 

LIBRARY

Michigan State
University

 

 

 

PLACE N RETURN 30X to roman this checkout from your record.

. TO AVOID FINES mum on at bdm dd. duo.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU I. An Afﬂmatlw MONEqud Oppommy Imam
7 7 Wm:

 

THE EFFECTS OF INDIVIDUAL DIFFERENCES, DISCOVERY LEARNING,
AND METACOGNITION ON LEARNING AND ADAPTIVE TRANSFER

By

Eleanor Marie Smith

A DISSERTATION
Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Psychology

1996

ABSTRACT

THE EFFECTS OF INDIVIDUAL DIFFERENCES, DISCOVERY LEARNING,
AND METACOGNITION ON LEARNING AND ADAPTIVE TRANSFER

By

Eleanor Marie Smith

[/0 psychologists have asserted that training research should incorporate concepts and
principles from cognitive psychology into the design of training environments to facilitate
transfer of training (Baldwin & Ford, 1988; Howell & Cooke, 1989; Tannenbaum & Yukl,
1992). This study focused on a particular type of transfer, adaptive transfer, deﬁned as the
extent to which individuals can adjust the knowledge and skills learned in training to novel
task demands. Two training interventions, derived from cognitive and instructional research,
were hypothesized to lead to greater ﬂexibility and adaptability of trained skills -- providing
opportunities for discovery learning; and metacognitive instruction. Tolerance for ambiguity
and mastery orientation were identiﬁed as individual differences that would inﬂuence learning
and transfer. Multiple learning outcomes were examined as intervening mechanisms in the
relationships between training interventions, individual differences, and adaptive transfer.

One hundred sixty-one undergraduate students participated in the study, learning how
to perform a complex, computer simulation of a radar tracking task. Results from a series of
hierarchical regression analyses provided support for limited portions of the conceptual model.
The discovery learning manipulation and metacognitive instruction interacted to inﬂuence
verbal knowledge and adaptive transfer. The guided discovery manipulation and a mastery
orientation to learning led to greater hypothesis-testing and self-regulation during training.
Mastery orientation and tolerance for ambiguity were related to self-efﬁcacy at the end of

training. Limitations and directions for future research are discussed.

ACKNOWLEDGEMENTS

I would like to thank my dissertation chair, Kevin Ford, for the helpful advice
and support he provided to me as I developed my ideas about training for adaptive
transfer. I would also like to thank my committee members, Richard DeShon, Steve
Kozlowski, and Neal Schmitt. Their insights and recommendations improved my
dissertation signiﬁcantly.

I thank the faculty for always challenging me in my class and project work. I
am particularly grateful to both Kevin Ford and Steve Kozlowski for mentoring me
during my graduate education at Michigan State. I have enjoyed and learned so much
from the research and consulting opportunities they provided.

I would also like to thank my fellow graduate students who helped me to see
the light at the end of the tunnel when graduate school became tough. My early years
in graduate school were much more enjoyable because I shared them with Stan, Dave,
Dennis, and Jen. Thanks also to Dan, Ken, Jean, and the rest of the graduate students
I worked with who made our research and consulting projects fun.

I am grateful to my family for their support throughout my graduate studies. I
would especially like to thank Sean for his love and encouragement, and for always

believing that I could succeed at my goals.

iii

TABLE OF CONTENTS

Page

LIST OF TABLES ............................................ vii
LIST OF FIGURES ........................................... ix
INTRODUCTION ............................................. l
TRANSFER OF TRAINING ...................................... 4
Types of Transfer ......................................... 8

Near and Adaptive Transfer .................................. 9
Training for Near and Adaptive Transfer ........................ 13
Discovery Learning ....................................... 18
Current research .................................... 21

Summary and limitations .............................. 27
Metacognition .......................................... 32
Research on metacognition ............................. 35

Research on interventions to promote metacognition ........... 39

Summary and limitations .............................. 44

Individual Differences and Learning ........................... 47
Research on individual differences and discovery learning ....... 48

Individual differences and I/O training research .............. 50

TRAINING FOR ADAPTIVE TRANSFER: A CONCEPTUAL MODEL ...... 52
Discovery Learning, Guided Discovery, and Procedural Instruction ...... 59
Metacognitive Instruction ................................... 61
Learning Activity ........................................ 65
Verbal Knowledge and Knowledge Structure ..................... 68
Self-Efﬁcacy ........................................... 75
Individual Difference Factors ................................ 78
Cognitive ability .................................... 78

Goal orientation factors ............................... 79

Tolerance for ambiguity .............................. 82

METHOD .................................................. 89
Sample ............................................... 89
Design ............................................... 89

Task ................................................. 90
Procedure ............................................. 91

iv

Discovery Learning Manipulation ............................. 94

Metacognitive Instruction ................................... 95

Measures .............................................. 96

Gender .......................................... 97

Age ............................................ 97

Video game experience ............................... 97

Cognitive ability .................................... 97

Tolerance for ambiguity ............................. 103

Mastery orientation ................................. 103

Performance orientation .............................. 103

Hypothesis-testing/self-regulatory activity .................. 106

Self-efﬁcacy ..................................... 106

Verbal knowledge .................................. 107

Knowledge structure ................................ 107

Adaptive transfer .................................. 108

RESULTS ................................................. 110

Data Analysis ......................................... 110

Verbal knowledge .................................. 1 1 1

Knowledge structure ................................ 113

Hypothesis-testing/self-regulatory activity .................. 115

Self-efﬁcacy ..................................... 1 17

Adaptive transfer .................................. 118

Follow-up analyses for verbal knowledge .................. 127

Tests for mediation ................................. 128

Demographic factors ................................ 133

Summary of results ................................. 147

DISCUSSION .............................................. 150

Limitations and Directions for Future Research ................... 160

Implications for Practice .................................. 163

APPENDIX A: Consent Form ................................... 165
APPENDIX B: Learning a Computerized Radar Simulation

Individual Training Manual ........................... 166

APPENDIX C: General Task Instructions ........................... 172

APPENDIX D: Discovery Learning Instructions ...................... 175

APPENDIX E: Metacognitive Instructions .......................... 182

APPENDIX F: Demographics ................................... 185

APPENDIX G: Tolerance for Ambiguity (adapted from Major, 1990) ........ 186

APPENDIX H: Mastery Orientation and Performance Orientation

(Button, Mathieu, & Zajac, 1995) ....................... 187
APPENDIX I: Hypothesis-Testing/Self-Regulatory Activity ............... 188
APPENDIX J: Self-Efﬁcacy .................................... 189
APPENDIX K: Verbal Knowledge Test ............................ 190
APPENDIX L: Knowledge Structure .............................. 194
APPENDIX M: Transfer Task Instructions .......................... 196
LIST OF REFERENCES ....................................... 197

vi

LIST OF TABLES

Page
Table l - Summary of Study Hypotheses ............................. 88
Table 2 - Means, Standard Deviations, Reliabilities, and Intercorrelations ....... 98
Table 3 - Rotated Factor Matrix for Individual Differences Items ........... 101

Table 4 - Rotated Factor Matrix for Learning Activity and Self-Efﬁcacy Items . . 105

Table 5 - Hierarchical Regression Analysis Results for Verbal Knowledge ..... 112
Table 6 - Hierarchical Regression Analysis Results for Knowledge Structure . . . . 114
Table 7 - Hierarchical Regression Analysis Results for
Hypothesis-Testing/Self-Regulation ........................... 1 16
Table 8 - Hierarchical Regression Analysis Results for Self-Efﬁcacy ......... 119
Table 9 - Hierarchical Regression Analysis Results for Adaptive Transfer ...... 121

Table 10 - Hierarchical Regression Analysis Results for Prioritization of Targets . 125
Table 11 - F ollow-Up Regression Analysis for Verbal Knowledge ........... 129

Table 12 - Hierarchical Regression Analysis Results for Adaptive Transfer
(Mediation Analysis) ..................................... 132

Table 13 - Hierarchical Regression Analysis Results for Prioritization of Targets
(Mediation Analysis) ..................................... 134

Table 14 - Hierarchical Regression Analysis Results for Verbal Knowledge
(Controlling for Demographic Factors) ........................ 136

Table 15 - Hierarchical Regression Analysis Results for Knowledge Structure
(Controlling for Demographic Factors) ........................ 138

vii

Table 16 - Hierarchical Regression Analysis Results for Hypothesis-Testing/
Self-Regulation (Controlling for Demographic Factors) ............. 139

Table 17 - Hierarchical Regression Analysis Results for Self-Efﬁcacy
(Controlling for Demographic Factors) ........................ 141

Table 18 - Hierarchical Regression Analysis Results for Adaptive Transfer
(Controlling for Demographic Factors) ........................ 142

Table 19 - Hierarchical Regression Analysis Results for Prioritization of Targets
(Controlling for Demographic Factors) ........................ 145

viii

LIST OF FIGURES

Figure 1 - A Conceptual Heuristic of the Impact of Individual Differences,
Discovery Learning, and Metacognitive Instruction on Learning
and Adaptive Transfer ...............................

Figure 2 - Predicted Cell Means for the Interactive Inﬂuence of Discovery
Learning and Metacognitive Instruction on Adaptive Transfer . . . .

Figure 3 - Predicted Cell Means for the Interactive Inﬂuence of Discovery
Learning and Metacognitive Instruction on Prioritization of Targets

Figure 4 - Predicted Cell Means for the Interactive Inﬂuence of Discovery
Learning and Metacognitive Instruction on Verbal Knowledge . . .

Figure 5 - Predicted Cell Means for the Interactive Inﬂuence of Discovery
Learning and Metacognitive Instruction on Verbal Knowledge
(Controlling for Demographic Factors) ...................

Figure 6 - Predicted Cell Means for the Interactive Inﬂuence of Discovery
Learning and Metacognitive Instruction on Adaptive Transfer
(Controlling for Demographic Factors) ...................

Figure 7 - Predicted Cell Means for the Interactive Inﬂuence of Discovery

Learning and Metacognitive Instruction on Prioritization of Targets
(Controlling for Demographic Factors) ...................

ix

Page

...... 57

..... 123

INTRODUCTION

The success of any training program depends on the extent to which knowledge
and skills are transferred back to the job (Baldwin & Ford, 1988; Goldstein, 1993;
Wexley & Latham, 1981). Baldwin and Ford (1988) reviewed research on transfer of
training and identiﬁed three factors that affect the extent to which transfer will occur:
trainee characteristics; training design; and the work environment. With regard to the
work environment, transfer is more likely to occur if organizational members are
supportive of the training and individual’s attempts to transfer, and if there are
sufﬁcient opportunities to use knowledge and skills back on the job. Recent studies
have examined the role of work environment factors such as the climate for transfer
(Rouillier & Goldstein, 1990) and the opportunity to perform trained skills (Ford,
Quinones, Sego, & Sorra, 1992).

In contrast, current research on transfer of training has paid less attention to the
learning principles incorporated into training programs to facilitate transfer back to the
job. Baldwin and Ford ( 1988) argued that learning principles to facilitate transfer have
come from a behavioral perspective on learning. Training design research has
emphasized observable behavior and its relation to environmental stimuli and
reinforcements (Howell & Cooke, 1989). Training principles derived from a

behaviorist tradition are relevant when individuals learn highly structured tasks that

2

require routine and stable responses. However, the tasks that people will be trained for
today and in the future impose greater cognitive demands in terms of inference,
diagnosis, and decision-making (Goldstein & Gilliam, 1990; Howell & Cooke, 1989).
Transfer of training requires not only the mimicking of trained skills on the job, but
also the adaptation of trained skills to different situations (Baldwin & Ford, 1988).
Research from the instructional and educational literatures identify types of
training transfer that differ in the extent to which adaptation of knowledge and skills is
required (Royer, 1979). Research has also suggested different design principles to
achieve each type of transfer (Clark & Voogel, 1985; Salomon & Perkins, 1989). In
the present study, adaptive transfer is identiﬁed as a critical type of transfer for
organizational training. Trainees should not only transfer skills to problems practiced
during training, they should also be able to adapt the knowledge and skills learned to
novel problems they may face on the job. The instructional and educational
psychology literatures suggest training methods that will facilitate the adaptive transfer
of skills. First, research suggests that allowing for discovery learning during
instruction will facilitate an individual’s capability to adapt skills to a novel transfer
task (McDaniel & Schlager, 1990). Second, researchers have noted that metacognitive
skills are important for the capability to adapt to novel tasks (Salomon & Perkins,
1989). Metacognitive instruction that provides individuals with skills in self-
monitoring and self-evaluating is suggested as a method to increase adaptive transfer.
Research that draws on these perspectives to training design will answer, in part, calls

to introduce a cognitive perspective into the training domain (Ford & Kraiger, 1995;

Howell & Cooke, 1989).

3

Baldwin and Ford (1988) noted a second limitation of training design research
in that it has failed to consider the effects of both trainee characteristics and the type
of training method used. Educational psychologists have long considered the role of
individual differences in learning and transfer (Cronbach & Snow, 1977). 1/0 training
research must also consider whether there are individual factors that inﬂuence learning
and transfer. Two factors relevant for learning and adaptive transfer identiﬁed in this
study are tolerance for ambiguity (Budner, 1962) and the goal orientation of the
learner (Dweck, 1986).

The focus of the present study is to advance research on training design by
considering design principles from a cognitive perspective on learning and transfer. In
addition, a speciﬁc type of transfer is examined in this study. Adaptive transfer is
identiﬁed as having particular relevance in present organizational contexts. Several
researchers have noted that technology has increased dramatically and thus requires
individuals to become ﬂexible learners who are willing to update their skills (Hesketh
& Bochner, 1994; Howell & Cooke, 1989). Finally, this study examines individual
differences in learners that may also inﬂuence what they will learn and transfer.

First, theoretical perspectives on transfer of training will be reviewed,
highlighting the contributions of cognitive psychology and the differentiation of types
of transfer. Second, adaptive transfer will be deﬁned, and design principles to
facilitate adaptive transfer will be identiﬁed. Third, research on learning and
individual differences will be discussed, and individual factors important for adaptive
transfer will be discussed. Finally, a conceptual model will be presented that

integrates and extends research on how to train for adaptive transfer.

TRANSFER OF TRAINING

Transfer of training is deﬁned as the extent to which individuals apply the
knowledge, skills, and abilities developed in training to the job (Baldwin & Ford,
1988; Goldstein, 1993). Initial conceptualizations of transfer can be traced to the
doctrine of "formal discipline" that was favored in the nineteenth century (Patrick,
1992). It was believed that there were general mental faculties such as reasoning and
memory skills that could be exercised through training in certain curricula such as
math and Latin. These mental faculties could transfer to a wide variety of tasks that
used these types of skills.

Thorndike and Woodworth (1901) reacted against this theory and instead
proposed a theory of identical elements to explain transfer. According to this theory,
the greater the amount of identical elements shared between two tasks, the greater the
transfer. Thorndike and Woodworth (1901) were not very clear in deﬁning what these
identical elements were, and behaviorists interpreted these identical elements to be
stimuli and responses (Butterﬁeld & Nelson, 1989; Patrick, 1992). In order for
transfer (i.e., the same response) to occur, the training and transfer environments had
to share identical stimulus elements. Behaviorists identiﬁed training principles that
were focused on how to shape the behavior of individuals during learning. These

principles concerned how to structure the learning environment through methods such

5

as reinforcement schedules, practice schedules, overlearning, stimulus variability, and
feedback. Through practice of the correct response and reinforcement of the response,
stimulus control of the response was said to occur (Salisbury, Richards, & Klein,
1985). The major educational approach used in current training programs is based on
these behaviorist principles (Howell & Cooke, 1989). In addition, most research in
1/0 psychology on the impact of the learning environment on transfer of training has
focused on these behaviorist learning principles (Baldwin & Ford, 1988; Weiss, 1990).

A second theory of transfer has been referred to as transfer through principles
(Goldstein, 1993; Patrick, 1992). Experiments by Judd (1908) and Hendrickson and
Schroeder (1941) demonstrated that teaching individuals theoretical principles
underlying skills facilitated transfer to a task that differed in stimulus features.
According to the theory of transfer through principles, instruction on theoretical
principles allows individuals to understand the rules underlying behavior and how to
apply the behavior to a transfer task that has different requirements. These studies
expanded research on transfer in two ways. First, these studies demonstrated that an
individual’s cognitive understanding of a skill, not just its behavioral execution and
reinforcement, is important for transfer. Second, these studies showed that transfer of
training can occur to tasks that are not identical to the training task.

A similar perspective developed with the rise of cognitive psychology in the
1960’s. New assumptions were made about the role of the learner during the process
of learning. The main assumptions of cognitive psychology are that mental processes
exist and humans are active information processors (Ashcraft, 1989). Information

processing theories began to identify how individuals attend to, encode, store, and

6

retrieve information. As researchers began studying these mental processes, cognitive
explanations for the theory of identical elements began to emerge. For Anderson
(1993), the common elements shared between training and transfer environments are
production rules. Production rules are If-Then statements specifying the conditions for
which an action is appropriate. Positive transfer between two tasks will occur to the
extent that the two tasks involve the same production rules. Cormier (1987) described
the encoding speciﬁcity principle as important to transfer. Stimulus cues in the
transfer environment must have been encoded with information learned during training
in order for the cues to aid later retrieval of that information during transfer.

Gick and Holyoak (1987) emphasized that it is the perceived similarity between
the training and transfer environments, not necessarily the actual similarity, that will
determine the amount of transfer. They also identiﬁed four determinants of transfer:
the structure of the task; encoding factors; retrieval of knowledge; and the learner’s
background knowledge. Viewing the learner as an active information processor has
led to learning principles focused on enhancing the organization, depth, and richness of
knowledge stored in memory. Cognitive psychologists have used the constructs of
schema, mental models, and knowledge structures to explain how information is stored
in memory. Training methods that link the learner’s existing knowledge to the to-be-
learned material should enhance transfer (Gick & Holyoak, 1987).

Several I/O psychologists have recommended that training research incorporate
principles from cognitive psychology when examining learning and transfer (Ford &
Kraiger, 1995; Goldstein, 1993; Howell & Cooke, 1989). For example, Kraiger, Ford,

and Salas (1993) emphasized that training research and practice must consider

7

evaluation criteria beyond traditional behavioral measures and simple knowledge tests.
They identiﬁed additional cognitive and affective outcomes that may be used to
evaluate the effectiveness of a training intervention. In addition, Ford and Kraiger
(1995) argued that research and practice must redirect attention to understanding the
learning process and the role of the learner during learning and transfer.

Researchers have begun to take a multidimensional perspective to learning by
developing more integrated models of complex skill acquisition. For example, Kanfer
and Ackerrnan (1989) developed and tested a theory of skill acquisition incorporating
individual differences in cognitive ability, motivation, and information-processing
demands. Kozlowski, Gully, Smith, Nason, and Brown (1995) developed and tested a
model of learning for complex tasks. They examined how individual difference factors
and training design interventions impacted outcomes such as declarative knowledge,
knowledge structure, training performance, and self—efﬁcacy. In addition, they
examined how these learning outcomes affected generalization to a more complex
version of the trained task. Smith, Ford, Weissbein, and Gully (1995) examined how
individual differences and learning activities affected multiple learning outcomes and
training transfer. In addition, they examined a learning environment where individuals
were responsible for choosing the exercises they would practice during training. Thus,
training research has begun to draw upon cognitive and instructional research to
develop more integrated models that examine how the learner plays an active role in
the learning process.

However, I/O training research has tended to treat transfer of training as a

unidirnensional construct. Instructional psychologists have made distinctions between

8

different types of transfer. Instructional research has also identiﬁed individual and
instructional factors that lead to these different types of transfer. The I/O training
literature can beneﬁt from instructional research by identifying more precisely the type
of transfer to be gained from training. Then appropriate instructional factors can be
designed into training to promote that type of transfer. The following section
describes various conceptualizations of transfer that have been developed in the
educational and instructional literatures.
Types of Transfer

Royer (1979) has identiﬁed several distinctions that have been made in the
instructional literature regarding transfer. For example, Gagne (1965) differentiates
vertical and lateral transfer. Vertical transfer takes place when knowledge or skill
facilitates the acquisition of superordinate knowledge or skill. Lateral transfer is
deﬁned as the generalization of knowledge or skill across a broad set of situations at a
similar level of complexity. The literature also distinguishes between speciﬁc and
nonspeciﬁc transfer. Speciﬁc transfer occurs when there is a clear similarity between
stimulus elements in the training and transfer tasks. Nonspeciﬁc transfer occurs when
there are no obvious shared stimulus elements in the training and transfer tasks, and is
demonstrated by learning to learn and warm-up effects in laboratory experiments. A
third distinction in the literature is made between literal and ﬁgural transfer. Literal
transfer occurs when knowledge or skill is transferred to a new learning task. Figural
transfer occurs when an individual uses a portion of their knowledge for thinking or
learning about a speciﬁc problem. An example of ﬁgural transfer is the use of a

metaphor or analogy to understand a current problem.

9

A distinction that has the greatest relevance for transfer of training to the job is
that of near v. far transfer. Near transfer occurs when the stimulus complex for the
transfer event is very similar to the stimulus complex for the learning event. In
contrast, far transfer occurs when the stimulus complex for the transfer event is
different from that for the original learning event. In an educational application, for
example, far transfer would involve the transfer of learning in school to real-world
problems or learning situations (Royer, 1979). However, in organizational settings,
one is usually interested in transfer of knowledge and skills back to the job; transfer to
another setting such as to life at home or in a social context is not usually necessary.
Therefore, the focus of the present study will be on a more speciﬁc form of far
transfer, which has been labelled adaptive transfer.

Royer (1979) noted that the previous distinctions between types of transfer are
not mutually exclusive. For example, he suggested that there is considerable overlap
between vertical v. lateral transfer and speciﬁc v. nonspeciﬁc transfer. In the present
study, the notion of adaptive transfer overlaps to some degree with the notion of
nonspeciﬁc transfer as well as the notion of far transfer. A description of adaptive
transfer and its distinction from near transfer is presented next.

Near and Adaptive Transfer

Researchers have noted that the distinction between near and far transfer
requires the concept of similarity, or in opposite terms, distance from the training task
(Butterﬁeld & Nelson, 1989; Salomon & Perkins, 1989). Researchers have also
described far transfer as transfer of training to a novel situation (McDaniel & Schlager,

1990). However, they have noted that it has been difﬁcult to deﬁne similarity,

10

distance, or novelty precisely or quantitatively (Butterﬁeld & Nelson, 1989). Gick and
Holyoak (1987) have identiﬁed two features of tasks that can be used to deﬁne and
understand the difference between near and far transfer. Training and transfer tasks
can differ on two dimensions -- surface components and structural components.
Surface components are features of a task or situation that are not related to
outcomes or goal attainment. Structural components are features of a task or situation
that are causally or functionally related to outcomes or goal attainment (Gick &
Holyoak, 1987). Near transfer can be deﬁned as transfer between two tasks that share
structural components, but that differ in minor ways in terms of surface components.
For example, students learning to apply the same mathematical formula across word
problems with different surface components (e.g., two trains converging v. two cars
converging) would be an example of near transfer. Training that exposes individuals
to the variety of instances for which the knowledge or skill is applicable would lead to
easy retrieval and application of the knowledge or skill to a familiar instance.
Researchers have described far transfer in terms of differences in both surface
features and structural features. When the surface features of two tasks are very
different, individuals have difﬁculty spontaneously transferring what was learned from
one task to the second task. For example, Gick & Holyoak (1980) found that training
on one example where a certain strategy was applicable led to very little transfer to a
second example where the surface features were different. In a subsequent study,
individuals were exposed to two examples where the strategy applied, and then had to

compare the two examples and write a description of their similarity. Gick and

11
Holyoak (1983) found that the quality of the description predicted transfer to a third

example that also varied on surface features.

The conceptualization of far transfer in terms of surface features conforms to
what Patrick (1992) deﬁned as the weakest form of novelty between a training and
transfer task. The weakest form of novelty involves a task that requires the same
method learned in training applied to a new exemplar. This can only be deﬁned as far
transfer if training failed to identify and train for all possible exemplars. However, if
the correct method or strategy is identiﬁed by the trainer, and the range of possible
exemplars is also identiﬁed and trained, then generalization of the method outside of
training becomes a near transfer situation. Individuals exposed to a sufﬁcient variety
of exemplars can quickly recognize that a situation falls within the range of those that
were trained. They will then retrieve and apply the appropriate method.

In contrast, far transfer can also be deﬁned in terms of differences in the
structural features of the training and transfer tasks. Patrick (1992) has identiﬁed two
additional forms of novelty that are examples of changes in structural components. A
stronger form of novelty requires the same methods learned in training but the
methods must be reconﬁgured to handle the novel task. For successful transfer to this
type of task to occur, trainees must learn a variety of methods, as well as learn how to
select and combine them in order to perform the novel task. Finally, the strongest
type of novel task requires different methods to those learned in training. In this case,
trainees must learn to identify when an existing strategy is not sufﬁcient, as well as

how to create a new and more appropriate one.

12

From this perspective, adaptive transfer can be deﬁned as the capability to
adjust knowledge, skills, or methods to successfully handle changes in the structural
features of a task. In this sense, adaptive transfer requires additional learning or
conscious processing on the part of the learner to adjust what is known from training
to what is required in the transfer environment. Adaptive transfer can take several
forms. For example, knowledge or skills learned in training must be combined in a
different way from the method learned in training. Butterﬁeld and Nelson (1991)
provided an example of this, which they labelled inventive transfer. Individuals
learned to combine two pieces of information through addition in the training task, but
the transfer task required them to combine the information multiplicatively. A second
form of adaptive transfer occurs when new knowledge or information is available in
the transfer setting that qualiﬁes what was learned in training. Adaptive transfer
occurs when individuals are successful in integrating this new knowledge into their
existing method or strategy. A third example of adaptive transfer occurs when
knowledge or skills learned in training are not appropriate for the transfer task, and a
new method for performing the task must be learned.

Several researchers have made similar distinctions between near and adaptive
transfer. For example, Hatano and Inagaki (1986) have identiﬁed two types of
expertise, routine and adaptive expertise. Routine expertise involves the development
of procedural knowledge that allows individuals to function effectively in familiar
environments. In contrast, adaptive expertise requires the development of conceptual
knowledge in addition to procedural knowledge. Individuals who become adaptive

experts try to understand more deeply the procedural skills they have developed. This

13

conceptual knowledge may allow individuals to develop new procedures and make
new predictions. Individuals who are adaptive experts are ﬂexible and able to adjust
their skills to new problems or situations (Hatano & Inagaki, 1986).

Similarly, Salomon and Perkins (1989) have differentiated between low-road
transfer and high-road transfer. Low-road transfer occurs when automatic, well-
learned behavior is triggered in a new context. This is consistent with the concepts of
routine expertise and near transfer. In contrast, high-road transfer occurs when an
individual purposefully and mindfully abstracts knowledge or skills from one context
and applies it to another. This is similar to the notion of adaptive transfer or expertise.
In fact, research in cognitive and instructional psychology has begun to identify
training methods that may lead to greater adaptive transfer. These methods are
somewhat different than prescriptions for training for near transfer.

Mg for Near and Adaptive Transfer

Cognitive research on the nature of expert performance has helped to identify
some of the mechanisms for achieving near and adaptive transfer. Holyoak (1991) has
distinguished three generations of theories on expertise. The ﬁrst generation of
theories on expertise are best captured by the work of Newell and Simon (1972).
They were interested in problem-solving as heuristic search strategies that could be
applied across a variety of domains. Thus, they considered skill at general heuristic
search to be the deﬁnition of expertise. However, examination of experts in
knowledge-rich domains such as chess and physics called this deﬁnition into question
(Holyoak, 1991). In addition, research provided little evidence that training in general

problem-solving skills transferred across content domains.

14

Research comparing the performance of experts and novices in particular
domains found that expertise was dependent on detailed domain knowledge (Holyoak,
1991) and the ability to represent and understand problems in terms of deeper,
structural features (Chi, Feltovich, & Glaser, 1981). In contrast to experts, novices
relied on the surface features of problems, and engaged in heuristic search strategies.
In addition, research indicated that expertise involved extensive time invested in
domain learning. Finally, experts were found to store more information, that was
better organized, in long-term memory (Anderson, 1993).

This research led to what Holyoak (1991) termed the second generation of
expertise theories. He cited Anderson’s ACT“ theory as a clear example of the second
generation theories. Because expertise was deﬁned as knowing how to do something
well, these theories focused on studying procedural learning in addition to declarative
knowledge (Holyoak, 1991). According to ACT“ theory, practice at a task leads to
compilation of declarative knowledge into procedural, condition-action rules.
Compilation also leads to larger, integrated chunks of procedural knowledge. With
continued practice, declarative knowledge becomes less accessible. Instead, speeded-
up "rule-ﬁring" leads to automatic and efﬁcient performance (Anderson, 1993;
Holyoak, 1991). Interestingly, this cognitive theory of skill acquisition relies on the
law of exercise that began with Thorndike (Anderson, 1993). Thus, what this theory
provides is a cognitive explanation for the long-standing ﬁnding in behaviorist research
that learning improves with practice. Holyoak (1991) likened this second generation

of expertise to the concept of routine expertise (Hatano & Inagaki, 1986).

15

In a similar vein, Clark and Voogel (1985) argued that the principles for
achieving near transfer come from behaviorist models of training. In general, design
principles from the behaviorist tradition tend to lead to near transfer, but are not
effective for achieving adaptive transfer. For example, behavioral models assume that
instruction is controlled by the instructional design. Methods for instruction typically
include behavioral objectives that guide instruction and evaluation, explicit directing
and monitoring of the learner’s progress, and shorter instructional sessions. In these
instructional sessions, language is standardized and simple, practice to criterion is
encouraged, feedback and reinforcement are provided, and tests immediately follow
instruction (Clark & Voogel, 1985).

Hatano and Inagaki (1986) and Salomon and Perkins (1989) also argue that
there are different mechanisms for achieving near and adaptive transfer. For near
transfer to occur, repeated practice is a critical requirement. Through repeated
experience, individuals learn to perform a skill more quickly and accurately, and
eventually automatically (Hatano & Inagaki, 1986). In addition, this extensive practice
must include sufﬁcient variability in surface features to allow for generalization of
skills to the range of contexts for which they are applicable (Salomon & Perkins,
1989)

Holyoak (1991) described this second generation of theories as a simple picture
of the development of expertise. He noted that theories of expertise must account for
the ability of experts to induce, retrieve, and instantiate schematic knowledge

structures. Serial production systems comprising speciﬁc condition-action rules (e.g.,

16

ACT“) cannot account for the ability of experts to simultaneously integrate multiple
sources of knowledge (Holyoak, 1991).

Holyoak (1991) introduced the notion of a third generation of expert theories
that corresponds to the notion of adaptive expertise (Hatano & Inagaki, 1986).
Although routine experts can solve familiar categories of problems quickly and
accurately, they have difﬁculty with novel problems. In contrast, adaptive experts can
invent new procedures based on their knowledge. The key to their ability to adapt to
novel problems is a deeper conceptual understanding of the target domain.

This means that individuals must not only learn procedural knowledge of what
to do, but also understand why procedures are appropriate for certain conditions.
Clark and Voogel (1985) argued that a cognitive model for instruction tends to lead to
greater adaptive transfer. Using a cognitive approach to design, training methods will
include encouragement of discovery strategies for learning, and will facilitate the
linking of previously acquired, abstract skills through the use of analogies,
paraphrasing, and advance organizers. For example, learners might be encouraged to
create their own organization for material to be learned (Clark & Voogel, 1985).

Similarly, Salomon and Perkins (1989) asserted that adaptive transfer depends
on the process of mindful abstraction -- the intentional, metacognitively guided
decontextualization of knowledge, principles, strategies, or procedures that then can be
transferred to a new setting. They also suggested that the processes for near and
adaptive transfer can occur at the same time; by both practicing skills and reﬂecting on
their execution, individuals not only improve their performance, but are also able to

apply their skills to new situations. Hatano and Inagaki (1986) suggested that the

17

deeper understanding required for adaptive expertise is more likely to occur when
learning involves tasks that are variable and somewhat unpredictable, and learners are
given opportunities to explore the task without explicit external rewards. Salomon and
Perkins (1989) suggested that the processes called upon during active or exploratory
learning are more leamer- than stimulus-controlled, and therefore engage the learner’s
previous knowledge structures in the understanding of the new material. For adaptive
transfer to occur, the abstractions that are learned must be understood at a deeper level,
not just learned as a formula. Active or exploratory learning settings allow individuals
to achieve abstractions through their own efforts, and this should lead to greater
adaptive transfer compared to learning settings where individuals learn abstractions by
rote (Salomon & Perkins, 1989).

In summary, training for adaptive transfer rests on several key learning
mechanisms. First, the learning environment must be one in which the learner is
required to engage in more active and exploratory learning processes (Clark & Voogel,
1985; Hatano & Inagaki, 1986). Providing opportunities to discover task strategies
will promote adaptive transfer for two main reasons. Discovery learning should lead
to a deeper conceptual understanding of the task because a variety of task strategies
may be attempted in trying to develop the appropriate task strategy. The learning
processes involved in more exploratory learning, namely hypothesis-testing and
problem-solving, are capabilities that are in themselves necessary in an adaptive
transfer environment. For example, individuals attempting to transfer trained skills
back to the job are likely to ﬁnd little guidance or assistance in these efforts. In

addition, problems faced on the job may be more ambiguous or ill-structured compared

18

to the examples practiced in training. The development of a deeper understanding of
the task, as well as practice at discovering task strategies should assist individuals
when faced with performing a novel task.

A second factor important for adaptive transfer is the capability to monitor and
regulate one’s comprehension (Holyoak, 1991; Salomon & Perkins, 1989). Salomon
and Perkins (1989) argued that the learning of knowledge and skills must be
metacognitively guided so that these knowledge and skills become abstracted and
decontextualized. This mindful reﬂection and evaluation during learning should lead
to a more complete understanding of the task being learned. This deeper conceptual
knowledge will improve one’s capability to adapt to novel tasks.

Two areas of research conducted in the cognitive, instructional, and educational
literatures provide suggestions for how to incorporate training design factors that
promote adaptive transfer. First, research on the discovery method of instruction has
focused on the beneﬁts of an exploratory approach to learning. Second, research on
metacognition suggests ways to guide learning activities so that individuals are
mindfully monitoring and evaluating the information they are learning. The literatures
on these constructs are reviewed with regard to their implications for adaptive transfer.
Discovegg Learning

Hatano and Inagaki (1986) argued that the opportunity to explore a task is
important for adaptive transfer. In fact, research in educational and instructional
psychology has focused on a method for promoting experimentation, discovery
learning. In discovery learning, individuals must explore and experiment with the task

to infer and learn the rules, principles, and strategies for effective performance. This

19

has often been described and operationalized as an inductive instructional method. In
contrast, explicit training on concepts, principles, or strategies that the individual then
applies to task performance is a deductive instructional approach.

Researchers have described several theoretical reasons for the beneﬁts of
discovery learning. First, individuals in a discovery learning setting are more
motivated because they are responsible for learning, and they are engaging in an active
learning process (Singer & Pease, 1976). In traditional instructional approaches,
individuals are explicitly taught the correct task procedures, and then they practice the
application of these procedures. In contrast, individuals in a discovery learning setting
must explore the task and generate correct task strategies on their own. Discovery
learning settings promote the use hypothesis-testing and problem-solving learning
strategies (McDaniel & Schlager, 1990; Veenman, Elshout, & Busato, 1994).
Individuals who use hypothesis-testing approaches during training will have these
strategies available to them when they face a novel transfer situation and must search
for a new response (McDaniel & Schlager, 1990). Thus, the speciﬁc learning
strategies used in a discovery learning setting may lead individuals to use similar
strategies when adapting to a novel transfer task (Singer & Pease, 1976).

