02
Change‘Pre-Post*Time l 4604 1 8.96 .00 .06
DV: Negartive Thoughts (14b)

Time 1 368 .3 1 .58 .03
Change 1 293 .73 .39 .05
Pre-Post 1 4597 21 .41 .00 .07
Change*Pre-Post 1 4597 53.51 .00 .1 1
Change*Time 1 368 .01 .91 .01
Pre-Post*Time 1 4596 6.12 .01 .04
Change‘Pre-Post*Time 1 4596 24.34 .00 .07

 

Notes: Num df = numerator degrees of freedom; Den df = denominator degrees of
freedom; r-equiv is the effect size for each source (sqroot(F/(F + Den df))

133

25,—.

8.9 8.w 8.5 8.0 86 8... 8.m 8N 8.. 8.
_ _ _ _ b _ h _ _ _

 

 

 

 

 

F3.
.- -/.
IE.—
W
a
B
u
m
1m: w
x
m
. m.
[cm .
@0895 ll
3822.: II
oﬁhcﬁucou lmw.

 

 

 

a... 3.3—3 33:— OMB-50$...—

xmmagbdxxt ..EMVSSN «.35 u§© MSBNmem oaks .8... @3585
mm 2%..

134

comcmao ..II
3830:: I

oghcﬁaou

95,—.
8H. 8_w_ 8.2 8mm. 8.2 8.3 8.2 8PN. 8”: 8m:

 

 

Inc.—

 

, IE.

I . Ion.—

 

-I la:

sulﬁnou user-yo "new

IQ...

 

1mg.

 

 

8.-.: 3.8.83.8... uniﬁes.

338.548xt 8.348.: «Ba KNQ M:...o...§..& 255 .28 mMSED
em as»...

135

comcﬁo -II
vowcﬁocb I

ugh—£30 U

 

 

-l I;

 

I

J»
N.
N

l
"'3
N

 

smﬂnou aApeﬁaN anew

lme

 

IVN

 

 

a... 38.6 2.8:. «M5293.—

«mM:§U-m.¢t 923M335 $53M»? MSBNEEK 2&5 has mwtgb
R 2:3

136

woman—#0 -II
38395 I

oghcﬂunsou

95,—.
8.2 8.w~ 8.2 8.2 8.3 8.3 8.2 8.2 8.: 8.2
_ _ _ . _ _ r _ _ _

 

 

1mm.~

 

 

é

smﬁnoql aApnﬁaN anew

 

v/
%~.
N

:5;

lmw.m

 

 

3?: 3923 33:— «Macaquoom

Emﬁzsécat mimzoﬁ 95:me wzzﬂhwgm 255 has “$235
mm 25m:

137

Summary of Results

The results detailed above are summarized in a model presented in Figure 22.
The solid lines represent signiﬁcant relationships. The top part of the ﬁgure depicts the
trait-level relationships with the main study DVs. It is clear from this ﬁgure that overall,
the trait-level hypotheses were not signiﬁcant. The detection hypotheses (9 and 10),
which were explored through different combinations of change and warning, showed
mixed results. The hypotheses focused on process outcomes (bottom half of the ﬁgure)

showed the most promise, with the majority of these hypotheses being supported.

138

 

 

 

25,—. «8%.:—

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

awo , I ..
£95.... 3.. z 3:: ..
I.“
3.3.3:... 33.55 4- a1:
E0695: :1 Ii K x \ ~ . 99:20
N:—
‘1
3:25:20.— 55 5333.5
58Em.....m 4/2:
03.—
53953; All «a: _ ...—.3 .35:
i 03.— ¢ =< .9 using?
‘ lllllll . IIIIIIIIIII
33!: . oEuwaasu
I. an:
a 3.3.3.— 325. - ......
_ . < - an: ..........
.._..._._._.._ ........................... .... ......... 8...
v ..................................................
.- lllllllllllllllllllllllllll “F: lllllll cum;
..4. n: m: ..................... 1|
_= o. I I I I lllllllllllll e: I I I 00“ L
IN...» .................................................... v: ......H. muocauaﬁ
855320.. - - - .NN/n ............ an». - - -
> .A/l. \\ u + ................................. 3: 3:23:
:—= .‘ su I\I\KI\\ \h\\‘. Iii—a
_ ---/l . Va... . . :8 m o out.» 8%:
uses?» - owns—U . «8%...— 25... a: m N B E 3 :

 

 

 

 

 

 

 

 

 

 

 

a 2:5

139

DISCUSSION

As organizations and jobs become more dynamic, employers need to hire and/or
train adaptive employees. Researchers studying adaptability and adaptation have made
great strides in identifying person and task characteristics and processes that impact
adaptive performance (e.g., Burke et al., 2006; Kozlowski & Bell, 2008; LePine,
Colquitt, & Erez, 2000). The current research sought to build on this research both
conceptually and methodologically. Speciﬁcally, the current study hoped to extend the
adaptation literature by: 1) exploring the role of detection in the adaptation process, 2)
examining the adaptation process as it unfolds over time, and 3) incorporating process
variables that may explain why individuals perform better or worse than others in an
adaptive context. In addition, the current research hoped to identify general person
characteristics (e. g., dependability, openness) that may impact the adaptation process,
thus having implications for selection. Behavioral (switching), outcome (performance)
and affective (process variables) variables served as adaptation outcomes. The
unconditional means models ran for each of these outcomes found that there was
signiﬁcant within- and between-person variance in these outcomes, supporting the
exploration of my hypotheses.

Several hypotheses were proposed to target each of these goals. In general,
hypotheses examining trait predictors (both main effects and interactions with the task
manipulations and time) of switching and performance were not supported. In addition,
the warning manipulation, which served as one of the mechanisms for teasing apart the
role of detection vs. behavioral change in the adaptation process seemed to be too weak

to elicit any effect on the key outcome variables. However, hypotheses exploring

140

predictors of affective (process) reactions were mainly supported, which partially
supported the belief that warnings (detection aid) are beneﬁcial when a change has
occurred. The mixed ﬁndings indicate that future research needs to be done to examine
these relationships to better our understanding of how the adaptation process unfolds. A

discussion of the ﬁndings is presented below.

Supported Predictions and Implications

One of the key ﬁndings in this study is that even in a simple task an abrupt change
in the task environment has effects on exploratory behavior, performance, and affective
reactions to the task. If this is replicated across different types and complexity of tasks, it
would have strong theoretical and practical implications. Theoretically, this ﬁnding
suggests that criteria other than just performance need to be considered in adaptation
research. In addition to performance, which is typically the focus of adaptation research,
important adaptive outcomes may include behavioral changes (such as exploration) and
affective reactions (e.g., satisfaction, frustration, self-efﬁcacy).

The present study found that a change led to lower levels of self-efﬁcacy and
satisfaction, while increasing exploration and negative or off-task thoughts. From a
resource allocation perspective (e. g., Kanfer et al., 1994), we have limited attentional
resources. If those resources are allocated to negative (e. g., rumination) or off-task
thoughts, the resources leﬁ to devote to the task at hand decrease, thus having
implications for performance. The ﬁndings suggest that an unexpected change may
trigger off-task related thoughts, suggesting that organizations should consider ways of

buffering the impact of a change, to reduce the shiﬁ of attentional resources away from

141

the task. Past research has also found that self-efﬁcacy and satisfaction with performance
have a positive relationship with actual performance (e.g., Stajkovik & Luthans, 1998).

Although no hypotheses examined the impact of these affective reactions on
switching and performance, supplementary analyses were conducted to explore whether
the expected relationships existed. Signiﬁcant positive main effects of self-efﬁcacy and
satisfaction with performance were found in predicting post-change performance. The
implications for the negative effect on satisfaction and self-efﬁcacy due to the change
then are even more important to consider, since lower levels of satisfaction and self-
efﬁcacy were shown to lead to lower levels of performance. If a change in a lab task
induces affective reactions that are related to performance changes, it is important to
consider what impact unexpected changes in a real world environment will have on
individual’s responses. Assuming that individuals are more committed and identiﬁed by
their work than they would be an artiﬁcial task, it is likely that these reactions will be
intensiﬁed in real world settings. Future research needs to examine how unexpected
changes in an organizational setting affect employees. If this relationship holds or
intensiﬁes, it would suggest there are practical implications that need to be considered.