In addition, because individuals must generate optimal methods on their own,
the knowledge they acquire becomes better integrated into their existing knowledge
(Egan & Greeno, 1973; Frese, Albrecht, Altrnann, Lang, Papstein, Peyerl, Prumper,
Schulte-Gocking, Wankmuller, & Wendel, 1988). This knowledge is also more
ﬂexible because the active processing during learning leads to acquisition of

knowledge at a higher level of regulation (Frese & Zapf, 1994). Students who are

20

ﬁrst presented with ambiguous material are given the opportunity to organize it in
terms of concepts developed through their own constructive abilities (Bruner, 1961).

A ﬁnal beneﬁt claimed for discovery learning is related to the type of
knowledge acquired during training. Researchers have argued that discovery learning
allows individuals to make errors and learn from them (Frese et al., 1988; Ivancic &
Hesketh, 1995/1996; Singer & Pease, 1976). While traditional behavioral approaches
have sought to minimize errors or incorrect responses (e.g., Skinner, 1987), some
researchers taking a cognitive perspective to learning have argued that error-making is
beneﬁcial to learning. Individuals learn not only from performing a task correctly but
also from making mistakes. Making mistakes leads to the development of a better
operative mental model of the task (Frese et al., 1988). In addition, the exploration of
dead ends and mistakes may provide individuals with a greater repertoire of strategies
to try out when faced with an adaptive transfer task (McDaniel & Schlager, 1990).

An early review of discovery learning studies indicated that, in general,
discovery learning (inductive learning) leads to greater transfer than rule-example
methods of instruction (deductive approach) (Hermann, 1969). In addition, there was
a slight tendency for discovery learning to be more effective for transfer over longer
time periods, and for transfer tasks that were more complex or novel. The rule-
example method led to greater retention than discovery instruction. However, there
were nonsigniﬁcant results in a number of studies comparing these two instructional
methods. Hermann (1969) noted a number of limitations of the existing literature.
For example, experimental groups had differed on additional factors such as the

amount of time spent learning, the number of examples presented, the meaningfulness

21

of the material, and the amount of activity during learning. An additional limitation
from an I/O training perspective is that this early literature on discovery learning had
focused on either relatively simple tasks such as coding or understanding word
relationships, or on academic tasks such as mathematics and other subjects. It is not
clear from this early literature whether or not a discovery learning method is effective
for more complex cognitive and/or motor skills.

Hermann (1969) noted that one of the key factors in the effectiveness of
discovery learning methods is the amount and type of guidance provided to learners
during instruction. In general, research suggests that a reasonable degree of guidance
is more effective than little guidance during discovery learning. Guidance can include
the following types: providing answers to problems; providing leading questions or
hints to learners either based on individual progress or constant across learners;
varying the size of steps in instruction; and providing prompts without giving solutions
(Hermann, 1969; Kamouri, Kamouri, & Smith, 1986). However, Hermann (1969)
noted that type of guidance had not been systematically examined in the literature.

Current research. Some contemporary research on discovery learning has
examined the theoretical mechanisms underlying the effectiveness of discovery
learning over procedural instruction for transfer of training. For example, Kamouri, et
a1. (1986) compared an exploration-based (discovery) condition, in which individuals
experimented with three analogical training devices, to an instruction condition, in
which individuals were presented with speciﬁc procedures (i.e., rules) to practice on
the devices. The exploration-based discovery condition led to greater transfer to an

analogous, novel transfer device two days later, compared to transfer to a disanalogous

22

device. This beneﬁt was not found for the instruction condition. In addition, the
discovery learning group reported greater similarity between the training devices and
analogous transfer device compared to the instruction group. Results suggested that
the exploration-based training promoted the use of analogical reasoning and facilitated
the induction of abstract device schema (Karnouri et al., 1986).

Other researchers have examined the role of errors during training. For
example, Prather (1971) compared trial-and-error training v. errorless training on a
range estimation task. In the trial-and-error learning condition, individuals had to
press the trigger when they estimated that a simulated target was at the correct open-
ﬁre range. In contrast, the errorless learning condition required individuals to press
the trigger when a green light came on. This light came on at precisely the correct
open-ﬁre range. Results indicated that trial-and-error learning led to greater transfer of
training to a photograph of a MIG-21 at various ranges.

Frese, Brodbeck, Heinbokel, Mooser, Schleiffenbaum, and Thiemann (1991)
examined error-free v. error-training conditions for learning word processing software.
The error-free group received written instructions on each step for the commands used
for solving each word processing task. In contrast, the error-training group did not
receive detailed instructions for each task. Instead, individuals in this group had to
discover which were the appropriate steps and commands to solve the tasks. Error-
training participants were provided a leaﬂet with all the necessary commands, and
were presented with heuristics on an overhead screen to counter the frustrating effects
of making errors (e.g., "I have made an error. Great!" and "There is a way to leave

the error situation").

23

Results indicated that the error-training group recalled signiﬁcantly more
commands than the error-free group. While the two groups were equally competent at
easy and medium difﬁculty tasks, the error-training group was signiﬁcantly more
competent at handling difﬁcult tasks. Surprisingly, the two groups did not differ on a
transfer task requiring them to solve three tasks not taught during training. Finally, the
correlation between the number of keystrokes and the performance rating on the
transfer task was higher in the error-training group than the error—avoidant group. The
authors suggest that the error-training group learned to explore the task (i.e., more
keystrokes), and those who explored the task also performed better (F rese et al., 1991).

Other research has been focused on the notion of guided discovery learning,
and various conceptualizations of guided discovery have been compared to pure
discovery and procedural instruction conditions. Guidance has been operationalized in
some studies as the degree of cues or prompting that are given to individuals in terms
of the correct task procedure. For example, in teaching a serial manipulation task,
Singer and Pease (1976) included a combination condition of prompted instruction and
discovery learning. In the ﬁrst four practice trials, cues appeared on a screen to
indicate each correct move in the sequence. Then for the remaining sixteen trials, the
learners had no cues available and had to practice the correct sequence without
prompts. This combination condition was compared to a pure discovery condition, in
which participants had to discover the correct move sequence, and a prompted
instructional condition, in which participants were cued on the correct moves for all
training trials. Dependent variables were measured as the time to achieve the correct

task procedure. Participants in the prompted instruction and combination instruction

24

conditions outperformed the discovery condition during the learning of the task.
However, the discovery and combination conditions performed better than the
prompted group on a retention test, and initial trials of a novel transfer task on the
following day (Singer & Pease, 1976).

Guidance has also been operationalized in terms of reducing the total variability
of possible responses. For instruction on a mirror tossing task, Greenockle and Lee
(1991) compared single guidance, variable guidance, and discovery conditions. The
single guidance condition allowed participants to practice throwing a volleyball to
rebound off a circle marked on a wall in order to hit a target on the ﬂoor hidden by a
partition. This circle was in the correct position for the ball to hit the ﬂoor target.
The variable guidance condition allowed participants to practice throwing the
volleyball at three circles at different locations on the wall. The discovery group had
no targets marked on the wall and they had to ﬁnd the correct area on the wall in
order to hit the ﬂoor target accurately.

During acquisition, participants in the single-guidance group performed
signiﬁcantly better than the discovery group, with the variable-guidance group falling
in between the other two. Participants were tested on a transfer task that required
them to throw the volleyball at the same target but from a greater distance. No
signiﬁcant difference for groups or for the groups X trial interaction was found on the
transfer task, but follow-up comparisons indicated that the variable-guidance group
outperformed the discovery group on initial transfer trials (Greenockle & Lee, 1991).

Researchers have also examined discovery learning of problem-solving tasks.

In these tasks, researchers have operationalized a guided discovery learning condition

25

as one in which the general strategy for accomplishing the task is known, but where
the speciﬁc moves or routines for accomplishing the strategy must be discovered
(Carlson, Lundy, & Schneider, 1992; McDaniel & Schlager, 1990). For example,
McDaniel and Schlager (1990) conducted an experiment to teach individuals how to
solve water-jar problems. These are transformation problems where individuals must
learn how to obtain a speciﬁc amount of water using a number of jars of different
sizes. The complete discovery condition required individuals to ﬁgure out both the
general strategy and the speciﬁc move sequence. The guided discovery condition
provided individuals with the general strategy and they had to discover the speciﬁc
move sequence. An example of a general strategy is 1 - 2 + 1, which indicates that
one jar must be ﬁlled with water, and then two of another jar subtracted and one of a
third jar added. For far transfer to a task requiring a different general strategy than
those learned in training, the pure discovery condition led to greater transfer (measured
by time to solution) than the guided discovery condition.

Carlson, Lundy, & Schneider (1992) examined four instructional conditions for
learning how to troubleshoot a simulated information network. They examined the
impact of these conditions on transfer to problems containing more complex
information networks. A discovery condition provided no guidance to learners. A
variable template condition allowed individuals to choose from a number of possibly
correct moves at each step. A ﬁxed template condition only allowed individuals to
choose the one correct move to make at each step. Finally, a procedural instruction
condition provided learners with instruction on the algorithm they should use to

troubleshoot the network. In addition, Carlson et a1. (1992) examined whether

26

guidance in the form of memory aiding (i.e., leaving the status of all tested
components on the screen) would lead to greater transfer compared to no memory
aiding. The procedural instruction group with memory aiding performed most
effectively on the transfer task. This procedural group was followed by the discovery
groups, the variable template group with memory aiding, and the procedural instruction
group without memory aiding which did not differ signiﬁcantly from one another.
Both ﬁxed template groups and the variable template group without memory aiding
performed signiﬁcantly worse than the discovery groups.

Andrews (1984) compared discovery learning and expository methods for
teaching students about a chemical concept. The discovery learning method differed
from traditional instruction in the order in which practice and concept introduction are
presented. In the discovery learning method, instruction begins with an exploration
phase which allows learners to work with the materials and experiment with different
ways to understand the material. Once the learners have been given the chance to
explore the material, clarifying and explanatory concepts are introduced. Thus, the
second step builds on what the learner has already discovered inductively, and adds
any additional concepts not discovered. This can be considered a form of guided
discovery, in that individuals are eventually taught the correct concepts after having a
chance to explore the materials. In contrast, the expository format begins with concept
introduction and explanation, followed by a phase of working with the materials. This
is a deductive approach where the practice becomes an application exercise. Andrews
(1984) found that the discovery learning format led to superior performance on an

immediate posttest.

27

Summgy and limitations. Current research on discovery learning has made
several advances since Hermann’s (1969) review of the literature. First, some research
has examined the processes theorized to account for the beneﬁts of discovery learning
on transfer of training. When using a discovery learning approach to learn how to use
several analogous devices, an analogical reasoning process assists individuals to
transfer this knowledge to another analogous device. Several studies have also shown
the positive function of errors during instruction. A second advance of the discovery
learning literature is that it has expanded its domain of tasks to include perceptual-
motor skills, problem-solving skills, and job-relevant tasks such as range estimation for
pilots and word-processing skills. Third, this literature has focused attention on
operationalizing guidance in a discovery learning environment.

However, results of this current research are not conclusive for the beneﬁts of
pure discovery and guided discovery learning. Some research has found that discovery
learning (Frese et al., 1991), guided discovery learning (Andrews, 1984), or both
methods (Singer & Pease, 1976) led to better performance on learning and retention
measures compared to more traditional, structured approaches. However, others have
found opposite results (Greenockle & Lee, 1991). Similarly, some research has found
that discovery learning leads to greater transfer of training to novel transfer tasks
(Kamouri et al., 1986) compared to procedural instruction. Other research has found
that procedural instruction with memory aiding leads to greater transfer than discovery
learning (Carlson et al., 1992). Finally, research has found that pure discovery is

better than guided discovery for far transfer (McDaniel & Schlager, 1990), while other

28

research has shown the two methods to have similar results (Singer & Pease, 1976), or
that guided discovery is more effective than pure discovery (Greenockle & Lee, 1991).

Reasons for these inconsistent results may be the different types of tasks used,
the different operationalizations of transfer, and the different operationalizations of
discovery, guided discovery, and traditional (i.e., procedural or expository) instruction.
In fact, the conceptualization and operationalization of guided discovery has been
limited in the literature. The guided discovery conditions have mainly focused on
ways to limit the range of responses a subject may attempt in learning the task, or
providing some information that limits the range of errors that could be made. One of
the major theoretical mechanisms for the effectiveness of discovery learning is the
opportunity for individuals to engage in active hypothesis-testing and problem-solving;
however, instructional research has not based operationalizations of guided discovery
on this theoretical foundation. It would seem that providing guidance to learners to
focus their attention on forming hypotheses about the learning material and testing
these ideas out would be an effective way to provide more structure to a discovery
learning environment. No research in the instructional literature was found that
examined how to guide individuals in hypothesis-testing and problem-solving during
discovery learning.

In fact, only one study was found that actually examined an instructional
condition employing a hypothesis-testing approach. Speciﬁcally, F rese et al. (1988)
were interested in comparing three instructional methods for teaching computer
software skills. The instructional conditions varied on two dimensions: sequential v.

integrated training; and passive v. active development of a mental model. One

29

condition was labelled the sequential group (sequential training, passive mental model).
The instruction for this group focused on the step by step keystrokes for commands
that individuals would use in word processing software. The chances to make errors
were minimized, and no explanations were given for why certain commands had to be
used. A second instructional condition was a hierarchical group (integrated training
process and a passive development of a mental model). In this group, individuals were
provided with a manual plus a hierarchical diagram presenting all the commands to be
learned. The materials also gave explanations and mnemonic aids. The third group
was called the hypotheses group (active training process plus an integrated mental
model). Individuals did not receive any written material because they were to develop
their own mental model of the system. Individuals were asked to develop hypotheses
about commands to use to correct a ﬂawed text, and were encouraged to try out
solutions they developed.

Results indicated a marginally signiﬁcant difference in recall across the groups
on the second day of instruction, which disappeared on the third day of training. On
the third day of training, participants had to type a copy of a text under speeded
conditions. Results indicated that the sequential group took longer than the
hierarchical group to correct errors made (marginally signiﬁcant). The sequential
group also took signiﬁcantly longer than the hypotheses group to correct errors made
in typing the text. Participants were also tested on a non-speeded performance test
where they had to correct errors in a text. A marginally signiﬁcant difference was
found between groups in the number of keystrokes used to make difﬁcult corrections.

Transfer to a relatively easy command not taught in training was signiﬁcantly different

30

for the sequential and hypotheses groups. The hypotheses group learned the new
command in signiﬁcantly less time than the sequential group (Frese et al., 1988).

The results of this study are suggestive of the beneﬁcial effects of guiding
individuals in a hypothesis-testing approach to learning. However, the very small
sample size (ﬁve participants per group) led to a number of effects reaching only
marginal levels of signiﬁcance. In addition, the study did not ﬁnd signiﬁcant
differences between the hypotheses group and the hierarchical group. One could argue
that the hierarchical group is in fact more representative of current training
prescriptions and practice compared to the sequential group. Therefore, it is not clear
from this study that the hypotheses instruction is better than current training practices.

From a theoretical standpoint, one would also need to assess whether providing
guidance in hypothesis-testing leads to greater learning and transfer than just providing
an environment of pure discovery. Finally, the transfer task used in the study was
limited from the standpoint of the notion of adaptive transfer. During training,
participants were focused on learning an interrelated set of commands for performing
the task of editing a ﬂawed text. However, the transfer task used in this study was the
amount of time necessary to learn one new command. These two tasks are not
comparable in their levels of complexity (Wood, 1986). A research question to be
answered is whether or not guidance in hypothesis-testing leads to greater adaptive
transfer to a task at the same or higher level of complexity and task demands
compared to the training task.

Thus, the Frese et a1. (1988) study was a positive step in conceptualizing and

operationalizing guidance consistent with the theoretical basis for discovery learning.

31

Research questions remain concerning its real beneﬁt compared to traditional
instruction and pure discovery learning, as well as its effect on adaptive transfer tasks.

A ﬁnal limitation with current research on discovery learning is that no attempt
has been made to assess the inﬂuence of this instruction on the structure of the
learner’s knowledge. Several researchers have suggested that guided discovery and
pure discovery learning will lead to better integration of new knowledge into the
learner’s existing cognitive structure. In addition, this knowledge has been
hypothesized to be more complete due to the understanding of errors and incorrect task
strategies. However, research has not examined learning outcomes that assess aspects
of the structure of the learner’s knowledge.

Allowing for exploration during learning was one condition identiﬁed as
important for adaptive transfer. Several researchers have noted that metacognitive
skills are critical for individuals to beneﬁt from opportunities to explore the task
(Leutner, 1993; Veenman, et al. 1994). An implicit assumption made in discovery
learning research is that the learners are actively engaged in planning, monitoring, and
correcting errors. However, individuals differ with respect to their metacognitive
skills. Some individuals may not know how to take advantage of the opportunities of
a discovery learning environment; individuals with insufﬁcient metacognitive skills
may not engage in systematic learning activities (Veenman et al., 1994). Novices in
particular may lack the skills for planning, monitoring, and revising task behavior
(Dorner & Scholkopf, 1991; Etelapelto, 1993). With a limited exception (Veenman et

al., 1994), research on discovery learning has not examined the beneﬁts of providing

32

metacognitive instruction. The next section reviews literature on metacognition to
identify ways to promote mindful learning.
Metacognition

An individual factor that has been identiﬁed as important for adaptive transfer
is metacognition (Salomon & Perkins, 1989). Metacognition is deﬁned as one’s
knowledge of and control over one’s cognitions (F lavell, 1979). It includes planning,
monitoring, evaluating, and revising goal-directed behavior (Karoly, 1993).
Metacognition is hypothesized to facilitate learning because individuals with greater
metacognitive skills can monitor their progress, determine when they are having
problems, and then adjust their learning or performance strategies accordingly. In
addition, the knowledge acquired through metacognitively guided learning is
hypothesized to be more abstract and decontextualized; therefore, it is more likely that
individuals will apply this knowledge to a novel task (Salomon & Perkins, 1989).

Nelson and Narens (1990) described two major components of metacognition:
monitoring and control. Cognitive processes are split into at least two interrelated
levels: the meta-level and the object-level. The object-level performs some process in
relation to a goal state deﬁned by the meta-level. Metacognitive control consists of
the modiﬁcation of the object level by the meta-level. Metacognitive monitoring
involves information from the object-level apprising the meta-level of the current state
or situation.

Other metacognitive theories also suggest that cognitive processing is
hierarchically organized (Carver & Scheier, 1982; Johnson-Laird, 1983). Johnson-

Laird’s (1983) theory of consciousness is based on three major components:

33

hierarchical parallel processing; recursive embedding of models; and the existence of a
higher-level model that controls the parallel hierarchy. The high-level processor
controls the overall goals of lower-level processors, which monitor the processors at
even lower levels. Similarly, Carver & Scheier (1982) posit that individuals possess a
hierarchy of control systems, including both superordinate and subordinate goals.
Goals for the higher level control systems are more abstract, while goals lower in the
hierarchy become increasingly concrete. In addition, the systems are linked in that the
behavioral output of the higher level system provides the reference value or goal for
the next lower level system (Carver & Scheier, 1982).

The notion of monitoring and control can be traced back to Miller, Galanter,
and Pribram (1960). In response to the behaviorist emphasis on stimulus and response,
Miller et a1. (1960) argued that individuals consciously monitored and controlled their
cognitive processes. They suggested the TOTE unit (Test-Operate-Test-Exit) as the
element of behavior. This idea of a feedback loop is also critical to current
conceptualizations of control theory (Carver & Scheier, 1982). This negative feedback
loop serves to reduce sensed deviations from one’s comparison value, or goal. An
individual senses information from the environment, compares that information to
his/her goal, and if there is a discrepancy, decides on a response to reduce the
discrepancy between the information and the goal (Carver & Scheier, 1982).

Self-regulation theories are broader than a strict deﬁnition of metacognitive
monitoring and control, in that self-regulation can occur at various levels of
consciousness (Lord & Keman, 1987). In contrast, metacognition is concerned with

conscious, executive-level monitoring and control processes. In this study, the

34

processes of metacognitive monitoring and control and self-regulation are considered
synonymous because the study is focused on this activity at a conscious level.

Social cognitive theories of self-regulation have identiﬁed similar processes that
regulate action: self-observation; self-judgment; and self-reaction (Bandura, 1991;
Schunk, 1989). In addition, social cognitive theories emphasize the role of self-
efﬁcacy during self-regulated behavior such as learning. Self-efﬁcacy is deﬁned as ". .
. judgments of how well one can execute courses of action required to deal with
prospective situations" (Bandura, 1982, p. 122). Bandura and Wood (1989) deﬁne
self-efﬁcacy as the ". . . beliefs in one’s capabilities to mobilize the motivation,
cognitive resources, and courses of action needed to meet given situational demands"
(p. 408). Self-efﬁcacy judgments are based on internal cues such as ability, past
experiences, attributions for one’s behavior, and level of anxiety, as well as external
cues such as task attributes, modeling, and verbal persuasion (Gist & Mitchell, 1992).
Thus, self-efﬁcacy judgments are based on more than just assessments of one’s ability;
factors such as the complexity of the task, level of arousal, and attributions for success
will also play a role.

Self-efﬁcacy and self-regulatory processes are hypothesized to reciprocally
inﬂuence one another during learning. For example, self-efﬁcacy affects the personal
goals that are set. It is also affected by the successful attainment of these goals. Even
if individuals negatively evaluate their goal attainment, they may feel efﬁcacious if
they believe they can improve their progress through increased effort (Schunk, 1989).

In all these conceptualizations, an individual actively monitors information

relevant to a goal or standard, and if there is a discrepancy between the goal and

35

behavior, engages in control processes to modify either thought or behavior.
Furthermore, cognitive processes are hierarchically organized, such that the highest-
level processor, the metacognitive component, monitors and controls processing at
lower levels. These self-regulatory theories have had the most impact on the I/O
literature in the motivation domain (e.g., Lord & Hanges, 1987; Klein, 1989), but have
received less attention as factors important during learning.

Research on the self-regulatory aspects of metacognition has occurred in a
number of psychological disciplines. Research has been focused on the developmental
and neurological aspects of metacognition, as well as the bases for metacognitive
monitoring judgments and their relationship to metacognitive control. Some research
has examined differences in metacognition between experts and novices in different
domains. The following section describes examples of this research, with a focus on
the processes of metacognitive monitoring and control.

Research on metacognition. Cognitive and developmental psychologists have
examined the development of metacognition and its speciﬁc components. A majority
of this research has focused on metamemory processes. Metamemory is the
knowledge and regulation of one’s memory (Nelson & Narens, 1990).

Developmental studies of metacognition indicate that some metacognitive
knowledge and monitoring develops earlier in childhood. Bisanz, Vesonder, and Voss
(1978) found that young children can accurately discriminate correct and incorrect
responses. However, they found a time lag between when individuals can make these
discriminations, and when they begin to use a metacognitive strategy of distributing

processing effort to previously correct and incorrect items. Kreutzer, Leonard, and

36
F lavell (1975) found that by the third and ﬁfth grades, children had a better

understanding of the variability of memory ability over occasions and people, and
showed more planning ability than younger children.

Cognitive psychologists have primarily focused on examining four types of
monitoring judgments that occur during various stages of memory. The majority of
research has examined F eeling-of-Knowing (FOK) judgments, which occur during or
after acquisition of items on a memory task, and are predictions about whether a
currently nonrecallable item is known and/or will be remembered on a subsequent
retention test (Nelson & Narens, 1990). Although some research shows that FOK
judgments are relatively accurate indicators of memory (Hart, 1965), their lack of
perfect prediction has led to various theories regarding the mechanisms underlying
these judgments (Nelson, Gerler, & Narens, 1984). Two theories that have received
attention are the trace-access mechanisms perspective (Nelson et al., 1984) and the
inferential mechanisms view (Costerrnans, Lories, & Ansay, 1992; Metcalfe, Schwartz,
& Joaquim, 1993; Reder & Ritter, 1992). Metcalfe et a1. (1993) provided evidence
that FOK judgments are related to familiarity of cues rather than to partial information
about the target in memory. Therefore, they argued that metamemory judgments are
frequently inaccurate because individuals use a cue-familiarity heuristic to make these
judgments rather than assessing the target itself.

Research has also examined the link between metamemory monitoring and
control processes. Zacks (1969) provided some early evidence that individuals actively
control and modify their learning strategies. She found that individuals engaged in

self-paced paired-associate learning allocated more study time to difﬁcult items and

37

less study time to easy items. Belmont and Butterﬁeld (1971) also found that
individuals allocated different amounts of study time to items in free recall tasks.
Bisanz et a1. (1978) found that, for college students, discrimination of correct and
incorrect responses (a monitoring function) was related to acquisition performance in
subsequent trials. Individuals discriminated their own correct and incorrect responses
on a given trial in learning a paired-associate list, and used this information for
distributing processing effort on the subsequent trial.

Pressley, Levin, and Ghatala (1984) showed that metacognitive experiences are
important for understanding the relative utility of the control strategies of repetition
and associative elaboration. They found that for adults, awareness of the greater utility
of the elaboration strategy only occurred following practice with the two strategies
along with a performance test. Pressley et al. (1984) explained their results as
evidence that metacognitive experiences change people’s metacognitive knowledge,
which leads to modiﬁcations of their goals and actions.

Finally, Justice and Weaver—McDougall ( 1989) found that adults possess
metacognitive knowledge about the effectiveness of potential memory strategies within
and across task situations. In addition, they found that participants tended to choose
the strategies they judged to be relatively effective when actually performing the
different memory tasks. Interestingly, individuals’ judgments of relative effectiveness
of task strategies changed after performing the memory tasks. These results provide
evidence that individuals use feedback from task performance to modify their
metacognitive knowledge and subsequent control processes. In sum, several

metacognitive monitoring judgments are related to control processes of strategy

38

selection and choice. In addition, research indicates that adults possess a repertoire of
memory strategies that they can effectively apply and modify in various task situations.

Research on expert-novice differences in metacognition have taken a broader
perspective to metacognition compared to metamemory research. First, this research
indicates that experts in a domain have superior metacognitive skills compared to
novices. For example, Larkin (1983) found that experts in physics would be more
likely to discontinue ineffective problem-solving strategies than novices. Etelapelto
(1993) found that expert computer programmers had superior metacognitive
understanding of the programming task, of ideal working strategies, and a better
awareness of their own performance strategies.

Domer and Scholkopf (1991) found that experts spent more of their time at the
outset setting their goals and establishing an understanding of the system. In contrast,
novices began taking action earlier by making decisions. Dorner and Scholkopf (1991)
characterize experts as individuals who analyze the facts of a situation, and program
their actions based on self-instructions. Experts do not follow these self-instructions
blindly; instead, they use self-reﬂection to continually check and recheck their actions.
Experts pay particular attention to failures, modifying their strategies when appropriate.

In summary, research on metamemory is somewhat mixed on how accurate
people are in making these metacognitive judgments. In contrast, research on experts,
which examines metacognitive processes more broadly, does provide evidence that
these processes improve with extensive experience. However, this research does not

provide clear prescriptions for how to build this metacognitive knowledge. Research

39

in instructional psychology has examined methods for instruction and promotion of
metacognition during learning.

Research on interventions to promote metacognition. Psychologists have
suggested that metacognitive activity may facilitate adaptive transfer through its impact
on the knowledge and skills acquired during training (Salomon & Perkins, 1989).
Individuals who plan their learning activities, and monitor and evaluate their
understanding are expected to develop more principled and abstracted knowledge about
the task. In addition, theories of self-regulation also implicate motivational processes
in the effectiveness of these activities (Bandura, 1991; Schunk, 1989). In particular,
researchers have hypothesized that instruction on monitoring and control processes can
increase beliefs in one’s self-efﬁcacy. Self-efﬁcacy has been identiﬁed as a critical
factor that provides resilience in the face of changing or difﬁcult task demands
(Bandura & Wood, 1989). Researchers in the areas of instructional psychology and
education have examined how metacognitive interventions can inﬂuence knowledge as
well as motivational outcomes such as self-efﬁcacy.

It should be noted that there are several problems with how metacognition has
been conceptualized and operationalized by researchers interested in learning. First,
the term metacognition is somewhat fuzzy in deﬁnition. In fact, it is often difﬁcult to
determine whether a phenomena of interest is metacognitive or just plain cognitive in
nature (Brown, 1987). Research in the areas of reading, writing, and studying provide
examples of this deﬁnitional problem. For example, F lavell (1976) pointed out that
activities such as asking oneself questions about the text being read might function to

improve one’s knowledge (a cognitive function) or to monitor it (a metacognitive

40

function). Brown (1987) conceded that some of this literature is calling any strategic
activity during reading as metacognitive in nature.

Nonetheless, it is possible to differentiate between the learning strategies
individuals use to act upon and understand learning materials, and the executive-level
processes and activities involved in monitoring and controlling cognition and behavior.
In this sense, metacognition can be limited to activities of planning, monitoring, and
evaluating cognition, learning strategies, and task strategies. These activities have
generally been labelled self-regulatory strategies or self-regulated learning in the
educational literature.

A number of studies in the educational literature have focused on teaching self-
regulatory strategies to young children and, speciﬁcally, to those with learning abilities
(Sawyer, Graham, & Harris, 1992). Results have generally been positive for training
these groups in self-regulatory skills. For example, Sawyer et al. (1992) examined the
effectiveness of teaching a multicomponent planning strategy for writing stories to
students with learning problems. Speciﬁcally, they examined whether this instructional
method, self-regulated strategy development (SRSD), would be more effective with the
addition of explicit instruction in goal setting, self-assessment, and self-recording (full
SRSD) than without this additional self-regulatory instruction (SRSD-WESR; i.e.,
without self-explicit regulation). A third condition removed any implicit self-
regulatory instructions from the SRSD strategy and was called a direct teaching
condition. One control group included students who wrote stories without instruction

(practice-control). A normally-achieving group served as another control condition.

41

Results of the three-week training were as follows: posttest stories written by
the full SRSD (with self-regulated strategy training), those in the SRSD-WESR group,
and the normally achieving group received signiﬁcantly higher grammar scores than
the practice-control group. In addition, there were no signiﬁcant differences between
the SRSD, SRSD-WESR, and direct teaching methods on the posttest story grammar
measure; however the direct teaching method also did not differ from the practice-
control condition. No signiﬁcant differences were found between the three
instructional conditions on the quality of the posttest story. In contrast, students in the
full SRSD condition did receive signiﬁcantly greater story grammar scores on a
generalization task (back in their classroom) compared to the SRSD-WESR and direct
teaching conditions, which did not differ from one another. However, no impact of
instructional condition was found on quality of the generalization story. Contrary to
hypotheses, instructional conditions and the practice-control group did not differ on
posttest self-efﬁcacy (Sawyer et al., 1992).

Seabaugh and Schumaker (1994) examined the impact of instruction on self-
regulation skills on the application of these skills to completing individual lessons.
Their sample of 11 students included learning-disabled and nondisabled students
enrolled in an alternative school for nonfunctional students. In this school, students
were responsible for setting their own pace through individualized lessons. Descriptive
results were presented on the mean number of lessons completed per day by each
subject. The instruction on behavior contracting, self-recording, self-monitoring, and

self-reinforcement led to increases in the rate of lesson completion after the

42

intervention was introduced. These increases occurred in the subject areas targeted by
each student for application of the self-regulatory skills.

Bandura and Schunk (1981) examined how setting proximal goals during
learning would affect the self-efﬁcacy of children with gross deﬁcits in mathematical
tasks. Individuals in the proximal goal setting condition reported signiﬁcantly greater
self-efﬁcacy than individuals in a distal goal condition or in a no goal condition.

Also, individuals in the proximal goal condition performed better on a math test
compared to the distal goal and no goal individuals. It should be noted that goals
were focused on the quantity of material to be studied during self-directed learning.

Research has also provided some evidence that incorporating metacognitive
activities into instruction will facilitate learning in older or adult samples as well
(Meloth, 1990; Veenman, Elshout, and Busato, 1994). Volet (1991) found that
undergraduate students who were taught metacognitive activities during an introductory
computer programming course received better grades at the end of the course
compared to the control group. The metacognitive treatment consisted of instruction
on a planning strategy that also included monitoring and evaluation components. In
addition, while a ﬁnal exam revealed no difference in programming knowledge for the
experimental and control groups, the experimental group was better at applying this
knowledge to solving new problems (Volet, 1991).

Lundeberg (1987) identiﬁed the strategies used by experts (lawyers and law
professors) in reading legal cases. She then provided guidelines based on these expert
strategies to novice law students, either with or without self-control training. The self-

control training consisted of training and practice in using the guidelines, feedback,

43

information on why the guidelines were useful, modeling of the use of guidelines, and
group discussion. Lundeberg (1987) found that instruction combining guidelines and
self—control training led to greater performance on a test compared to training on just
the guidelines, which lead to greater performance compared to no training.

Greiner and Karoly (1976) conducted a study to teach introductory psychology
students a standard study method to improve their study habits. Students were
instructed on this study method alone or in combination with instruction on various
components of self-regulation, as follows: (1) self-monitoring, (2) self-monitoring
with self-reward, or (3) self-monitoring, self-reward and planning strategies. Results
of the training indicated that training on self-monitoring or self-monitoring with self-
reward did not lead to signiﬁcantly different learning outcomes compared to training
on the study method alone. In contrast, the group receiving instruction on self-
monitoring, self-reward, and planning strategies spent signiﬁcantly more time studying
than the other two self-regulatory groups, and they spread their study time out more
evenly over the academic quarter.

There were no signiﬁcant differences between groups on the ﬁrst quiz, but
there were signiﬁcant differences between groups on the second quiz. The planning
group performed signiﬁcantly better on the second quiz compared to the ﬁrst, and the
other groups did not change signiﬁcantly. In terms of study habits, posttreatrnent
results indicated that the planning strategy group performed signiﬁcantly better than
the other groups. No signiﬁcant group effects were found using GPA in the

introductory psychology course (Greiner & Karoly, 1976).

44

Zimmerman and Bandura (1994) conducted a study of self-regulatory
inﬂuences on achievement in a college-level writing course. Results of a path model
indicated that self-regulatory efﬁcacy for writing was positively related to self-efﬁcacy
for academic achievement, as well as positively related to self-evaluative standards.
Self-efﬁcacy for academic achievement and self-evaluative standards were in turn
positively related to grade goals. Finally, both self-efﬁcacy for academic achievement
and grade goals were positively related to ﬁnal grades.

Summm and limitations. Research on instruction to teach metacognitive
monitoring and control has generally shown it to be an effective intervention for
increasing acquisition of knowledge and skills. Studies have shown that the
combination of teaching a speciﬁc learning strategy for the task being learned and self-
regulatory strategies are important for test performance (Lundeberg, 1987) and
generalization of the learning strategy back to the classroom (Sawyer et al., 1992).
Seabaugh and Schumaker (1994) found that instruction in self-regulatory strategies was
important for increasing productivity of individuals engaged in self-initiated learning
activities. Bandura and Schunk (1981) found that setting proximal goals during self-
directed learning enhanced self-efﬁcacy beliefs. Finally, there is also evidence that
individuals trained in self-regulatory skills are better able to apply their knowledge to
solving new problems (Volet, 1991).

Most research has examined self-regulation as a complete system; in other
words, the self-regulatory instruction included all three components of planning,
monitoring, and evaluating goal-directed behavior. One study that examined the effect

of adding each component to instruction found the addition of the planning component

45

to the monitoring and evaluation components to be particularly important (Greiner &
Karoly, 1976). Thus, from the research to date, instruction to improve self—regulation
of learning should include guidance on the multiple components of planning,
monitoring, and evaluation.

For the most part, research has examined the role of metacognitive instruction
in relatively structured or traditional learning environments (Seabaugh and Schumaker,
1994 is an exception). It is likely that metacognitive skills may be even more critical
when individuals are faced with unstructured learning environments such as those
involved in discovery instruction. The capability to plan, monitor, and diagnose one’s
behavior in a discovery learning setting may enhance a more systematic and effective
learning approach. One study provides a limited examination of the role of
metacognitive prompting in a discovery learning environment.

Veenman et a1. (1994) examined the impact of metacognitive prompting and
guided discovery learning on the acquisition of electricity principles. Their task was a
computer simulation of an electricity lab. They compared an unguided discovery
learning condition with a condition that included two components: guidance in terms
of telling individuals certain experiments to perform to learn the relationships between
concepts; and metacognitive prompts for raising their working method to a more
conscious level of processing. Their results indicated that individuals in the
metacognitive-mediated/guided discovery condition exhibited a better working method
than individuals in the unguided discovery condition. However, no differences were

found between the groups on a retention test three weeks later.

46

One limitation of the study is that it did not test metacognitive prompting in
both unguided and guided discovery conditions (i.e., it did not use a completely-
crossed design). Therefore, the effects of guided discovery and metacognitive
prompting cannot be separated. In addition, this study did not examine the effects of
metacognitive prompting on adaptive transfer as deﬁned in the present study. The
retention test used in the study by Veenman et a1. (1994) consisted of quantitative and
comprehension problems similar to those given to individuals on a pretest.

In summary, discovery learning and metacognitive training are two methods to
get the learner more actively involved in the learning process. These interventions are
likely to lead to greater adaptive transfer of skills. First, discovery learning
environments allow individuals to engage in active learning processes of hypothesis-
testing and problem-solving. These strategies may prove beneﬁcial when the
individual must identify the critical aspects of an adaptive transfer task. Second,
discovery learning environments allow individuals the opportunity to make errors and
learn from them; this should lead to a better conceptual understanding of the task. In
addition, it has been suggested that this conceptual knowledge will be better integrated
into the individual’s knowledge structure.

Metacognitive instruction should lead to greater adaptive transfer because it
requires individuals to reﬂect upon their errors in understanding and develop plans to
improve them. Thus, metacognitive instruction should also lead to knowledge about
the task that is more principled and structured. Also, the capability to monitor and
evaluate one’s comprehension should be an important skill for being successful on a

novel task. Learning how to engage in metacognitive activities may also impact an

47

individual’s self-efﬁcacy, and this self-efﬁcacy should produce resilience and
persistence in the face of an adaptive transfer task.

In addition to these design factors, one must consider how individual
differences may inﬂuence the capability to learn and adapt knowledge and skills. The
next section brieﬂy reviews research in educational psychology that has examined how
individual differences impact learning, and highlights individual differences that may
be critical when training for adaptive transfer.

Individg Differences and Lear_nir_1g

Educational and instructional psychology researchers have recognized the role
of individual differences in learning and transfer. People differ in what they do during
learning, and in their capability to succeed in particular types of learning situations
(Snow, 1989). For example, research on aptitude-treatment interactions indicates that
certain instructional methods are more or less effective for different individuals.

Cronbach and Snow (1977) reviewed the literature and described several
general results. First, they concluded that general cognitive ability entered into
interactions more frequently than other aptitudes. In particular, ability was found to
interact with the amount of structure and completeness of the instructional method.
Speciﬁcally, high ability individuals learned more effectively in low structure
environments, while low ability individuals learned more effectively in high structure
environments. Low structure environments are those where individuals must act more
independently and rely on their own efforts to structure and ﬁll in the gaps in their
understanding; inductive, discovery-oriented, and leamer-controlled methods are

examples of these. In contrast, high structure environments contain a high level of

48

external control of learning activities, pacing, feedback, and reinforcement; teacher-
controlled instruction and drill and practice are examples of high structure learning
settings (Snow, 1989).

A second strong ATI result was found for personality and motivational
aptitudes. Speciﬁcally, anxiety and two achievement motives, achievement via
independence and achievement via conformance, were found to interact with the
structure of the learning environment. Anxious and conforming students learn more in
high structure environments; in contrast, nonanxious or independent students learn
more in low structure environments (Cronbach & Snow, 1977).

Research on individual differences and discoveg learning. Several studies have
examined the role of cognitive ability in discovery learning environments. Results are
consistent with the conclusions of Cronbach and Snow (1977). Egan and Greeno
(1973) compared learning how to solve joint probability problems through discovery or
rule learning methods. They also examined the role of individual differences in
conceptual, arithmetic, and permutation skills. Results indicated that participants lower
in permutations and concepts abilities learned more in the rule learning than the
discovery learning condition. In addition, they found a signiﬁcant interaction between
ability (weighted composite of arithmetic, conceptual, and permutation abilities) and
instructional method on the posttest. A graph of the results indicated that low ability
individuals committed more errors on the posttest when in the discovery condition
compared to the rule condition. In contrast, little difference was found for moderate
and high ability individuals. Shute, Glaser, and Raghavan (1989) reported on a study

conducted by the ﬁrst author examining the relationship between cognitive ability and

49

the learning of economics concepts in a computer-simulated discovery learning
environment, Smithtown. Shute found that cognitive ability was signiﬁcantly related
to learning more concepts in the 3.5 hour training session (Shute et al., 1989).

Research has also begun to examine more dispositional factors such as
cognitive styles and personality factors. Andrews (1984) examined the interaction
between dependent and independent learning styles, and inductive and deductive
learning sequences. Independent learners are those who prefer self-directed learning,
while dependent learners rely more on external structure. Andrews (1984) found that
ﬁeld dependent learners outperformed ﬁeld independent learners after a deductive
learning sequence (i.e., exposition on key concepts, followed by application of
concepts). In contrast, ﬁeld independent learners outperformed ﬁeld dependent
learners after an inductive sequence where learners ﬁrst explored the task on their own
before instruction on key concepts. However, an inductive sequence led to better
performance for both types of learners compared to a deductive sequence.

F rese et al. (1991) examined how a stable personality factor moderated the
effectiveness of error—avoidant and error training methods. The individual difference
factor they were interested in was cognitive failures, measured by the Cognitive
Failure Questionnaire (CF Q; Broadbent, Cooper, Fitzgerald, & Parkes, 1982). This
construct seemed to be a general tendency to commit errors, or a greater sensitivity to
stress. Unfortunately, this construct was not well-deﬁned in the study. Frese et al.
(1991) correlated scores on the CF Q with learning and transfer performance in each
training group. Results indicated that signiﬁcant negative correlations were found

between CF Q and performance variables only in the error-avoidant group. The

50

difference between the correlations for the two groups was signiﬁcant for transfer
performance. They interpreted these results to suggest that "scatty" people proﬁt from
error training, but do not proﬁt from error-avoidant training (Frese et al., 1991).

Individual differences and I/O training research. I/O research has also
examined the role of individual differences in learning and training effectiveness. For
example, Ackerman (1988) found that initial skill acquisition is best predicted by
general cognitive ability, intermediate skills are best predicted by perceptual speed
ability, and late stages of skilled performance are best predicted by psychomotor
abilities. Mathieu, Tannenbaum, and Salas (1992) found that pretraining motivation
predicted greater learning and reactions to training. Gist, Schwoerer, and Rosen
(1989) found that self-efﬁcacy prior to training was positively related to performance
on a test at the end of training.

Thus, training research has considered the role of ability and motivation in the
learning and transfer of skills. In their review of training research, Tannenbaum and
Yukl (1992) suggested that certain personality or dispositional factors may also impact
learning and transfer. For example, certain "big ﬁve" personality factors may
inﬂuence training effectiveness. The goal orientation of the learner has also been
considered in research on training transfer (Kozlowski, et a1. 1995; Smith, et al. 1995).

When the focus of training is adaptive transfer, one must identify individual
differences that may be critical for the capability to be ﬂexible in novel or complex
circmnstances. Tolerance for ambiguity is a personality factor that may be relevant for
the capability to adapt one’s skills to a novel task. Tolerance for ambiguity can be

deﬁned as a tendency to perceive ambiguous situations as desirable (Budner, 1962).

51

Individuals who are high on tolerance for ambiguity tend to welcome the challenge of
ambiguous, complex, or novel tasks. Therefore, tolerance for ambiguity may affect
how much is learned and applied to an adaptive transfer task.

In addition, Dweck (1986) has identiﬁed two goal orientations that individuals
may take to a learning situation: mastery orientation and performance orientation. A
mastery orientation includes the belief that effort leads to improved outcomes, and that
ability is malleable. Individuals with a mastery orientation are focused on developing
new skills and successfully achieving self-referenced standards for mastery (Ames,
1992; Dweck, 1986). In contrast, individuals with a performance orientation to
learning believe that ability is demonstrated by surpassing normative-based standards,
or by succeeding with little effort (Ames, 1992). These motivational dispositions must
be considered when examining learning and transfer processes.

A conceptual model is presented next that integrates the literatures on discovery
learning and metacognition. This model raises hypotheses concerning individual
differences and instructional design factors that may inﬂuence adaptive transfer. In
addition, this model identiﬁes learning outcomes that serve as intervening factors

between individual differences, design factors, and adaptive transfer.

TRAINING FOR ADAPTIVE TRANSFER: A CONCEPTUAL MODEL

In reviewing l/O research on transfer of training, Baldwin and Ford (1988)
identiﬁed three training input factors that will affect learning and transfer: trainee
characteristics; training design; and work environment. With regard to situational
factors, it is acknowledged that the work environment will affect whether knowledge
and skills will be adapted to novel task situations. However, the conceptual model to
be presented is focused on the design of the training environment and how to structure
it to facilitate adaptive transfer. In addition, the model also considers individual
characteristics of the learner and how they may affect learning and transfer. Finally,
this model identiﬁes learning outcomes that mediate the relationship between
individual differences, training input factors, and adaptive transfer.

Current I/O research on design principles to achieve training transfer has been
limited in the types of independent variables examined, and in the deﬁnition and
operationalization of transfer of training. First, training research has focused
predominantly on the effects of design principles that have been developed from a
behaviorist paradigm. As the review suggested, these principles lead to well-learned,
and efﬁcient procedures, but not necessarily to ﬂexibility of skills. Second, the
training literature has provided little differentiation between types of transfer and how

to train for them. It is argued that adaptive transfer is of critical importance in the

52

53

rapidly changing organizational environment. Principles for instruction when jobs
were relatively stable and well-deﬁned may be insufﬁcient in today’s increasingly
complex, and cognitively demanding jobs.

With advances in technology, individuals will be required to learn more
complex skills, and they will need the capability to learn new skills in the future
(Hesketh & Bochner, 1994). This places a greater emphasis on the ﬂexibility and
adaptability of skills learned in training to new or unforeseen contexts. Hesketh and
Bochner (1994) argued that training for jobs in the future require that individuals take
responsibility for their own learning.

Several programs of research at Michigan State University are focused on how
to engage the learner more actively during training. In addition, this research is
focused on factors that inﬂuence the acquisition of complex skills. For example,
Smith et a1. (1995) examined an integrated model of learning in an environment where
individuals had responsibility for choosing their own practice exercises. They
predicted that the goal orientation of the learner would inﬂuence the activities they
engaged in during training to master a complex, decision-making task. These learning
activities were expected to impact multiple learning outcomes, which would in turn
inﬂuence the transfer of training to a more difﬁcult version of the trained task.

Smith et a1. (1995) found that mastery orientation was positively related to
meatcognitive activities during training. Metacognition was positively related to the
learning outcomes of ﬁnal training performance and self-efﬁcacy. In addition, mastery
orientation was found to be positively related to self-efﬁcacy, while performance

orientation was negatively related to self-efﬁcacy. The use of speciﬁc practice

54

strategies during training were positively related to several learning outcomes. Finally,
the three learning outcomes of knowledge, ﬁnal training performance, and self-efﬁcacy
were all positively related to transfer of training.

Kozlowski et al. (1995) examined how individual differences and training
design factors affected multiple learning outcomes and adaptive transfer. Individual
differences included cognitive ability, mastery orientation, and performance orientation.
This study focused on two training interventions to orient individuals to learning
objectives for skills critical to the development of expertise. Speciﬁcally, they
examined how sequenced mastery training goals and advance organizers inﬂuenced
multiple learning outcomes. Distinct theoretical mechanisms were conceptualized for
these two interventions, and different learning outcomes were expected to be
inﬂuenced by each. An outcome of particular interest in this study was the structure
of the knowledge developed during learning. It was expected that providing mastery
goals to trainees would be especially beneﬁcial in building effective knowledge
structures. With few exceptions (Kraiger & Salas, 1993; Nason, Gully, Brown, &
Kozlowski, 1995), training research has tended to assess knowledge in a simpler form
through verbal knowledge tests. This study extended training research by examining
the unique impacts of declarative and structural knowledge, in addition to skills and
self-efﬁcacy, on transfer to a more complex version of the trained tasks.

Results of this study indicated that transfer of training was independently
predicted by declarative knowledge, knowledge structure, training performance, and
self-efﬁcacy. Advance organizers were found to affect more traditional learning

outcomes of declarative knowledge and ﬁnal training performance. Mastery training

55

goals led to more rapid acquisition of declarative knowledge, and also affected the
development of knowledge structures across training sessions. Mastery goals also
inﬂuenced the development of self—efﬁcacy over training sessions.

These studies provided the basis for the form of the conceptual model
examined in the present study. Speciﬁcally, similar to these previous studies, the
present model examines how individual differences and training design interventions
inﬂuence learning and transfer of training. A general conceptual heuristic is presented
in Figure 1. As in previous research, this study takes a multi-dimensional perspective
to learning by identifying several learning outcomes that will inﬂuence adaptive
transfer. The model in Figure 1 is a general overview and does not display the
speciﬁc hypotheses to be tested.

In reviewing cognitive and instructional research, discovery learning methods
and metacognitive processes were identiﬁed as two factors that will increase the
responsibility of the learner for acquiring knowledge and skills through their own
interpretive and constructive processes. This active participation by the learner is
conceptualized to lead to greater adaptive transfer of this knowledge and skill. The
research conducted by Kozlowski et al. (1995) indicated that mastery goal training is
another method for building expert knowledge that will lead to greater transfer. In
addition, the study by Smith et al. (1995) indicated that metacognitive activity is also
relevant when learners have control over instructional events. Although mastery goals
and learner control are operationalized differently than the training interventions
examined in this study, it is clear that each of these training interventions inﬂuence

learning and transfer through engaging the learner in more active processing of

56

training material. Hypothesis-testing, problem-solving, and self-regulation are
processes that are consistent with more active engagement in learning.

The present study is focused on how to structure training materials to facilitate
adaptive transfer. However, training events are controlled by the instructional medium
in this study. Learners are not given control over the sequence, complexity, or number
of exercises to complete, although this is acknowledged as another possible method to
achieve learning and transfer. In addition, mastery goal training has traditionally
focused more generally on the motivational effects of particular goals on learning
regardless of the structure of the training environment (e.g., Dweck & Leggett, 1988).
In contrast, the interventions in the present study are focused on the speciﬁc training
content and skills being learned. Individuals are expected to learn differently based on
the amount of information and guidance provided during training (discovery learning
intervention) and whether or not they are instructed on how to monitor, diagnose, and
evaluate speciﬁc skills learned during training (metacognitive instruction).

Several learning outcomes are identiﬁed as mediators of the effects of these
instructional methods on adaptive transfer. A multidimensional perspective on learning
outcomes (Kraiger, et al., 1993) is used to hypothesize the learning mechanisms that
lead to adaptive transfer. For example, allowing individuals to explore and experiment
with tasks should lead to the development of more detailed verbal knowledge about the
task. Also, discovery learning methods and metacognitive instruction should lead to
knowledge structures that are better developed and more integrated compared to more

passive learning methods.

57

3.29... o>=amu< ucm mEEmo.

:0 5.82.2. o>=EmooEm2 ucm .mEEmo. E9085 .woocmEtﬁ .9636... .0 SEE. 9.. .o 0.623... .maamocoo < .w 9:9”.

 

 

53cm; o>=amn<

 

 

 

 

 

‘I

 

68553 -
3.2.9... bagsmmmaaw -
mc_.mo...-m_mo£oa>1 -
93025 $3.265. .. /,
$8.265. .mn.o> - /
moEooSO 3.53..

 

 

5.62.2. o>.._:moom.o.>. oz
E03258. 93.38922 ..

.5385 San.

5303.0 8.230
Ease—Soc. .Esuoooi -
wco_.m_iq.cms. uEElmmH

 

 

 

/

 

 

3.35:2. 8.. 8:990... -
cos—35:0 roams. -
coszotO mocmEEtmd -
3...“? 82.280 -

Boﬁlmn. 8:935 .9636...

 

 

 

58

Second, researchers have suggested that discovery learning methods and
metacognitive training provide individuals with learning strategies that enable them to
adjust their task knowledge and skills to handle novel circumstances (McDaniel &
Schlager, 1990). The two learning activities identiﬁed as critical for adaptive transfer
are hypothesis-testing and self-regulatory skills.

Finally, self-efﬁcacy is identiﬁed as a learning outcome that should have a
motivational impact on adaptive transfer. Self-efﬁcacy is expected to lead to greater
persistence in trying to succeed on the adaptive transfer task. Self-efﬁcacy is a central
concept in models of self-regulated learning (Schunk, 1989; Zimmerman & Bandura,
1994). Self-efﬁcacy and self-regulated learning processes are expected to be
reciprocally related to one another during learning activities. Researchers have found
that aspects of self-regulation during learning lead to increased beliefs in self-efﬁcacy
(Bandura & Wood, 1981). The model examines how metacognitive instruction may
lead to greater self-efﬁcacy at the end of training that then affects adaptive transfer.

In addition, this model examines the impact of both individual differences and
instructional interventions on learning and transfer. Baldwin and Ford (1988)
criticized training research for examining individual differences and design factors in
isolated studies. Recent training research provides some exceptions (Gist, Rosen, &
Schwoerer, 1988; Kanfer & Ackerman, 1989). In the present model, tolerance for
ambiguity and mastery orientation are identiﬁed as theoretically important individual
difference factors to consider when examining learning outcomes and adaptive transfer.
Cognitive ability and performance orientation are also included in the model to control

for any impacts they may have on learning and transfer.

59

The ﬁnal outcome examined in this model is adaptive transfer. In this model,
adaptive transfer refers to the adjustment of knowledge and skills learned in training to
a novel task. Novelty is deﬁned as new stimuli in the transfer task that require
individuals to modify task strategies learned in training to achieve successful transfer.

This study will expand research on discovery learning and metacognition by
generalizing it to a complex, decision-making task that has relevance for organizational
tasks. The task used in this study is a computer simulation of a radar tracking task.
In addition, this study will build on traditional training principles by including
sufﬁcient opportunities for practice of skills, feedback on performance, and a training
sequence that builds from simple to complex skills. In this way, this study will be
examining how design elements such as discovery learning opportunities and
metacognitive processing can add beneﬁt to traditional training approaches. This
perspective is also consistent with the argument that individuals can be taught to
actively process and interpret knowledge and skills while practicing and reﬁning them
at the same time (Salomon & Perkins, 1989). The training manipulations, individual
differences, and learning and transfer outcomes are described in detail next.
Discovery Learning, Guided Discoveg, and Procedural Instruction

One major focus of this model is the opportunity for individuals to engage in
discovery learning. Allowing individuals to explore and experiment during training
should lead to greater adaptive transfer than traditional instructional methods.
Unfortunately, complete discovery learning may not be an efﬁcient way to train since

it usually takes more time for individuals to inductively learn rules or principles than

60

to practice and apply instructor-provided rules. An early review (Hermann, 1969)
suggested that guided discovery is more effective than pure discovery learning.

In the present study, guided discovery learning is deﬁned as directing
individuals to engage in hypothesis-testing. During initial practice, individuals will be
provided with a description of a simple task situation and the strategy they should use
to respond to the situation. In addition, they will be instructed to explore the task on
the ﬁrst simple trial so that they can see how a scenario unfolds. This will provide
initial structure to the task and an introduction to the task parameters that may require
attention. Next, individuals will have to make predictions about what parameters may
change in the task to make it more complex. They will then have practice
opportunities to explore more complex situations, discover if their predictions were
correct, and develop the best strategies to handle the more complex situations.

In contrast, a pure discovery condition will provide little structure to learners in
how to approach the task. Individuals will be presented with the same task situations
in sequence of increasing complexity, but they will be left completely on their own to
learn strategies for performing the task under different task situations. The guided and
pure discovery conditions will be compared to a procedural instruction condition as
well. This condition will be consistent with current training practices, as individuals
will be presented descriptions of the different task situations and the strategies they
should use to effectively perform the task. Then participants will be provided with
opportunities to practice and apply these speciﬁc procedures.

It is expected that guided discovery learning will lead to greater adaptive

transfer compared to pure discovery and procedural instruction. In addition, pure

61

discovery learning is expected to lead to greater adaptive transfer than procedural
instruction. The two discovery conditions should lead to greater adaptive transfer than
procedural instruction because individuals are provided with opportunities to explore
and experiment with the task, and they are responsible for developing their own
understanding of the task and their own strategies for performing it. In addition, the
guided discovery condition should be more effective than the pure discovery condition,
because individuals are directed to aspects of the task in the low complexity situation
and prompted to test hypotheses in more complex situations. Individuals may differ in
their capability to develop a systematic learning approach when given total
responsibility to learn about a task. This may lead to inefﬁcient exploration of the
task. In contrast, individuals who are provided some guidance in generating
predictions and thinking about task strategies will still achieve the beneﬁts of
exploration, but in a more systematic way.

Hypothesis 1. Guided discovery learning will lead to greater adaptive transfer

compared to pure discovery learning and procedural instruction. Pure discovery

learning will also lead to greater adaptive transfer than procedural instruction.
MetagLnitive Instruction

Metacognitive instruction is a second focus of this study. Metacognition is
deﬁned as the self-regulatory processes of planning, monitoring, and evaluating task
behavior. Metacognitive instruction will be operationalized as having individuals set a
learning goal for practice, devising a plan for achieving the goal, and monitoring and
evaluating their progress relative to the goal. A concern in the literature on these self-
regulatory processes is that individuals may not have the attentional resources required

to self-regulate early during skill acquisition (Kanfer & Ackerman, 1989). Initial

62

learning of an unfamiliar task requires attention that may not be available for
monitoring and evaluating this learning. Therefore, this study will introduce the
metacognitive instruction after individuals have had initial opportunities to practice
simple task situations. In addition, the prompts for individuals to set a learning goal
and plan will occur after individuals have had one trial to experience the moderate task
situation. Individuals will be asked to use the feedback they have received from this
ﬁrst trial, as well as their experience in performing it, to diagnose any difﬁculties they
are having in learning the task. They are then asked to develop a goal and plan for
the next trial to improve their understanding. A similar process will occur when
individuals move to the most complex training situation.

It is expected that instruction on metacognitive planning, monitoring and
evaluation of task strategies will lead to greater adaptive transfer. Individuals who are
instructed to plan, attend to, and evaluate their understanding should notice any errors
in their performance and ﬁnd ways to avoid these errors in the future. Thus, by
attending to their learning and trying to improve it, these individuals are more likely to
be ﬂexible in adapting their skills to novel circumstances.

Hypothesis 2. Individuals who receive metacognitive instruction should

perform signiﬁcantly better on an adaptive transfer task compared to

individuals who do not receive metacognitive instruction.

The ﬁrst two hypotheses are general predictions based on theory and research
from the instructional and educational psychology literatures. Previous research has
not examined the interaction between discovery learning manipulations and
metacognitive instruction. It is expected that there will be an interaction between these

two instructional conditions in predicting adaptive transfer. Speciﬁcally, it is expected

63

that metacognitive instruction will lead to greater adaptive transfer in the guided
discovery and pure discovery conditions compared to no metacognitive instruction;
however, metacognitive instruction is expected to have little or no added beneﬁts in
the procedural instruction condition.

Metacognitive instruction may be especially important in helping learners to
cope with more exploratory learning environments. Discovery learning environments
put high metacognitive demands on novice learners (Veenman et al., 1994). In
addition, individuals in discovery learning situations are likely to commit errors that
require greater diagnosis compared to individuals in procedural instruction conditions.
Frese and Altmann (1989) have distinguished between different types of errors, and
two that are relevant to discovery v. procedural instruction are slips and mistakes.
Slips are errors which result from wrong plans but right intentions. Mistakes occur
when the intentions were wrong but the plan conformed to the intention (Norman,
1984). Individuals receiving procedural instruction should be less likely to commit
errors overall, as they are provided with information on the correct strategy to use to
perform the task. In addition, if they do commit errors, they are more likely to
commit slips. They are presented with the correct strategy for task performance, but
they may apply it incorrectly. In contrast, individuals in discovery learning conditions
must discover the correct strategy for task performance. Discovery learning
environments are designed so that individuals will commit more errors overall. In
addition, they are more likely to commit mistakes where they choose and apply an

incorrect task strategy.

64

When individuals receiving procedural instruction perform poorly during
practice, it should be fairly obvious to them what errors they may have committed.
They have the correct task strategy to study, and they can determine to what extent
their actions during practice matched the strategy they should have been practicing. In
contrast, when individuals in discovery learning conditions commit errors, the reasons
for the errors are not as obvious. The individual may have committed a mistake in
terms of trying out the wrong task strategy, or the individual may have committed a
slip in terms of an error in applying the correct task strategy. Thus, discovery learning
environments should beneﬁt more from instruction and guidance on how to monitor
and diagnose difﬁculties during learning. This should help them increase their
understanding of correct and incorrect task strategies that will then lead to greater
adaptive transfer.

Hypothesis 3. An interaction between discovery learning conditions and
metacognitive instruction will signiﬁcantly inﬂuence adaptive transfer. Metacognitive
instruction will lead to greater adaptive transfer compared to no metacognitive
instruction for guided and pure discovery conditions; there will be little or no
difference between the presence or absence of metacognitive instruction in the
procedural learning condition.

In addition, there are several learning outcomes expected to play intervening
roles in the relationships between discovery learning, metacognitive instruction and
adaptive transfer. These learning outcomes provide the rationale for the general

effectiveness of guided discovery learning and metacognitive instruction on adaptive

transfer. These more complex relationships are described next.

65
Learning Activity

The two instructional manipulations to be examined in this study serve the
purpose of teaching individuals effective methods for learning new tasks. Not only
does the instruction provide a better method for learning during training, it also
provides individuals with practice on more general learning skills that can be applied
to performance on the adaptive transfer task. First, guided discovery learning teaches
individuals an approach to learning new information that consists of making
hypotheses or predictions about the task, exploring the task to test these predictions,
and experimenting with task strategies to develop those that are optimal for task
performance. It is expected that this type of hypothesis-testing/problem-solving
approach to learning is an effective skill for adaptive transfer. Individuals who are
successful at developing their own systematic approach to learning in the pure
discovery condition are also likely to get some practice in this hypothesis-testing
activity. This is in contrast to the less active approach of receiving explicit instruction
on task strategies and applying them to the situation at hand. Thus, individuals in the
procedural instruction condition will have little opportunity to engage in this
hypothesis-testing activity.

Therefore, individuals in the guided discovery condition are more likely to
report that they practiced and used a hypothesis-testing approach during instruction
compared to the pure discovery and guided discovery conditions. One may question
the validity of self-reports of learning activities, but it is expected that self-reported
hypothesis-testing is a satisfactory measure of learning activities. Previous research

has shown that self-reports of learning strategies are related to important learning

66
outcomes. For example, Pokay and Blumenfeld (1990) found that self-reported use of
geometry-speciﬁc learning strategies and effort management strategies predicted grades
early in the semester in a geometry class.

Thus, it is expected that individuals who are instructed on a hypothesis-testing
approach in the guided discovery condition will report greater practice and use of
hypothesis-testing methods during training compared to the pure discovery and
procedural instruction conditions. In addition, because they are left on their own to
develop an understanding of the task, individuals in the pure discovery condition
should report greater use of a hypothesis-testing approach to practice compared to
individuals receiving procedural instruction. It is also expected that the self-reported
use of hypothesis-testing will be positively related to adaptive transfer.

Hypothesis 4. Individuals in the guided discovery condition will report

signiﬁcantly greater use of hypothesis-testing activities during practice

compared to individuals in the pure discovery condition, who will report
signiﬁcantly greater use of these activities compared to the procedural

instruction condition.

Hypothesis 5. Self-reported use of hypothesis-testing activities will be
positively related to adaptive transfer.

Metacognitive instruction provides individuals with practice on a second
learning activity that should be effective for adaptive transfer. Individuals in the
metacognitive instruction condition will learn how to pay attention to and monitor their
task performance, to identify any difﬁculties, and to develop learning goals and plans
to improve their performance. The opportunity to practice these skills will make it
more likely that they will apply these skills to an adaptive transfer task. In contrast,

individuals in the condition without metacognitive instruction will have no explicit

67

opportunities to practice this self-regulatory approach to learning. Thus, individuals in
the metacognitive instruction condition should report greater use of self-regulatory
activity compared to the condition without metacognitive instruction.

Research has provided some evidence for the validity of self-reports of self-
regulatory activity during learning. For example, Pintrich and DeGroot (1990) found
that self-reports of self-regulated learning were positively related to academic
performance in terms of classwork, exams/quizzes, essays/reports, and average grade.
Pokay and Blumenfeld (1990) found that self-reported metacognition was negatively
related to achievement early in the semester, but positively related to achievement late
in the semester. Finally, Smith, et al. (1995) found that self-reported metacognition
during learning was positively related to performance on a ﬁnal training trial.
Therefore, it is expected that metacognitive instruction will lead to reports of greater
use of self-regulatory activities during practice, and this self-regulation should be
positively related to adaptive transfer.

Hypothesis 6. Individuals who receive metacognitive instruction should report

signiﬁcantly greater use of self-regulatory activities compared to individuals

who do not receive metacognitive instruction.

Hypothesis 7. Reported use of self-regulation during learning will be positively
related to adaptive transfer.

In addition to providing individuals with learning methods that will assist them
during transfer, it is expected that discovery learning and metacognitive instruction will

inﬂuence other learning outcomes as well. These relationships are explained next.

68

Verbal Knowledge and Knowledge Structure

A fundamental outcome of any training program is the development of
knowledge about the training task. As individuals practice a task, they develop
declarative knowledge about the task and its features, and they also develop procedural
knowledge on how to perform task components. Researchers have hypothesized that
more principled knowledge is one mechanism through which guided discovery
instruction leads to better adaptive transfer compared to more traditional instruction.
Because individuals are allowed to make errors and learn from them under discovery
learning conditions, their knowledge of the task will include not only an understanding
of when strategies are appropriate, but also when they will not be successful. This
greater depth of understanding should assist individuals when they must transfer their
knowledge and skills to a novel task.

It is expected that individuals under guided discovery and pure discovery
conditions will develop greater verbal knowledge of the task compared to individuals
receiving procedural instruction. In addition, guided discovery should lead to greater
knowledge compared to pure discovery learning. Because individuals in the guided
condition are prompted to develop and test hypotheses about the task, they should be
more likely to develop systematic and principled knowledge about the task. Some
support has been found for these hypotheses in previous research; for example, Frese
et a1. (1991) found that individuals in an error-training condition recalled signiﬁcantly
more word processing commands compared to an error-free training group. Singer

and Pease (1976) found that a pure discovery and combination group outperformed a

69

procedural instruction group on a retention test. Therefore, similar results are expected
for a knowledge test in the present study.
Hypothesis 8. Guided discovery learning will lead to signiﬁcantly better verbal
knowledge compared to pure discovery learning, which will in turn lead to
signiﬁcantly better verbal knowledge compared to procedural instruction.
Theory also indicates that metacognitive instruction should lead to greater
knowledge about the training task. Greater metacognitive skills should facilitate
learning because they allow individuals to monitor their progress, identify any
problems they are having during training, and then adjust their learning strategies to
try to overcome any difﬁculties. Research has found that metacognitive instruction
leads to greater performance on posttests of knowledge compared to groups receiving
no metacognitive instruction (Greiner & Karoly, 1976; Lundeberg, 1987). Therefore,
it is expected in this study that metacognitive instruction will lead to greater verbal

knowledge compared to a no metacognitive instruction condition.

Hypothesis 9. Metacognitive instruction will lead to signiﬁcantly better verbal
knowledge compared to training without metacognitive instruction.

Finally, it is expected that greater declarative and procedural knowledge about
the task, as assessed on a knowledge test, will be positively related to adaptive
transfer. Individuals who develop a greater depth of knowledge about the task, by
learning when certain strategies and actions are appropriate and inappropriate, should
be better able to apply and adapt this knowledge to a novel task. Research on
discovery learning and metacognitive training has not tended to examine the
relationship between knowledge at the end of training and transfer; instead, it has

treated them as different outcomes of instructional interventions. However, in the

70

present study, knowledge about the task is seen as a prerequisite for the ability to
successfully perform in an adaptive transfer task.

Hypothesis 10. Verbal knowledge will be positively related to performance on
an adaptive transfer task.

As individuals gain experience with a task, their declarative knowledge not
only becomes compiled into procedural rules, it also becomes meaningfully structured
in memory (Kraiger et al., 1993). Thus, a second method for assessing the quality and
depth of knowledge acquired from training is to determine how individuals relate and
link various concepts and actions in their memory. Researchers have developed
various concepts to describe this knowledge organization, such as scripts, schema,
mental models, and cognitive maps. A general term referring to the organization of
knowledge is knowledge structure. Research on experts and novices has found that
they differ in the organization of their knowledge structures. For example, experts
possess knowledge structures that contain both problem deﬁnitions and solutions,
whereas novices tend to possess separate knowledge structures for problem deﬁnition
and problem solution (Glaser & Chi, 1989).

Patel and Groen (1991), in a study of medical expertise, also found distinct
differences between novices, intermediates, and experts. They distinguish between
experts and intermediates in terms of generic expertise. "A distinguishing trait of
experts, even outside their domain of specialization, is knowledge of what not to do"
(p. 121). Intermediates possess the speciﬁc domain knowledge that experts possess
(i.e., what to do given a particular task situation), but they do not know what not to

d_o. This leads them to conduct irrelevant searches, be distracted by irrelevant clues,

71

and access unnecessary knowledge from their memory. Novices do not conduct these
irrelevant searches because they do not know what to do. This research suggests that
experts possess more complex knowledge structures that not only contain information
on correct task strategies given a particular task situation, but also information on
strategies that would be errors. From their considerable experience with the task,
experts are likely to have made errors and developed an understanding of these errors.
They are then able to avoid these errors in the future.

Researchers have suggested that allowing individuals to engage in discovery
learning processes will lead to better developed and integrated knowledge structures.
By exploring the task and making mistakes, individuals should develop a better
operative mental model of the task (Frese et al., 1988). For example, Egan and
Greeno (1973) interpret the results of their study to indicate that learning by discovery
leads to "external connectedness" of one’s knowledge structure. Because individuals
learning by discovery performed better at problems involving more interpretation, the
new structural components developed during learning became well integrated into
existing cognitive structure. Individuals who learn by discovery must use their
previous knowledge to help them understand the task to be learned, whereas
individuals learning under rule application conditions can merely add these new
elements to their cognitive structure without reorganizing it (Egan & Greeno, 1973).

While an integrated knowledge structure has been described as an outcome of
discovery learning methods, research has not explicitly measured knowledge structures
at the end of training to test this hypothesis. Instead, recall or retention tests have

been used to assess the amount of knowledge acquired (e.g., Singer & Pease, 1976).

72

However, cognitive psychologists have examined a number of ways to assess
knowledge structure. Measurement of knowledge structure has included card sorting
tasks, as well as making judgments of similarity or relatedness among a number of
predeﬁned elements. Structural assessment is the strategy of submitting judgments of
similarity to a clustering or scaling algorithm (Kraiger et al., 1993).