Although the warning manipulation overall did not appear to be effective, the cell
comparison between those individuals experiencing a change without any warning versus
all other conditions, suggests that there is some support for providing a detection aid to
reduce the unexpectedness of changes in one’s environment. Individuals experiencing a
change without a warning experienced even greater jumps in frustration and bigger drops
in self-efﬁcacy and satisfaction with performance than those individuals either not

experiencing a change or those experiencing a change, but given some warning about it.

142

As was discussed in the previous section, collapsed over warning conditions, a change
leads to negative affective reactions. However, this comparison suggests that those not
given any type of warning experience these reactions even more strongly than those
given some level of warning. This ﬁnding suggests that dynamic organizations should
provide even a general warning to employees that changes may occur within their work
environments, to prepare them for the possibility of change. Failure to do so may lead to
more negative reactions, which could have implications for performance. Linking this
back to one of the key goals of this study, these ﬁndings suggest that the detection
process, without an aid available, is an important stage to consider in the adaptation
process. These results suggest that the act of change detection has negative effects on
affective outcomes, more so than changes accompanied by a warning. However, future
research should be done to replicate this ﬁnding across different types of tasks and in
different environments to see if these conclusions can be made.

Overall, these results support the proposition that unexpected changes in an
otherwise routine environment can have a substantial impact on an individual’s affect,
behaviors, and performance. However, it is important to consider the very small effect
sizes reported for the supported relationships. Although they reached statistical
signiﬁcance, the effect sizes were all smaller than .2, with the majority being less than .1.
The small effect sizes suggest that these relationships should be interpreted with caution.
Future research should try to replicate and extend these ﬁndings to other tasks and
environments using different types of changes. The impact of an abrupt change may

have different effects than that of a gradual change or a change occurring in an already

143

turbulent environment. Exploring the boundary conditions for the supported

relationships above is an important next step in the adaptation literature.

Unexpected Findings and Implications

Although many of the hypotheses related to the impact of change on one’s
reactions and behaviors were supported, the hypotheses relating to early performance
predicting post-change switching (Hypothesis 1), individual differences predicting
behavioral outcomes (i.e., performance and switching), and the interaction of the
manipulations (i.e., change X warning) predicting performance were not supported. A
summary of these ﬁndings is provided below, along with possible implications and future
research suggestions.

Performance. The ﬁrst of these, early performance predicting later switching, was
proposed based on the expectation that higher performance before a change would allow
for easier detection of the change. The change was designed to lead to a drop in
performance, which would be more noticeable if performance levels were high.
However, I did not ﬁnd support for this relationship. It is possible that the discrepancy
between behavior and performance, due to the underlying probabilities linking these two,
may have created noise in this relationship. That is, there wasn’t a perfect one-to-one
correspondence between behaviors and performance outcomes, which may have clouded
an individual’s ability to determine whether it was normal task noise causing the
performance drop or a real change in the task enviromnent. Even those individuals
consistently picking the same agency experienced ﬂuctuations in their scores due to the
probabilities of payoff associated with the agency, and it is possible that teasing apart

when a change indicated the need to explore versus riding it out may have been difﬁcult.

144

Another possible explanation for this unexpected result is that those performing well
before a change quickly found the appropriate new strategy for coping with the change.
It is possible that an individual understood the task well enough to perform well early on,
that understanding may have led to a quick exploration period to ﬁnd the new solution,
followed by an immediate decrease in exploration. If few switches between agency
choices are needed to decide on the new strategy, the effect will not be strong enough to
detect. Future research should explore this hypothesis using a task with more complex
underlying properties to see if the expected relationship emerges.

Individual Differences. Several hypotheses explored how individual differences
interacted with the change over time to predict adaptive outcomes. Although the
individual differences chosen were theoretically-relevant to the study, none of the
individual difference hypotheses were supported. Dependability, openness, and goal
orientation have been examined in previous adaptation research, but the current study
sought to understand the impact of these traits on the adaptation process in more depth,
by exploring their interaction with task changes over time. The following sections
discuss the relationship of these traits with performance, switching, and affective
outcomes.

LePine and colleagues (2000) found that while dependability was beneﬁcial for
performance in stable, routine situations, it was detrimental to performance in adaptive
contexts. Dependability has been associated with rigidity and persistence, suggesting that
once dependable individuals have found an effective strategy, they will resist changing
even when an environment makes the strategy ineffective (Costa & McCrae, 1992).

Openness, on the other hand, has been demonstrated to be more beneﬁcial in adaptive

145

contexts than routine contexts (LePine et al., 2000). Openness refers to the tendency to
be more explorative, curious, and comfortable in novel environments (Barrick & Mount,
1991; Costa & McCrae, 1992). Similarly, Ieaming goal orientation represents an
individual’s desire to master and learn their new environment (V andewalle, 1997),
suggesting that those higher in LGO will be more open to exploring their environment,
without experiencing the negative affective reactions associated with negative feedback
(e. g., performance drops). Conversely, performance prove and performance avoid
individuals are more likely to react poorly to negative feedback, attributing a drop in
performance to their lack of ability, rather than something about the task that can be
learned and changed. Based on the research bases for these variables, I expected that
dependability, PPGO, and APGO would negatively impact adaptive outcomes (switching,
performance and/or affective reactions) when faced with a change, whereas openness and
LGO would exert a more positive impact on these same outcomes. Unexpectedly, the
data did not support these hypotheses. Past research ﬁndings and the theoretical
relevance of these individual differences in an adaptive context suggests that something
about the task may have prevented these relationships from reaching signiﬁcance. The
most likely explanations for these results are the simplicity of the task and the nature of
the change manipulation. However, the initial conditions explored suggested that there
was adequate between person variance in these outcome variables to test for predictors of
that variation, suggesting that the task was not so simple as to wipe out the effect of
individual differences. It is possible though that the variation between individuals was
mostly accounted for by the change manipulation, leaving little variance left for

individual differences to explain. Future research needs to explore these relationships

146

using more complex tasks to see if individual differences interact with task characteristics
to impact adaptive outcomes over time.

Manipulation Interaction. Another unexpected result stemmed from the
hypothesis examining the effect on performance by the interaction of the change and
warning manipulations. I expected that the closer and more speciﬁc a detection aid was
to the actual change, the less negative the impact of the change on performance would be.
That is, individuals receiving a warning immediately prior to a change in the task would
not have to waste time detecting that a change was there, and instead begin exploring for
the new strategy right away. Conversely, when a detection aid is presented when no
change occurs, I expected that the more speciﬁc the warning was about the change, the
more negative the impact on performance would be, causing them to explore when they
should continue to exploit. The primary purpose of this hypothesis was to examine one
of my main goals for the study: what is the role of detection in adaptation? The results
failed to ﬁnd an interaction of change and warning, suggesting that providing a detection
aid, whether it indicates a real change or not, does not affect performance. However, I
hesitate to accept this as a conclusion. Two possible explanations should be considered
ﬁrst. First, the strength of the warning manipulation was likely reduced when the data set
was reduced. Two of the four warning levels presented warnings early on in the task,
during trials that were excluded from analysis. Although the effects of these warnings
still may have impacted later performance, the immediate effect of these warnings was
not able to be explored. Second, the effectiveness of the warning manipulation overall
was not clear, as indicated by the mixed reports in the manipulation check. There was a

large portion of individuals who reported being warned of a change when they were not

147

and a number of people who were warned of a change, but who said they were not. It is
possible that detection aids, like warnings, can be used to tease apart detection and
behavioral adjustment, but the speciﬁc warnings used in this study were not strong
enough to see an impact. Future research should incorporate different types and strengths
of warning manipulations in their studies to test the role of detection in the adaptation

process.

Additional Implications

Although some speciﬁc implications of this research were embedded in the
discussion of the ﬁndings presented above, a few broad considerations are discussed
below.

Theoretical Implications. The model proposed and tested in the present study
serves as a building block for future research. Several theoretically-derived person, task,
and environmental factors have been proposed that may impact different stages of the
adaptation model. As empirical support is found for some of these relationships, a more
comprehensive model of adaptation can be developed. The manipulations and variables
explored in the current study are only the tip of the iceberg, leaving many research
avenues open for further exploration. In addition, the model opens up the door to
exploring how to measure behaviors occurring at each stage. The current study provided
examples of how some of the stages may be assessed, but alternative ways are possible.
The research questions that stem from the proposed model will have implications for how
adaptation theories and models are developed in the future.

Practical Implications. In addition to the theoretical implications of my study, a

few key practical implications need to be noted. First, I expected that my results would

148

inform organizations on whom to select and/or train for jobs that are likely to experience
changes requiring adaptation. Unfortunately, the results of the current study did not ﬁnd
support for any of the proposed individual differences I explored. However, if future
research ﬁnds empirical support for a set of individual difference characteristics that
make individuals more or less adaptive in response to a change, it would have practical
implications for organizations. Selecting and/or training individuals based on traits that
allow for better responses to change will likely have an effect on employee well-being
and performance.

Another potential implication stemming from the results of this research is that
providing advanced warning, at any level of generality, may buffer the impact of a
change on individuals. In environments where a level-head and quick reaction to change
is needed, either informing individuals of the possibility of a sudden disruption in their
environment or training individuals on how to respond to possible changes that may
occur in their environment may improve adaptive behavior. Granted, this effect is based
on results from an artiﬁcial task, but it is expected that these reactions would be
intensiﬁed in a real world environment. However, ﬁiture research needs to explore

whether this effect is replicated across different conditions and in more realistic contexts.

Limitations and Future Research

As with any study, the current research is not without limitations. First, I was
unable to test my study hypotheses with the full data set because the manipulations only
caused short-term effects, which were missed when examining outcomes at the 50 trial
level. The decisions on how to reduce and transform the data set to be able to adequately

test the hypotheses were based on the nature of the task and the general questions I was

149

trying to address. However, these changes had consequences that need to be discussed.
Speciﬁcally, cutting out the ﬁrst 350 and last 350 trials reduced the variation in
performance and other outcome variables that may be meaningful. For example, for my
ﬁrst hypothesis, early performance was expected to impact switching behavior. When I
cut out the ﬁrst 350 trials, this reduced the variation within the early performance blocks.
My failure to ﬁnd support for this hypothesis may be at least partially due to the data
reduction. However, using only the 100 trials immediately preceding the change allowed
for a cleaner test of this relationship. That is, theoretically, I would expect that
performance immediately prior to the change is what is important in detecting the change,
and thus likely to have the effect on switching behavior. Another implication of the
reduced data set is its impact on my warning manipulation. The told-general beginning
and told-general after block 5 warning manipulations were presented during the trials that
were removed from the data set for the actual analyses. Thus, the immediate impact of
these warnings on outcomes was not able to be examined in the study. Warning timing
did not lead to signiﬁcant differences in performance or switching, either due to the
weakness of the manipulation or the reduction in the data set. The manipulation check
data, which depicts that the warning was not salient to all individuals, supports the
former. However, all but one of the hypotheses collapsed over warning conditions, so the
weakened effect of warning did not impact my ability to test the majority of my
hypotheses. Future research should consider how to make stronger warning
manipulations and test how these warnings directly or indirectly (interactions with

change) impact adaptive performance outcomes. Exploring warnings is crucial to

150

disentangling detection and behavioral responses, which is one of the main theoretical
contributions of this study.

Another limitation of the current research involves the change manipulation. The
short time period in which most of the action took place in response to the change
manipulation suggests one of three things. One possibility is that the task employed for
the current study was too simple to ﬁnd longer term effects of a change. The second
possibility is that the nature of the change was too weak to see longer lasting effects on
task outcomes. The small effect sizes found for all of the hypotheses support this
possibility. A weak manipulation is likely to result in a small effect. Future research
needs to vary the strength of the manipulation to explore whether stronger manipulations
ﬁnd stronger effect sizes.

A third possibility is that both the simplicity of the task and the nature of the
change manipulation contributed to the short-term effects found in the data. Some may
argue that such a short-term impact on outcomes suggests that studying how changes in
an environment impact outcomes is not important. However, considering what even the
smallest impact of a change may have in a high risk environment, like a military context,
reminds us of the importance of understanding the adaptation process. A simple task and
manipulation were needed for this study to test the hypotheses in an environment where I
could see the impact, rather than having too much complexity to pull the effects apart.
Even in the simplest of environments, I found some affective and behavioral effects of a
change. In the future, research should explore unexpected changes in a more complex
environment to see if they substantially inﬂuence affective or behavioral reactions, and

whether individual differences predict those reactions.

151

Several choices were made in this study that may limit the generalizability of the
ﬁndings. With any study, decisions on how to operationalize variables have to be made.
Those decisions should be justiﬁed, with the acknowledgement that other ways of
operationalizing the variables exist. That said, the way I chose to operationalize
exploitation/exploration (as the number of switches) may not be the only way to capture
this type of behavior. Theoretically, the operationalization was appropriate given my
understanding of these two behaviors. However, I recognize that other methods of
capturing this behavior are possible. Additionally, I also made a decision to set the 50
trial goal to 30 out of 50 for everyone. The rationale for setting the goal at 60% was
based on the underlying probabilities of the agency choices as well as what I have learned
from goal-setting theory (e. g., Locke & Latharn, 2002). I wanted to set a difficult,
speciﬁc goal for everyone, to give them something to strive for, that wouldn’t easily be
reached. The probability of the “best” agency was set to .6, so reaching the 60% goal
would be difﬁcult to reach unless the correct agency was chosen repeatedly. However,
because the agencies rewarded individuals based on probabilities, even picking the best
agency for all 50 trials, did not always lead to goal attainment. It is possible that this
confused the participants and encouraged switching when switching was not the best
strategy. However, outside factors operate in real world environments as well, such that
even when an individual makes the best decisions possible, sometimes these factors
prevent them from reaching performance goals. Future research should explore different
goal levels in this environment to see whether the difﬁculty level of the goal impacts how

individuals respond to changes.

152

One of the biggest issues concerning generalizability is the globally small effects
found for the hypothesized relationships. While the small effects help explain why
several of the hypotheses were not supported, it also suggests that the supported
relationships need to be interpreted cautiously. The small effects may be attributed to
aspects of the task, the manipulation, or the lack of an actual relationship. Future
research exploring variations in the task and manipulations used can help address this
question.

Theoretical limitations. In addition to the methodological limitations proposed
above, a few key theoretical limitations need to be addressed. First, I developed a
behavioral model of adaptation, which does not include some of the cognitive
components that have traditionally been explored in the adaptation literature. Although I
defend the validity of a behavioral model of adaptation, I do acknowledge that some of
the cognitive processes that have been shown to be important to successful adaptation
were not included. For example, cognitive Ieaming outcomes (e.g., declarative and
procedural knowledge) were not measured. The adaptive transfer literature identiﬁes
declarative and procedural (or strategic) knowledge as a key predictor of an individual’s
ability to perform in more complex or changed environments (e.g., Bell & Kozlowski, in
press). Instead of using cognitive indices of Ieaming, I relied on the patterns in the
behavioral data to indicate learning. For example, if an individual begins to repeatedly
select the best agency, it suggests that they have learned the demands of the task. The
behavioral model of adaptation ﬁt well with the behavioral task individuals performed.
However, there may be cases where cognitive processes need to be incorporated into the

model. Future research can explore how to combine the behavioral stages of my model

153

with the cognitive processes found in other conceptualizations to model the adaptation
process.

In my review of the theoretically-relevant person, task and performance factors
that I suggested may predict behaviors at different stages of the adaptation process, I
included a much more comprehensive list of factors than I was able to explore. It is
possible that had I included some of these predictors in the current study, I would have
been able to predict the different patterns in behavioral and affective responses to the
change.