Recently, interest has been focused on different statistical methods to represent
the structural properties of knowledge. Goldsmith, Johnson, and Acton (1991)
examined a new algorithm for representing knowledge structure, Pathﬁnder. The
Pathﬁnder scaling algorithm transforms a matrix of relatedness (proximity) ratings into
a network structure. In this structure, each element is represented by a node in the
network, and the relatedness between the elements is depicted by how closely they are
linked. Based on this network representation, they developed an index of closeness
(C) that measured how similar a student’s knowledge structure was to that of an expert
(i.e., the instructor) in a statistics course. This was compared to correlational measures
of similarity derived from the raw proximity data and to structures derived from MDS.
These three measures of similarity were related to course performance on exams and
papers in the statistics course. The correlation between C and course performance was
a better predictor than the raw rating data or the MDS measure. Results indicated
that when the other indices were held constant, the C index still captured unique
predictive variance in course grades (Goldsmith et al., 1991).

Subsequent research has provided further construct validity for the Pathﬁnder
method of measuring knowledge structure. Several studies have found instructional

interventions to predict knowledge structure. For example, Kraiger and Salas (1993)

73

found C scores to discriminate between individuals who participated in a Naval
aircrew coordination training program and a group of control participants. Kraiger
(1993) found that providing training goals before training led to greater C scores than
presenting training goals after training. Finally, Kozlowski, et al. (1995) used the
Pathﬁnder algorithm to assess the coherence of knowledge structures after various
training interventions. Coherence measures the extent to which relations among pairs
of concepts are logically consistent across all concepts that were rated. Kozlowski et
a1. (1995) found that the coherence of knowledge structures increased signiﬁcantly
over training sessions for individuals provided advance organizers or mastery goals.
Thus, having individuals make relatedness ratings among a set of concepts and
submitting these ratings to the Pathﬁnder algorithm is an effective way to measure
knowledge structure. This is the operationalization of knowledge structure to be used
in the present study. Participants’ knowledge structures will be compared to an expert
knowledge structure to create the index of closeness, C. It is expected that the
discovery learning and metacognitive instruction interventions will affect C scores.
Consistent with theoretical explanations (Egan & Greeno, 1973; Frese et al.,
1988), it is expected that guided discovery and pure discovery learning will lead to
more accurate knowledge structures compared to procedural instruction. Individuals in
the two discovery conditions must actively generate their own understanding of the
task and how to respond to it. This should lead to a better integrated and more
complex knowledge structure. Because they are likely to make errors in their
responses to the task, they are likely to develop a more elaborated knowledge structure

as well. In contrast, individuals in a procedural instruction condition do not have to

74

actively construct an understanding of the task; rather, they are explicitly provided a
description of each task situation and how they should respond. Therefore, they will
be applying this knowledge in a more passive manner that does not require them to
generate their own explanations of the task. It is expected that guided discovery will
also result in a better knowledge structure compared to pure discovery learning. While
both groups must develop their own understanding of the task, the guided discovery
group is directed to the important features of the task on which they should focus their
attention. They should develop more accurate knowledge structures than the pure
discovery learners, who may develop a less systematic understanding of the task.

Hypothesis 11. Guided discovery learning will lead to a signiﬁcantly better

knowledge structure compared to pure discovery learning, which will in turn

lead to a signiﬁcantly better knowledge structure compared to procedural
instruction.

Research on metacognition has not explicitly considered the role of self-
regulatory processes in the development of well-structured knowledge. However, it
has been hypothesized that metacognitive activity should lead to more principled
knowledge (Salomon & Perkins, 1989). It can be hypothesized that individuals who
engage in self-regulatory activities during learning should develop a more integrated
and elaborated understanding of the task. By planning their activities, and monitoring
and diagnosing errors in their performance, they should develop an operative
knowledge structure that tells them when certain task strategies are appropriate, when

they will lead to errors, and why this is so.

Hypothesis 12. Metacognitive instruction will lead to a signiﬁcantly better
knowledge structure compared to training without metacognitive instruction.

75

Finally, it is expected that knowledge structure will be positively related to
performance on the adaptive transfer task. Previous research has shown the positive
relationship between knowledge structure and class performance (Goldsmith et al.,
1991), and between knowledge structure and skill generalization to a more complex
transfer task (Kozlowski et al., 1995). Individuals who have developed a more richly
integrated knowledge structure concerning the task and the range of possible strategies
will have a better repertoire of responses to try out when faced with a task in which
they must adapt this knowledge.

Hypothesis 13. Knowledge structure will be positively related to performance
on the adaptive transfer task.

Self-Efﬁcacy

Self-efﬁcacy is one’s conﬁdence that one can successfully perform a given task
(Bandura, 1977). Self-efﬁcacy can be differentiated from similar constructs such as
self-esteem because self-efﬁcacy perceptions are task-speciﬁc. In contrast, self-esteem
is usually considered a trait that reﬂects an individual’s global, affective evaluation of
the self (Gist & Mitchell, 1992). Social cognitive theories of self-regulation highlight
the role that self-efﬁcacy plays during self-regulated learning (Schunk, 1989). Self-
efﬁcacy inﬂuences self-regulatory processes, and is inﬂuenced by them as well
(Bandura, 1991). Research on self-efﬁcacy and self-regulation has usually examined
self-efﬁcacy as an antecedent variable that affects how individuals self-regulate their
performance. For example, Zimmerman and Bandura (1994) found self-efﬁcacy to

positively inﬂuence goals set by students in a writing course.

Of

WC

76

However, a few studies have also examined how self-regulatory activity
inﬂuences self-efﬁcacy as a learning outcome (Bandura & Schunk, 1981; Sawyer et
al., 1992). These studies provided conﬂicting results. It is hypothesized in this study
that providing individuals with skills for monitoring their performance, noticing errors,
and identifying ways to correct these errors should enhance their conﬁdence that they
can succeed at the task. For example, Gist (1989) found that including a cognitive
modeling component during a training course did improve self-efﬁcacy over lecture
and practice alone. This cognitive modeling intervention included the self-monitoring
and self-correcting features of self-regulation.

Of particular interest in this study is the individual’s self-efﬁcacy for handling
the task if it changes and becomes more complex. Thus, self-efﬁcacy in this study is
targeted at an individual’s conﬁdence that he or she can successfully perform the task
under circumstances not experienced before. This deﬁnition of self-efﬁcacy was
chosen to tap into the idea that self-efﬁcacy may have a resiliency component
(Bandura & Cervone, 1986). Bandura and Wood (1989) argued that self-efﬁcacy
provides resilience in the face of failures and setbacks, and predisposes individuals to
view obstacles as challenges rather than reﬂections of personal inadequacies. This
resilience and persistence is expected to be an important inﬂuence on the capability to
succeed in the adaptive transfer task. Thus, self-efﬁcacy for unseen or novel task
situations should enable individuals to persist in the challenging adaptive transfer task.

It is expected that individuals who have received instruction on self-regulation
of learning will develop greater self-efﬁcacy compared to individuals who have not

received this guidance. Individuals who have developed proﬁciency in monitoring and

77

diagnosing their performance should believe that they have certain skills that will assist
them in adapting to novel and complex circumstances. Therefore, they should feel
more conﬁdent that they can adapt to changing task demands compared to individuals
who have not received instruction on these self-regulatory skills.

Hypothesis l4. Metacognitive instruction should lead to signiﬁcantly greater
self-efﬁcacy at the end of training compared to no metacognitive instruction.

The discovery learning literature has not examined self-efﬁcacy as an outcome
of instruction. Therefore, no clear hypotheses can be derived from past literature on
possible relationships between the two variables. In fact, it is not clear that the two
guided discovery conditions will lead to greater self-efﬁcacy compared to the
procedural instruction group. For example, the guided discovery group and the
procedural instruction group may develop similar self-efﬁcacy beliefs by the end of
training. However, their self-efﬁcacy beliefs are based on different task experiences.
The guided and pure discovery groups are likely to base their self-efﬁcacy perceptions
on skills that are more useful for adaptive transfer. The procedural instruction group
may feel self-efﬁcacious as well at the end of their training, because they have been
explicitly guided through strategies to deal with the task at various levels of
complexity. However, their perceptions of self-efﬁcacy are likely to be based on
insufﬁcient skills for adaptive transfer. Nonetheless, no clear difference in the level of
self-efﬁcacy is predicted for different discovery learning groups.

Finally, self-efﬁcacy is expected to lead to greater adaptive transfer.
Individuals who are more conﬁdent in their capability to handle novel circumstances

Should show greater resilience when faced with the demands of the adaptive transfer

78

task. Previous research provides support for the prediction that self-efﬁcacy will
positively affect transfer to a more complex task (Kozlowski et al., 1995).

Hypothesis 15. Self-efﬁcacy will be positively related to adaptive transfer.

In summary, it is generally expected that guided discovery learning and
metacognitive instruction will be effective instructional interventions for achieving
various learning outcomes and adaptive transfer. In addition to these design factors,
the conceptual model examines how individual differences may also impact learning
outcomes and adaptive transfer. These factors are described next.

Individual Difference Factors

This study will examine how individual differences in ability, personality and
motivation for learning affect learning outcomes and adaptive transfer. Cognitive
ability has been shown to interact with the amount of structure in the learning
environment to affect learning outcomes. In addition, mastery and performance
orientations to learning were identiﬁed as motivational factors that may impact
learning and transfer. Finally, tolerance for ambiguity was identiﬁed as a personality
construct that is relevant to situations where individuals must adapt to complex or
novel circumstances.

Cognitive ability. Research has shown that cognitive ability affects early skill
acquisition (Ackerman, 1988). Therefore, the present study will measure and control
for the effects of ability on adaptive transfer. Ability should also impact the learning
outcomes of knowledge, knowledge structure, hypothesis-testing, self-regulatory

activity, and self-efﬁcacy.

79

Of greater interest in this study is whether certain dispositional factors will
affect learning and adaptive transfer after controlling for the effects of ability. The
goal orientation of the learner and tolerance for ambiguity were identiﬁed as individual
differences that may affect learning and transfer of knowledge and skills.

Goal orientation factors. Mastery and performance orientations to learning
represent different ideas of success and different reasons for engaging in learning
(Ames, 1992; Dweck, 1986). A mastery orientation includes the belief that effort
leads to improved outcomes, and that ability is malleable. Individuals with a mastery
orientation engage in learning activities with the purpose of trying to understand new
tasks (Ames, 1992; Dweck, 1986). In contrast, individuals with a performance
orientation to learning believe that ability is demonstrated by performing better than
others; therefore, a learning situation is a means through which the individual can
publicly achieve greater success compared to others (Ames, 1992).

Research indicates that a performance orientation is related to the belief that
success requires high ability, while a mastery orientation is related to the belief that
success requires interest, effort, and collaboration (Duda & Nicholls, 1992). Kroll
(1988) found that mastery orientation is positively related to tolerance for ambiguity,
thoughtfulness, and open-mindedness. Performance orientation is negatively related to
tolerance for ambiguity, thoughtfulness, complexity, and individualism.

Classroom settings emphasizing mastery goals lead students to use more
effective learning strategies, to prefer challenging tasks, to have a more positive
attitude towards the class, and to have a stronger belief that success follows from

effort. In contrast, classrooms emphasizing performance goals lead students to focus

80

on their ability, to evaluate their ability negatively, and to attribute their failures to
lack of ability (Ames & Archer, 1988). Mastery goals lead to greater persistence in
the face of difﬁculties, whereas performance goals led to the avoidance of challenging
tasks (Dweck, 1986; Dweck & Leggett, 1988; Elliott & Dweck, 1988).

Two studies have shown the impact of mastery and performance orientations on
the learning of a complex, decision-making task. Smith et al. (1995) found mastery
orientation to be positively related to metacognitive activity during training.
Kozlowski et a1. (1995) and Smith et a1. (1995) found mastery orientation to be
positively related to self-efﬁcacy at the end of training. In addition, Smith et al.
(1995) found performance orientation to be negatively related to self-efﬁcacy at the
end of a training program where individuals were responsible for choosing practice
exercises. The two goal orientation factors were not related to transfer of training in
either study after controlling for instructional factors and learning outcomes. It should
be noted that performance orientation was negatively correlated with transfer
performance in the Smith et a1. (1995) study, but this relationship became
nonsigniﬁcant after controlling for learning strategies and learning outcomes.

Based on theoretical distinctions between the two goal orientation factors, it is
expected that mastery orientation will be positively related to both hypothesis-testing
activity and self-regulatory activity during training. Previous research has shown that
mastery orientation is related to various learning strategies for trying to understand
new tasks such as metacognitive activity during training (Ames & Archer, 1992; Smith
et al., 1995). Individuals with a mastery orientation tend to use deep processing

strategies that require cognitive effort but lead to understanding; in contrast,

81

performance orientation is related to more short-term and surface-level processing
strategies (Meece, 1994). Individuals with a mastery orientation to learning are more
likely to formulate hypotheses about the task and test them out during practice. They
are also more likely to monitor and evaluate their progress to assess their level of
understanding. Consistent with previous ﬁndings, performance orientation is not
expected to inﬂuence the use of hypothesis-testing and self-regulatory activities.

Hypothesis l6. Mastery orientation will be positively related to hypothesis-
testing activity.

Hypothesiéll Mastery orientation will be positively related to self-regulatory
activity.

Mastery orientation is also expected to be positively related to general
knowledge about the task, as well as the individual’s knowledge structure. Because
mastery-oriented individuals are focused on learning new skills and understanding the
task, they should develop a better understanding of the task overall, as well as a better
understanding of the relationships among concepts and actions they are presented
during training. Consistent with previous research, performance orientation is not
expected to inﬂuence an individual’s knowledge or knowledge structure.

Hypothesis 18. Mastery orientation will be positively related to verbal
knowledge.

Hypothesis 19. Mastery orientation will be positively related to knowledge
structure.

Mastery goal orientation is also expected to impact self-efﬁcacy at the end of
training. Individuals who are focused on developing new skills during training are

likely to develop greater conﬁdence in their abilities at the end of training. In fact,

82

previous research has found a positive relationship between mastery orientation and
self-efﬁcacy (Smith et al., 1995).

Hypothesis 20. Mastery orientation will be positively related to self—efﬁcacy.

Previous research has shown inconsistent results for the impact of performance
orientation on self-efﬁcacy. While Kozlowski et a1. (1995) found little or no impact
of performance orientation on self-efﬁcacy, Smith et al. (1995) found a negative
relationship between performance orientation and self-efﬁcacy. In the Smith et al.
study, individuals were required to choose the level of complexity and sequence of
exercises they were to practice. Individuals with a higher performance orientation may
have had difﬁculty in taking this active responsibility for learning activities, and thus
were less conﬁdent in their capabilities by the end of training. In the present study,
the learning activities and sequence of exercises are prescribed for the learner;
therefore, performance orientation is not expected to have an impact on self—efﬁcacy.
No speciﬁc hypotheses are made concerning the role of performance orientation in
learning and transfer. Instead, any effects for performance orientation will be
controlled for prior to testing study hypotheses.

Tolerance for ambiguity. The term tolerance for ambiguity can be traced to
Frenkel-Brunswik (1949), who used this concept to describe an individual’s emotional
and cognitive orientation towards life. She was interested in whether individuals who
were prejudiced or rigid in their attitudes would show similar rigidity on perceptual
tasks. Budner (1962) was interested in tolerance for ambiguity as a construct in its
own right, and developed a deﬁnition of this construct in terms of several component

dimensions. He deﬁned intolerance of ambiguity as ". . . the tendency to perceive (i.e.

83

interpret) ambiguous situations as sources of threat," and tolerance of ambiguity as ". .
. the tendency to perceive ambiguous situations as desirable" (Budner, 1962, p. 29).

He described three types of situations that deﬁne ambiguity: (1) a new situation where
there are no familiar cues; (2) a complex situation where a great number of cues must
be considered; and (3) a contradictory situation in which different elements suggest
different structures. In other words, ambiguous situations are characterized by novelty,
complexity, or insolubility (Budner, 1962).

Budner (1962) developed a scale to measure intolerance for ambiguity, and
found it to be positively related to measures of authoritarianism and conventionalism,
and negatively related to Machiavellianism. He also found a marginally signiﬁcant
tendency for medical students intolerant of ambiguity to choose relatively structured
specialties, and those tolerant of ambiguity to choose relatively unstructured specialties.

Crandall (1968) provided additional evidence for the task preferences of
individuals tolerant and intolerant of ambiguity. Participants in the study were
required to learn pairs of CVC syllables. Crandall (1968) found that individuals
intolerant of ambiguity preferred stimuli that were familiar and had the strongest
conﬁrmation value. In contrast, individuals tolerant of ambiguity lost interest in
stimuli at higher levels of repetition. Crandall (1968) interpreted the results to indicate
that individuals intolerant of ambiguity prefer stimuli that are useful for providing
closure or consolidating knowledge, rather than expanding it. Individuals tolerant for
ambiguity welcome extension of their cognitive system, whereas individuals intolerant

of ambiguity prefer deﬁnition of their systems.

84
Blake, Perloff, Zenhausen, and Heslin (1973) found that individuals intolerant

of ambiguity viewed atypical consumer products as newer than individuals tolerant of
ambiguity. Perceived product newness was positively related to willingness to buy
among tolerant participants, but negatively related to willingness to buy among
intolerant participants (Blake et al., 1973). Frone (1990) conducted a meta-analysis to
examine intolerance of ambiguity as a moderator of the occupational role stress-strain
relationship. He expected that the role stress-strain relationship would be stronger
among high intolerant of ambiguity individuals (IOA) than low IOA individuals.
Results supported the moderating role of tolerance for ambiguity (Frone, 1990).

Surprisingly, very little research has been conducted that examines the
relationship between tolerance for ambiguity, instructional methods, and learning
outcomes (Jonassen & Grabowksi, 1993). Ebeling and Spear (1980) did examine
whether tolerance for ambiguity would impact performance on two tasks of varying
ambiguity (a decoding task v. a creativity task of thinking of different uses for
common objects). They found that individuals high on tolerance for ambiguity
performed better on both levels of task ambiguity compared to individuals low on
tolerance for ambiguity. Even though previous research has shown that tolerance for
ambiguity is in fact related to preferences for novel and ambiguous stimuli, Ebeling
and Spear’s (1980) study suggests that high tolerance for ambiguity individuals may
perform tasks better regardless of their task preferences.

Tolerance for ambiguity is an important factor to consider when examining
adaptive transfer. Individuals high on tolerance for ambiguity welcome the challenge

of ill-structured, novel, and complex situations. It is expected that tolerance for

85

ambiguity will be positively related to adaptive transfer. Individuals who view
ambiguous situations as desirable are likely to perform well in a transfer situation that
requires them to learn about and adapt to a new feature of the task. Individuals low
on tolerance for ambiguity may have difﬁculty adjusting to new task demands.

Hypothesis 21. Tolerance for ambiguity will be positively related to adaptive
transfer.

The literature on tolerance for ambiguity has not examined what mechanisms
lead to greater learning or transfer by individuals high on tolerance for ambiguity.
Researchers have suggested that individuals tolerant of ambiguity will perform well
when faced with situations that are novel or unstructured, because they are able to
hypothesize well and provide their own structure. It is assumed that individuals high
on tolerance for ambiguity are better able to make logical inferences concerning task
cues, and that these inferences facilitate success in task performance (Chapelle &
Roberts, 1986). In contrast, individuals low on tolerance for ambiguity may have
difﬁculty providing their own structure to ambiguous learning situations (Jonassen &
Grabowski, 1993). Thus, one mechanism that may account for the positive
relationship between tolerance for ambiguity and adaptive transfer is hypothesis-testing
skill. Individuals high on tolerance for ambiguity may engage in greater hypothesis
testing while learning tasks that enables them to perform well when faced with novel
circumstances. Previous research has not tested the learning mechanisms that facilitate
the success of high tolerance for ambiguity learners (Chapelle & Roberts, 1986), and
the present study will examine whether tolerance for ambiguity does in fact lead to

greater hypothesis-testing during learning.

86

Hypothesis 22. Tolerance for ambiguity will be positively related to
hypothesis-testing skill.

It is expected that tolerance for ambiguity will be related to knowledge
acquired during training. Researchers suggest that individuals low on tolerance for
ambiguity may be more detail-oriented and unable to view situations in global terms
(Jonassen & Grabowski, 1993). Chapelle and Roberts (1986) provided some indication
that tolerance for ambiguity is related to knowledge. They found that tolerance for
ambiguity was positively related to proﬁciency in English as a second language for
foreign students. After controlling for initial proﬁciency, anxiety, and motivation,
tolerance for ambiguity was positively related to the acquisition of English structure
and listening comprehension by the end of the semester. Therefore, it is expected that
tolerance for ambiguity will be related to verbal knowledge about the task as assessed
on a test at the end of training. A tentative hypothesis is made that tolerance for
ambiguity will be related to the quality of one’s knowledge structure for a new task.
Individuals low on tolerance for ambiguity may develop a less-integrated knowledge
structure because they have difﬁculty viewing a complex task in global terms.

Hypothesis 23. Tolerance for ambiguity will be positively related to verbal
knowledge.

Hypothesis 24. Tolerance for ambiguity will be positively related to knowledge
structure.

A ﬁnal learning outcome that may be inﬂuenced by tolerance for ambiguity is
self-efﬁcacy. It is suggested that tolerance for ambiguity is related to motivation and
persistence in the face of ambiguity (Jonassen & Grabowski, 1993). Individuals who

welcome and are challenged by novelty, ambiguity, and complexity are likely to feel

87

more self-conﬁdent in their capabilities to adapt to these types of situations. In
contrast, individuals low on tolerance for ambiguity are characterized as those who do
not enjoy taking risks (Birckbichler & Omaggio, 1978). Thus, they may report low
self-efﬁcacy for being able to succeed in ambiguous situations.

Hypothesis 25. Tolerance for ambiguity will be positively related to self-
efﬁcacy.

Table 1 summarizes the hypotheses to be tested in this study. Although each
hypothesis could be tested separately as a univariate relationship between factors, this
study will test sets of hypotheses in a more integrated fashion. Speciﬁcally,
hypotheses concerning the same dependent variable will be tested in one overall
analysis. In this way, a particular hypothesis will be tested after controlling for all
other relevant factors in the model. For example, hypotheses 4, 16, and 22, which are
concerned with factors that inﬂuence hypothesis-testing activity, will be tested in one
overall analysis. This will provide a rigorous test of speciﬁc relationships in the
conceptual model.

In addition, the conceptual model presented in Figure 1 indicates that the
learning outcomes are expected to mediate the relationship between individual
differences, training manipulations, and adaptive transfer. While not stated as an

explicit hypothesis, this mediation will also be tested for in the analyses.

88
Table 1

Summary of Study Hypotheses

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Hypothesis Independent Variables Dependent Variable
1 Discovery Learning (Pure discovery v. guided Adaptive transfer performance
discovery v. procedural instruction)
2 Metacognitive Instruction (yes or no) Adaptive transfer performance
3 Discovery Learning X Metacognitive Instruction Adaptive transfer performance
4 Discovery Learning (Pure discovery v. guided Hypothesis-testing activity
discovery v. procedural instruction)
5 Hypothesis-testing activity Adaptive transfer
6 Metacognitive Instruction (yes or no) Self—regulatory activity
7 Self-regulatory activity Adaptive transfer
8 Discovery Learning (Pure discovery v. guided Verbal knowledge
discovery v. procedural instruction)
9 Metacognitive Instruction (yes or no) Verbal knowledge
10 Verbal knowledge Adaptive transfer
1 1 Discovery Learning (Pure discovery v. guided Knowledge structure
discovery v. procedural instruction)
12 Metacognitive Instruction (yes or no) Knowledge structure
13 Knowledge structure Adaptive transfer
14 Metacognitive Instruction (yes or no) Self-efﬁcacy
15 Self-efﬁcacy Adaptive transfer
16 Mastery orientation Hypothesis-testing activity
17 Mastery orientation Self-regulatory activity
18 Mastery orientation Verbal knowledge
19 Mastery orientation Knowledge structure
20 Mastery orientation Self-efﬁcacy
21 Tolerance for ambiguity Adaptive transfer
22 Tolerance for ambiguity Hypothesis-testing activity
23 Tolerance for ambiguity Verbal knowledge
24 Tolerance for ambiguity Knowledge structure
'1 25 Tolerance for ambiguity Self-efﬁcacy

 

 

 

 

METHOD

Sample

Participants were undergraduate students at Michigan State University enrolled
in psychology courses who received extra credit in their course for participation in the
experiment. One hundred sixty—nine individuals participated in this experiment.
However, eight participants were eliminated from the sample. Six participants were
eliminated because of extreme difﬁculty in understanding the task. Two other
participants were eliminated after it was discovered that they did not engage any
targets on the transfer task (it appeared that they were looking up information
randomly but not making a single decision during the nine-minute transfer trial).
Therefore the sample used in this study included 161 participants, with cell sizes
ranging from 26 to 29 individuals.
136—sign

The study was a 3 (discovery instruction) X 2 (metacognitive instruction) fully-
crossed factorial design. The three levels of discovery instruction were pure discovery,
guided discovery, and procedural instruction. The second factor was the presence or
absence of metacognitive instruction. A power analysis was conducted to determine
the sample size necessary to detect a moderate effect size with power of .80, and a

signiﬁcance level of .05 (Cohen, 1977). For a 3 X 2 factorial design, cell sizes of 25

89

90

(i.e., a total sample size of 150) would result in power of .78 for detecting an
interaction using p < .05 as the level of signiﬁcance for rejecting the null hypothesis.
Based on this power analysis, the sample size of 161 participants was judged to be
sufﬁcient for examining the study hypotheses.
Las_k

The task individuals learned was a revised version of TANDEM (Tactical
Naval Decision Making System; Dwyer, Hall, Volpe, Cannon-Bowers, & Salas,
1992). TANDEM is a simulation software program that depicts targets on a screen.
Trainees were placed in the role of Radar Operator of a US. Navy Aegis-class cruiser.
The Operator was to "hook" a target on the radar screen and then collect information
to classify the target’s Type, Class, and Intent. Then the Operator decided to shoot
hostile targets and clear peaceful targets from the screen. The goal of the task was to
correctly identify and process each target in the shortest amount of time possible.
Individuals learned how to prevent targets from entering critical zones surrounding
their own ship. If individuals allowed targets into these "penalty circles," they would
lose points. Individuals learned to check the speed and range for each target to
prioritize the targets they would engage ﬁrst. Trainees practiced task scenarios that
varied in the number of targets, the proportion of dangerous targets surrounding each
penalty circle, and their order of entry into the penalty circles. Three different
scenarios were presented, sequenced from low to moderate to high complexity.

The design of the scenarios focused on increasing the component and
coordinative complexity of the task (Wood, 1986) as they increased from low to

moderate to high complexity. Speciﬁc descriptions of the scenarios are provided in

91

Appendix D (under Procedural Instruction section). As scenarios increased in
complexity, the number of targets surrounding the penalty circles was greater to
increase the component complexity of the task. By increasing the number of targets,
the number of distinct acts required by participants to assess these targets was also
increased. In addition, the coordinative complexity of the task was increased over
scenarios by making the sequence of the prosecution of targets more important. As
scenarios increase in complexity, a greater proportion of targets were dangerous, and
individuals had to zoom in and out more frequently to prosecute targets in the correct
sequence. After training, individuals were tested on an adaptive transfer task that was
different on the dimension of dynamic complexity (Wood, 1986).
Procedure

The experiment was conducted over two consecutive days. The ﬁrst day
comprised the training manipulation. When participants ﬁrst arrived, they read and
signed a consent form that described the experiment (see Appendix A). They then
completed an individual differences questionnaire that assessed their tolerance for
ambiguity, mastery orientation, and performance orientation, as well as certain
demographic and experience variables. Next, participants watched a brief
demonstration by the experimenter on how to perform the basic functions of the task
(hooking targets, accessing information from the menus, making decisions about
targets, and zooming in and out on the radar screen). Participants had ten minutes to
read a short manual describing these functions in more detail (see Appendix B). Then

participants performed a 9 minute trial and a 4.5 minute trial to practice hooking

92

targets, calling up information about these targets to classify their Type, Class, and
Intent, and making Final Engagement decisions.

Participants learned a simpliﬁed version of the task in which they only needed
to access one information cue to make each of the Type, Class, and Intent decisions.
They used a table from the manual to determine the correct classiﬁcation to choose for
each of the Type, Class, and Intent decisions. This portion of the task was simpliﬁed
so that the decision criteria and decision sequence could be learned quickly. The
training manipulations focused on a more complex skill, prioritization of targets to
prevent them from entering two penalty circles.

After individuals practiced the basic rules for classifying targets, the discovery
learning manipulation was introduced. All participants were told that the goal of the
next portion of training was to learn how to prevent targets from entering two penalty
circles surrounding their ship, and that they would practice scenarios at low, medium,
and high levels of complexity (see General Task Instructions in Appendix C). After
this general introduction to the training, participants received instructions for the pure
discovery, guided discovery, or procedural instruction condition (explained in more
detail in the Discovery Learning Manipulation section below). Participants then
practiced the low complexity scenario for three trials. The number of practice trials
that were sufﬁcient for individuals to learn each scenario were determined in a pilot
study. Each practice trial lasted 4.5 minutes. For the low complexity trials, the score
was turned off during the scenario. This was done so that all groups were focused on

a mastery—oriented approach to learning and were not too focused on their performance

9

'JJ

during initial learning. At the end of each practice trial. participants received feedback
concerning the number of targets they allowed to enter each penalty circle.

After practicing the low complexity scenario. participants in the metacognitive
instruction condition received training on how to self-regulate their learning. and why
it was important (see Metacognitive Instruction section for more details). In addition.
participants were presented with instructions for the medium complexity scenario that
varied by discovery learning condition. For the moderate and high complexity
scenarios, the score was turned back on during the practice trials so that individuals
could monitor their progress.

Participants completed one trial of the medium complexity scenario. For the
metacognitive instruction group, participants were presented with the ﬁrst set of
questions they were to answer to regulate their learning. Participants completed three
additional practice trials, with metacognitive questions answered between trials for the
metacognitive instruction group.

After the medium complexity scenario, participants received a short break.
Finally, participants were presented with the high complexity task scenario. Before
practice, participants received instructions consistent with their discovery learning
condition. After the ﬁrst practice trial, participants in the metacognitive instruction
condition began answering questions to guide them in self-regulating their learning.
Participants completed a total of 5 trials to practice the high complexity scenario.

When participants ﬁnished practicing the high complexity scenario. they
completed a questionnaire concerning their use of hypothesis-testing and self-regulation

during the training session. This ﬁrst day of training lasted for 3 hours.

94

Participants returned one day later to perform on the adaptive transfer task. On
the second day of the experiment, participants ﬁrst completed a self-efﬁcacy
questionnaire. They then completed ratings to assess their knowledge structure,
followed by a short, multiple-choice verbal knowledge test. Participants next
completed a test of general cognitive ability. Finally, they performed the adaptive
transfer task, which was 9 minutes in length. The second experimental session took
one hour to complete.

Discovery Learning Manipulation

The discovery learning manipulation determined the type of learning strategy or
approach used by the learner to develop an understanding of the task. Participants in
the pure discovery condition received instructions that told them they would face a
scenario of low complexity. They were instructed to explore the task in order to
discover the best strategy to deal with the situation and prevent targets from entering
the penalty circles. Similar instructions were presented prior to practice on the
medium and high complexity scenarios as well.

Participants in the guided discovery condition were presented with an explicit
description of the low complexity scenario, and the task strategies they should use to
handle it. However, for the ﬁrst low complexity scenario trial, participants were asked
to just check speed and range on targets and learn where the penalty circles were.
They were told not to make decisions about targets so that targets remained on the
screen and participants would be able to learn what happens when targets entered the
penalty circles. Then for the remaining two low complexity trials, participants were

asked to practice the task strategy appropriate for that scenario.

95

For the medium and high complexity scenarios, the guided discovery group was
directed to hypothesize what aspects of the task might change to increase the
complexity of the task. Then they were asked to identify how they would respond to
these changes in their strategy for performing the task. They recorded their answers to
these questions on a sheet provided to them. They were instructed to explore the task
to test out their hypotheses and their task strategies. In this way, they could determine
what were the important features of the scenario and how they should respond to them.
Thus, individuals in the guided discovery condition were provided with explicit
procedural instruction on the low complexity scenario to provide a knowledge base for
their own hypothesis-testing on the medium and high complexity scenarios.

Participants in the procedural instruction condition received explicit description
of the scenario and the task strategy to use for each of the low, medium, and high
complexity scenarios. They were instructed to focus their attention on practicing the
task strategy outlined for them. The speciﬁc instructions provided to each instructional
condition are detailed in Appendix D.

Metacognitive Instruction

The metacognitive guidance provided learners with a way to improve the
effectiveness of their performance strategies to handle the different task situations they
faced. Individuals in the metacognitive instruction condition received training on
planning, monitoring, and evaluating their task strategies. Metacognitive instruction
was introduced before practice of the moderate complexity task. Participants were
taught what it meant to plan, monitor and evaluate their task performance in order to

ﬁnd ways to improve it. Consistent with previous research on metacognitive

96

instruction, participants were also told why these self-regulatory processes are
important. The speciﬁc instructions provided to participants can be found in Appendix
B. After the ﬁrst trial of practicing the moderate complexity task, participants in the
metacognitive instruction condition were asked to record how many intrusions they
allowed in each penalty circle on a sheet provided to them. Participants were asked
to think about their feedback from the last trial, and identify any difﬁculties or
problems upon which they could improve. They were then asked to develop a
learning goal for the next trial to improve their understanding, and a plan for
accomplishing that goal. They recorded these on the sheet provided (see Appendix E).
They were told to monitor their practice with regard to their goal. This process was
repeated three times for the moderate complexity scenario and four times for the high
complexity scenario trials.

Participants in the condition not receiving metacognitive instruction were
instructed that they could do as they like to prepare for the next practice trial, and they
sat quietly between trials for several minutes. For initial trials, where the self-
regulation questions took longer to answer, participants in the no-metacognition
instruction group were given three minutes to sit quietly. For later trials, where the
self-regulation questions were answered more quickly, participants in the no-
metacognition instruction condition had two minutes to sit quietly between trials.
Measures

Participants completed several survey measures at the beginning and end of the
training program. Cognitive ability was assessed with a computerized test on the

second day of the experiment. Two measures of knowledge were also collected the

97

day after the training portion of the experiment. Finally, performance on the adaptive
transfer test was assessed on the second day of the experiment. Table 2 presents the
means, standard deviations, reliabilities, and intercorrelations of the measures used in
this study. The speciﬁc measures and their development are described next.

Participants completed a questionnaire at the beginning of the experiment,
which included questions with regard to their age, sex, GPA, and previous video game
experience. A problem with the GPA measure arose during the study, as many
participants were freshmen in their ﬁrst semester of college so that they did not yet
have a GPA. This variable, therefore, was not included in the study. Speciﬁc
questions can be found in Appendix F.

GLder. Participants were asked to indicate their gender. Gender was coded as
1 = male and 2 = female.

Agp. Participants indicated one of ten categories for their age. See Appendix
F for the speciﬁc categories which ranged from "17 or younger" to "26 or older."

Video game experience. Participants answered the item "how often do you
play with video games" using a 5-point scale ranging from (1) never to (5) always.