Person factors. Locus of control may have been an important variable to explore,
speciﬁcally when examining how people interpreted and responded to the change.
Internal locus of control is associated with the belief that you can control what is
happening in your environment (e. g., Spector, 1988). When a change occurs, individuals
with an internal locus of control would likely attempt to change their behaviors to
respond to that change. However, those with an external locus of control may notice that
something has changed when their performance level drops, but decide that taking action
would not help their scores. When examining the qualitative data collected for the
manipulation check, some of the comments alluded to this feeling of “helplessness”, with
individuals reporting that no matter what they did they still could not reach the goal.
Other individual difference variables, such as fear of failure and threshold for change,
may have been important factors to explore. Future research should include additional,
theoretically-relevant variables when exploring the adaptation process.

Task and performance factors. Another direction for future research is

manipulating some of the task- and performance—related factors that may impact the

154

adaptation process. For example, manipulating the level of goal set, giving the individual
the responsibility of setting the goal for each block, or varying the feedback speciﬁcity
and/or frequency are likely to inﬂuence individual’s experience in an adaptive
environment. Goal-setting theory (e. g., Locke & Latham, 2002) studies have provided
empirical support for the impact of goal and feedback manipulations on performance. In
the current study, performance feedback was provided after each set of 50 trials, with a
running score available throughout the 50 trials. This feedback was the source of
information they used to determine whether their behavioral decisions were effective.
The performance feedback was also the information they had available to determine
whether the task had changed. Providing different types of feedback, like the percentage
of times they chose each agency, may have changed how they interpreted changes in the
task. The more feedback an individual has about their environment, the more likely they
are to detect a real change. Feedback manipulations may be a different way to explore

the role of detection in the adaptation process.

Summary of Limitations and Future Research Directions

As highlighted in the discussion above, many of the relationships explored were
not supported. The limitations of the current study may explain the lack of signiﬁcant
relationships. Key takeaways from this study, in terms of suggestions for future research,
targeted at addressing these limitations, are summarized below.

The ﬁrst takeaway from this study is that a longer window may be needed for
adaptation to be explored, and speciﬁcally for the role of detection to be examined. While
a simple task was purposely chosen for the current study based on the types of data it

could provide, the simplicity of the task may have made testing the relationships more

155

difﬁcult. The nature of the task allowed for adaptation to occur very quickly, reducing
the variance in response times and behavioral reactions to the change across individuals.
For example, it was hypothesized that the change manipulation would have an impact on
response time, such that after a change, response times would increase (that is,
individuals would slow down). This increase could be considered an indicator of change
detection; however, response times were not impacted signiﬁcantly by the change. The
nature of the task was so simple that picking another alternative did not require response
time changes. While using a different, more complex, task would be one way of
addressing this limitation, future research could also make alterations to the task used in
the current study to increase the complexity embedded in the task. Related to this, the
nature of the change used in this task may also have been problematic in uncovering
relationships that actually exist. The change itself was an abrupt, fairly obvious change
that led to noticeable performance decrements. Whereas individuals may not have
understood or been able to attribute what to do in response to the change, the change
itself was designed to stand out. To understand detection more clearly, embedding more
ambiguity around the change may help tease apart detection from behavioral adjustment
in response to the change. Using the current task, ambiguity could be manipulated by
making the underlying probabilities of the optimal action more similar to the probabilities
of the less optimal responses. Additionally, incorporating multiple changes within the
task, or a more gradual change, may have increased ambiguity, allowing for more
variance in individuals’ responses to the change.

Given what was learned from the current study, another important consideration is

whether or not individuals could actually learn in this task based on the instructions they

156

were given, and relatedly, whether or not the task was actually perceived as a detection
task by participants. The degree to which participants understood the task and what they
should be focused on during the task (e. g., sampling from the different agencies) has
implications for understanding the (lack of) effectiveness of both the change and warning
manipulations, and thus the ability to explore the role of detection. In the current study, it
was assumed that the task was simple enough that individuals would learn it without
training focused on developing a mental model of the task and the relationships between
behaviors and performance. One takeaway from this study is that Ieaming cannot be
assumed, and even for simple tasks, training participants on the deeper structural features
of the task is needed to ensure they understand the task. “Learning” was indirectly
assessed in the current study through the examination of the performance graphs across
individuals. However, given the characteristics of the task, it was possible for individuals
to perform well without actually understanding what they were doing to achieve the
scores they received. Without a deep understanding of the task, manipulations of the task
environment will be ineffective, as individuals will not be able to attribute accurate
meaning to those manipulations. Along the same path, the instructions presented up ﬁont
about the task did not frame it as a detection task, and without ensuring that the task was
learned, it is unlikely that individuals perceived the task as a “detection task”. That is,
individuals were focused on their performance levels only, and given the normal
ﬂuctuations in performance levels (not due to any changes in their environment) and a
lack of task knowledge, individuals were not likely to focus on detecting task changes.
Rather, an examination of the qualitative reports provided by participants suggests that

performance level shiﬂs were attributed to a variety of non-task related factors or factors

157

that were perceived as out of their control (e. g., the task was unfair, performance levels
were random, rather than linked to speciﬁc behaviors). Future research should address
this limitation in the current research by training individuals on the task features and
testing their mental representations of the task prior to collecting experimental data.
Given the limitations in the current study in terms of studying detection,
alternative ways of framing the patterns in the data could be considered. For example,
while speciﬁc hypotheses were not proposed in terms of linking the main study DVs to
the process variables assessed, the patterns in the correlation table suggest that
performance and behavior are related to affective reactions. From a self-regulatory
perspective (e. g., Bandura, 1986; Carver, 2004), affective reactions may serve an
evaluative function in the regulation of performance and behavior, and the relationship
between affect and performance (and behaviors) may be dynamic, with each inﬂuencing
the other in turn. If this is the case, then it may be that effective adaptation is a result of
the self-regulation of performance and behaviors in response to discrepancies driven by
the introduction of a change into the environment. Future research should consider how
self-regulatory processes can be embedded in and/or differentiated from the adaptation

process as deﬁned in the literature.

Conclusion

The purpose of the present study was to develop and test a behavioral adaptation
model, which was built on existing adaptation research. Testing the adaptation process
over time, examining theoretically relevant individual difference predictors and process
outcomes, and incorporating a manipulation to tease out detection from behavioral

change were the key contributions of this study. The results of this study demonstrated

158

that even in a simple task environment, an unexpected change can impact affect (e. g.,
satisfaction with performance, frustration), beliefs (e.g., self-efﬁcacy), behaviors
(switching) and performance. Unexpectedly, the study failed to ﬁnd individual
differences that predicted these reactions to change. In addition, the warning
manipulation did not have the impact that was expected. However, this study serves as a
ﬁrst step toward exploring this adaptation model. The directions for future research
suggested above are targeted at improving our understanding of the adaptation process
and addressing the unexpected ﬁndings I have discussed. The model I proposed can
serve as a useful tool in guiding future thinking about how the adaptation process unfolds

within and between individuals over time.

159

APPENDICES

160

APPENDIX A

Informed Consent

Electronic Informed Consent: Psychological Predictors of Decision-Making
Project Title: Psychological Predictors of Decision-Making
Investigators' Names: Dr. Rick DeShon and Tara Rench

Description and Explanation of Procedure:

This research study is about how people make choices in playing games. You will be
asked in the lab session to make choices in a game task. You will also be asked to answer
questionnaires to help us understand your characteristics and how you interact with the
game tasks.

If you agree to participate, you will next ﬁll out a set of questionnaires online and then
participate in a lab session doing a game task at the time you selected. Filling out the
online questionnaires will begin immediately upon entering the requested information
below and will take approximately 30 minutes to complete [1 credit]. It includes
questions about your demographic information, your SAT/ACT scores, and other
characteristics related to the game task you will learn. You will then go to the lab in room
80 Psychology Building at your scheduled time to participate in the game task simulation,
which will take 1 hour to complete [2 credits]. You will receive basic training in the
games and will then perform the games over a number of trials. You will be asked
questions about your reactions to the games as you play them.

Those not interested in this research can seek other alternatives and research studies for

subject pool credit by consulting their instructor or the Department of Psychology subject
pool web site.