Cognitive ability. General cognitive ability was measured with a computerized
version of the Wonderlic Personnel test. This was a short form test of general
cognitive ability, consisting of 50 items arranged in order of difﬁculty. The items
included word comparisons, disarranged sentences, sentence parallelism, following
directions, number comparisons, number series, analysis of geometric ﬁgures, and

math or logic story problems. Participants were given 12 minutes to complete as

98

 

3.364. Eo.m.=wom....om

 

83 8. :8. 8.-....8. tom. :8. S.- 8.- :8. 8. agave-£358»: .:
S. 8. 8. :8. 8. .8. 8.- S.- 8. 232:8 8.8.55. .o.
98 8.- ~.. :8. 8. S. 8. S. :8. uwco.3o=v..8._o> .8
5.3.48. 8.- 8.- o..- 8.- 2.- 8. 88.5.5 005.585.. .w
83 .2. :R. 2.- 8. 8. 8. 8.3.5.5 bass. .5
8. 8. 8. S. :8. be? .8
A8,. 3.- .o.- 5.- :. b.8555. 8:803 .m
1. S. .or vo. acesosbmﬁo>£=woom62 .v
--- Iomr 3. 100585 2?. .m
8.- .5885 82.6 .m
t- Boom comment. 3333‘ ..
2 m. E m. N. : o. a w s o m a. m N _

 

22.22.0885 was .8....Em:om £83359 2.6.55 .28.).

N 035.

99

.o.nZ

.o n 5.8.585 o>...:woo$o$. oz can . H 5.888.: o>...:w882>. 88 ounce 8. 5.83.8:— o>_..:woo§o$. ..
.o u 5.83.2: .8338... can .. u Eczema and .o n @3885 32:0 .8... cm cocoa-88:5“.
8. 8.3085 2:“. .o u 5.8385 85.8805 use d u b83085 2.5 J u b30085 32:0 28:3 838...; cocoa->862. a 8. b.2685 “882:0 .

 

 

8.8 8.. 8... 88 8.8 8.8 8.8 8.8 E... 8.8 8... 8.8 88.8 8.8 :8 8.88.. am
88 8.8 88.. 8.: 8.8 .88 :8 88.8 .88 88.8 8.8 t... .8... 88.8 8.8 8.88 s.
8.-....8.-t8. E. 3. 8. :8. :.- 8... 8. .2. 8. :. 8.-....8. 85:89.8 2:8 885.8.
...:.-:8.- 8. 8. E. 8.- 8. :8. 8. 8.- z.- 8. 8.- :88.- 88.8.8.
:.- .2.- 8..- 8. 8.- 8.. 8.- 8.- .8..- 8.- .8. 8. .8..- 585.:
8.. 8. 8. :88. .8. 8. :8. 8. 8. 8.. 8.-....8. 3885.8 8:38.588.
.88....8. 8. 8. E. :8. .8. :8. 8.- 8.- 8. 8.. 8888-88....

8. 8. E m. N. : 8. 8 8 8 8 8 8 m m .

 

8.88. m 8.888

100

many items as possible. Scores on the Wonderlic test have been shown to be highly
related to scores on longer tests of cognitive ability. In addition, test-retest reliabilities
have ranged from .82 to .94, and internal consistency reliabilities (based on correlating
odd and even items) have ranged from .88 to .94 (Wonderlic Personnel Test &
Scholastic Level Exam User’s Manual, 1992). Participants completed this test on the
second day of the experiment.

Participants completed a questionnaire at the beginning of the experiment to
assess three individual difference factors: tolerance for ambiguity (10 items); mastery
goal orientation (8 items); and performance goal orientation (8 items). These
individual differences are conceptually distinct. However, to test the empirical
independence of these three constructs, the 26 items making up these three measures
were subject to a common factor analysis with varimax rotation. A three-factor
solution was speciﬁed. Table 3 displays the rotated factor matrix.

The three factors accounted for 42 percent of the total variance. The predicted
factor structure was maintained to some extent. The ﬁrst factor was made up of
mastery orientation items. The third factor was made up of tolerance for ambiguity
items. However, the second factor was made up of all performance orientation items
along with four items from the tolerance for ambiguity scale. Item 6 from the
tolerance for ambiguity scale also loaded on both the ﬁrst factor and the third factor.

In addition, the scree plot indicated the possibility of a six-factor solution. A
factor analysis was run with a six-factor solution as well; results of the six factor
solution did not conform well to the conceptualization of these constructs. The

mastery orientation and performance orientation items separated into four factors, and

101
Table 3

Rotated Factor Matrix for Individual Differences Items

Factor

1 2 3
Mastery Orientation 5 ,7_ .21 .02
Mastery Orientation 6 .__2._ .13 -.03
Mastery Orientation 4 i -.14 .24
Mastery Orientation 7 =_6_4 .01 .28
Mastery Orientation 3 ._6_1_ .25 -.01
Mastery Orientation 8 ;5_8_ .19 -.05
Mastery Orientation 1 i8 .03 .19
Mastery Orientation 2 i .01 .25
Performance Orientation 5 -.06 ._65 -.14
Performance Orientation 8 .26 ._6; -.13
Tolerance for Ambiguity 7a -.01 -_.__l .39
Performance Orientation 6 .15 Q .20
Tolerance for Ambiguity 8a .08 :._5_ .39
Tolerance for Ambiguity 3a .07 -._2 .28
Performance Orientation 4 .16 ._4_9 .03
Performance Orientation 3 .07 ._4_6_ .04
Performance Orientation 2 .12 ,fﬁ .15
Performance Orientation l .03 .__8 .19
Performance Orientation 7 .19 £6 .02
Tolerance for Ambiguity 2" .11 i .20
Tolerance for Ambiguity 10 .14 -.01 ﬂ
Tolerance for Ambiguity 9 .24 -.27 ﬂ
Tolerance for Ambiguity 1 -.06 .00 £0
Tolerance for Ambiguity 6a .38 -.15 4‘1
Tolerance for Ambiguity 4 .03 .05 A;
Tolerance for Ambiguity 5 .17 .13 £6

3 These items were dropped from ﬁnal scales.

102

four tolerance for ambiguity items separated from the performance orientation items.
The ﬁve tolerance for ambiguity items that loaded together in the ﬁrst factor analysis
remained a clear factor in the six-factor solution.

It was decided that the scales would be developed based on the results of the
three-factor solution. The tolerance for ambiguity scale was made up of the ﬁve items
from the third factor in Table 3 that loaded cleanly on this one factor. The ﬁve items
that loaded on other factors were dropped from the scale. The eight items making up
the mastery orientation scale and the eight items making up the performance
orientation scale were retained. This decision was based on several previous studies
that had shown the construct validity and psychometric properties of these two scales
(Button, Mathieu, & Zajac, 1995; Kozlowski et al., 1995; Ford et al., 1995). In
addition, reliability analyses for these scales indicated adequate internal consistency for
the mastery and performance orientation scales.

Table 1 displays the intercorrelations between these three scales. As found in
previous research (Kroll, 1988), tolerance for ambiguity was positively related to
mastery orientation (r = .27, p < .01). Tolerance for ambiguity was not signiﬁcantly
related to performance orientation (r = -.02, p > .05). In addition, consistent with
some research (Kozlowski et al., 1995), mastery orientation and performance
orientation also exhibited a signiﬁcant, positive correlation (r = .26, p < .01).

It should be noted that other research has not found a signiﬁcant relationship
between mastery and performance orientation (Smith et al., 1995). When corrected for
attenuation, the intercorrelations between tolerance for ambiguity, mastery orientation,

and performance orientation were as follows: tolerance for ambiguity and mastery

103

orientation, rtrue = .37; tolerance for ambiguity and performance orientation, rtrue = -.O3;
and mastery orientation and performance orientation, rtrue = .34. These small to
moderate correlations after correcting for attenuation indicate that the scales measure
distinct constructs.

Also, while tolerance for ambiguity and performance orientation were not
signiﬁcantly related to cognitive ability, mastery orientation did exhibit a signiﬁcant,
positive correlation with ability (r = .17, p < .05). These scales also exhibited
differential relationships with gender, age, and previous video game experience. The
differential relationships provide evidence for the construct validity of these three
scales.

Tolerance for ambiguity. Tolerance for ambiguity was measured with a 10-item
scale adapted from Major (1990). Items can be found in Appendix G. Participants
responded to these items on a seven-point scale ranging from (1) strongly disagree to
(7) strongly agree. Based on factor analysis results, the ﬁnal scale was made up of
ﬁve items. Coefﬁcient alpha for this scale was .66.

Mastery orientation. Participants completed an eight-item scale developed by
Button, et al. (1995). Items can be found in Appendix H. Participants responded to
these items on a seven-point scale ranging from (1) strongly disagree to (7) strongly
agree. Coefﬁcient alpha for the scale was .80.

Performance orientation. Participants completed an eight-item scale developed
by Button et a1. (1995). Items can be found in Appendix H. Participants responded to
these items on a seven-point scale ranging from (1) strongly disagree to (7) strongly

agree. Coefﬁcient alpha for the scale was .75.

104

At the end of training, participants completed two scales measuring learning
activities, hypothesis-testing activity (6 items) and self-regulatory activity (6 items). In
addition, when participants returned on the second day of the experiment, they
completed an eight-item self-efﬁcacy scale. These twenty items were subjected to a
common factor analysis. A three-factor solution was speciﬁed and rotated to a
varimax criterion. The rotated factor matrix is presented in Table 4. The three factors
accounted for 63 percent of the total variance. However, the scree plot indicated that
a two-factor solution was also appropriate, and the ﬁrst two factors accounted for 57
percent of the variance.

Table 4 shows that factor 1 included all self-efﬁcacy items. Factor 2 included
all self-regulation items and three hypothesis-testing items. Factor 3 was made up of
the remaining three hypothesis-testing items. However, the items from Factor 3 also
loaded fairly strongly on factor 2 as well. These results suggested that self-regulation
and hypothesis-testing were not distinct constructs. In fact, the correlation between the
two original scales was found to be .74. When corrected for attenuation, this
correlation became .90. Therefore, it was decided to combine these 12 items into one
scale that measured hypothesis-testing/self-regulatory activity.

The correlation between hypothesis-testing/self-regulatory activity and self-
efﬁcacy was .43 (p < .01). After correcting for attenuation, the correlation between
hypothesis-testing/self-regulatory activity and self-efﬁcacy was .49. Although these
scales were signiﬁcantly correlated, Table 2 shows that the hypothesis-testing/self-

fegulation and self-efﬁcacy scales were differentially related to the discovery learning

manipulation. In addition, hypothesis-testing/self-regulatory activity was positively

1 05
Table 4

Rotated Factor Matrix for Learning Activity and Self-Efﬁcacy Items

Factor
1 2 3
Self-Efﬁcacy 6 316 15 .22
Self-Efﬁcacy 5 ._85 14 .08
Self-Efﬁcacy 8 £4 23 .04
Self-Efﬁcacy 3 ,83 11 .02
Self-Efﬁcacy 4 £3 22 -.03
Self-Efﬁcacy 1 ._7_8 12 .12
Self-Efﬁcacy 7 ,1; 21 .23
Self-Efﬁcacy 2 J_2 16 .26
Self-Regulation 1 .16 ﬂ .11
Self-Regulation 4 .00 14 .05
Self-Regulation 5 .16 ﬂ .20
Self-Regulation 6 .20 Q .19
Hypothesis-Testing 4 .32 ﬂ .22
Self-Regulation 2 .23 ‘56 .30
Self-Regulation 3 .21 ii .30
Hypothesis-Testing 5 .16 A6 .23
Hypothesis-Testing 6 .07 ,4_5_ .32
Hypothesis-Testing 3 .15 .47 ,&
Hypothesis-Testing 2 .20 .35 g
Hypothesis-Testing 1 .06 .31 ﬂ

106

related to verbal knowledge (r = .25, p < .01), while self-efﬁcacy was not signiﬁcantly
related to verbal knowledge (r = .05, p > .05). These differential relationships provide
evidence for the construct validity of these scales.

Hypothesis-testing/self-regulatorv activity. Self-reported hypothesis-testing and
self-regulation during training was assessed with 12 items developed for this study.
Although originally conceptualized as separate constructs and scales, factor analysis
and correlational results suggested a single construct. Several items on this scale were
adapted from a metacognition measure developed by Smith et a1. (1995). This scale
measured the extent to which individuals made and tested predictions about changes in
scenario complexity, and explored the task and experimented with different task
strategies. This scale also measured the extent to which individuals set learning goals
and plans for their practice trials, and monitored and evaluated their understanding.
Participants in this study rated items on a 7-point Likert—type scale ranging from (1)
strongly disagree to (7) strongly agree. Items for this scale can be found in Appendix
I. Coefﬁcient alpha for this scale was .90.

Self-efﬁcacy. Self-efﬁcacy was measured on the second day of the experiment
by a modiﬁed version of an 8-item scale developed by Smith, et a1. (1995) and
Kozlowski, et al. (1995). Items were adapted to reﬂect a focus on the subject’s
conﬁdence that they could handle the task if the situation they faced was complex and
one that they had never experienced before. Participants rated the 8 items on a 7-point
Likert-type scale ranging from (1) strongly disagree to (7) strongly agree. Self-

efﬁcacy items are presented in Appendix J. Coefﬁcient alpha for the scale was .85.

107

Verbal knowledge. Participants completed a 14-item multiple choice test to
assess their declarative and procedural knowledge of the task. Questions assessed their
understanding of the location of the penalty circles, the use of the zoom function, and
how to prioritize the order of prosecuting targets. The test questions can be found in
Appendix K. A reliability analysis was conducted to assess the internal consistency of
the test items. The coefﬁcient alpha for the original 14-item scale was .58. Based on
an analysis of internal consistency statistics, 4 items were dropped from the scale
which exhibited low corrected item-total correlations and low squared multiple
correlations. These four items were eliminated from the test to form the ﬁnal 10—item
knowledge test used in the study. Coefﬁcient alpha for this lO-item test was .68.

Knowledge structure. The knowledge structure measure was collected at the
end of training. The organization of each subject’s knowledge was measured by
having them make relatedness ratings between 13 concepts and actions important for
task performance. Thus, participants made a total of 78 ratings indicating how related
they thought two items are on a 9-point scale ranging from (1) unrelated to (9)
related. These ratings were submitted to the Pathﬁnder algorithm to derive a
Pathﬁnder network. Each subject’s network was compared to an expert structure
derived from the primary experimenter. A closeness index, C, was computed for each
subject’s network. This index examined the degree to which the same item in each of
the two networks (subject and expert) was surrounded by a similar neighborhood of
items. This neighborhood comparison was made for each item in the network, and the

results were averaged across all items to compute the overall index of closeness

108

(Goldsmith et al., 1991). This closeness index can range from 0.0 to 1.0. Instructions
and items that were rated can be found in Appendix L.

Reliability information for this measure was not available. However, research
does indicate that individuals who are trained on a task exhibit knowledge structures
that are signiﬁcantly closer to an expert structure compared to a control group that did
not receive training (Kraiger & Salas, 1993). Also, Kozlowski et a1. (1995) found
that, with training, these knowledge structures become more coherent over time.

We mfer. Adaptive transfer was measured by the score achieved on a
novel transfer task. As in training, participants gained 40 points for accurate decisions,
and gained fewer points or lost up to 40 points when some or all the decisions were
incorrect. More importantly, participants lost 100 points for each target allowed into
either of the two penalty circles. While the training trials lasted 4.5 minutes, the
transfer task was 9 minutes in duration. This transfer task required adaptability
because participants were presented with "pop-up" targets that suddenly appeared on
the screen. During training trials, all targets were present on the screen for the whole
scenario, so that once participants assessed each target’s speed and range they would
know the situation that they faced. In contrast, the pop-up targets appearing
throughout the transfer task dynamically altered the situation over the course of the
scenario. In the beginning of the scenario, the targets on the screen indicated that the
most dangerous targets were on the outer penalty circle and that individuals should
focus their attention on prosecuting targets there ﬁrst. However, targets suddenly

appeared at both the inner and outer penalty circles for the duration of the scenario,

Which required individuals to continually assess new targets and adjust their strategy

1

109
for prioritizing targets. These "pop-up" targets required participants to frequently
zoom in and out to prosecute targets on the inner and outer penalty circles. Thus, this
task included a new stimulus feature, pop-up targets, that required individuals to
continually assess and re-assess the situation at each penalty circle. This scenario was
higher on dynamic complexity (Wood, 1986) compared to the training scenarios. A
task high on dynamic complexity requires individuals to frequently adapt to changes in
cause-effect relationships during performance of the task. Instructions for the adaptive

transfer task are found in Appendix M.

RESULTS
Data Analysis

The psychometric properties of the tolerance for ambiguity, hypothesis-
testing/self-regulatory activity, and self-efﬁcacy scales were assessed prior to testing
the conceptual model. In addition, it was found that two participants failed to
complete the six items assessing self-regulatory activity. Their value on the
hypothesis-testing/self-regulation scale was based on their answers to the six
hypothesis-testing items.

The hypotheses in this study were tested using hierarchical regression analyses
(Cohen & Cohen, 1983). In all analyses, cognitive ability and performance orientation
were entered in the ﬁrst step as control variables. In addition, variables prior to or
concurrent with variables in the model were entered to control for the effects of all
other factors when assessing the hypothesized relationships. Variables were entered in
an order consistent with the conceptual model in Figure 1. Individual difference
factors were entered into the regression ﬁrst because they were conceptualized as
stable characteristics that individuals brought to the training. The discovery learning
and metacognitive instruction manipulations were entered next to examine the role of
training on the dependent variable. The discovery learning manipulation was
represented by two dummy-coded variables. First, guided discovery was a variable
coded 1 for the guided discovery group and 0 for the other two groups. Pure

discovery was coded 1 for the pure discovery group and 0 for the other two groups.

110

lll

Metacognitive instruction was also coded as 1 for the presence of the instruction and 0
for its absence from the training.

Verbal knowledge. Hierarchical regression analysis was used to examine how
individual differences, discovery learning, and metacognitive instruction inﬂuenced the
acquisition of verbal knowledge about the task. The ﬁrst step in the regression
controlled for the effects of cognitive ability and performance orientation. The second
step in the regression contained the individual differences of tolerance for ambiguity
and mastery orientation. The third step assessed the main effects for the discovery
learning and metacognition manipulations. Table 5 presents results of the analysis.

Results indicated that cognitive ability and performance orientation accounted
for a signiﬁcant amount of variance in verbal knowledge (R2 = .21, F = 20.93, p <
.01). Beta-weights indicated that cognitive ability was positively related to verbal
knowledge (B = .45 at step 1). Hypotheses 18 and 23 predicted that mastery
orientation and tolerance for ambiguity, respectively, would positively affect verbal
knowledge. The second step in the regression, containing these individual difference
factors, did not account for a signiﬁcant amount of variance in verbal knowledge.
Hypotheses 18 and 23 were not supported by these results.

Hypothesis 8 predicted that guided discovery would lead to greater verbal
knowledge compared to pure discovery, which would lead to greater verbal knowledge
compared to procedural instruction. Hypothesis 9 predicted that metacognitive
instruction would lead to greater verbal knowledge compared to no metacognitive
instruction. The third step in the regression, which contained the main effects for

these training manipulations, did not account for a signiﬁcant amount of variance in

l 12
Table 5

Hierarchical Regression Analysis Results for Verbal Knowledge

 

 

Step Predictors [3 at Step Final [3 R2 AR2
1 Ability .45“ .43"

Performance Orientation -.04 -.04 .21** .21"
2 Tolerance for Ambiguity -.01 .01

Mastery Orientation .07 .06 .21“ .00
3 Guided Discovery .10 .10

Pure Discovery .07 .07

Metacognitive Instruction .05 .05 .22" .01

 

*ps.05;**ps.01

113

verbal knowledge. Therefore, hypotheses 8 and 9 were not supported by the
regression results. Overall, the variables in the regression accounted for 22 percent of
the variance in verbal knowledge (F(7,153) = 6.30, p < .01).

Knowledge structure. A hierarchical regression analysis was run to examine
how individual difference factors, discovery learning manipulations, and metacognitive
instruction inﬂuenced the structure of knowledge developed about the training task.
The effects of cognitive ability and performance orientation were controlled for in the
ﬁrst step in the regression analysis. Tolerance for ambiguity and mastery orientation
were entered in the second step, followed by the training manipulations in the third
step in the regression.

Results are presented in Table 6. Cognitive ability and performance orientation
accounted for a signiﬁcant amount of variance in knowledge structure (R2 = .06, F =
4.91, p < .01), and beta-weights indicated that ability was the inﬂuential factor (B =
.24). Hypotheses 19 and 24 predicted that mastery orientation and tolerance for
ambiguity would be positively related to knowledge structure. The second step in the
regression, containing these factors, did not account for a signiﬁcant amount of
variance in knowledge structure. These results did not support hypotheses 19 and 24.

Hypothesis 11 predicted that guided discovery would lead to a better knowledge
structure compared to pure discovery learning, which would lead to a better knowledge
structure than procedural instruction. Hypothesis 12 predicted that metacognitive
instruction would lead to a better knowledge structure compared to no metacognitive
instruction. The step containing the training manipulations did not account for

signiﬁcant variance in knowledge structure, which did not support hypotheses 11 and

1 14
Table 6

Hierarchical Regression Analysis Results for Knowledge Structure

 

 

Step Predictors B at Step Final B R2 AR2
1 Ability .24" .24"

Performance Orientation .05 .04 .06" .06“
2 Tolerance for Ambiguity .04 .03

Mastery Orientation -.01 .00 .06* .00
3 Guided Discovery -.04 -.04

Pure Discovery -.07 -.07

Metacognitive Instruction .02 .02 .06 .00

 

*p_<_.05;**ps.01

115

12. When all factors were entered into the regression, they no longer accounted for a
signiﬁcant amount of variance in knowledge structure (F(7,153) = 1.48, p > .05).

Hypothesis-testingzself-regulatory activity. A hierarchical regression analysis
was run to examine the impact of individual differences and training manipulations on
hypothesis-testing/self-regulatory activity during training. Cognitive ability and
performance orientation were controlled for in the ﬁrst step in the regression analysis.
Tolerance for ambiguity and mastery orientation were entered in the second step,
followed by the training manipulations in the third step. Results are presented in
Table 7.

Cognitive ability and performance orientation accounted for a signiﬁcant amount
of variance in the dependent variable (R2 = .04, F = 3.20, p < .05). The beta-weights
indicated that cognitive ability was positively related to hypothesis-testing/self-
regulation (B = 20 at step 1). Hypotheses 16/17 predicted a positive relationship
between mastery orientation and hypothesis-testing/self-regulatory activity. Hypothesis
22 predicted a positive relationship between tolerance for ambiguity and hypothesis-
testing. The second step in the regression analysis accounted for a signiﬁcant amount
of variance in the dependent variable (AR2 = .13, AF = 12.16, p < .01). Beta-weights
indicated that mastery orientation was positively related to hypothesis-testing/self-

regulation (B = .32 at step 2). Although tolerance for ambiguity exhibited a signiﬁcant

zero-order correlation with hypothesis-testing/self-regulation (r = .22, p < .01), this
relationship became nonsigniﬁcant once the inﬂuence of the other individual
differences were controlled. These results did not support hypothesis 22, but they did

support hypotheses 1 6/1 7.

116

Table 7

Hierarchical Regression Analysis Results for Hypothesis-Testing/Self-Regulation

 

 

Step Predictors B at Step Final B R2 AR2
1 Ability .20* .13
Performance Orientation -.01 -.07 .04* .04*
2 Tolerance for Ambiguity .13 .13
Mastery Orientation .32M .31** .17M .13"
3 Guided Discovery .25“ .25“
Pure Discovery .03 .03
Metacognitive Instruction -.02 -.02 .22**‘ .06*

 

*ps.05;**ps.01
' R2 values do not add up due to rounding of numbers.

117

Hypothesis 4 predicted that guided discovery would lead to greater hypothesis-
testing compared to pure discovery, which would lead to greater hypothesis-testing
compared to procedural instruction. Hypothesis 6 predicted that metacognitive
instruction would lead to greater self-regulatory activity compared to no metacognitive
instruction. The third step in the regression, containing these training manipulations,
accounted for a signiﬁcant amount of variance in hypothesis-testing/self-regulation
(AR2 = .06, AF = 3.71, p < .05). Beta-weights indicated that guided discovery led to
greater hypothesis-testing/self-regulation compared to pure discovery and procedural

instruction (B = .25). The predicted cell means after controlling for individual

differences were: guided discovery, X = 5.79; pure discovery, X = 5.38; procedural
instruction, X = 5.33. These results provided partial support for hypothesis 4. Guided
discovery did lead to greater reported use of hypothesis-testing/self-regulation
compared to pure discovery and procedural instruction. However, the pure discovery
group did not report greater use of hypothesis-testing/self-regulation compared to the
procedural instruction group. Hypothesis 6, predicting an effect for metacognitive
instruction, was not supported by the regression results. Overall, variables in the
equation accounted for 22 percent of the variance in hypothesis-testing/self-regulation
(F(7,153) = 6.34, p < .01).

Self-efﬁcag. A hierarchical regression analysis was run to examine the
relationships between individual differences, training manipulations, and self-efﬁcacy.
The effects of cognitive ability and performance orientation were controlled for in the

ﬁrst step in the regression. Tolerance for ambiguity and mastery orientation were

118

entered in the second step, followed by discovery learning and metacognitive
instruction in the third step. Results are presented in Table 8.

The ﬁrst step in the regression did account for signiﬁcant variance in self-
efﬁcacy (R2 = .06, F = 5.27, p < .01). Beta-weights for the step indicated that both

cognitive ability (B = .20 at step 1) and performance orientation (B = .15 at step 1)

were positively related to self-efﬁcacy. Hypotheses 20 and 25 predicted that mastery
orientation and tolerance for ambiguity would be positively related to self-efﬁcacy.

The second step in the regression, containing these factors, did account for signiﬁcant
variance in self-efﬁcacy (AR2 .15, AF = 15.17, p < .01). In addition, beta-weights for

the step indicated that both tolerance for ambiguity (B = .29 at step 2) and mastery
orientation (B = .20 at step 2) were positively related to self-efﬁcacy. These results

supported hypotheses 20 and 25.

Hypothesis 14 predicted that metacognitive instruction would lead to greater
self-efﬁcacy compared to no metacognitive instruction. The third step in the
regression, which contained the metacognitive instruction factor as well as the
discovery learning manipulations, did not add a signiﬁcant increment in R2.
Hypothesis 14 was not supported by these results. The full regression equation
accounted for 22 percent of the variance in self-efﬁcacy (F(7,153) = 6.16, p < .01).

Adaptive transfer. A regression analysis was run to examine how individual
difference factors, discovery learning, metacognitive instruction, and learning outcomes
were related to adaptive transfer. Results are presented in Table 9. Cognitive ability
and performance orientation were entered ﬁrst to control for their effects on adaptive

transfer. In the second step, tolerance for ambiguity and mastery orientation were

l 19
Table 8

Hierarchical Regression Analysis Results for Self-Efﬁcacy

 

 

Step Predictors B at Step Final B R2 AR2

1 Ability .20" .15*

Performance Orientation .15* .11 .06M .06“
2 Tolerance for Ambiguity .29" .30“

Mastery Orientation .20* .21" .22**' .15"
3 Guided Discovery -.01 -.01

Pure Discovery -.04 -.04

Metacognitive Instruction .06 .06 .22M .00

 

*ps.05;**ps.01
' R2 values do not add up due to rounding of numbers.

120

entered into the regression. The main effects for the discovery learning manipulation
and metacognitive instruction were entered in the third step. The fourth step contained
the interaction between discovery learning and metacognitive instruction. The four
learning outcomes were entered in the ﬁfth step of the regression.

Results indicated that cognitive ability and performance orientation accounted
for a signiﬁcant amount of variance in adaptive transfer (R2 = .07, F = 5.61, p < .01).
The beta-weights revealed that cognitive ability was the inﬂuential factor which was
positively related to adaptive transfer (B = .26 at step 1). Hypothesis 21 predicted that
tolerance for ambiguity would be positively related to adaptive transfer. The second
step, containing tolerance for ambiguity and mastery orientation, did not account for a
signiﬁcant amount of variance in adaptive transfer. These results did not support
hypothesis 21.

Hypothesis I predicted that guided discovery would lead to greater adaptive
transfer compared to pure discovery, which would lead to greater transfer compared to
procedural instruction. Hypothesis 2 predicted that metacognitive instruction would
lead to greater adaptive transfer than no metacognitive instruction. The third step in
the regression, containing the main effects for discovery learning and metacognitive
instruction, did not add a signiﬁcant increment in R2 to the prediction of adaptive
transfer. Therefore, hypotheses 1 and 2 were not supported by the regression.

Hypothesis 3 predicted that the discovery learning manipulations would interact
with metacognitive instruction to affect adaptive transfer. The fourth step in the
regression, containing the interaction terms, did account for a signiﬁcant amount of

variance in adaptive transfer (AR2 = .05, AF = 4.03, p < .05).

121

Table 9

Hierarchical Regression Analysis Results for Adaptive Transfer

 

 

Step Predictors B at Step Final B R2 AR2
1 Ability .26M .17
Performance Orientation .03 .04 .07** .07**
2 Tolerance for Ambiguity .12 .12
Mastery Orientation -.06 -.07 .08* .01
3 Guided Discovery .00 -.22
Pure Discovery .1 1 -.06
Metacognitive Instruction .04 -.23 .09* .01
4 Guided Discovery X Metacognitive Instr. .38" .32*
Pure Discovery X Metacognitive Instr. .28* .25 .14” .05*
5 Verbal Knowledge .17 .17
Knowledge Structure .03 .03
Hypothesis-testing/Self-regulation .00 .00
Self-Efﬁcacy .04 .04 . 16* .02

 

*ps.05;**ps.01

122

Figure 2 plots the predicted means for each group to examine this interaction.
These means are based on the regression equation containing the individual difference
factors and the training manipulations (i.e., before entering learning outcomes into the
regression). The ﬁgure reveals that, for the guided discovery group, metacognitive
instruction led to greater adaptive transfer compared to no metacognitive instruction.
This was consistent with hypothesis 3. Surprisingly, the ﬁgure reveals that for the
procedural instruction group, metacognitive instruction led to poorer adaptive transfer
compared to no metacognitive instruction. This was not as hypothesized, as it was
expected that metacognitive instruction would have little impact on the procedural
instruction group. In fact, the interaction reveals that the metacognitive instruction had
the least impact on the pure discovery group. The results also show that the
procedural instruction group without metacognitive instruction, and the guided and
pure discovery groups with metacognitive instruction performed at a similar level on
the transfer task. Possible explanations for these results are presented in the
discussion.

Hypotheses 5, 7, 10, 13, and 15 predicted that adaptive transfer would be
inﬂuenced by hypothesis-testing, self-regulation, verbal knowledge, knowledge
structure, and self-efﬁcacy, respectively. Hypotheses 5 and 7 were combined in the
analysis as hypothesis-testing/self-regulation was examined as a single construct. The
ﬁfth step in the regression analysis, which contained these learning outcomes, did not
account for signiﬁcant variance in adaptive transfer. It should be noted that verbal
knowledge exhibited a signiﬁcant zero-order correlation with adaptive transfer;

however, when entered into the regression, it did not add a signiﬁcant increment in R2

123

acumen...- m>=nau< co auto-.52.. 33.33822 new 9:58.. ba>oom5
._o cocoa—E. 3.623... 2: .8 means. :00 62052.". .~ 2:9“.

.5385 05a .5385 2650 c2327.... 35385
.- . o

.

 

om

oow

om?

 

 

:26:sz m>=Emoo£o§+ com

:28:sz 02:38.32). 02+

 

 

0mm

com

. 0mm

 

 

 

00..

31003 .Iajsueil anndepv

124

after controlling for prior factors in the causal model. Therefore, results did not
support hypotheses 5/7, 10, 13, and 15. The variance in adaptive transfer accounted
for by all variables in the equation was 16 percent (F(13,147) = 2.17, p < .05).

Follow-up analyses were conducted to try to understand how the interaction of
metacognitive instruction and discovery learning inﬂuenced adaptive transfer. In
addition, these analyses were conducted to better understand the zero-order correlation
between verbal knowledge and adaptive transfer. A second measure of adaptive
transfer was assessed which focused solely on the extent to which individuals were
able to prevent targets from entering the penalty circles. The total score on the
adaptive transfer task included both points for making correct decisions about targets,
as well as points lost for targets entering the penalty circles. Performance with regard
to the penalty circles was isolated from the total score because this skill was the focus
of the training manipulations in the study. The second adaptive transfer outcome was
labelled prioritization of targets. This outcome assessed the number of targets that
individuals prevented from entering the penalty circle on the adaptive transfer task.
This outcome was strongly related to the total adaptive transfer score (r = .90). The
same regression analyses presented for adaptive transfer score were run with the
second transfer outcome. Results are presented in Table 10.

In the ﬁrst regression analysis, results indicated that the interaction between
discovery learning and metacognitive instruction accounted for a signiﬁcant amount of
variance in prioritization of targets in the adaptive transfer task (AR2 = .05, AF = 3.89,
p < .05). Figure 3 plots the predicted cell means for this dependent variable to

examine the nature of the interaction. Again, this plot is based on the regression

125

Table 10

Hierarchical Regression Analysis Results for Prioritization of Targets

 

 

Step Predictors B at Step Final B R2 AR2
1 Ability .25" .11
Performance Orientation .02 .02 .06" .06"
2 Tolerance for Ambiguity .08 .08
Mastery Orientation -.02 -.04 .07* .01
3 Guided Discovery .00 -.21
Pure Discovery .09 -.04
Metacognitive Instruction .03 -.20 .08 .01
4 Guided Discovery X Metacognitive Instr. .39M .29*
Pure Discovery X Metacognitive Instr. .22 .18 .12“ .05*
5 Verbal Knowledge .27“ .27"
Knowledge Structure .03 .03
Hypothesis-testing/Self-regulation -.02 -.02
Self-Efﬁcacy .06 .06 . 17* * .05b

 

*ps.05;**ps.01
‘ R2 values do not add up due to rounding of numbers.

bp=.06

126

.3093 .0 00:05:32; :0 00:02.00.
05.300205. 0:0 05500.. E03005 .0 00:00:... 02620.:— 05 .2 0000.2 =00 00.0.0000 .n 059....

 

 

 

 

 

E03005 050 b03005 00050 00:02.0... 3500090
. . 4. mm
or
0.0—
M
m.
u.
n.
a
n.
o
C u
00:03.05 0>=Emoo£02+ m...
02.02sz 025680.02 02+ H
m
9
«m
m. S
. N.

 

 

 

ll. 2.

127

equation before learning outcomes are entered. Results are similar to that found for
the adaptive transfer score. Metacognitive instruction led to better prioritization of
targets for individuals in the guided discovery group. However, metacognitive
instruction led to poorer prioritization of targets for the procedural instruction group
compared to the absence of metacognitive instruction. For the pure discovery group,
metacognitive instruction had no signiﬁcant effect on prioritization of targets.

Step 5 in the regression analysis contained the four learning outcomes.
Learning outcomes accounted for 5 percent of the variance in prioritization of targets
after controlling for all prior factors, but this step just missed statistical signiﬁcance

(AF = 2.28, p = .06). The beta-weights indicated that verbal knowledge was positively
related to the dependent variable ([3 = .27). When all variables were entered into the

regression, they accounted for a total of 17 percent of the variance in prioritization of
targets (F(13,147) = 2.34, p < .01).

Follow-up analyses for verbal knowledge. Based on the post-hoc analyses with
prioritization of targets as the dependent variable, an additional regression analysis was
conducted with verbal knowledge as the dependent variable. A fourth step was entered
in the regression that contained the interaction between discovery learning and
metacognitive instruction. It was expected that this interaction term would account for
signiﬁcant variance in verbal knowledge. This expectation was consistent with the
theoretical rationale for the hypothesized effect for adaptive transfer. It was expected
that individuals in the discovery learning groups would be able to learn from their

mistakes and incorrect task strategies when they were provided metacognitive

128

instruction. This rationale suggests that they would develop greater knowledge about
correct and incorrect task strategies, and about important features of the task.