Estimated time required:
30 minutes for the online questions [1 Psychology subject pool credit];
1 hour for the lab based game task [2 Psychology subject pool credits]

Risks and discomforts:
None anticipated

Beneﬁts:

You will gain experience completing several standard psychological measures on-line.
You will also learn some task skills in the games you play in the lab. Finally, you will
learn about the process of conducting psychological research. This research will also
beneﬁt society. It will help us to better understand task performance, and research by

161

Hunter & Hunter (1984) calculates that an increase in overall American workforce task
performance of just .02% would raise our GNP by approximately ﬁve billion dollars.

Please DO NOT use the "Back" or "Forward" features of your browser! Use only the
links to go to and from pages within this study. DO NOT open up other web sites from
within this browser window!

Agreement to Participate

Participation in this study is completely voluntary. You also give permission to the
experimenters to access or verify your ACT/SAT score from the University Registrar.
You have been fully informed of the above-described procedure with its possible beneﬁts
and risks. You are free to withdraw this consent and discontinue participation in this
project at any time without penalty. If you choose to withdraw from the study prior to its
completion, you will receive credit for the time you have spent in the study (1 credit per
30 minutes).

The investigators will be available to answer any questions you may have. If, at any time,
you feel your questions have not been adequately answered or you want to discuss the
research, please contact the investigators (Dr. Rick DeShon, deshon@msu.edu , 353—
4624; Tara Rench, renchtar@msu.edu, 353-9166) or the Head of the Department of
Psychology (Neal Schmitt, 353-9563). If you have questions or concerns regarding your
rights as a study participant, or are dissatisﬁed at any time with any aspect of this study,
you may contact—anonymously, if you wish— Peter Vasilenko, Ph.D., Director of
Human Research Protections, (517)355-2180, fax (517)432-4503, e-mail irb@msu.edu,
mail 202 Olds Hall, Michigan State University, East Lansing, MI 48824-1047

If you agree to participate, please choose that you agree and then continue on to answer
the online survey. The reason you are asked for your name and PID in the online survey
is to ensure that you receive credit for participating in the study. Participants' identity data
will be kept secure and conﬁdential. Your identity will not be associated with your

responses for any data analyses. Your privacy will be protected to the maximum extent
allowable by law.

Do you agree to participate in this study?

[I I agree to participate in this study. [Please continue to the online survey]
E] I do not agree to participate in this study.

162

APPENDIX B

Online Trait Survey Measures

(T o be administered online through the HPR system prior to in person session)

Background Information

Directions: Please respond to the following questions to the best of your ability.
Remember, your responses will be conﬁdential.

1. What is your overall GPA?

2. What is your age?

3. What is your gender?

3.
b.

Male
Female

4. What is your year in school?

.09-99‘s»

Freshman
Sophomore
Junior
Senior

5th year +

5. What ethnicity do you consider yourself to be?

rump rm 9.0 9‘9»

Mexican American

Puerto Rican

Other Hispanic

American Indian or Alaskan native

Asian

Black/Aﬁican American
White/Caucasian/Not of Hispanic origin
Native Hawaiian or other Paciﬁc Islander
Other

6. What was your overall SAT score?

7. What was your overall ACT score?

163

Trait Anxiety (as measured by IPIP-NEO Facet N1: Anxiety)

Directions: Please use the rating scale provided to indicate how much you agree or
disagree that each of the statements below describes how you are in general. This is
a survey of typical behavior, so there are no right or wrong answers. Please respond
as accurately as possible.

Scale: 1 = Strongly Disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 = Strongly Agree

NI: Anxiety

1. Worry about things.

2. Fear for the worst.

3. Am afraid of many things.

4. Get stressed out easily.

5. Get caught up in my problems.
6. Am not easily bothered by things.
7. Am relaxed most of the time.
8. Am not easily disturbed by events.
9. Don't worry about things that have already happened.
10. Adapt easily to new situations.

Need for Control (self-constructed)

Directions: Please use the rating scale provided to indicate how much you agree or
disagree that each of the statements below describes how you are in general. This is

a survey of typical behavior, so there are no right or wrong answers. Please respond
as accurately as possible.

Scale: I = Strongly Disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 = Strongly Agree

I like to maintain control over my environment.
I do not like when 1 cannot control what is happening in my environment.
I like to have control over what I do and how I do it.

When I face situations that I do not have control over, I try my best to regain
control.

PPS”?

164

Goal Orientation (V andewalle, 1997)

Directions: Please use the rating scale provided to indicate how much you agree or
disagree that each of the statements represents your general orientation toward

work.

There are no right or wrong answers. Please respond as accurately as

possible.

Scale: I = Strongly Disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 = Strongly Agree

1.

2.

10.

11.

12.

13.

I am willing to select a challenging assignment that I can learn a lot from.
I often look for opportunities to develop new skills and knowledge.

I enjoy challenging and difﬁcult tasks where I’ll learn new skills.

For me, development of my ability is important enough to take risks.

I prefer situations that require a high level of ability and talent.

I’m concerned with showing that I can perform better than others.

I try to ﬁgure out what it takes to prove my ability to others.

I enjoy it when others are aware of how well I am doing.

I prefer projects where I can prove my ability to others.

1 would avoid taking on a new task if there were a chance that I would appear
rather incompetent to others.

Avoiding a show of low ability is more important to me than Ieaming a new skill.

I’m concerned about taking on a task at work if my performance would reveal that
I had low ability.

I prefer to avoid situations where I might perform poorly.

165

IPIP-NEO Facets

Directions: Please use the rating scale provided to indicate how much you agree or
disagree that each of the statements below describes how you are in general. This is
a survey of typical behavior, so there are no right or wrong answers. Please respond
as accurately as possible.

Scale: I = Strongly Disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 = Strongly Agree
IPIP-NEG Conscientiousness Facets

CI: Self-eﬂicacy*
1. Complete tasks successfully.
2. Excel 1n what I do.
3. Handle tasks smoothly.
4. Am sure of my ground.
5. Come up with good solutions.
6. Know how to get things done.
7. Misjudge situations.
8. Don't understand things.
9. Have little to contribute.
10. Don't see the consequences of things.

C2: Orderliness **
11. Like order.
12. Like to tidy up.
13. Want everything to be "just right."
14. Love order and regularity.
15. Do things according to a plan.
16. Oﬁen forget to put things back in their proper place.
17. Leave a mess in my room.
18. Leave my belongings around.
19. Am not bothered by messy people.
20. Am not bothered by disorder.

C3 : Dutifulness **
21. Try to follow the rules.
22. Keep my promises.
23. Pay my bills on time.
24. Tell the truth.
25. Listen to my conscience.
26. Break rules.
27. Break my promises.
28. Get others to do my duties.
29. Do the opposite of what is asked.
30. Misrepresent the facts.

166

C4: Achievement-striving*
31. Go straight for the goal.
32. Work hard.
33. Turn plans into actions.
34. Plunge into tasks with all my heart.
35. Do more than what's expected of me.
36. Set high standards for myself and others.
37. Demand quality.
38. Am not highly motivated to succeed.
39. Do just enough work to get by.
40. Put little time and effort into my work.

C5: Self-discipline*
41. Get chores done right away.
42. Am always prepared.
43. Start tasks right away.
44. Get to work at once.
45. Carry out my plans.
46. Find it difﬁcult to get down to work.
47. Waste my time.
48. Need a push to get started.
49. Have difﬁculty starting tasks.
50. Postpone decisions.

C6: Cautiousness/Deliberation **
51. Avoid mistakes.
52. Choose my words with care.
53. Stick to my chosen path.
54. Jump into things without thinking.
55. Make rash decisions.
56. Like to act on a whim.
57. Rush into things.
58. Do crazy things.
59. Act without thinking.
60. Oﬁen make last-minute plans.

Note. *Achievement-related facets; "Dependability-related facets

167

IPIP-NEO Openness Facets

01: Imagination

l.
2
3
4
5
6.
7
8
9.
l

0.

Have a vivid imagination.