Results of this follow-up analysis are presented in Table 11. Results revealed
that the interaction between the discovery learning and metacognition manipulations
did account for signiﬁcant variance in verbal knowledge (AR2 = .04, AP = 4.15, p <
.05). Figure 4 plots the cell means in order to examine this relationship. The plot
reveals a similar pattern of results to those found for Adaptive Transfer Score and
Prioritization of Targets. Individuals in the guided discovery group acquired more
verbal knowledge when also provided metacognitive instruction compared to no
metacognitive instruction. In contrast, individuals in the procedural instruction group
performed worse on the verbal knowledge test if they also received metacognitive
instruction. Metacognitive instruction had no impact on verbal knowledge for the pure
discovery group.l

Tests for mediation. It was expected in this study that the learning outcomes
would mediate the relationships between the individual differences, training
manipulations and adaptive transfer. However, tolerance for ambiguity and mastery
orientation did not exhibit signiﬁcant zero-order correlations with adaptive transfer.
Therefore, there was not a signiﬁcant relationship between these factors and adaptive

transfer that could be mediated by learning outcomes.

 

‘ Previous research has suggested that cognitive ability and the degree of discovery learning
during training will interact to affect learning (Cronbach & Snow, 1977). Therefore, regression
analyses assessing the outcomes of adaptive transfer score, prioritization of targets, verbal
knowledge, and knowledge structure were re-run with the addition of this interaction term. The
interaction between cognitive ability and the discovery learning manipulation did not account for
signiﬁcant variance in any of the four outcomes.

129

Table l 1

F ollow-Up Regression Analysis for Verbal Knowledge

 

 

Step Predictors B at Step Final [3 R2 AR2
1 Ability .45“ .43"
Performance Orientation -.O4 -.06 .21“ .21"
2 Tolerance for Ambiguity -.Ol .02
Mastery Orientation .07 .07 .21** .00
3 Guided Discovery .10 -.14
Pure Discovery .07 -.04
Metacognitive Instruction .05 -.19 .22" .01
4 Guided Discovery X Metacognitive Instr. .37" .37M
Pure Discovery X Metacognitive Instr. .18 .18 .26" .04*

 

*ps.05;**ps.01

130

 

 

00:02.00. 0>£000080§+
00:02.00. 0200000205. 02+

 

 

00300.0 0.00

0000.305. _0€0> 00 0000.500.
0200000505. 000 050.00.. E03005 .0 00:02.0. 02.00.20. 0... .0. 0000.2 =00 030.00.“. .0 0.00.“.

00002.00. .0500020

 

 

02,805 0330

 

 

m0

m.m

m6

ms

md

aﬁpalmoux |qua/\

131

On the other hand, results did show that the training manipulations interacted to
affect adaptive transfer. In addition, one learning outcome, verbal knowledge, was
signiﬁcantly correlated with adaptive transfer. However, this relationship became
nonsigniﬁcant once all other factors were controlled for in the regression analysis (see
Table 9). These results did not satisfy the requirements for mediation (James & Brett,
1984). Nonetheless, a hierarchical regression analysis was run with adaptive transfer
as the dependent variable. The order of entry was reversed so that the learning
outcomes were entered prior to the training manipulations and their interaction. This
allowed for an assessment of whether the variance accounted for by the interaction
between the manipulations became nonsigniﬁcant after controlling for the learning
outcomes. Results are presented in Table 12. The results indicated that the interaction
between the discovery learning groups and metacognitive instruction no longer
accounted for signiﬁcant variance in adaptive transfer after controlling for the learning
outcomes.

A similar analysis was conducted for prioritization of targets. One condition for
mediation was satisﬁed as the training manipulations interacted to inﬂuence
prioritization of targets. In addition, verbal knowledge also showed a positive
relationship with the dependent variable when all factors were controlled for in the
regression (although the step containing the learning outcomes just missed statistical
signiﬁcance). Finally, the follow-up analyses for verbal knowledge indicated that the
interaction between discovery learning and metacognitive instruction inﬂuenced verbal

knowledge.

132

Table 12

Hierarchical Regression Analysis Results for Adaptive Transfer

(Mediation Analysis)

 

 

Step Predictors B at Step Final [3 R2 AR2
1 Ability .26" .17
Performance Orientation .03 .04 .07** .07**
2 Tolerance for Ambiguity .12 .12
Mastery Orientation -.06 -.O7 .08* .01
3 Verbal Knowledge .22* .17
Knowledge Structure .01 .03
Hypothesis-testing/Self-regulation -.04 .OO
Self-Efﬁcacy .06 .04 . 12* .04
4 Guided Discovery -.01 -.22
Pure Discovery .10 -.06
Metacognitive Instruction .02 -.23 .13* .01
5 Guided Discovery X Metacognitive Instr. .32* .32*
Pure Discovery X Metacognitive Instr. .25 .25 .l6* .03

 

*ps.05;**p.<_.01

133

A test for mediation was conducted using hierarchical regression. The order of
entry was reversed so that the learning outcomes were entered before the training
manipulations in the regression predicting adaptive transfer. If the learning outcomes
mediated the relationship between the training manipulations and adaptive transfer, the
step containing the interaction between the manipulations should become
nonsigniﬁcant once the learning outcomes are controlled (James & Brett, 1984).
Results are presented in Table 13. The interaction between the discovery learning
groups and metacognitive instruction no longer accounted for signiﬁcant variance in
prioritization of targets after controlling for the learning outcomes. Although this
suggests that learning outcomes might be mediators, the fact that learning outcomes
did not account for signiﬁcant variance when entered last in the regression did not
provide deﬁnitive results for this prediction. These follow-up analyses did provide
some limited support for hypothesis 10 that verbal knowledge would signiﬁcantly
affect adaptive transfer.

Demographic factors. Participants in this study indicated their gender, age, and
video game experience at the beginning of the experiment. No speciﬁc relationships
or predictions were made about these factors; these variables were collected as
descriptive information about the sample used in the study. As Table 3 shows, these
demographic factors were signiﬁcantly correlated with some learning outcomes,
adaptive transfer score, and prioritization of targets. These relationships were not
conceptualized as part of the theoretical model examined in this study. However,
because these relationships were found, post-hoc analyses were conducted in which

these three demographic factors were controlled for in each regression analysis. These

134

Table 13

Hierarchical Regression Analysis Results for Prioritization of Targets

(Mediation Analysis)

 

 

Step Predictors [3 at Step Final [3 R2 AR2
1 Ability .25M .11
Performance Orientation .02 .02 .06M .06M
2 Tolerance for Ambiguity .08 .08
Mastery Orientation -.O2 -.O4 .07* .01
3 Verbal Knowledge 3]" .27M
Knowledge Structure .02 .03
Hypothesis-testing/Self-regulation -.O6 -.02
Self-Efﬁcacy .07 .06 .14" .07*
4 Guided Discovery .00 -.21
Pure Discovery .09 -.O4
Metacognitive Instruction .01 -.20 .15“ .01
5 Guided Discovery X Metacognitive Instr. .29* .29*
Pure Discovery X Metacognitive Instr. .18 .18 .17** .02

 

*ps.05;**ps.01

135

analyses provided a very rigorous test of the conceptual model, as they were not
considered in the hypotheses or in the power analysis for the study.

Tables 14 through 19 present regression analyses that repeat the analyses
presented earlier with the demographic factors entered in the ﬁrst step. Table 14
presents a hierarchical regression analysis in which verbal knowledge is regressed on
the independent variables. Results were similar to those presented in Table 11.
Although the demographic factors did not account for signiﬁcant variance in verbal
knowledge, beta-weights indicated that video game experience was positively related to
verbal knowledge ([5 = .17). Figure 5 plots the interaction effect for the training
manipulations. This ﬁgure reveals similar results to those obtained without controlling
for demographics.

Table 15 presents analyses with knowledge structure as the dependent variable.
These results show similar effects to those reported in Table 6. In addition, the three
demographic factors did not account for a signiﬁcant amount of variance in knowledge
structure. It should be noted that the beta-weight for the relationship between age and
knowledge structure was positive and statistically signiﬁcant.

Table 16 presents analyses with hypothesis-testing/self-regulatory activity as the
dependent variable. Results are slightly different compared to analyses presented in
Table 7. Speciﬁcally, the step containing ability and performance orientation no
longer accounted for signiﬁcant variance in the dependent variable. Although the
addition of the demographic factors changed the regression results somewhat, these

factors did not account for signiﬁcant variance in hypothesis-testing/self-regulation.

136
Table 14
Hierarchical Regression Analysis Results for Verbal Knowledge

(Controlling for Demographic Factors)

 

 

Step Predictors B at Step Final B R2 AR2
1 Gender -.01 .03
Age -.02 -.03
Video Game Experience 20* .17“ .04 .04
2 Ability .44" .42"
Performance Orientation -.02 -.OS .24" .20"
3 Tolerance for Ambiguity -.05 -.02
Mastery Orientation .10 .10 24*" .01
4 Guided Discovery .10 -.13
Pure Discovery .05 -.O8
Metacognitive Instruction .04 -.20 .25" .01
5 Guided Discovery X Metacognitive Instr. .35M .35"
Pure Discovery X Metacognitive Instr. .20 .20 .29" .04"

 

*ps.05;**p$.01
' R2 values do not add up due to rounding of numbers.

137

 

 

02.02.00. 0>.._00000.0_2|I|
00002.00. 0200000905. 02+

 

 

00.0.00... 02.00.00.000 .0. 05:05:00. 0000.305. .00.0> 00 02.03.00.
0>...00000.0s_ 000 00.0.00. E05005 .0 00002.0. 03.00.20. 00. .0. 0000s. =00 030.005 .0 0.00.”.

003005 0.0.".

003005 000.30

00002.00. .0.:0000.n.

 

 

F
0

 

 

0.0

mm

:90

:08

ind

abpamoux |quaA

138

Table 15

Hierarchical Regression Analysis Results for Knowledge Structure

(Controlling for Demographic Factors)

 

 

Step Predictors B at Step Final B R2 AR"
1 Gender .08 .09

Age .16* .17*

Video Game Experience .10 .09 .03 .03
2 Ability .23** .24“

Performance Orientation .04 .04 .08* .05*
3 Tolerance for Ambiguity .06 .06

Mastery Orientation -.03 -.02 .08* .00
4 Guided Discovery -.04 -.04

Pure Discovery -.08 -.O8

Metacognitive Instruction .04 .04 .09 .Ol

 

*ps.05;**ps.01

Hierarchical Regression Analysis Results for Hypothesis-Testing/Self-Regulation

139
Table 16

(Controlling for Demographic Factors)

 

 

Step Predictors B at Step Final B R2 AR2
1 Gender -.08 —.01
Age .06 .02
Video Game Experience .10 .13 .03 .03
2 Ability .18* .12
Performance Orientation .Ol -.05 .06 .03
3 Tolerance for Ambiguity .11 .ll
Mastery Orientation .33M .31” .18" .12"
4 Guided Discovery .25“ .25"
Pure Discovery .02 .02
Metacognitive Instruction -.03 -.03 .24" .06M

 

’ps.05;**ps.01

140

Table 17 presents analyses with self-efﬁcacy as the dependent variable. The
demographic variables did not account for signiﬁcant variance in self-efﬁcacy, and the
results of this follow-up analysis were similar to those found in Table 8.

Table 18 presents analyses with adaptive transfer score as the dependent
variable. The results were similar to those found in Tables 9 and 12. Results also
indicated that age was negatively related to adaptive transfer score (B = -.38), and

video game experience was positively related to adaptive transfer score (B = .20).

These demographic factors accounted for 23 percent of the variance in adaptive
transfer score. When all factors were entered in the regression, they accounted for 35
percent of the variance in adaptive transfer (F(16,144) = 4.77, p < .01).

Figure 6 plots the predicted cell means for the interaction effect before learning
outcomes are entered into the regression. The pattern of results is similar to that
presented in Figure 2, except that the guided discovery and pure discovery groups with
metacognitive instruction are not predicted to perform quite as well.

Table 19 presents results with prioritization of targets as the dependent variable.
The results in this analysis were mostly similar to those found in Tables 10 and 13.
When gender, age, and video game experience were controlled for, the learning
outcomes no longer accounted for signiﬁcant variance in the dependent variable when
entered before the training manipulations. This change in the results limits the
implications of the results presented earlier. The demographic factors accounted for 16

percent of the variance, with age negatively related to prioritization of targets (B = -

.30) and video game experience positively related to prioritization of targets (B = .19).

Hierarchical Regression Analysis Results for Self-Efﬁcacy

141
Table 17

(Controlling for Demographic Factors)

 

 

Step Predictors B at Step Final B R2 AR2
1 Gender -. 13 -.08
Age -.02 -.04
Video Game Experience .07 .07 .03 .03
2 Ability .l9* .14*
Performance Orientation .19* .13 .10" .07**
3 Tolerance for Ambiguity .26** .26**
Mastery Orientation .21“ .22" .23" .13”
4 Guided Discovery .00 .00
Pure Discovery -.05 -.05
Metacognitive Instruction .04 .04 .24"‘*‘l .00

 

*ps.05;**ps.01
" R2 values do not add up due to rounding of numbers.

142

Table 18

Hierarchical Regression Analysis Results for Adaptive Transfer

(Controlling for Demographic Factors)

 

 

Step Predictors B at Step Final B R2 AR2
1 Gender -.14 -.15
Age -.36** -.38**
Video Game Experience .24" .20* .23" .23"
2 Ability .24** .18*
Performance Orientation .09 .06 .29** .06**
3 Tolerance for Ambiguity .00 .Ol
Mastery Orientation .05 .05 .29** .00
4 Guided Discovery -.01 -.20
Pure Discovery .08 -.08
Metacognitive Instruction -.03 -.27* .30M .01
5 Guided Discovery X Metacognitive Instr. .32* .29*
Pure Discovery X Metacognitive Instr. .25* .24 .34M .03*
6 Verbal Knowledge .10 .10
Knowledge Structure .07 .07
Hypothesis-testing/Self-regulation -.Ol -.01
Self-Efﬁcacy -.01 -.Ol .35” .01

 

143
Table 18 (cont’d)

 

Step Predictors B at Step Final B R2 AR2
1 Gender -.14 -.15
Age -.36** -.38**
Video Game Experience .24M .20* .23** .23M
2 Ability .24M .18*
Performance Orientation .09 .06 .29M .06M
3 Tolerance for Ambiguity .00 .01
Mastery Orientation .05 .05 .29" .00
4 Verbal Knowledge .14 .10
Knowledge Structure .06 .07
Hypothesis-testing/Self-regulation -.04 -.Ol
Self-Efﬁcacy .00 -.01 .3 l * * .02
5 Guided Discovery -.01 -.20
Pure Discovery .08 -.08
Metacognitive Instruction -.04 -.27* .32M .01
6 Guided Discovery X Metacognitive Instr. .29* .29*
Pure Discovery X Metacognitive Instr. .24 .24 .35" .03'

 

05; ** p 5.0]

*ps.
‘p=.056

144

20.0.00".
0.000.000.00 .0. 00:32:00. .2000... 02.000< 00 00000200. 0200000205. 000 050.00.. 20300.0
.0 00002.0. 02.00.20. 00. .0. 0000.... =00 020.020 .0 0.00.0

 

 

 

 

 

00300.0 0.00 205005 000.00 00000000. .0.:0000.0
4_ w n O
. on

. 8.
V
- om. m.
d
R
A
a
I.
00000000. 0200000202|I| com m
S
02.00000. 0200000202 02+ m.
.0
.. 000 m
9

com

. omm

 

 

 

cow

145

Table 19

Hierarchical Regression Analysis Results for Prioritization of Targets

(Controlling for Demographic Factors)

 

 

Step Predictors B at Step Final B R2 AR2

1 Gender -.05 -.05

Age -.28** -.30**

Video Game Experience .24“ .l9"‘ .16** .16"
2 Ability .23** .12

Performance Orientation .06 .02 .22" .06“
3 Tolerance for Ambiguity -.02 .00

Mastery Orientation .08 .07 .22“ .01
4 Guided Discovery -.01 -.19

Pure Discovery .05 -.07

Metacognitive Instruction -.03 -.23 .23" .00
5 Guided Discovery X Metacognitive Instr. .34* 26*

Pure Discovery X Metacognitive Instr. .21 .18 .26" .O3*
6 Verbal Knowledge .21* .21*

Knowledge Structure .06 .06

Hypothesis-testing/Self-regulation -.03 -.03

Self-Efﬁcacy .02 .02 .29" .03

 

146

Table 19 (cont’d)

 

Step Predictors B at Step Final B R2 AR2
1 Gender -.05 -.05

Age -.28** -.30**

Video Game Experience .24** .l9* .16" .16“
2 Ability .23 ** . 12

Performance Orientation .06 .02 .22" .06"
3 Tolerance for Ambiguity -.02 .00

Mastery Orientation .08 .07 .22"‘8 .01
4 Verbal Knowledge .24** 21*

Knowledge Structure .05 .06

Hypothesis-testing/Self-regulation -.06 -.O3

Self-Efﬁcacy .03 .02 .27" .04b
5 Guided Discovery -.02 -.19

Pure Discovery .04 -.O7

Metacognitive Instruction -.04 -.23 .27” .00
6 Guided Discovery X Metacognitive Instr. .26* 26*

Pure Discovery X Metacognitive Instr. .18 .18 .29** .02

 

*ps.05; **p$.01
’ R2 values do not add up due to rounding of numbers.
b

p = .06

147

When all factors were entered into the regression equation, they accounted for 29
percent of the variance in prioritization of targets (F( 16,144) = 3.69, p < .01).

Figure 7 plots the predicted cell means for the interaction effect. Similar to the
results found for adaptive transfer score, the guided discovery and pure discovery
groups with metacognitive instruction performed a little more poorly after demographic
factors are controlled.

Summa_ry of results. The results of this study provided limited support for the
conceptual model that was tested. Analyses provided strict tests of study hypotheses
by controlling for prior factors in the model before assessing a particular effect.
Because hypothesis-testing and self-regulation were examined as a single construct, the
analyses tested a total of 23 hypotheses. Three of these hypotheses received clear
support from the analyses. Speciﬁcally, mastery orientation was found to be positively
related to hypothesis-testing/self-regulatory activity during training. Both mastery
orientation and tolerance for ambiguity were positively related to self-efﬁcacy.

Two hypotheses received partial support from the regression analysis. First, the
discovery learning manipulation and metacognitive instruction were found to have a
signiﬁcant interactive effect on adaptive transfer. However, the nature of the
interaction was somewhat different than hypothesized. Second, guided discovery was
found to lead to signiﬁcantly greater hypothesis-testing/self-regulation compared to
pure discovery and procedural instruction, although the latter two groups were not
found to differ on this variable.

The remaining hypotheses were not supported by the regression analysis. There

was the suggestion that verbal knowledge mediated the relationship between the

148

 

 

00002.00. 0220000222+
00002.00. 0220000205. 02101

 

 

20.0.00". 020200.000 .0. 05:02:00. 0.00.0 h .0 000020.000 00 02.00200.

0200000205. 000 050.00.. E03005 .0 00002.0. 02.00.20. 0... .0. 0000s. =00 020.020 K 0.00.0

20.6005 0.00

003005 000.00

»

00002.05 0500020

 

 

4.

if
4

m0

:0.

,. m6.

2.

0.:

.N—

 

 

9N.

$106121 30 uonezmioud

149

interaction of the training manipulations and prioritization of targets. The regression
analyses did not support a strict interpretation of mediation; however, the plot of the
means for each training group did show a similar pattern for verbal knowledge and
adaptive transfer score.

There were some interesting results for factors not identiﬁed in explicit
hypotheses. For example, cognitive ability was found to be positively related to verbal
knowledge, knowledge structure, and self-efﬁcacy. It was also positively related to
adaptive transfer score after demographic factors were controlled. Surprisingly,
demographic factors had inﬂuences on several learning outcomes and adaptive transfer.
Age was found to be negatively related to adaptive transfer score and prioritization of
targets. Video game experience was positively related to these two transfer outcomes.
Video game experience positively inﬂuenced verbal knowledge, and age positively

inﬂuenced knowledge structure.

DISCUSSION

The purpose of this study was to examine how individual difference factors,
discovery learning, and metacognitive instruction would inﬂuence multiple learning
outcomes and adaptive transfer. Tolerance for ambiguity and mastery orientation were
identiﬁed as key individual difference factors when examining learning and adaptive
transfer. Cognitive ability and performance orientation were also included as control
variables based on previous research linking these variables to learning processes.

Two training interventions were examined as likely methods for facilitating learning
and the capability to adapt to novel task situations: discovery learning and
metacognitive instruction. Learning outcomes hypothesized to mediate the inﬂuence of
the individual differences and training manipulations on adaptive transfer were also
identiﬁed. Verbal knowledge, knowledge structure, hypothesis-testing/self-regulatory
activity, and self-efﬁcacy were each expected to inﬂuence adaptive transfer. A
conceptual model was tested in which individual differences and training manipulations
led to particular learning outcomes, which then affected performance on an adaptive
transfer task.

Results provided support for limited portions of the model. Results indicated
that the discovery learning and metacognitive instruction manipulations interacted to

affect performance on the adaptive transfer task. Similar results were obtained for

150

151

overall score on the adaptive transfer task, as well as when performance on the
prioritization skill was isolated. The nature of the interaction was somewhat different
than hypothesized. As expected, metacognitive instruction did lead to greater adaptive
transfer compared to no metacognitive instruction for the guided discovery group.
Prompting individuals to make hypotheses and predictions about the task in the guided
discovery manipulation was not enough to help them adapt their skills to the transfer
task. Individuals may have pursued incorrect or less important hypotheses about the
task, or were unsystematic in their testing of predictions when not required to
explicitly evaluate their understanding. Without guidance on monitoring and
evaluation, individuals may have had difﬁculty recognizing errors they were making
during practice. Asking individuals to reﬂect on their practice trials seems to have
made the exploration process more systematic in the guided discovery condition.

The impact of metacognitive instruction on the procedural instruction group was
not as expected. Metacognitive instruction led to poorer performance on the adaptive
transfer task compared to no metacognitive instruction. It is difﬁcult to explain why
instructing individuals to attend to their understanding of the task would lead to poorer
transfer. The different effect of metacognitive instruction on the guided discovery and
procedural instruction conditions may have depended on what learners were focusing
on in their self-regulatory efforts. It may be that individuals in the procedural
instruction group felt that they already comprehended the task strategies for prioritizing
targets because this information was provided to them before practice trials. They may
have then directed their attention to improving other aspects of task performance and

thus spent less time and effort on improving their prioritization skill. A second

152

possibility is that individuals who were provided the correct task procedure and then
asked to self-regulate their learning focused their attention on developing very efﬁcient
and streamlined performance of the strategies. This focus on effectively applying a
known strategy may have made it difﬁcult for these individuals to adapt to a task that
required more dynamic processing and change in task strategies. This second
explanation, however, does not explain why the procedural instruction group that did
not receive metacognitive instruction performed well on the adaptive transfer task.

Another reason for the detrimental effect of metacognitive instruction may deal
with attentional processes. For example, Kanfer and Ackerman (1989) found that
prompting individuals to engage in self-regulatory processes led to decrements in
performance early in skill acquisition. Because of this research, the present study did
not introduce the metacognitive instruction until after the individuals had practiced the
low complexity scenario and thus had some exposure to the nature of the skills they
were learning. However, it is possible that individuals were still in very early stages
of skill acquisition where asking them to focus on self-regulation may have presented
additional processing requirements that were overwhelming to them.

Yet Kanfer and Ackerman’s (1989) theory that there are limited resources
available for self-regulation early in skill acquisition does not seem to explain why the
metacognitive instruction hurt the procedural instruction group, but not the guided
discovery group or the pure discovery group. It would seem that, from this
perspective, the two discovery groups would have less attention available to engage in
self-regulatory activities because they had to engage in more active learning to develop

their own understanding of the task.

153

However, Allport (1989) has proposed a different perspective on the role of
attentional processes which he calls selection-for-action. His theory is opposed to the
notion of attention as a limited resource. Basically, this perspective states that we are
faced with multiple goals in any particular situation, and that these goals are assigned
different priorities. The purpose of the attentional system is to ensure coherence of
behavior under multiple and often conﬂicting goals. Attention is selectively engaged
for goals with high priority, and attention to lower priority goals is inhibited. In the
case of the guided and pure discovery groups, the metacognitive instruction is likely to
have been compatible with the overall goal of exploring the task and experimenting
with strategies for dealing with it. In fact, the self-regulatory activities may have
focused this activity and made it more systematic. In contrast, the instruction to
consciously monitor and evaluate one’s understanding may have been in conﬂict with
the overall goal of the procedural instruction group to apply and practice the explicit
strategies for task performance. The application of known strategies was likely a more
rote and passive learning process (Frese et al., 1988), and adding the self-regulatory
activity may have confused trainees or provided them with additional work that did not
increase their understanding.

Overall, the results suggested that there were several methods for training
individuals to achieve a similar level of performance on the adaptive transfer task.
Procedural instruction without metacognitive instruction, guided discovery with
metacognitive instruction, and pure discovery with metacognitive instruction all
produced similar levels of adaptation. The results, therefore, do not resolve some of

the varying results for discovery learning instruction in previous research (e.g., Carlson

154
et al., 1992; Karnouri et al., 1986; Singer & Pease, 1976). Alternatively, this may

suggest problems with the nature of the adaptive transfer task used in this study.
Adaptive transfer was operationalized at a moderate level of novelty where a new
feature of the task required individuals to reconﬁgure the strategies learned in training.
This task may not have been novel enough to detect differences in adaptability across
the discovery learning groups.

Results did not support hypotheses predicting that the learning outcomes, would
inﬂuence performance on the adaptive transfer task. Knowledge structure, hypothesis-
testing/self-regulation, and self-efﬁcacy were not found to be signiﬁcantly related to
adaptive transfer. Although verbal knowledge was positively correlated with adaptive
transfer, this relationship became nonsigniﬁcant once all the other factors were entered
in the regression analysis. Nonetheless, this positive correlation was intriguing and led
to some post-hoe analyses to try to understand the role of verbal knowledge in the
conceptual model.

In fact, the follow-up analysis suggested that the interaction of the training
manipulations was signiﬁcantly related to verbal knowledge. The pattern of means for
the training groups was similar to that found for the adaptive transfer outcomes.
Metacognitive instruction was especially beneﬁcial for the guided discovery group.
Prompting individuals to make predictions about the task and to test out different
strategies was not enough guidance to help individuals understand the task. They also
needed to be taught to explicitly monitor and evaluate their progress in developing

effective task strategies.

155

Contrary to hypotheses, the training interventions did not impact knowledge
structure. Also, knowledge structure did not inﬂuence performance on the adaptive
transfer task. The lack of results for knowledge structure is likely due to a deﬁciency
in the measure itself. For example, one would have expected some correlation
between the verbal knowledge measure and the knowledge structure measure, as both
measures tap into the content of trainee’s knowledge. A signiﬁcant correlation was
not found between the verbal knowledge and knowledge structure measures used in the
present study, although verbal knowledge was correlated with adaptive transfer. This
suggests that the knowledge structure measure may not have been tapping into critical
task concepts or the linkages among them. The low mean closeness index (C = .17)
found between the study participants and the expert knowledge structure suggests that
the structure used as the measure of a high-quality structure may not have been
tapping into an effective representation of the task.

The representation used as the expert structure in this study was based on how
the experimenter designed the task. It was expected that certain aspects of the task
would be most critical for understanding the task and performing it well. However,
due to the complex and dynamic nature of the task, trainees may have developed
alternate knowledge representations and task strategies that proved just as effective for
adaptive transfer.

A second possibility for the deﬁciency of the knowledge structure measure is
the range of concepts and actions that were rated by trainees. In the present study, the
task was simpliﬁed to some extent so that individuals were focused on learning about

a single domain of task performance. Thus, the concepts and actions that were rated

156

in the assessment of knowledge structure all tapped into one task domain. It may be
that individuals rated all concepts and actions as highly related because they all
referred to this particular aspect of task performance.

In a previous study, a measure of knowledge structure was found to differentiate
instructional groups as well as to affect performance on a transfer task (Kozlowski et
al., 1995). However, the difference between the Kozlowski et al. (1995) study and the
present one is that Kozlowski and colleagues trained individuals on two important
aspects of task performance that were distinct from one another. Therefore,
individuals in that study who learned the task well would have clearly differentiated
these domains by having two separate clusters of concepts in their knowledge
representation. In the present study, the nature of the knowledge structure would not
have led to distinct clusters of concepts, but rather to a structure that differed on the
degree to which certain concepts and actions were important for task performance.
The method used to assess knowledge structure in this study may not have been
sensitive enough to capture this distinction.

Individual differences in tolerance for ambiguity and mastery orientation were
not found to inﬂuence verbal knowledge and knowledge structure after training. Also,
tolerance for ambiguity was not related to performance on the adaptive transfer task.
One issue with the tolerance for ambiguity measure was its low internal consistency
reliability. This may have limited the extent to which a relationship could be detected
in the study. In terms of mastery orientation, the results suggested that being

motivated by the desire to learn new things did not lead individuals to acquire more

157

knowledge about the task. Instead, cognitive ability was found to be an important
individual difference factor in this study for predicting knowledge.

Results showed that individual differences and the discovery learning
manipulation did affect the learning activities that trainees engaged in during practice.
Mastery orientation was related to reports of greater hypothesis-testing and self-
regulatory activity during training. Also, guided discovery learning led to greater
hypothesis-testing/self—regulation compared to pure discovery and procedural
instruction. Thus, a mastery orientation to learning and the discovery learning
manipulation did inﬂuence how individuals approached the learning task. These
individuals were more willing to test out hypotheses about the task, monitor their
understanding of the task, and set learning goals to improve their knowledge. It
should be noted, however, that even the procedural and pure discovery groups reported
a relatively high degree of hypothesis-testing/self-regulation during learning.

In contrast to hypotheses, this learning activity was not found to be related to
performance on the adaptive transfer task. One rationale for the beneﬁts of discovery
learning was that it allows individuals to use more analytic learning strategies that then
may be available to use to learn how to handle a novel transfer task (e.g., Singer &
Pease, 1976). Greater practice at these activities did not help individuals when faced
with the novel transfer task. Yet hypothesis-testing/self-regulatory activity was
positively correlated with verbal knowledge. While not an effect examined in the
study, this does suggest that engaging in these learning activities will lead to greater

knowledge about the task.

158

As predicted, results indicated that tolerance for ambiguity and mastery
orientation were positively related to self-efﬁcacy. Cognitive ability was also
positively related to self-efﬁcacy. Thus, dispositional factors and ability led
individuals to have greater conﬁdence in their capability to handle changing task
demands. However, the metacognitive instruction did not lead to higher levels of self-
efﬁcacy compared to no metacognitive instruction. This is in contrast to research by
Bandura and Schunk (1981), but it is consistent with results found by Sawyer et a1.
(1992). The lack of a relationship may have been due to the results of the self-
regulatory activities for different individuals. If individuals were unable to improve
their understanding through self-regulatory activities, then this activity was probably
unrelated or possibly negatively related to self-efﬁcacy. It is possible that the
relationship between metacognitive instruction and self-efﬁcacy was moderated by the
success of these metacognitive activities for facilitating learning.

Self-efﬁcacy was not found to be related to adaptive transfer. This result is in
contrast to several studies that have found self-efﬁcacy to inﬂuence transfer of training
(Kozlowski et al., 1995; Smith et al., 1995). This result may be explained in some
part by some issues brought up in the introduction. It was not expected that the
discovery learning manipulations would lead to differential self-efﬁcacy by the end of
training. In fact, it was suggested that the procedural instruction, guided discovery,
and pure discovery groups may have developed similar levels of self-efﬁcacy by the
end of training. However, these self-efﬁcacy beliefs would be developed based on
different task experiences. For example, the explicit rules provided to the procedural

group were expected to be less effective strategies for dealing with the adaptive

159

transfer task. In addition, although the training groups in this study did not differ in
their self-efﬁcacy, they did differ in the amount of knowledge acquired and their
capability to transfer their knowledge and skills to the adaptive transfer task. Yet self-
efﬁcacy was unrelated to performance on the verbal knowledge test in the present
study. In previous research that found self-efﬁcacy to predict transfer of training, self-
efﬁcacy did exhibit a positive relationship with knowledge (Kozlowski et al., 1995) or
performance on a posttest (Gist, 1989). It may be that self-efﬁcacy had no impact on
adaptive transfer because individuals were being persistent in applying both effective
and ineffective strategies to the task.

Surprisingly, demographic factors of age, gender, and previous video game
experience inﬂuenced several learning outcomes and adaptive transfer. For example,
age was found to be negatively related to performance on the adaptive transfer task,
and video game experience was positively related to adaptive transfer. The results for
previous video game experience suggested that these individuals were able to use their
previous knowledge and experience with a different task and apply it to performing
this computer-simulated task. They were able to recognize similar features of previous
video games and likely used analogical reasoning processes to learn this new task
(Kamouri et al., 1986). The results for age are more difﬁcult to interpret. The range
of ages in this study was not varied enough to suppose that there were information-
processing differences across people. Also, age was not related to previous video
game experience, and affected adaptive transfer after controlling for this experience.
One might speculate on another experience-related difference between older and

younger participants. It may be that the older participants in this sample had less

160

experience in using personal computers compared to the younger participants, which
affected what they learned and transferred during the study. However, the study did
not provide evidence to support this speculation.

L_ir_n_itations and Directions for Future Research

One limitation raised earlier was the deﬁciency in the measure of knowledge
structure. Future research should consider more fully the extent to which the method
used to measure the knowledge structure can capture meaningful differences in
knowledge representations. The range of knowledge representations that might be
equally effective should also be considered, as should the nature of the domain of
knowledge to be assessed. In addition, future research examining this construct could
use a different approach to develop a representation of an expert structure. For
example, an expert structure could be developed based on the structures from
individuals who learn and perform very well on this task. Some average across a
group of expert performers could be derived and used as an expert representation. If
there are several equally effective knowledge structures for a particular task, then an
average representation would not be meaningful.

A second issue that can be debated is the nature of the adaptive transfer task in
this study. The task was operationalized at a moderate level of task novelty, in which
a new feature of the task was introduced that had not been experienced during
practice. This new task feature made the task more dynamic in nature. This change in
the task was designed so that individuals would continually have to reprioritize targets
in the task. This is in contrast to the practice trials where initial assessment of the task

led to a relatively stable understanding of how the task would unfold. However, a

161

transfer task at the far end of the novelty continuum might require the development of
strategies never used in training. Future research should examine whether results are
more consistent with the conceptual model when a task of greater novelty is used to
assess transfer of training.

A broader concern for future research is the nature of transfer of training and
the elements that are necessary for transfer to occur. The present study focused on
multiple learning outcomes that were expected to be necessary conditions for adaptive
transfer. For example, it was expected that an individual must have extensive
knowledge about the task (verbal knowledge) and that this knowledge should be well-
organized in memory (knowledge structure). It was also expected that practice in
using more active learning approaches such as hypothesis-testing and problem-solving
during training would assist individuals when they had to learn what was different
about the adaptive transfer task and how they should respond. Finally, it was expected
that motivational processes play a part in transfer as well; speciﬁcally, it was
hypothesized that self-efﬁcacy would allow people to be resilient and to persist in the
face of the demands of the adaptive transfer task.

However, theories of transfer also highlight the role of similarity in the
successful transfer of knowledge and skills. For example, theories presented by
Thorndike and Woodworth (1901), Anderson (1993), and Gick and Holyoak (1987) all
emphasize that there are certain features that are similar between the training and
transfer tasks that will promote transfer. Studies that have examined how this
similarity Operates have focused on the processes through which individuals recognize

similar features and then apply their knowledge and skills to a new task. For example,

162

Gick and Holyoak (1980, 1983) examined the role of analogical reasoning in the
transfer of trained strategies to a novel task.

In this study, individuals were told that they had to apply the knowledge and
skills learned in training to a more complex task. Therefore, they did not have to
engage in the recognition processes that would occur in more natural transfer situations
back on the job. This decision was made so that the study could focus on how
individuals prepared themselves during training to learn as much as they could before
the transfer task. However, future research should examine whether the training
manipulations examined in this study would have similar impacts if individuals were
not aware that they were being assessed for their capability to transfer knowledge and
skills to a novel task.

Another issue for future research is to understand the somewhat puzzling
interaction between the discovery learning manipulations and metacognitive instruction.
Examining how metacognitive instruction inﬂuences what individuals focus on during
practice may reveal why metacognitive instruction had a negative impact on
knowledge and transfer for the procedural instruction group. The metacognitive
instruction in this study allowed individuals to direct their own self-regulatory
activities. Therefore, they may not have chosen the best areas of the task to focus on
improving their understanding. In addition, the study provided general feedback on
their success at prioritizing targets so they would have to diagnose their own errors in
performing the task. This may have been difﬁcult for some individuals to handle.
One area for future research is to examine the provision of diagnostic feedback that

gives the learner more directed advice for where they are having problems and thus

163

how they should focus their self-regulatory activity. A related issue is the fact that
self-reported level of hypothesis-testing/self-regulation did not inﬂuence adaptive
transfer. Both of these results suggest that future research should expand our
understanding of metacognitive processes during learning to focus on the content and
quality of these self-regulatory activities in addition to their presence or quantity.

The results of this study may also have implications for how individuals focus
their attention during learning. The different results for the procedural and guided
discovery group may suggest that self-regulatory training may create conﬂicting goals
when the task and what is to be done are clearly speciﬁed to trainees. Research
should focus on understanding how this metacognitive training may have led trainees
to attend to different content in the procedural and discovery learning groups.

A ﬁnal issue for future research is the nature of the task being learned.
Previous research on discovery learning and metacognitive instruction have tended to
use simpler tasks or focused on academic settings. In this study, a complex and
dynamic task was used that was more generalizable to the types of jobs people
perform in organizations. Research should examine the effects of these training
interventions using other complex tasks to determine the extent to which these
interventions have similar impacts across levels of task complexity.

Implications for Practice

Results of this study do not provide very clear implications for training for
adaptive transfer. If a training program is developed using a guided discovery learning
environment such as the one examined in this study, then there should also be

instruction on metacognitive activity and the opportunity to monitor and reﬂect on

164

practice activities. In addition, the results suggest that if one is providing explicit
procedural training, do not add additional pressures or competing goals by also
requiring self-regulatory activity during training. However, the results also showed
that procedural instruction with no metacognitive instruction, and pure discovery
learning with or without metacognitive instruction may produce similar levels of
performance on adaptive transfer tasks. This study cannot resolve some of the
inconsistent results found in previous research on discovery learning and training
transfer. In addition, the differential effects of the particular metacognitive
manipulation in this study need to be understood before making recommendations for
including this type of intervention in organizational training programs.

In sum, this study was designed to examine how individual differences,
discovery learning, and metacognitive instruction inﬂuenced multiple learning
outcomes and adaptive transfer. Results found that the interaction between discovery
learning manipulations and metacognitive instruction did inﬂuence both verbal
knowledge and adaptive transfer. However, the nature of the interaction was
somewhat different than hypothesized. The individual difference factors of tolerance
for ambiguity and mastery orientation were found to have limited inﬂuence in the

conceptual model.

APPENDICES

APPENDIX A

Consent Form

The study in which you are about to participate is designed to examine how
individuals learn and transfer complex, decision-making skills. You will be asked to
learn and perform a computer-simulated, radar tracking task in which you will measure
the attributes of targets on your screen. You will be asked to collect information about
the targets, classify them, and decide what action should be taken for each target.

Also, you will learn how to use information about each target to prioritize which
targets should be acted upon ﬁrst. In addition to learning this task, you will be asked
to complete a short test of general cognitive ability. You will also answer
questionnaires about yourself, as well as provide ratings that will measure your
understanding of the task.

This experiment will occur over two consecutive days. The ﬁrst day of the
experiment is expected to last about 3 hours. The second day of the experiment will
last approximately one hour. You will receive extra credit in your psychology course
for your participation in this experiment. At the end of the experiment, you will be
provided with written feedback explaining the purpose of this research in more detail.

Your participation in this research is strictly voluntary. You are free to
discontinue participation in this experiment at any time for any reason without penalty.
Your responses will be kept completely conﬁdential. In addition, individuals will
remain anonymous in an report of research ﬁndings. You are free to ask any
questions you might have about this experiment at any time. You may ask questions
about the outcome of the study at any time by contacting Eleanor Smith through the
Department of Psychology by mail or you may call directly at 353-9166.

By signing this sheet, you agree that the researcher has explained the study to

you, that you understand the procedures to be used, and that you freely consent to
participate.

Date: Name:

 

Signature:

 

165

166
APPENDIX B

Learning a Computerized Radar Simulation
Individual Training Manual

During this experimental session investigating how people learn, you will be
operating a naval war ﬁghting simulation.

Scenario

You are the Chief Radar Operator of the USS Intrepid, a US. Navy Aegis-class
cruiser. You are seated on the bridge of the USS Intrepid where you can receive
information from all the ship’s sensors on your computerized radar screen. Your mission
is to protect your ship and crew from hostile enemy vessels and to avoid destroying
peaceful vessels. The decisions you make are critical. You are advising the Captain of
the ship on a course of action to take when confronted with other vessels.

Your ship is in the center of the radar scope on your screen. Surrounding your
ship are a number of "asterisks" called targets. The sensors on your ship provide you the
information you need to classify these targets.

Your goal is to protect the USS Intrepid by correctly identifying the Type, Class,
and Intent of a target, and deciding on the correct Engage response corresponding to the
target’s Intent. You must decide what action the USS Intrepid should take toward each
target by deciding whether the Type of the target is Air/Sub/Surface, the Class of the
target is Civilian/Military, and the Intent of the target is Peaceful/Hostile. You then
Engage the target by advising your Captain to "Clear" Peaceful Targets from the area or
"Shoot" Hostile Targets. All the information you collect must be correct for the Captain
to be able to respond to the targets on your screen. The Type, Class, and Intent decisions
you recommend allow the Captain to order which weapons and communication devices to
use to carry out the Shoot or Clear decision. All four decisions you make must be correct
to allow the Captain to order the correct response. For example, you might identify a
target as Surface, Military, and Hostile, and then recommended that the Captain have the
crew shoot the target. But if the target was really an Air target instead of a Surface target,
the Captain would order the wrong weapons be used to shoot the target down. You must
make sure that all four decisions you make are accurate so that your ship can effectively
respond to other vessels in the area.

167
APPENDIX B

Your Radar Console

The radar console looks and operates like a computer. On the upper left comer of
your console, you see the Time, in minutes and seconds, remaining in your simulation.
When the Time counts down to zero (0), your simulation will be ﬁnished. In the upper
right of your console, you see your Score. Your job is to protect your ship by correctly
Engaging as many targets as possible to get the highest possible score.

In the center of your console, you see your computerized radar screen with the
USS Intrepid in the center surrounded by targets. In the lower left comer of your screen,
you see the radius at which your radar console is currently set. This indicates the distance
from your ship your console displays. You may enlarge that radius to see targets outside
your viewing area by clicking on the Zoom_Out function on the OPER menu.

In the lower right comer of your console, you see the Hooked Track #. Each of
the Targets on your console is assigned a Track number. When you "Hook" a Target, by
placing the mouse pointer on the Target and clicking the left button, the Hooked Track #
changes to correspond to the Target number. When you gather information from your
ship’s sensors, that information will be given for the target you currently have Hooked.
Each target retains the same track # throughout the simulation.

On the far upper right comer of your console, you see OPER, TYPE, CLSS, and
ITNT. These are pull-down menus. These menus allow you to gather the information
you need to make the Air/Sub/Surface, Civilian/Military, Peaceful/Hostile, and Shoot/Clear
decisions. Your console will not allow you to make a Shoot/Clear decision before you
have determined a target’s Type, Class, and Intent. You must make the four decisions in
the following order:

(1) ID Air/Sub/Surface;
(2) ID Civilian/Military;
(3) ID Peaceful/Hostile;
(4) ENGAGE Shoot/Clear.

Once you make the Engage decision, the target will disappear from the screen and
you cannot change any of your decisions. If you want to go back and change any decision
before you make the Engage decision, you need to make that decision and all the decisions
that follow it. For example, if you think you made the wrong Type decision, you can go
back and change that decision. But then you have to make the Class decision again,
followed by the Intent decision as well. Then you can Engage the target.

168
APPENDIX B

To display the contents of afmenu. use the mouse to place the pointer on the menu
label and click the Lith mouse button. To display the information gathered by your
sensors, place the pointer on the item and press and hold down the right mouse button.
The sensor information is displayed in the lower right comer of your console.

In using your mouse, it is more efﬁcient to use one ﬁnger for the left mouse
button (e.g., your pointer ﬁnger), and a second ﬁnger to use the right mouse button (e.g.,
your ring ﬁnger). This will help you to avoid clicking on the wrong mouse button to
hook targets v. calling up the menu information.

Your console menus contain the following items:

 

 

 

 

 

 

 

 

OPER TYPE CLSS ITNT
End_Simulation Altitude/Depth Initial_Bearing Missile_Lock
Start_Simulation ID_Air/Sub/Surface ID_Civilian/Military ID_Peaceful/Hostile
Zoom_ln
Zoom_Out

 

ENGAGE_Shoot/Clear

 

Range

 

Speed

 

 

 

The ﬁrst item in the TYPE, CLSS, and ITNT menus provide you the information
from your sensors to make the ID_Air/Sub/Surface, ID_Civilian/Military, and
ID_Peaceful/Hostile, decisions. When you choose each item, you receive a piece of
information in the lower right comer of your console. There are rules to follow to
interpret each piece of information and make the correct decision for each target. These
rules are outlined in the table on the following page. Use these rules to classify the
target’s Type, Class, and Intent.

1 69
APPENDIX B

INFORMATION VALUES

Type = Air/Surface/Sub

l I AIR SURFACE SUB

Altitude/Depth > 0 feet 0 feet < 0 feet

 

 

Class - Civilian/Military

 

CIVILIAN MILITARY

 

Initial Bearing 091 - 270 degrees 000 - 090 degrees

 

Intent - Peaceful/Hostile

 

PEACEFUL HOSTILE

 

Missile Lock Locked

 

For example, the TYPE menu allows you to gather the information needed to
determine whether the Target is Air/Sub/Surface. We will use information about
Altitude/Depth to illustrate the rules for making the ID_Air/Sub/Surface decision.

You hook a target and pull down the Type menu to look at the Altitude/Depth.
The Altitude/Depth is 100 feet. Compare this value to the table above to
determine the target’s Type. Since this value is > 0, it indicates the Target Type is
Air. You would then select the ID_Air/Surface/Sub option from your menu. A
list of choices will appear, and you will select the choice corresponding to the
decision you have made by clicking the right mouse button on the decision. In
this example, you would classify the target as Air.

Once you make the Type decision, the symbol representing the Target on the

screen will change. Air Targets are represented by upper-half symbols (n). Submarines
are represented by lower-half symbols (u), and surface vessels are represented by whole
symbols ([1).

170
APPENDIX B

When you have made the Class decision, the symbols will change again to reﬂect
Military or Civilian. Civilian targets are represented by curved symbols. Military targets
are represented by angular or pointed symbols. These symbols are outlined in the table

 

 

 

below.
Symbols
Air Surface Submarine
Civilian n O 0
Military A 0 V

 

 

 

 

 

 

The ITNT menu contains the Missile Lock information necessary to make the
Intent decision. When you have examined this information, you may choose the
ID_Peaceful/Hostile item from the bottom of the ITNT menu and label the Target as
"Peaceful" or "Hostile."

Engage Decision

After making decisions about the Type, Class, and Intent of a Target, you are now
able to Engage the Target. Go to the OPER menu and click on ENGAGE_Shoot/Clear.
Then choose "Clear" for Peaceful Targets or "Shoot" for Hostile Targets. Remember, you
must make all four decisions (ID_Air/Sub/Surface, ID_Civilian/Military,
ID_Peaceful/Hostile, and ENGAGE_Shoot/Clear) correctly to successfully engage a
Target.

When you choose Shoot or Clear, you will receive feedback on your decisions if
you hold your right mouse button down on your choice. Feedback will appear in the
lower right comer of your screen, indicating whether your decision was correct or not, the
points gained or lost, and the correct Type, Class, and Intent of the target chosen.

Targets disappear from your console after you Engage them so you cannot change
your decision after Engaging a target. You can go back and correct your Type, Class, or
Intent decisions if you do it before you make the ﬁnal Engage decision.

171
APPENDIX B

Score

Your score will increase, remain the same, or decrease depending on the
correctness of your decisions. Remember that you must make all four decisions correct
(Type, Class, Intent, and Engage) to get the maximum score. Points for each completed
target are as follows:

All 4 decisions correct 40 points
3 decisions correct 20 points
2 decisions correct 0 points
1 decision correct -20 points
0 decisions correct -40 points

For example, you hooked a target and made the Type decision correct, but were incorrect
on the Class, Intent, and ﬁnal Engage decision. You would lose 20 points for that target
because only one decision was correct. It is important to be accurate on each decision in
order to do well on this task.

Some Target combinations may seem strange, such as Air-Civilian-Hostile. In
times of war, any type of target can be used for hostile purposes (for example, a kamikaze
civilian aircraft or a research sub).

Summary

In sum, to perform your job as Chief Radar Operator, you must ﬁrst hook a target
by pointing to the target with the mouse pointer and clicking the left mouse button. Then,
you gather information from your sensors to make decisions about the target’s Type,
Class, and Intent. Then, you Engage the Target by choosing to Clear it away or Shoot it.
You should continue identifying and engaging Targets until you have Engaged all the
Targets in your area. Remember to Zoom_Out to check for Targets that have traveled
outside the radius of your radar screen.

1 72
APPENDIX C

General Task Instructions

You have had an opportunity to practice the basic procedures for performing this task.
You have practiced hooking targets, calling up information on the menus, and making
the Type, Class, Intent, and Engage decisions. Now you will be faced with a more
complex task skill to learn.

In the next portion of your training, you will be learning how to prioritize the order in
which you classify and make decisions about targets on your screen. Certain targets
are more important than others because of the presence of two penalty circles on your
screen.

Penalty circles refer to the presence of one or more circles surrounding your ship. It
is a matter of policy that no targets should enter these zones. If a target enters one of
these circles, you will lose points. For example, you don’t want targets getting too
close to your own ship where they can ﬁre on you. There is a visible penalty circle
that indicates this critical range. There is also a second penalty circle that is invisible
and further away that will also cost you points if you let targets enter it. This invisible
penalty circle indicates a safety zone for the rest of your ﬂeet.

Location of Penalty Circles

The ﬁgure on the third page shows you the two penalty circles you will face. The
picture on the left hand side shows your radar screen when the scenario begins. The
inner penalty circle is the circle at 10 n.m. that is shaded with grid lines. You
cannot see the second penalty circle from this screen.

The second, outer penalty circle is located at 256 n.m. If you use the zoom function
and zoom out four times, your radar screen will show a radius of 512 n.m. This is
displayed in the ﬁgure on the right hand side of the page. The second, invisible
penalty circle falls halfway between the outer perimeter of your radar screen and your
ship. This is highlighted in the ﬁgure by the dotted circle. Remember, though, that
you will not see this penalty circle in the actual scenario. Also, remember that you
need to zoom your radar out to 512 n.m. in order to see targets that are close to
entering the outer penalty circle at 256 n.m. You should zoom out and look for
targets that are about halfway between your ship and the edge of your screen when
you zoom out to 512 n.m.

Using the zoom function will help you to see what is going on close to your inner
penalty circle, and to look out in the distance at targets near your outer penalty circle.

173
APPENDIX C

Penalty Points

You will lose 100 points for each target that enters the outer penalty circle, and
will also lose 100 points for each target that enters the inner penalty circle. So
these two penalty circles are equally important for you to defend. Remember, once a
target enters a penalty circle, you have already lost points and there is nothing you can
do about regaining them. For example, the ﬁgure on the right side of the page shows
a target with a range of 152. This target is already inside the outer penalty circle (256
n.m.) and far away from the inner penalty circle (10 n.m.). So this target would be a
very low priority to choose. But you can try to prevent other targets from entering a
penalty circle by choosing targets close to the outside of the penalty circle.

Speed and Range of Targets

You may have several targets near a penalty circle at the same time. How will you
decide which target to make decisions about? Speed and Range information are
available on the OPER menu. The Speed information tells you how fast the target is
travelling towards you -- the higher the number, the greater the speed. The Range
information indicates how far the target is from your ship. For example, a target with
a range of 20 n.m. would be 20 n.m. away from your ship.

You can hook a target and look up its Speed and Range to determine how soon it will
enter a penalty circle. Comparing the Speed and Range for several targets will help
you identify and prioritize which targets to make decisions about ﬁrst. You should
ﬁrst choose and engage targets that are close in range and moving towards you
quickly. By making the four decisions about a target and removing it from the screen,
you will prevent it from entering the penalty circle.

You will now have opportunities to practice these new skills. You will ﬁrst practice a
scenario at a low level of complexity. After this, you will practice a scenario at a
moderate level of complexity, followed by a scenario at a high level of complexity.

 

 

_ .5. .z N... u 00.000 _ _ .5. .2 00 u 00.000 _

 

 

 

 

 

 

174

    
   

00.0
2.0000 .000.

0.2.0 .5. .z 000.

3.0000 .200 \ - l / - /

 

 

 

 

 

/\\ ¥ //
.4. \ v/wmm.u0000.4

* .
A 0%... _.<.
0%

 

  
 

:2 .z o:
/ ......a§mmw§::. *

yo

 

_————_—____ '

1 75
APPENDIX D

Discovery Learning Instructions

Low Complexity Scenario

Pure Discoveg Learning

You will ﬁrst be facing a low complexity scenario. An effective method for learning
skills is to explore a task and develop your own understanding of it. As you practice
the scenario, explore the task to understand what is occurring in the scenario, and to
discover the best strategy to deal with the situation. Experiment with different
strategies for dealing with the low level of complexity. Remember that your task is to
prioritize targets to make decisions about to prevent them from entering the inner
(located at 10 n.m.) and outer (located at 256 n.m.) penalty circles.

Guided Discoveﬂ Learning

You will ﬁrst be facing a low complexity scenario. The scenario is described as
follows: There will be a moderate amount of targets on the screen, with ﬁve targets
surrounding the inner penalty circle and seven targets surrounding the outer penalty
circle. In general, the targets are not very critical, with the majority of targets moving
pretty slowly and not too close in range to the penalty circles. However, there will be
two critical targets surrounding the inner penalty circle, and one critical target
surrounding the outer circle that you must prosecute to prevent yourself from losing
penalty points. You will have plenty of time to identify these targets and prosecute
them before they enter the penalty circle. In addition, these three critical targets will
enter the penalty circles at about the same time.

Your strategy to deal with this is as follows: First, you should check the speed and
range values of targets around the inner (located at 10 n.m.) and outer (located at 256
n.m.) penalty circles. It is important for you to check all targets around both circles
ﬁrst to develop a good assessment of the overall situation you will be facing. You
will see the targets around your inner circle when the scenario ﬁrst starts. Make sure
that you do zoom out to 512 n.m. to check the targets on the outer circle as well.
Determine which targets are critical on the inner and outer circles. Because these three
targets will enter the two circles at about the same time, the order of prosecuting these
three targets is less important in this scenario. But you must make sure to prosecute
these critical targets ﬁrst. Once you have prosecuted these three targets, prosecute as
many additional targets as you can, prioritizing them based on their speed and range
values.

176
APPENDIX D

Low Complexity Scenario (cont ’d)

To learn how to handle this scenario, you should spend time checking the speed and
range values of all targets around the inner and outer penalty circles. Take time to
explore the task and understand what this low complexity scenario is like. Try to
ﬁgure out what features of this low complexity scenario are important, and why they
lead to the strategy described above. Once you are comfortable with your
understanding of the scenario, then try out the strategy described above. It is
important for you to focus your attention on exploring the task, and trying to
understand its features and the strategy you must use to handle it.

Guided Discovegg Learning - Trial 1 Instructions

On this ﬁrst practice trial, I want you to take time to explore the task and to watch
how the scenario unfolds. You should practice checking the speed and range
information for targets around the inner and outer penalty circles. Focus on
understanding what these two pieces of information mean for each target. First, check
the speed and range values of the targets on the inner circle. Next, zoom out to 512
n.m. on your screen, and look for your outer penalty circle. Check the speed and
range values of all targets around the outer penalty circle. Get a good idea of which
targets you think will enter ﬁrst on the inner circle, and the target that will enter ﬁrst
on the outer circle.

Don’t make decisions on these targets on this ﬁrst trial (or they will be removed from
the screen). Instead, watch how the scenario unfolds, and notice when the targets
enter the penalty circles. For the inner penalty circle, it is apparent on your screen
when the targets enter the penalty circle. For the outer penalty circle, you must check
the range of targets to notice when they have entered the 256 range of the outer circle.

You should explore the task on this ﬁrst trial to develop a good understanding of the
penalty circles, and what the speed and range values mean. You will have several
more trials to experiment with and practice getting the targets before they enter the
circles later.

177
APPENDIX D

Low Complexity Scenario (cont ’6!)
Procedural Instruction

You will ﬁrst be facing a low complexity scenario. The scenario is described as
follows: There will be a moderate amount of targets on the screen, with ﬁve targets
surrounding the inner penalty circle and seven targets surrounding the outer penalty
circle. In general, the targets are not very critical, with the majority of targets moving
pretty slowly and not too close in range to the penalty circles. However, there will be
two critical targets surrounding the inner penalty circle, and one critical target
surrounding the outer circle that you must prosecute to prevent yourself from losing
penalty points. You will have plenty of time to identify these targets and prosecute
them before they enter the penalty circle. In addition, these three critical targets will
enter the penalty circles at about the same time, so it does not matter the order in
which you prosecute these targets. But you must make sure to prosecute these critical
targets ﬁrst.

Your strategy is as follows: First, you should check the speed and range values of
targets around the inner (located at 10 n.m.) and outer (located at 256 n.m.) penalty
circles. Range can be estimated somewhat by looking at targets around the inner
circle, but looking up the exact range can help you when it is hard to see which target
is closer. On the outer circle, you must look at the range of targets on the outer circle
to assess their distance from the outer penalty circle. It is important for you to check
all targets around both circles ﬁrst to develop a good assessment of the overall
situation you will be facing. You will see the targets around your inner circle when
the scenario ﬁrst starts. Make sure that you do zoom out to 512 n.m. to check the
targets on the outer circle as well.

Determine which targets are critical on the inner and outer circles. Because the three
critical targets will enter the two circles at about the same time, you can choose to
prosecute these targets in any order. Once you have prosecuted these three targets,
prosecute as many additional targets as you can, prioritizing them based on their speed
and range values.

An effective method for learning skills is to practice them as often as possible. It is
important for you to focus your attention on practicing the strategy described above
during the practice trials.

1 78
APPENDIX D

Moderate Complexity Scenario

Pure Discoveg Learning

You will now face a task that has increased in complexity to a moderate level. An
effective method for learning skills is to explore a task and develop your own
understanding of it. As you practice the scenario, explore the task to understand what
is occurring in the scenario, and to discover the best strategy to deal with the situation.
Experiment with different strategies for dealing with the moderate level of complexity.
Remember that your task is to prioritize targets and prevent them from entering the
penalty circle.

Guided Discoveg Learning

You will now face a task that has increased in complexity to a moderate level. You
must identify how the situation has changed, and how you must adjust your task
strategy to deal with the changes.

1. Identify at least three aspects of the moderate complexity scenario that might be
different from the low complexity scenario. Write your predictions below.

1

 

2

 

3

 

2. Identify how you would respond to these changes. In other words, identify
what task strategy you would use to handle each of the changes you identiﬁed
in the ﬁrst question. Write these new task strategies below.

1

 

 

 

 

 

 

1 79
APPENDIX D

Moderate Complexity Scenario (cont ’d)

An effective method for learning skills is to explore a task and develop your own
understanding of it. As you practice the next scenario, you should explore the task to
try to identify what is going on and what makes it a moderate complexity task. Watch
and discover what is happening on both the inner and outer penalty circles. As you
explore the task, check whether or not your predictions about the changes that might
occur are correct. Try to understand what these changes mean for how you should
perform the task. Once you understand what is going on in the scenario, experiment
with and try out different strategies for dealing with the situation. Remember that you
must prevent targets from entering the penalty circles so you do not lose points.

Procedural Instruction

You will now face a task that has increased in complexity to a moderate level. The
scenario is described as follows: There will be six targets surrounding the inner
penalty circle and seven targets surrounding the outer penalty circle. A moderate
proportion of the targets are critical. Two targets surrounding the inner penalty circle
and three targets surrounding the outer circle will be critical, and you must prosecute
these targets to prevent yourself from losing penalty points. The two critical targets on
the inner circle will enter the circle pretty quickly. The three critical targets on the
outer penalty circle will enter towards the end of the scenario.

Your strategy is as follows: First, you should check the speed and range values of
targets around the inner and outer penalty circles. Determine which targets are critical
on the inner and outer circles. Then prosecute the two targets that are critical on the
inner circle. Once these targets have been dealt with, move to the outer circle and
prosecute the three critical targets on the outer circle. Once you have prosecuted these
ﬁve targets, prosecute as many additional targets as you can, prioritizing them based
on their speed and range values.

An effective method for learning skills is to practice them as often as possible.
Therefore, it is important for you to focus your attention on practicing the strategy
described above during the practice trials.

l 80
APPENDIX D

High Complexity Scenario
Pure Discovery Learnirg

You will now face a task that has increased in complexity to a high level. An
effective method for learning skills is to explore a task and develop your own
understanding of it. As you practice the scenario, explore the task to understand what
is occurring in the scenario, and to discover the best strategy to deal with the situation.
Experiment with different strategies for dealing with the moderate level of complexity.
Remember that your task is to prioritize targets to prosecute to prevent them from
entering the penalty circle.

Guided Discoveg Learning

You will now face a task that has increased in complexity to a high level. You must
identify how the situation has changed, and how you must adjust your task strategy to
deal with the changes.

1. Identify at least three aspects of the high complexity scenario that might be
different from the moderate complexity scenario. Write your predictions below.

1

 

2

 

3

 

2. Identify how you would respond to these changes. In other words, identify
what task strategy you would use to handle each of the changes you identiﬁed
in the ﬁrst question. Write these new task strategies below.

1

 

 

 

 

 

 

l 8 1
APPENDIX D

High Complexity Scenario (cont ’d)

An effective method for learning skills is to explore a task and develop your own
understanding of it. As you practice the next scenario, you should explore the task to
try to identify what is going on and what makes it a high complexity task. Watch and
discover what is happening on both the inner and outer penalty circles. As you
explore the task, check whether or not your predictions about the changes that might
occur are correct. Try to understand what these changes mean for how you should
perform the task. Once you understand what is going on in the scenario, experiment
with and try out different strategies for dealing with the situation. Remember that you
must prevent targets from entering the penalty circles so you do not lose points.

Procedural Instruction

You will now face a task that has increased in complexity to a high level. The
scenario is described as follows: There will be six targets surrounding the inner
penalty circle and seven targets surrounding the outer penalty circle. A majority of the
targets are critical. Four targets surrounding the inner penalty circle and four targets
surrounding the outer circle will be critical, and you must prosecute these targets to
prevent yourself from losing penalty points. The critical targets on the inner and outer
penalty circles will enter the two penalty circles in an alternating fashion. In other
words, two of the critical targets on the outer penalty circle will enter ﬁrst, followed
by two targets entering on the inner circle. Then two targets will enter on the outer
circle, and two more will follow on the inner circle.

Your strategy is as follows: First, you should check the speed and range values of
targets around the inner and outer penalty circles. Determine which targets are critical
on the inner and outer circles. Then prioritize the targets and prosecute them in the
order outlined below:

1. two on the outer circle;

2. two on the inner circle;

3. two on the outer circle;

4. two on the inner circle.

With any time remaining, prioritize and prosecute the remaining targets on the screen.
An effective method for learning skills is to practice them as often as possible.

Therefore, it is important for you to focus your attention on practicing the strategy
described above during the practice trials.

182
APPENDIX E

Metacognitive Instructions

Now that you have had some experience with practicing this task, it is important for you
to focus your attention on improving your understanding of the task, and your strategies
for accomplishing it. One learning method that is beneﬁcial for identifying and correcting
errors in your understanding of a task is called self-regulation. Self-regulation is a
process of setting a learning goal, developing a plan for achieving the goal, and
monitoring and evaluating your progress with reference to the goal. When people are
active in self-regulating their learning they are able to continuously improve their
understanding.

The ﬁrst step in self-regulation is setting a learning goal. A learning goal tells you what
you should focus your attention on while you practice the task. To set a learning goal,
you need to identify the speciﬁc information or skills that you need to learn more about.

Once you have set a learning goal, you need to develop a plan for how you will
accomplish that goal. In other words, you will identify the speciﬁc thoughts or actions
you will engage in during practice to try to achieve that goal.

While you practice the task, you will also be active in monitoring your improvement with
regard to your learning goal. Are you making progress towards that goal, or have you
started focusing on less important issues?

Finally, when you are ﬁnished with the practice session, you need to evaluate yourself in
terms of your progress towards your learning goal. Did you improve your understanding
of the task, and did it lead to better outcomes? Once you evaluate your progress, the self-
regulatory process begins again. You may still need to work on the same learning goal, or
you may identify some new problem that you want to tackle.

Example of Self—Regulation

Let’s say I am focused on improving my capability to make accurate Type, Class, and
Intent decisions about each target. I have noticed that I am pretty good at remembering
the initial bearing information for the Class decision, and the missile lock information for
the Intent decision. But I have had some difﬁculty remembering the ranges of
altitude/depth for the Type decision. I decide that my learning goal for the next trial is to
focus on understanding and memorizing the altitude/depth information. Notice that a
learning goal is focused on the speciﬁc skills that need improvement, not on overall
performance of the task.

183
APPENDIX E

My plan for achieving that goal is to hook ten targets and just make the Type decision on
them ﬁrst. I ﬁgure that by practicing that skill ten times in a row without interruptions
from the Class and Intent decisions, I may improve my memorization of the altitude/depth
information. I will also try to make these ten decisions without looking at the information
in the manual. Once I have made these ten decisions, I will go back and make the Class
decision and the Intent decision on each of the ten targets.

I believe that this strategy of practicing many Type decisions in a row will improve my
understanding. I will monitor if this is so by checking the feedback I get when I make
the Final Engagement decision. The feedback should match what I decided the Type of
the target to be. When the scenario is over, I can also check my ﬁnal score to see if it has
improved due to my focus on the Type decision. I need to remember, though, that the
overall score may not be very high because I decided to take extra time to focus on the
Type decision. But in general, I can evaluate whether my understanding and progress on
the task has improved. If I feel that it has, I may set a new learning goal focused on a
different aspect of the task.

Instructions for Practice

In your practice sessions, you should be focused on improving your understanding of how
to prioritize targets so that they do not enter the penalty circles. So you need to identify
any difﬁculties you are having in this prioritization skill, and come up with a learning goal
and plan to improve your mastery of this skill. For self-regulation to be successful, you
need to think carefully about your practice exercises, your feedback, and your
understanding of the task in order to identify the speciﬁc areas that need improvement. A
superﬁcial or hasty approach to self-regulation will not help you to improve your learning.

For the next trial, focus on learning the medium complexity scenario. After the trial, you
will then have some experience with the task to help you identify difﬁculties that you need
to correct through self-regulation.

Metacognitive Questions

Participants in the metacognitive instruction condition will be asked to answer questions
after the ﬁrst trial practicing the medium complexity scenario, and any subsequent trials to
practice this scenario. A similar procedure will occur for the high complexity scenario,
where participants will have the ﬁrst trial to gain an initial understanding of the task, and
after this ﬁrst trial and subsequent trials they will answer questions to guide their self-
regulatory learning.

The speciﬁc sheet that participants will ﬁll out after each trial is found on the following
page.

184

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

APPENDIX E
Self-Regulation
Subject #
Trial #
1. How many targets did you allow into the inner and outer penalty circles?
2. Based on the information above, do you think you have learned about and understand
what is occurring in the scenario? Why or why not?
3. Where did you have the most difﬁculties in understanding the scenario? Be speciﬁc.
4. What is your learning goal on the next trial to improve your learning? In other words,
what are the knowledge or skills that you want to master?
5. What is your plan to accomplish your goal? In other words, what will you be thinking

about or what will you be doing to work towards your learning goal?

 

 

 

 

 

On your next practice trial, remember to periodically monitor your progress towards your
learning goal and plan. Adjust your behavior if necessary if you do not feel you are making
progress towards your goal.

185
APPENDIX F

Demographics

Please answer the following questions about yourself:
1. What is your gender?

1 = Male
2 = Female

2. What is your age?

1 =17 or younger
2=18

3=19

4=20

5=21

6:22

7=23

8=24

9=25

10:26 or older

3. What is your overall grade point average?

1 = 0.00 - 0.50
2 = 0.51 - 1.00
3 = 1.01 - 1.50
4 = 1.51 - 2.00
5 = 2.01 - 2.50
6 = 2.51 - 3.00
7 = 3.01 - 3.50
8 = 3.51 - 4.00

4. How often do you play with video games?

1 = Never

2 = Rarely

3 = Sometimes
4 = Frequently
5 = Always

1 86
APPENDIX G

Tolerance for Ambiguity (adapted from Major, 1990)

Instructions

Please answer the following questions about yourself. Use the following rating scale
to indicate your agreement with each of the following statements:

1 2 3 4 5 6 7
strongly moderately slightly neither agree slightly moderately strongly
disagree disagree disagree nor disagree agree agree agree

1. I prefer situations that are unpredictable.

2 I dislike supervisors who expect me to ﬁgure out my work assignments on my
own.a

3. I prefer explicit instructions when I’m learning a new task.a

4. Jobs that have a lot of change and uncertainty are more desirable than jobs with

little change and uncertainty.

I enjoy learning tasks that are ambiguous.

I like working on problems that have more than one solution.a

7. A good job is one where what is to be done and how it is to be done are always
clear.3

8. I prefer work assignments with speciﬁc directions to those with vague directions
that require my own interpretation.

9. I like trying to ﬁgure out tasks when the directions are not very clear.

10.1 enjoy dealing with unexpected situations.

5"?"

" Based on factor analysis results, these items were dropped from the ﬁnal tolerance
for ambiguity scale.

1 87
APPENDIX H

Mastery Orientation and Performance Orientation
(Button, Mathieu, & Zajac, 1995)

Instructions

Please answer the following questions about yourself. Use the following rating scale
to indicate your agreement with each of the following statements:

1 2 3 4 5 6 7
strongly moderately slightly neither agree slightly moderately strongly
disagree disagree disagree nor disagree agree agree agree

Mastery Orientation Items

ﬂ
0

I do my best when I’m working on a fairly difﬁcult task.

When I have difﬁculty solving a problem, I enjoy trying different approaches to
see which one will work.

I try hard to improve on my past performance.

The opportunity to do challenging work is important to me.

The opportunity to extend the range of my abilities is important to me.

The opportunity to learn new things is important to me.

I prefer to work on tasks that force me to learn new things.

When I fail to complete a difﬁcult task, I plan to try harder the next time I work
on it.

1"

”>19???”

Performance Orientation Items

The things I enjoy the most are the things I do the best.