. Enjoy wild ﬂights of fantasy.
. Love to daydream.

. Like to get lost in thought.

. Indulge in my fantasies.

Spend time reﬂecting on things.

. Seldom daydream.
. Do not have a good imagination.

Seldom get lost 1n thought.
Have difﬁculty imagining things.

02: Artistic Interests

11.
12.
l3.
14.
15.
16.
17.
18.
19.
20.

Believe in the importance of art.
Like music.

See beauty in things that others might not notice.
Love ﬂowers.

Enjoy the beauty of nature.

Do not like art.

Do not like poetry.

Do not enjoy going to art museums.

Do not like concerts.

Do not enjoy watching dance performances.

03: Emotionality

04:

21 .
22.
23.
24.
25.
26.
27.
28.
29.
30.

Experience my emotions intensely.

Feel others' emotions.

Am passionate about causes.

Enjoy examining myself and my life.

Try to understand myself.

Seldom get emotional.

Am not easily affected by my emotions.

Rarely notice my emotional reactions.
Experience very few emotional highs and lows.
Don't understand people who get emotional.

Adventurousness *

31. Prefer variety to routine.

32. Like to visit new places.

33. Interested in many things.

34. Like to begin new things.

35. Prefer to stick with things that I know.
36. Dislike changes.

37. Don't like the idea of change.

168

38. Am a creature of habit.
39. Dislike new foods.

40. Am attached to conventional ways.

05: Intellect*
41. Like to solve complex problems.
42. Love to read challenging material.
43. Have a rich vocabulary.
44. Can handle a lot of information.
45. Enjoy thinking about things.
46. Am not interested in abstract ideas.
47. Avoid philosophical discussions.
48. Have difﬁculty understanding abstract ideas.
49. Am not interested in theoretical discussions.
50. Avoid difﬁcult reading material.

06: Liberalism
51. Tend to vote for liberal political candidates.
52. Believe that there is no absolute right or wrong.
53. Believe that criminals should receive help rather than punishment.
54. Believe in one true religion.
55. Tend to vote for conservative political candidates.
56. Believe that too much tax money goes to support artists.
57. Believe laws should be strictly enforced.
58. Believe that we coddle criminals too much.
59. Believe that we should be tough on crime.
60. Like to stand during the national anthem.

Note. *Key facets from openness.

169

APPENDIX C

Verbal Bandit Task Instructions

“Okay, we will go ahead and get started with the instructions for the session. If you have
any questions as we move along, please raise your hand and I will do my best to answer
your questions.” [Continue. . .]

“The computerized task you will be completing is an Employment Decision Making task.
The use of temporary employees is becoming increasing prevalent in today’s economy.
For many companies, hiring temporary employees is part of their daily routine. This
experiment is interested in studying how this process works. You will play the role of
someone in charge of making daily temporary employment decisions.”

“At each of your desks, you will ﬁnd a picture of the decision making task.” [Give them a
minute to look at the picture and then continue. . .] “As the person in charge of hiring
temporary workers you will hire one temporary worker each day from one of 4 temporary
agencies labeled on the screen as A, B, C, D. Each day you will get a different temp
worker. If the temporary worker is successful for that day you get 1 point, if he is not you
get 0 points. You will make these decisions in blocks of 50 days. Each temp agency may
have a different rate of success and failure, so take that into account when making your
decisions.”

“I will now describe the process of this task. You will engage in blocks of 50 hiring
decisions.

After each block of hiring decisions, you will be answering a series of questions. We will
all be moving through the blocks at the same pace so: DO NOT GO FORWARD
UNLESS INSTRUCTED TO DO SO. People that do so will have to do trials again
and/or be asked to leave based on the frequency of doing so.”

[Make sure you read the General Task Instructions below verbatimll

“For this task, you will be making decisions about which Agency you would like to make
a hiring attempt from. To make a hiring attempt, click on the Agency button that
represents your decision about which Agency you would like to hire from. Your goal is to
make 30 successful hiring attempts out of every set of 5 0 hiring attempts. This goal will
be displayed at the top of the screen for each trial. Each set of 5 0 hiring attempts
represents 1 block of this task. You will be completing several blocks of this task during
today’s experiment session. ”

170

APPENDIX D

Process Measures
(T o be administered after every block of 5 0 hiring attempts on the computer)

Directions: This set of questions asks you to describe how you felt during the last set
of 50 choices. Please respond as honestly as possible.

Boredom
During the last set of 50 choices, I found the task boring.

I = Strongly Disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 = Strongly Agree
Frustration

During the last set of 50 choices, I found the task frustrating.

I = Strongly Disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 = Strongly Agree
Anxiety
During the last set of 50 choices, I found the task anxiety-producing.

I = Strongly Disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 = Strongly Agree
Performance
During the last set of 50 choices, I was satisﬁed with my performance.

I = Strongly Disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 = Strongly Agree
Personal Goal

The goal you are assigned for this task is 30 out of 50. What is your personal goal for the
next set of 50 choices of this task?

outof50
Goal Commitment (adapted from Kanfer et al, 1994)
I intend to put in effort to reach the assigned goal (30 out of 50) during the next set of 50

choices.

171

I = Strongly Disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 = Strongly Agree
Self-Efﬁcacy

How likely is it that you will be able to reach the assigned goal (30 out of 50) during the
next set of 50 choices?

I = Not at all likely, 2 = A little likely, 3 = Somewhat likely, 4 = Very likely, 5 =
Extremely likely

I am conﬁdent that I will be able to reach the assigned goal (30 out of 50) during the
next set of 50 choices.

I = Not at all confident, 2 = A little confident, 3 = Somewhat confident, 4 = Very
confident, 5 = Extremely confident

Threat and Challenge Appraisals
How demanding do you expect the next set of hiring attempts to be?

I = Not at all demanding, 2 = A little demanding, 3 = Somewhat demanding, 4 = Very
demanding, 5 = Extremely demanding

How able are you to cope with the demands of the next set of hiring attempts?

I = Not at all able, 2 = A little able, 3 = Somewhat able, 4 = Very able, 5 = Extremely
able

Off-task thoughts (adapted from Kanfer, Ackerman, Murtha, Dugdale, & Nelson,
1994)

I took "mental breaks" during the task.

I = Never, 2 = Once or twice, 3 = Sometimes, 4 = Often, 5 = Constantly
I thought about the difﬁculty of the task.

1 = Never, 2 = Once or twice, 3 = Sometimes, 4 = Often, 5 = Constantly

Affective (negative) thoughts (adapted from Kanfer, Ackerman, Murtha, Dugdale,
& Nelson, 1994)

I thought about how poorly I was doing.

I = Never, 2 = Once or twice, 3 = Sometimes, 4 = Often, 5 = Constantly

172

APPENDIX E

Manipulation Check

Warning check:

I was informed that a change may or would occur in the task environment.
1 = Yes or 0 = No

If yes, at what time point during the task were you informed that a change may occur?
1 = Never, 2 = Before the task began, 3 = Middle of the task

If yes, brieﬂy describe what you were told about a change.

 

 

 

Change check:

Thinking back over the 18 blocks of trials, did any aspects of the task change?
1 = Yes or 0 = No

If yes, brieﬂy describe what you think changed.

 

 

 

If yes, did you attempt to change your strategy for performing the task?
1 = Yes or 0 = No

If you attempted to change your strategy, which (if any) of the following did you try to
do:
1 = Explored all of the agencies to see which agency is best at the time.
2 = Quickly chose another agency to make hiring attempts from that appeared to
be the best at that time.
3 = Continued with current strategy, but put more effort (e. g., time, attention) into
each decision.
4 = Other (please describe):

 

 

 

173

APPENDIX F

Debrieﬁng

Psychological Predictors of Decision-Making
Investigators' Names: Dr. Rick DeShon and Tara Rench

The research study you just participated in examined how people behave and
perform in tasks in changing and variable situations. It looked at how goals and relevant
task factors effected how you went about doing the task at hand. We will also be
examining how your personality characteristics affected the choices and behaviors you
did in this experimental task. This research can give us valuable insights into how people
perform tasks and are motivated toward certain behaviors.

If you have any questions about this study or would like to receive a copy of the
results when they are complete, please notify the investigator now. If, in the future, you
have any questions about the study or would like to receive the results when they are
complete, please call the investigator listed below. Finally, thank you for participating in
this study. If you have any other questions or comments please do not hesitate from
contacting the experimenters.

Inveﬂgators:

Dr. Rick DeShon, deshon@msu.edu, 353-4624
Tara Rench, renchtar@msu.edu, 353-9166

174

REFERENCES

175

REFERENCES

Ackerman, PL. (1987). Individual differences in skill learning: An integration of
psychometric and information processing perspectives. Psychological Bulletin,
102, 3-27.

Anderson, C]. (2003). The psychology of doing nothing: Forms of decision avoidance
result from reason and emotion. Psychological Bulletin, 129, 139-167.

Anderson, M. H. & Nichols, M. L. (2007). Information gathering and changes in threat
and opportunity perceptions. Journal of Management Studies, 44, 367-3 87.

Bandura, A. (1986). Social foundations of thought and action: A social-cognitive view.
Englewood Cliffs, NJ: Prentice Hall.

Barcikowski, R. S. (1981). Statistical power with group mean as the unit of analysis.
Journal of Educational Statistics, 6, 267-285.

Barrick, M. R. & Mount, M. K. (1991). The big ﬁve personality dimensions and job
performance: A meta-analysis. Personnel Psychology, 44, 1-26.

Barrick, M. R., Mount, M. K. & Judge, T. A. (1991). Personality and performance at the
beginning of the new millennium: What do we know and where do we go next?
International Journal of Selection and Assessment, 9, 9-30.

Beck, M. R., Levin, D. T., & Angelone, B. (2007). Change blindness blindness: Beliefs
about the roles of intention and scene complexity in change detection.
Consciousness and Cognition, 16, 31-51.

Bell, B. S., Kanar, A. M., & Kozlowski, S. W. J. (2008). Current issues and future
directions in simulation-based training in North America. The International
Journal of Human Resource Management, I 9, 1416-1434.

Bell, B. S. & Kozlowski, S. W. J. (2008). Active learning: Effects of core training design

elements on self-regulatory processes, learning, and adaptability. Journal of
Applied Psychology, 93, 296-316.

Bell, B. S. & Kozlowski, S. W. J. (in press). Toward a theory of learner centered training
design: An integrative framework of active learning. In S. W. J. Kozlowski & E.
Salas (Eds.), Learning, Training, and Development in Organizations.

Burke, C. S., Stagl, K. C., Salas, E., Pierce, L., & Kendall, D. (2006). Understanding

team adaptation: A conceptual analysis and model. Journal of Applied
Psychology, 9] , 1 189-1207.

176

Campbell, D. J. (1988). Task complexity: A review and analysis. The Academy of
Management Review, 13, 40-52.

Campbell, J. P., McCloy, R. A., Oppler, S. H., & Sager, C. E. (1993). A theory of
performance. In N. Schmitt & W. C. Borrnan (Eds), Personnel selection in
organizations (p.35-70). San Francisco: Jossey-Bass.

Carver, C. S. (2004). Self-regulation of action and affect. In R.F. Baumeister &. K.D.
Vohs (Eds), Handbook of Self-Regulation: Research, Theory and Applications
(p.13-39). New York: The Guilford Press.

Chan, D. (1996). Criterion and construct validation of an assessment center. Journal of
Occupational and Organizational Psychology, 69, 167-181.

Chan, D. (2000). Understanding adaptation to change in the work environment:
Integrating individual difference and Ieaming perspectives. Research in Personnel
and Human Resources Management, 18, 1—42.

Chen, G., Thomas, B. & Wallace, J. C. (2005). A multilevel examination of the
relationships among training outcomes, mediating regulatory processes, and
adaptive performance. Journal of Applied Psychology, 90, 827-841.

Costa, P. T., Jr. & McCrae, R. R. (1992). The Revised NEO Personality Inventory
(N EO-PI-R) and NBC Five-Factor Inventory (N EO-FFI) Professional Manual.
Psychological Assessment Resources, Odessa, FL.

Csikszentmihalyi, Mihaly (1990). F low: The Psychology of Optimal Experience. New
York: Harper and Row.

Dweck, C. S. (1986). Motivational processes affecting Ieaming. American Psychologist,
41, 1040-1048.

Eysenck, M. W., Derakshan, N., Santos, R. & Calvo, M. G. (2007). Anxiety and
cognitive performance: Attentional control theory. Emotion, 7, 336-3 53.

Ford, J. D. (1985). The effects of causal attributions on decision makers’ responses to
performance downturns. The Academy of Management Review, 10, 770-786.

Ford, J. D. & Baucus, D. A. (1987). Organizational adaptation to performance
downturns: An interpretation-based perspective. The Academy of Management
Review, 12, 366-380.

Ford, J. K., Smith, E. M., Weissbein, D. A., Gully, S. M., & Salas, E. (1998).

Relationships of goal orientation, metacognitive activity, and practice strategies
with Ieaming outcomes and transfer. Journal of Applied Psychology, 83, 218-233.

177

Fowler, H. (1965). Curiosity and Exploratory Behavior. The MacMillan Company: New
York.

Garcia, R., Calantone, R. & Levine, R. (2003). The role of knowledge in resource
allocation to exploration versus exploitation in technologically oriented
organizations. Decision Sciences, 34, 323-349.

Georgesdottir, A. S. & Getz, I. (2004). How ﬂexibility facilitates innovation and ways to
manage it in organizations. Creativity and Innovation Management, 1 3, 166-175.

Gully, S. M., Payne, S. C., Koles, K. L. K, & Whiteman, J. K. (2002). The impact of
error training and individual differences on training outcomes: An attribute-
treatment interaction perspective. Journal of Applied Psychology, 87, 143-155.

Hollenbeck, J. R., Ilgen, D. R., Sego, D. J ., Hedlund, J., Major, D. A. & Phillips, J.
(1995). Multilevel theory of team decision making: Decision performance in

teams incorporating distributed expertise. Journal of Applied Psychology, 80, 292-
316.

Hunter, J. E., & Hunter, R. F. (1984). Validity and utility of alternate predictors of
performance. Psychological Bulletin, 96, 72—98.

Hyland, M. E. (1987). Control theory interpretation of psychological mechanisms of
depression: Comparison and integration of several theories. Psychological
Bulletin, 102, 109-121.

Hyland, M. E. (1988). Motivational control theory: An integrative framework. Journal of
Personality and Social Psychology, 55, 642-651.

Ilgen, D. R. & Pulakos, E. D. (1999). Employee performance in today’s organizations. In
D.R. Ilgen & E.D. Pulakos (Eds), The changing nature of work performance:

Implications for staﬂing, motivation, and development (pp. 1-20). San Francisco:
Jossey-Bass.

International Personality Item Pool: A Scientiﬁc Collaboratory for the Development of
Advanced Measures of Personality Traits and Other Individual Differences
(http://ipip.ori.org/). lntemet Web Site.

Ivancic, K. & Hesketh, B. (2000). Learning from errors in a driving simulation: Effects
on driving skill and self-conﬁdence. Ergonomics, 43, 1966-1984.

Judge, T. A., Jackson, C. L., Shaw, J. C., Scott, B. A. & Rich, B. L. (2007). Self-efﬁcacy

and work-related performance: The integral role of individual differences. Journal
of Applied Psychology, 92, 107-127.

178

Kaehlbling, L. P., Littman, M. L. & Moore, A. W. (1996). Reinforcement Ieaming: A
survey. Journal of Artiﬁcial Intelligence Research, 4, 237-285.

Kahneman, D. & Tversky, A. (1979). Prospect theory: An analysis of decision under risk.
Econometrica, 47, 263-292.

Kanfer, R. & Ackerman, P. L. (1989). Motivation and cognitive abilities: An

integrative/aptitude-treatment interaction approach to skill acquisition. Journal of
Applied Psychology, 74, 657-690.