I feel smart when I do something without making any mistakes.

I prefer to do things that I can do well rather than things that I do poorly.

I like to be fairly conﬁdent that I can successfully perform a task before I attempt
it.

5. I am happiest at work when I perform tasks on which I know that I won’t make
any errors.

I feel smart when I can do something better than most other people.

The opinions others have about how well I can do certain things are important to
me.

8. I like to work on tasks that I have done well on in the past.

ewwr

$9

l 88
APPENDIX I

Hypothesis-Testing/Self-Regulatory Activity

Instructions

Please answer the following questions about your activities during the practice
sessions. Use the following rating scale to indicate your agreement with each of the
following statements:

1 2 3 4 5 6 7
strongly moderately slightly neither agree slightly moderately strongly
disagree disagree disagree nor disagree agree agree agree

Hypothesis-Testing Items

1. Before practicing a new scenario, I made predictions concerning how the scenario
might be different due to its increased complexity.

2. I identiﬁed strategies for dealing with changes that might occur in a scenario of

greater complexity.

I explored the task to test my predictions about what the scenario would be like.

4. I watched what occurred at both the inner and outer penalty circle to develop a good
understanding of how each scenario would unfold.

5. I experimented with different task strategies to handle the level of complexity I faced
in a scenario.

6. I thought about how well my predictions about a scenario matched the actual events
that occurred during the scenario.

DJ

Self-Regulation Items

1. I thought carefully about my performance on the previous practice trial to develop a
learning goal for the next practice trial.

2. I developed a speciﬁc plan of action to achieve my learning goal during practice.

3. While practicing a scenario, I monitored how well I was learning its requirements.

4. During practice, I monitored closely the areas where I needed the most practice.

5. I noticed where I made the most mistakes during practice and focused on improving
those areas.

6. I evaluated the feedback at the end of the practice trial to determine how I would

approach the task on the next trial.

Note: These items were originally conceptualized as separate constructs, but factor and
correlational analyses indicated that these items tapped into a single construct.
Therefore, these 12 items were combined into a single scale for study analyses.

1 89
APPENDIX J

Self-Efﬁcacy

Instructions

Please answer the following questions about your activities during the practice
sessions. Use the following rating scale to indicate your agreement with each of the
following statements:

H
0

9°."

1 2 3 4 5 6 7
strongly moderately slightly neither agree slightly moderately strongly
disagree disagree disagree nor disagree agree agree agree

I can successfully meet the challenges of a scenario I have never experienced.

I am conﬁdent that my knowledge of this task can be adapted to unforeseen
situations in the future.

I can deal with a task scenario I have never handled before.

I am certain I can handle the initial difﬁculties of a scenario I have not performed
before.

I believe I will perform well if the task requires me to develop a more complex
task strategy.

I am conﬁdent that I can cope with this simulation if it becomes more
complicated.

I believe I can develop methods to handle changing aspects of this task.

I am certain I can cope with task components competing for my time.

1 90
APPENDIX K

Verbal Knowledge Test

If a target is outside of the current radius of your screen, you can view it by doing
what?

a. there is nothing you can do

b. wait for the target to enter you screen a

c. Zoom_Out '~ I.
d. Zoom_ln

What information should you use to prioritize targets to engage?

Speed and Initial Bearing
Speed and Range

Range and Initial Bearing
Speed

9.0.61»

The inner penalty circle is characterized by which of the following features:

a located at 10 n.m. and visible
b. located at 32 n.m. and visible
c located at 32 n.m. and invisible
(I located at 16 n.m. and visible

If you zoom out to 512 n.m.:

a. the outer Penalty Circle is visible (is marked on the screen)

b. the outer Penalty Circle is invisible and half way between the center
and the screen’s edge

c. the inner Penalty Circle is invisible, and the outer Penalty Circle is
visible

(1. the outer Penalty Circle is around the edge of the screen but invisible

At the start of the scenario, there are two targets, one by each Penalty Circle.
Both targets are traveling at 250 knots. One target has a range of 14 n.m. and the
other has a range of 261 n.m. Which target should be engaged ﬁrst?

the target near the inner Penalty Circle

the target near the outer Penalty Circle

it does not matter

engage the target with the highest initial bearing ﬁrst no matter what

9.09:1»

1 91
APPENDIX K

6. How many points would you lose if you allowed three targets to enter the inner
penalty circle and two targets to enter the outer penalty circle?

7.

8.

a.
b.
c.
d.

200
300
500
1000

If you have three targets,
target (a) with Range = 255 n.m., Speed = 150 knots
target (b) with Range = 260 n.m., Speed == 100 knots, and
target (c) with Range = 25 n.m., Speed = 120 knots,
which target should be engaged ﬁrst?

9.09:1»

target a
target b
target c
engage in any order

If you Zoom_Out to ﬁnd three targets clustered together near the Outer Penalty
Circle, how would you determine which to engage ﬁrst?

9.09:1»

check the Ranges

check the Speeds

check both Range and Speed for each target

zoom to 256 n.m. and engage the ﬁrst target closest to the center
(own ship)

You have been checking targets around the inner circle and zoom out to look at
your outer penalty circle. There are two targets surrounding your outer circle.
You check the range on one target and it is 254 n.m. What do you do next?

a.

b.
c.

Make the four decisions for that target so it won’t travel near the
inner penalty circle.

Check the target’s speed to determine how critical it is overall.
Zoom back in to the inner penalty circle.

Hook the second target on the outside circle.

 

 

1 92
APPENDIX K

10. You have checked the speed and range values on targets on your inside and
outside penalty circles. Their values are as follows:

Speed Range
Target 001 380 22
Target 002 229 26
Target 003 428 45
Target 004 412 262
Target 005 150 261
Target 006 390 280

What four targets would you engage ﬁrst, and in what order?

999‘?”

004, 001, 006, 002
004, 005, 001, 002
003, 004, 006, 001
004, 006, 001, 002

11. In which of the following situations would looking up the range of targets be most

critical?

a.
b.
c.

d.

To determine which of two targets is closer to the inner penalty circle.
To compare how close targets are to the outer penalty circle.

To identify targets that are an equal distance between the inner and
outer penalty circles.

Range is not important to look up in any situation because speed is
the critical factor.

12. When a scenario ﬁrst begins, what should you do?

Prosecute the target that looks closest to the inner penalty circle.
Choose a penalty circle, compare the speed on 2 or 3 targets around
it, and choose the fastest of the targets to prosecute.

Identify the targets with a Hostile Intent and prosecute them ﬁrst.
Check targets on both the inner and outer penalty circle to get an
overall assessment of the scenario.

1 93
APPENDIX K

13. When dealing with a high complexity scenario, what is the best strategy for
preventing targets from entering your penalty circles?

a. As you look up the speed and range of each target, immediately
engage any fast and close ones that you see before checking the rest
of the targets.

b. Check just the speed of the targets to save time and then prosecute the
fastest targets ﬁrst.

c. Frequently zoom in and out to your two penalty circles to
continuously monitor what is happening.

d. Quickly get rid of all of the critical targets around one circle to have

more time to devote to the other penalty circle.

14. When a scenario begins, the radius of the screen is 32 n.m. If you hit the
zoom_out button three times, what will be the radius of the radar screen?

a. 126 n.m.
b. 128 n.m.
c. 256 n.m.
d. 512 n.m.

Note: Items 8, 12, 13, and 14 were dropped from the ﬁnal knowledge test based on
an analysis of the internal consistency of the test.

194
APPENDIX L

Knowledge Structure
Instructions

Your goal is to make "relatedness" ratings about actions you perform in the radar simulation
task. We want you to think about how strongly or weakly pairs of actions "go together" or
"are connected" in a meaningful way IN YOUR MIND when you perform the task.

Pairs of actions will be presented by the computer in random order, so the order of
presentation is not important -- whether an action is presented as the ﬁrst or second item in a
pair should not affect your ratings of relatedness. We are only interested in how strong or
weak YOU THINK the overall connection is between each pair of aspects.

Press SPACE BAR to continue

As you make your ratings, bear in mind that the actions can go together in different ways.
For example, your ratings should indicate HOW STRONGLY OR WEAKLY you think taking
one action relates to taking another action.

Each pair of actions will be presented on the screen along with a "relatedness" rating scale.
Pairs of actions that you think are weakly related should be rated with 1, 2, or 3; moderately
related pairs should be rated 4, 5, or 6; and highly related pairs should be rated 7, 8, or 9 by
pressing the number on the keyboard -- a bar marker will move directly above the number
you pressed. If you wish to change your response, simply press another number. When you
are satisﬁed with your rating, press the SPACE BAR to enter your response. The next pair of
items to be rated will be then be displayed.

Press SPACE BAR to continue

Because the task is complex, there are many pairs of actions to be rated -- please make your
ratings carefully. It is very important to us that you think about HOW RELATED THEY
SEEM TO YOU based on your experience with the task.

If at any time you feel like taking a break tell the experimenter you are doing so and leave the
machine running. Thanks for helping us with this part of the experiment.

Now the complete list of actions to be rated will be presented. This is done to give you a
general idea of the types of actions you will asked to rate. Thank you again for your help.

Press SPACE BAR to continue

195

APPENDIX L

List of concepts and actions to be rated:

Check Speed

Check Range

Zoom Out

Zoom In

Monitor Penalty Circles
Proportion of Dangerous Targets
Number of Targets around Circles
Order of Target Entry

Prioritize Targets

Minimize Point Loss

Monitor Score

Assess Scenario Complexity
Change Performance Strategy

If you have any questions about your task, please ask the experimenter.

Press the SPACE BAR to begin the ratings.

 

 

+ + + + + + + +
1 2 3 4 5 6 8 9
UNRELATED RELATED

Monitor Penalty Circles

Check Speed
Enter <1 through 9> followed by <SPACE>

77 Ratings to go.

I u'

I-r"

196

APPENDIX .\1

Transfer Task Instructions

Yesterday you had a number of opportunities to practice dealing \\ith penalty circles.
Now it is time for you to perform on a more complex version of this task. Use the
knowledge and skills you have gained during the practice exercises to Show how well

you can perform this task. F]

In this ﬁnal task, your score counts. You should concentrate on achieving the best
score you can. This session will last nine minutes. You need to engage as many
targets as you can. You will earn 40 points when you make all four decisions
correctly about a target, and you will gain fewer points or even lose points if some or v "
all of the decisions are incorrect.

 

More importantly, you will lose 100 points for each target that you allow into either of
your two penalty circles. These penalty circles are located at 10 n.m. and 256 n.m.

Your score will be -700 when this task begins. This happens due to the placement of
targets on the screen. We will adjust your score at the end of the session by adding
those 700 points to the score you achieve.

This scenario will be more challenging than those that you have practiced. The
scenario will last twice as long as the practice trials. You will no longer have
feedback on the accuracy of your decisions after you make the Engage decision. In
addition, the scenario will be more dynamic than any you have faced so far.

Do you have any questions? If not, let’s begin the ﬁnal session.

LIST OF REFERENCES

LIST OF REFERENCES

Ackerman, P. L. (1988). Determinants of individual differences during skill
acquisition: Cognitive abilities and information processing. Journal of
Experimengl Psychology: General, 117. 288-318.

Allport, A. (1989). Visual attention. In M. I. Posner (Ed.), Founﬁtions of cognitive
science (pp. 631-681). Cambridge, Massachusetts: The MIT Press.

 

Ames, C. (1992). Classrooms: Goals, structures, and student motivation. Journal of
Educational Psychology, 8_4, 261-271.

Ames, C., & Archer, J. (1988). Achievement goals in the classroom: Students’
learning strategies and motivation processes. Journal of Educational
Psychology, 80, 260-267.

Anderson, J. R. (1993). Problem solving and learning. American Psvchologi_s_t, 48,
35-44.

Andrews, J. D. W. (1984). Discovery and expository learning compared: Their
effects on independent and dependent students. Journal of Educational
Research E, 80-89.

 

Ashcraft, M. H. (1989). Human memog and cognition. New York: Harper Collins.

Baldwin, T. T., & Ford, J. K. (1988). Transfer of training: A review and directions
for future research. Personnel Psychology. 41, 63-105.

Bandura, A. (1977). Self-efﬁcacy: Toward a unifying theory of behavioral change.
Psychological Review. 84, 191-215.

Bandura, A. (1982). Self-efﬁcacy mechanism in human agency. American
Psychologist. 32, 122-147.

Bandura, A. (1991). Social cognitive theory of self-regulation. Organizational
Behavior and Human Decision Proce§s_e§, 5_O_, 248-287.

Bandura, A., & Cervone, D. (1986). Differential engagement of self-reactive

inﬂuences in cognitive motivation. Organizational Behavior and Human
Decision Processes. 38, 92-113.

197

198

Bandura, A., & Schunk, D. H. (1981). Cultivating competence, self-efﬁcacy, and
intrinsic interest through proximal self-motivation. Journal of Person_alitv and
Social Psychology. 4_1, 586-598.

Bandura, A., & Wood, R. (1989). Effect of perceived controllability and performance
standards on self-regulation of complex decision making. Journal of
Person_alitv and Social Psychology. 56, 805—814.

Belmont, J. M., & Butterﬁeld, E. C. (1971). Learning strategies as determinants of
memory deﬁciencies. Cognitive Psychology. 2, 411-420.

Birckbichler, D. W., & Omaggio, A. C. (1978). Diagnosing and responding to
individual learner needs. Modern Language Journal, Q, 336-345.

Bisanz, G. L., Vesonder, G. T., & Voss, J. F. (1978). Knowledge of one’s own
responding and the relation of such knowledge to learning. Journal of
Exmrimental Child Psychology. _2_5_, 116-128.

Blake, B. F., Perloff, R., Zenhausen, R., & Heslin, R. (1973). The effect of
intolerance for ambiguity upon product perceptions. Jourpal of Applied
Psychology, ﬂ, 239-243.

Broadbent, D. E., Cooper, P. F ., Fitzgerald, P., & Parkes, K. R. (1982). The
cognitive failures questionnaire (CFQ) and its correlates. British Journal of
Clinical Psychology. a, 1-16.

Brown, A. (1987). Metacognition, executive control, self-regulation, and other more
mysterious mechanisms. In F. Weinert & R. Kluwe (Eds.), Metacogpition,
motivationLand understanding (pp. 65-116). Hillsdale, NJ: Erlbaum.

Bruner, J. S. (1961). The act of discovery. Harvard Educational Review, 3_1, 21-32.

Budner, S. (1962). Intolerance of ambiguity as a personality variable. Journal of
Personality, 30, 29-50.

Butterﬁeld, E. C., & Nelson, G. D. (1989). Theory and practice of teaching for
transfer. Educational Technology Research and Development. 3_7, 5-38.

Butterﬁeld, E. C., & Nelson, G. D. (1991). Promoting positive transfer of different
types. Cognition and Instruction. 8, 69-102.

Button, S. B., Mathieu, J. E., & Zajac, D. M. (1995). The development and

psychometric evaluation of measures of learning goal and performance goal
orientation. Manuscript submitted for publication.

199

Carlson, R. A., Lundy, D. H., & Schneider, W. (1992). Strategy guidance and
memory aiding in learning a problem-solving skill. Human Factors, 3_4, 129-
145.

Carver, C. S., & Scheier, M. F. (1982). Control theory: A useful conceptual
framework for personality-social, clinical, and health psychology.
Psychological Bulletin. 22, 111-135.

Chapelle, C., & Roberts, C. (1986). Ambiguity tolerance and ﬁeld independence as
predictors of proﬁciency in english as a second language. Language Learning,
3_6, 27-45.

Chi, M. T. H., Feltovich, P. J., & Glaser, R. (1981). Categorization and
representation of physics problems by experts and novices. Cognitive Science.
_5_, 121-152.

Clark, R. E., & Voogel, A. (1985). Transfer of training principles for instructional
design. ECTJ, ﬂ, 113-123.

Cohen, J. (1977). Statistical power analysis for the behavioral sciences (Rev. ed.).
New York: Academic Press.

 

Cohen, J., & Cohen, P. (1983). Applied multiple regression/correlation a_n_alysis for
the behavioral sciences (Second Edition). Hillsdale, NJ: Erlbaum.

Cormier, S. M. (1987). The structural processes underlying transfer of training. In S.
M. Cormier & J. D. Hagman (Eds), Transfer of learning: Contemporm
research and applications (pp. 152-182). San Diego: Academic Press.

Costerrnans, J., Lories, G., & Ansay, C. (1992). Conﬁdence level and feeling of
knowing in question answering: The weight of inferential processes. Journal
of Experimental Psychology: Learning. Memory. and Cognition, 3, 142-150.

Crandall, J. E. (1968). Effects of need for approval and intolerance of ambiguity
upon stimulus preference. Joumal of Personalig, 3_6, 67-83.

Cronbach, L. J., & Snow, R. E. (1977). Aptitudes and iastructional methods. New
York: Irvington.

Dorner, D., & Scholkopf, J. (1991). Controlling complex systems; or, Expertise as
"grandmother’s know—how." In K. A. Ericsson, & J. Smith, Toward a general
theory of expertise (pp. 218-239). Cambridge: Cambridge University Press.

Duda, J. L., & Nicholls, J. G. (1992). Dimensions of achievement motivation in
schoolwork and sport. Journal of Educational Psychology, &, 290-299.

200

Dweck, C. S. (1986). Motivational processes affecting learning. American
Psychologist, ﬂ, 1040- 1 048.

Dweck, C. S., & Leggett, E. L. (1988). A social-cognitive approach to motivation
and personality. Psychological Review. 25, 256-273.

Dwyer, D. J., Hall, J. K., Volpe, C., Cannon-Bowers, J. A., & Salas, E. (September,
1992). A performance assessment task for examining tactical decision making
under stress (Special Report 92-002). Naval Training Systems Center, Human
Factors Division, Orlando, FL.

Ebeling, K. S., & Spear, P. S. (1980). Preference and performance on two tasks of
varying ambiguity as a function of ambiguity tolerance. Austrzian Journal of
Psychology, 32, 127-133.

Egan, D. E., & Greeno, J. G. (1973). Acquiring cognitive structure by discovery and
rule learning. Journal of Educational Psychology, ﬂ, 85-97.

Elliott, E. S., & Dweck, C. S. (1988). Goals: An approach to motivation and
achievement. Journal of Personality and Social Psychology, i4, 5-12.

Etelapelto, A. (1993). Metacognition and the expertise of computer program
comprehension. Scandinavian Journal of Educational Research 3_7, 243-254.

 

Flavell, J. H. (1976). Metacognitive aspects of problem solving. In L. Resnick (Ed.),
The nature of intelligence. Hillsdale, NJ: Lawrence Erlbaum Associates.

F lavell, J. H. (1979). Metacognition and cognitive monitoring: A new area of
cognitive-developmental inquiry. American Psychologist, _3_4_, 906-911.

Ford, J. K., & Kraiger, K. (1995). The application of cognitive constructs to the
instructional systems model of training: Implications for needs assessment,
design and transfer. In C. L. Cooper & I. T. Robertson (Eds), _T_ha

International Review of Industrial and Organizational Psychology (pp. 1-48).
Chichester: Wiley.

Ford, J. K., Quinones, M., Sego, D., & Sorra, J. (1992). Factors affecting the
opportunity to perform trained tasks on the job. Personnel Psychology, _4_5_,
511-527.

F renkel-Brunswik, E. (1949). Intolerance of ambiguity as an emotional and
perceptual personality variable. Jourrall of Personaligy, 1_8., 108-143.

201

Frese, M., Albrecht, K., Altmann, A., Lang, J., Papstein, P. V., Peyerl, R., Prumper,
J., Schulte-Gocking, H., Wankmuller, I., & Wendel, R. (1988). The effects of
an active development of the mental model in the training process:
Experimental results in a word processing system. _Be_lraviour and Information
Technology, 1, 295-304.

F rese, M., & Altmann, A. (1989). The treatment of errors in learning and training.
In L. Bainbridge & S. A. R. Quintanilla (Eds), Developing Skills with
Information Technology (pp. 65-86). Chichester: Wiley.

Frese, M., Brodbeck, F ., Heinbokel, T., Mooser, C., Schleiffenbaum, E., & Thiemann,
P. (1991). Errors in training computer skills: On the positive function of
errors. Human-Computer Interaction, 6, 77-93.

Frese, M., & Zapf, D. (1994). Action as the core of work psychology: A German
approach. In H. C. Triandis, M. D., Dunnette, & L. M. Hough (Eds),
Handbook of industrial and organizational psychology (Volume 4, pp. 271-
340). Palo Alto, CA: Consulting Psychologists Press.

F rone, M. R. (1990). Intolerance of ambiguity as a moderator of the occupational
role stress-strain relationship: A meta-analysis. Journal of Organizational
Behavior ﬂ, 309-320.

 

Gagne, R. M. (1965). The conditions of learning. New York: Holt, Rinehart, &
Winston.

Gick, M. L., & Holyoak, K. J. (1980). Analogical problem solving. Cognitive
Psychology, Q, 306-355.

Gick, M. L., & Holyoak, K. J. (1983). Schema induction and analogical transfer.
Cognitive psychology, 15, 1-38.

Gick, M. L., & Holyoak, K. J. (1987). The cognitive basis of knowledge transfer. In
S. M. Cormier & J. D. Hagman (Eds), Transfer of training: Contemporag
research and applications (pp. 9-46). New York: Academic Press.

Gist, M. E. (1989). The inﬂuence of training method on self-efﬁcacy and idea
generation among managers. Personnel Psychology. 42, 787-805.

Gist, M. E., & Mitchell, T. R. (1992). Self-efﬁcacy: A theoretical analysis of its
determinants and malleability. _A_cademy of Ma_n_agement Review. 11, 183-211.

Gist, M. E., Rosen, B., & Schwoerer, C. (1988). The inﬂuence of training method
and trainee age on the acquisition of computer skills. Personnel Psychology,
4_1, 255-265.

202

Gist, M. E., Schwoerer, C., & Rosen, B. (1989). Effects of alternative training
methods on self-efﬁcacy and performance in computer software training.
Journal of Applied Psychology. 14, 884-891.

Glaser, R., & Chi, M. T. (1989). Overview. In M. T. Chi, R. Glaser, & M. J. Farr
(Eds), The riﬂe of expertise (pp. xv-xxviii). Hillsdale, NJ: Lawrence
Erlbaum.

Goldsmith, T. E., Johnson, P. J., & Acton, W. H. (1991). Assessing structural
knowledge. Journal of Educational Psychology, 83, 88-96.

Goldstein, I. L. (1993). Training in organizations. Paciﬁc Grove, CA: Brooks/Cole.

Goldstein, I. L., & Gilliam, P. (1990). Training system issues in the year 2000.
American Psychologist. 4_5_, 134-143.

Tvt‘g 474 T '

Greenockle, K. M., & Lee, A. (1991). Comparison of guided and discovery learning
strategies. Perceptual and Motor Skills, Z_2_, 1127-1130.

Greiner, J. M., & Karoly, P. (1976). Effects of self-control training on study activity
and academic performance: An analysis of self-monitoring, self-reward, and

systematic-planning components. Journal of Counseling Psychology. a, 495-
502.

Hart, J. T. (1965). Memory and the feeling-of-knowing experience. Journal of
Educational Psmhology. &, 208-216.

Hatano, G., & Inagaki, K. (1986). Two courses of expertise. In H. Stevenson, H.
Azuma, & K. Hakuta (Ed.), Child development and education in Japan (pp.
262-272). San Francisco: Freeman.

Hendrickson, G., & Schroeder, W. H. (1941). Transfer of training in learning to hit a
submerged target. Journal of Educational Psychology. ﬂ, 205-213.

Hermann, G. (1969). Learning by discovery: A critical review of studies. Journal of
Experimental Education. 3, 58-72.

Hesketh, B., & Bochner, S. (1994). Technological change in a multicultural context:
Implications for training and career training. In H. C. Triandis, M. D.,
Dunnette, & L. M. Hough (Eds), Handboak of industrial and organizational
psychology (Vol. 4, pp. 191-240). Palo Alto, CA: Consulting Psychologists
Press.

203

Holyoak, K. J. (1991). Symbolic connectionism: Toward third-generation theories of
expertise. In K. A. Ericsson, & J. Smith, Toward a general theory of
expertise (pp. 301-336). Cambridge: Cambridge University Press.

Howell, W. C., & Cooke, N. J. (1989). Training the human information processor:
A review of cognitive models. In I. L. Goldstein (Ed.), Training and
development in organizations (pp. 121-182). San Francisco: Jossey-Bass.

Ivancic, K, & Hesketh, B. (1995/1996). Making the best of errors during training.
Training Research Journal, 1, 103-125.

James, L. R., & Brett, J. (1984). Mediators, moderators, and tests for mediation. 7
Journal of Applied Psychology. _62, 307-321. '

Johnson-Laird, P. N. (1983). A computational analysis of consciousness. Cognition
and Brain Theory. 6. 499-508.

 

Jonassen, D. H., & Grabowski, B. L. (1993). Handbﬂ of iralividual differences.
Hillsdale, NJ: Lawrence Erlbaum.

Judd, C. H. (1908). The relation of special training and general intelligence.
Educational Review. 36, 42-48.

Justice, E. M., & Weaver-McDougall, R. G. (1989). Adults’ knowledge about
memory: Awareness and use of memory strategies across tasks. Journal of

Educational Psychology, 8_1_, 214-219.

Kamouri, A. L., Kamouri, J., & Smith, K. H. (1986). Training by exploration:
Facilitating the transfer of procedural knowledge through analogical reasoning.
International Journal of Man-Machine Studies. 23, 171-192.

Kanfer, R., & Ackerman, P. L. (1989). Motivation and cognitive abilities: An
integrative aptitude-treatment interaction approach to skill acquisition. Journal

of Applied Psychology, 73, 657-690.

Karoly, P. (1993). Mechanisms of self-regulation: A systems view. Annual Review
of Psychology. 441, 23-52.

Klein, H. J. (1989). An integrated control theory model of work motivation.
Academy of Management Review. E, 150-172.

204

Kozlowski, S. W. J., Gully, S. M., Smith, E. M., Nason, E. R., & Brown, K. G.

(1995, May). Learning orientation and learning objectives: The effects of

sequenced masteg goals and advance organizers on performance, knowledge,
meta-cognitive structure. self-efﬁcacy. and skill generalization for complex

tasks. Paper presented at the Tenth Annual Conference of the Society for
Industrial and Organizational Psychology, Orlando, FL.

 

Kraiger, K. (1993, August). Further evidence of structurzﬂtssessment as a valid
measure for training evaluation. Paper presented at the Annual Meeting of the
American Psychological Association, Toronto, CA.

Kraiger, K., Ford, J. K., & Salas, E. (1993). Application of cognitive, skill-based,
and affective theories of learning outcomes to new methods of training
evaluation. Journal of Applied Psychology. 7_8, 311-328.

Kraiger, K., & Salas, E. (1993, April). An empirical test of two cognitively-based

measures of learning during training. Paper presented at the Annual Meeting of
the Society for Industrial/Organizational Psychology, San Francisco, CA.

Kreutzer, M. A., Leonard, S. C., & Flavell, J. H. (1975). An interview study of

children’s knowledge about memory. Monographs of the Society for Research
in Child Development. ﬂ, 1-57.

Kroll, M. D. (1988). Motivational orientations, views about the purpose of education,
and intellectual styles. Psychology in the Schools. 2_5, 338-343.

Larkin, J. H. (1983). The role of problem representation in physics. In D. Gentner &

A. L. Stevens (Eds), Menﬂ models (pp. 75-98). Hillsdale, NJ: Lawrence
Erlbaum Associates.

Leutner, D. (1993). Guided discovery learning with computer-based simulation

games: Effects of adaptive and non-adaptive instructional support. Learning
and Instruction. 3, 113-132.

Lord, R. G., & Hanges, P. J. (1987). A control system model of organizational

motivation: Theoretical development and applied implications. Behavioral
Science, Q, 161-178.

 

Lord, R. G., & Kernan, M. C. (1987). Scripts as determinants of purposeful behavior
1n organrzatrons. Academy of Management Review. 2, 265-277.

Lundeberg, M. A. (1987). Metacognitive aspects of reading comprehension:

igdiing understanding in legal case analysis. Reading Research Quarterly, 22,
- 32.

 

205

Major, D. A. (1990). Differential effects of interaction content and proaction on
organizational socialization outcomes. Unpublished master’s thesis, Michigan
State University, East Lansing, MI.

Mathieu, J. E., Tannenbaum, S. I., & Salas, E. (1992). Inﬂuences of individual and
situational characteristics on measures of training effectiveness. Academy of
Management Journal, 3_5_, 828-847.

McDaniel, M. A., & Schlager, M. S. (1990). Discovery learning and transfer of
problem-solving skills. Cognition and Instruction, 1, 129-159.

Meece, J. L. (1994). The role of motivation in self-regulated learning. In D. H. La
Schunk & B. J. Zimmerman (Eds), Self-regulation of learning and performance
(pp. 25-44). Hillsdale, NJ: Erlbaum. 1

Meloth, M. S. (1990). Changes in poor readers’ knowledge of cognition and the
association of knowledge of cognition with regulation of cognition and reading
comprehension. Journaa of Educational Psychology, _8_2_, 792-798.

Metcalfe, J., Schwartz, B. L., & Joaquim, S. G. (1993). The cue-familiarity heuristic

in metacognition. Journal of Experimental Psychology: Learning, Memog,
and Cognition, _1__9_, 851-861.

Miller, G. A., Galanter, E., & Pribram, K. H. (1960). Plans and the structure of
behavior. New York: Adams, Bannister, Cox.

Nason, E. R., Gully, S. M., Brown, K. G., & Kozlowski, S. W. J. (1995). SM
acquisition and declarative knowledge: Where structural knowledge ﬁts. Paper
presented at the 10th Annual Conference of the Society for Industrial and
Organizational Psychology, Orlando, FL.

 

Nelson, T. O., Gerler, D., & Narens, L. (1984). Accuracy of feeling-of-knowing
judgments for predicting perceptual identiﬁcation and releaming. J_ourn_alo_f
Experimental Psychology: General, m, 282-300.

Nelson, T. O., & Narens, L. (1990). Metamemory: A theoretical framework and
new ﬁndings. The Psychology of Learning and Motivation. 2_6, 125-141,

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

Norman, D. A. (1984). Working papers on errors and error detection. San Diego,
CA: University of California.

206

Patel, V. L., & Groen, G. J. (1991). The general and speciﬁc nature of medical
expertise: A critical look. In K. A. Ericsson, & J. Smith, Towarcg general
theory of expertise (pp. 93-125). Cambridge: Cambridge University Press.

Patrick, J. (1992). Training: Research and practice. London: Academic Press.

Pintrich, P. R., & DeGroot, E. V. (1990). Motivational and self-regulated learning
components of classroom academic performance. Journal of Educational
Psychology, 8;, 33-40.

Pokay, P., & Blumenfeld, P. C. (1990). Predicting achievement early and late in the
semester: The role of motivation and use of learning strategies. Journal of
Educational Psychology, 8_2_, 41-50.

Prather, D. C. (1971). Trial-and-error versus errorless learning: Training, transfer,
and stress. American Journal of Psychology. 8_4, 377-386.

Pressley, M., Levin, J. R., & Ghatala, E. S. (1984). Memory strategy monitoring in

adults and children. Journal of Verbal Learning and Verbal Behavior. 23, 270-
288.

Reder, L. M., & Ritter, F. E. (1992). What determines initial feeling of knowing?
Familiarity with question terms, not with the answer. Journal of Experimental
Psychology: Learning, Memog, and Cognition, ﬂ, 435-451.

Rouillier, J. Z., & Goldstein, I. L. (1990). Determinants of the climate for transfer of
training. Paper presented at the meeting of the Society for Industrial and
Organizational Psychology, St. Louis, MO.

Royer, J. M. (1979). Theories of transfer of learning. Educational Psychologist. M,
53-69.

Salisbury, D. F ., Richards, B. F ., & Klein, J. D. (1985). Designing practice: A
review of prescriptions and recommendations from instructional design theories.
Joumal of Instructional Development. 8, 9-19.

Salomon, G., & Perkins, D. N. (1989). Rocky roads to transfer: Rethinking
mechanisms of a neglected phenomenon. Educational Psychologist. 24, 113-
142.

Sawyer, R. J., Graham, 8., & Harris, K. R. (1992). Direct teaching, strategy
instruction, and strategy instruction with explicit self-regulation: Effects on the
composition skills and self-efﬁcacy of students with learning disabilities.
Jomal of Educational Psychology, 851, 340-352.

 

207

Schunk, D. H. (1989). Social cognitive theory and self-regulated learning. In B. J.
Zimmerman & D. H. Schunk (Eds), Self-regulated learning and academic
achievement: Theory. research. an_d practice (pp. 83-110). New York:
Springer-Verlag.

Seabaugh, G. O., & Schumaker, J. B. (1994). The effects of self-regulation training

on the academic productivity of secondary students with learning problems.
Journal of Behavioral Education. 4, 109-133.

Shute, V. J., Glaser, R., & Raghavan, K. (1989). Inference and discovery in an
exploratory laboratory. In P. L. Ackerman, R. J. Stemberg, & R. Glaser

(Eds), Learning and individual differences: Advances in theory aad research
(pp. 279-326). New York: W. H. Freeman and Company.

Singer, R. N., & Pease, D. (1976). A comparison of discovery learning and guided
instructional strategies on motor skill learning, retention, and transfer. T_ha
Research Quarterly. 41, 788-796.

Skinner, B. F. (1987). What every happened to psychology as the science of
behavior? American Psychologiat, 4;, 780-786.

Smith, E. M., Ford, J. K., Weissbein, D. A., & Gully, S. M. (1995). The effects of

ggil orientation. metacognition, and practice strategies on learning and transfer.
Paper presented at the Tenth Annual Conference of the Society for Industrial

and Organizational Psychology, Orlando, FL.

 

Snow, R. E. (1989). Aptitude-treatment interaction as a framework for research on
individual differences in learning. In P. L. Ackerman, R. J. Sternberg, & R.

Glaser (Eds), Learning and individual differences: Advances in theog and
research (pp. 13-59). New York: W. H. Freeman and Company.

Tannenbaum, S. I., & Yukl, G. (1992). Training and development in work
organizations. Annual Review of Psychology. 48, 399-441.

Thorndike, E. L., & Woodworth, R. S. (1901). The inﬂuence of improvement in one
mental function upon the efﬁciency of other functions. Psychological Review.
8, 247-261.

Veenman, M. V. J., Elshout, J. J., & Busato, V. V. (1994). Metacognitive mediation

in learning with computer-based simulations. Computers in Human Behavior.
m, 93-106.

Volet, S. E. (1991). Modelling and coaching of relevant metacognitive strategies for
enhancing university students’ learning. Learning and Instruction. 1, 319-336.

208

Weiss, H. M. (1990). Learning theory and industrial and organizational psychology.
In M. D. Dunnette & L. M. Hough (Eds), Handbook of industrial and
organizational psychology (Vol. 1, pp. 171-222). Palo Alto, CA: Consulting
Psychologists Press.

Wexley, K. N., & Latham, G. P. (1981). Developing and training human resources in
organizations. Glenview, IL: Scott Foresman.

Wonderlic Personnel Test & Scholastic Level Exam User’s Manual. (1992).
Libertyville, IL: Wonderlic Personnel Test, Inc.

Wood, R. E. (1986). Task complexity: Deﬁnition of the construct. Organizational
Behavior and Human Decision Processes. 3_7_, 60-82.

Zacks, R. T. (1969). Invariance of total learning time under different conditions of
practice. Jouraal of Experimental Psychology. 8_2, 441-447.

Zimmerman, B. J., & Bandura, A. (1994). Impact of self-regulatory inﬂuences on
writing course attainment. American Educational Research Journal, 81, 845-
862.