Kanfer, R. & Ackerman, P. L. (1996). A self-regulatory skills perspective to reducing
cognitive interference. In LG. Sarason, B.R. Sarason, & G.R. Pierce (Eds),
Cognitive interference: Theories, methods, and ﬁndings (p. 153-171). Mahwah,
NJ: Lawrence Erlbaum Associates, Inc.

Kanfer, R., Ackerman, P. L., Murtha, T. C., Dugdale, B. & Nelson, L. (1994). Goal
setting, conditions of practice, and task performance: A resource allocation
perspective. Journal of Applied Psychology, 79, 826-835.

Kozlowski, S. W. J. (2008). An active learning approach for the development of soldier
cognitive competencies. White Paper prepared for the Consortium Research
Fellows Program.

Kozlowski, S. W. J. & Bell, B. S. (2008). Team Ieaming, development, and adaptation. In
V. I. Sessa & M. London (Eds), Group learning (pp. 15-44). Mahwah, NJ: LEA.

Kozlowski, S. W. J ., Gully, S. M, Brown, K. G., Salas, E., Smith, E. M., & Nason, E. R.
(2001 ). Effects of training goals and goal orientation traits on multidimensional
training outcomes and performance adaptability. Organizational Behavior and
Human Decision Processes, 85, 1-31.

Kozlowski, S. W. J ., Gully, S. M., Nason, E. R., & Smith, E. M. (1999). Developing
adaptive teams: A theory of compilation and performance across levels and time.
In D. R. Ilgen & E. D. Pulakos (Eds), The changing nature of work performance:
Implications for staffing, personnel actions, and development (pp. 240-292). San
Francisco: Jossey-Bass.

Kozlowski, S. W. J., Watola, D. J ., Nowakowski, J. M., Kim, B. H., & Botero, I. C. (in
press). Developing adaptive teams: A theory of dynamic leadership. In E. Salas,
G. F. Goodwin, & C. S. Burke (Eds), Team eﬂectiveness in complex

organizations: Cross-disciplinary perspectives and approaches (SIOP Frontier
Series). Mahwah, NJ: LEA. ’

Lazarus, R. & Folkman, S. (1984). Stress, appraisal, and coping. New York: Springer.

179

LePine, J. A., Colquitt, J. A., & Erez, A. (2000). Adaptability to changing task contexts:
Effects of general cognitive ability, conscientiousness, and openness to
experience. Personnel Psychology, 53, 563-593.

LePine, J. A. (2003). Team adaptation and postchange performance: Effects of team
composition in terms of members’ cognitive ability and personality. Journal of
Applied Psychology, 88, 27-39.

LePine, J. A. (2005). Adaptation of teams in response to unforeseen change: Effects of
goal difﬁculty and team composition in terms of cognitive ability and goal
orientation. Journal of Applied Psychology, 90, 1153-1167.

Levin, D. T. & Simons, D. J. (1997). Failure to detect changes to attended objects in
motion pictures. Psychonomic Bulletin & Review, 4, 501-506.

Lewin, K. (1951). Field theory in social science. New York: Harper & Row.

Linderrnan, K., Schroeder, R. G., Zaheer, S. & Choo, A. S. (2003). Six sigma: A goal-
theoretic perspective. Journal of Operations Management, 21, 193-203.

Locke, E. A., & Latham, G. P. (2002). Building a practically useful theory of goal setting
and task motivation: A 35-year odyssey. American Psychologist, 5 7, 705-717.

Maner, J. K. & Gerend, M. A. (2007). Motivationally selective risk judgments: Do fear
and curiosity boost the boons or the banes? Organizational Behavior and Human
Decision Processes, 103, 256-267.

March, J. & Simon, H. (1958). Organizations. New York: Wiley.
Marks, M. A., Zaccaro, S. J ., & Mathieu, J. E. (2000). Performance implications of leader
brieﬁngs and tearn-interaction training for team adaptation to novel environments.

Journal of Applied Psychology, 85, 971-986.

Miles R, Snow C. 1978. Organizational Strategy, Structure, and Process. McGraw-Hill:
New York.

Mitchell, V. & Walsh, G. (2006). Gender differences in German consumer decision-
making styles. Journal of Consumer Behaviour, 3, 331-346.

Newell, A. & Simon, H. A. (1972). Human Problem Solving. Englewood Cliffs, NJ:
Prentice-Hall.

Payne, S. C., Youngcourt, S. S. & Beaubien, J. M. (2007). A meta-analytic examination

of the goal orientation nomological net. Journal of Applied Psychology, 92, 128-
150.

180

Ployhart, R. E. & Bliese, P. D. (2006). Individual ADAPTability (I-ADAPT) Theory:
Conceptualizing the antecedents, consequences, and measurement of individual
differences in adaptability. In S. Burke, L. Pierce, & E. Salas (Eds),
Understanding Adaptability: A Prerequisite for Effective Performance within
Complex Environments. Elsevier Science.

Porter, G. & Tansky, J. W. (1999). Expatriate success may depend on a “learning
orientation”: Considerations for selection and training. Human Resource
Management, 38, 47-60.

Pulakos, E. D., Arad, S., Donovan, M. A., & Plamondon, K. E. (2000). Adaptability in
the workplace: Development of taxonomy of adaptive performance. Journal of
Applied Psychology, 85 , 612-624.

Pulakos, E. D., Schmitt, N., Dorsey, D. W., Arad, 8., Hedge, J. W., & Borman, W. C.
(2002). Predicting adaptive performance: Further tests of a model of adaptability.
Human Performance, 15, 299-323.

Rosenthal, R. & Rubin, D. B. (2003). requivalenti A simple effect size indicator.
Psychological Methods, 8, 492-496.

Rosin, P. L. (1997). Thresholding for change detection.
(httpz/Amvw.bmva.ac.uk/bmvc/ l 997/papers/007/thresh2.html). Internet Web Site.

Siegler, R. S. & Lemaire, P. (1997). Older and younger adults’ strategy choices in
multiplication: Testing predictions of ASCM using the choice/no-choice method.
Journal of Experimental Psychology: General, 126, 71-92.

Smilek, D., Eastwood, J. D., Reynolds, M. G., & Kingstone, A. (2007). Metacognitive
errors in change detection: Missing the gap between lab and life. Consciousness
and Cognition, 16, 52-57.

Smith, E. M., Ford, J. K., & Kozlowski, S. W. J. (1997). Building adaptive expertise:
Implications for training design. In M.A. Quinones & Ehrenstein (Eds), Training
for the 21“" century technology: Applications of psychological research (pp. 89-
118). Washington, DC. American Psychological Association.

Spector, P. E. (1988). Development of the work locus of control scale. Journal of
Occupational Psychology, 61, 335-340.

Stajkovik, A. D. & Luthans, F. (1998). Self-efﬁcacy and work-related performance: A
meta-analysis. Psychological Bulletin, 124, 240-161.

Staw, B. M., Sandelands, L. E. & Dutton, J. E. (1981). Threat rigidity effects in

organizational behavior: A multilevel analysis. Administrative Science Quarterly,
26, 501-524.

181

Stewart, G. L. & Nandkeolyar, A. K. (2006). Adaptation and intraindividual variation in
sales outcomes: Exploring the interactive effects of personality and environmental
opportunity. Personnel Psychology, 59, 307-332.

Terborg, J. & Miller, H. (1978). Motivation, behavior and performance: A closer
examination of goal-setting and monetary incentives. Journal of Applied
Psychology, 63, 29-39.

Vandewalle, D. (1997). Development and validation of a work domain goal orientation
instrument. Educational and Psychological Measurement, 5 7, 995-1015.

Weiner, B. (1985). An attributional theory of achievement motivation and emotion.
Psychological Review, 92, 548-573.

Weiss, H. M. & Cropanzano, R. (1996). Affective events theory: A theoretical discussion
of the structure, causes and consequences of affective experiences at work.

Research in Organizational Behavior, 18, 1-74.

Wood, R. E. (1986). Task complexity: Deﬁnition of the construct. Organizational
Behavior and Human Decision Processes, 36, 60-82.

182

   

”'illllllllllnlllillillllﬂill“