Citation
Common instructional interaction strategies and their effects on the construction of fragile knowledge in flight simulation

Material Information

Title:
Common instructional interaction strategies and their effects on the construction of fragile knowledge in flight simulation
Creator:
Mullas, Michael William
Publication Date:
Language:
English
Physical Description:
xii, 112 leaves : illustrations, maps ; 28 cm

Subjects

Subjects / Keywords:
Human information processing ( lcsh )
Learning, Psychology of ( lcsh )
Problem solving ( lcsh )
Human information processing ( fast )
Learning, Psychology of ( fast )
Problem solving ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 108-112).
General Note:
Submitted in partial fulfillment of the requirements for the degree, Doctor of Philosophy, Educational Leadership and Innovation.
General Note:
School of Education and Human Development
Statement of Responsibility:
by Michael William Mullas.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
36436027 ( OCLC )
ocm36436027
Classification:
LD1190.E3 1996d .M85 ( lcc )

Full Text
COMMON INSTRUCTIONAL INTERACTION STRATEGIES AND THEIR
EFFECTS ON THE CONSTRUCTION OF FRAGILE KNOWLEDGE IN FLIGHT
SIMULATION -
by
Michael William Mullas
B.S., Embry-Riddle Aeronautical University, 1982
M.S.A., Central Michigan University, 1988
A thesis submitted to the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
Educational Leadership and Innovation
1996


1996 by Michael William Mullas
All rights reserved.


This Thesis for the Doctor of Philosophy
degree by
Michael William Mullas
has been approved
by
John A. Williams


Mullas, Michael William (Ph.D., Educational Leadership and Innovation)
Common Instructional Interaction Strategies and Their Effects on the Construction of
Fragile Knowledge in Flight Simulation
Thesis directed by Associate Professor R. Scott Grabinger
ABSTRACT
The construction of fragile knowledge hinders the learning process- in some
cases it can produce tragic consequences. This study examined the effects of common
instructional interaction strategies on fragile knowledge construction in learners
during procedural skill acquisition. It also examined whether explicit instruction on
executive control strategies can eliminate or minimize fragile knowledge
construction.
The experimental design was a 4 X 2 factorial pretest, posttest, control group
design. The two independent variables were instructional interaction strategies and
executive controls. There were four levels of interaction strategies: no instructional
interaction (the control group), directive interaction, question-based interaction, and
mixed (directive and question-based) interaction. There were two levels of executive
controls: no instruction on executive control processes and explicit instruction on
executive control processes.
IV


One hundred forty-seven undergraduate students received eight hours of
classroom instruction on basic air navigation, a pretest, two 2-hour simulator missions
(instructed according to their particular treatment group), and a posttest.
An ANCOVA compared posttest scores with pretest scores as the covariate to
examine them for main effects or interaction. A MANCOVA was performed to
determine whether certain interaction strategies produced particular types of fragile
knowledge. Results showed that the mixed interaction strategy exhibited a one
standard deviation advantage over all other strategies in reducing fragile knowledge
construction. However, there were no significant differences for executive control
processes or types of fragile knowledge construction.
There are several implications: first, use a mixed interaction strategy that
begins with a directive approach and changes to a questioning style while the learner
performs the procedural task; second, avoid using a entirely directive or questioning
approach; finally, use questioning strategies that focus on where we get information,
how we process information, and what we do with the results.
This abstract accurately reflects the content of the candidates thesis. I recommend its
publication.
Signed
R. Scott Grainger
v


DEDICATION
To the men and women who educate and train tomorrows air and space leaders
and
To my wife Karen who is educating and training four of tomorrows air and space
leaders: Emily, Beth, Amy, and Jenna


CONTENTS
Figures...........................................................xi
Tables............................................................xii
CHAPTER
1. STATEMENT OF THE PROBLEM........................................1
Introduction.................................................1
Definition of Fragile Knowledge..............................2
Purpose of the Study.........................................4
Experimental Intervention....................................5
Research Hypotheses..........................................8
Significance of the Study....................................9
Definition of Terms..........................................9
2. REVIEW OF LITERATURE...........................................13
Introduction.....................[..........................13
Theoretical Limitations.....................................14
Philosophical Assumptions...................................14
Definition of Fragile Knowledge.............................16
Definition of Cognitive Skill...............................19
Acquisition of Cognitive Skill..............................20
vii


Origins and Development of Procedural Skill............... 22
Executive Control Processes................................25
Teacher Behavior and Student Achievement...................26
Instructional Interaction Strategies.......................28
Hypotheses.................................................31
Hypothesis 1 The Effect of Interaction Strategies on
Fragile Knowledge Construction.......................32
Hypothesis 2 The Efficacy of Executive
Control Strategies...................................32
Hypothesis 3 The Interaction of Instructional Interaction
Strategy and Control Processes.......................33
3. EXPERIMENTAL DESIGN, METHODOLOGY,
AND MATERIALS..................................................35
Methodology................................................35
Participants...............................................36
Design................................................... 38
Variables Studied..........................................40
Independent Variables................................40
Instructional Interaction Strategy............40
Executive Control Processes...................40
Dependent Variables..................................41
Treatments.................................................41
viii


Materials...................................................42
Topic Selection.............................................43
Instruments............................................... 45
Pretest and Posttest.................................45
Procedure...................................................46
Pilot Study..........................................46
Pre-Treatment Training of Instructors................47
Classroom Pretest....................................48
Assignment to Treatment..............................48
Instructional Intervention...........................49
Data Collection......................................50
Data Analysis........................................50
Assumptions.................................................51
4. RESULTS........................................................55
Effect of Instructional Interaction Strategy:
Research Question#l.........................................57
Types of Fragile Knowledge Construction:
Research Question#la........................................58
Effect of Executive Controls:
Research Question#2.........................................59
5. SUMMARY, DISCUSSION, AND RECOMMENDATIONS.......................61
IX


Summary of the Study.
61
Discussion......................................63
Limitations.....................................65
Recommendations.................................67
Future Studies..................................68
Conclusion......................................70
APPENDIX
A. DATA SUMMARY FOR ANCOVA...........................72
B. DATA SUMMARY FOR MANCOVA..........................76
C. PRESTUDY QUESTIONNAIRE............................81
D. AERONAUTICAL KNOWLEDGE TEST #1 (PRETEST)..........82
E. AERONAUTICAL KNOWLEDGE TEST #2 (POSTTEST).........92
F. STEM AND LEAF PLOTS OF STUDY DATA................102
G CONSENT FORM......................................107
BIBLIOGRAPHY.
108


FIGURES
Figure
4.1. Posttest Scores by Instructional Interaction Strategy....................58
4.2. Instances of Specific Types of Fragile Knowledge by Interaction Strategy..59
4.3. Posttest Scores by Executive Controls....................................60
XI


TABLES
Table
3.1 Basic Navigation Procedural Skills.....................................35
3.2 Experimental Design....................................................39
4.1 Summary of Means and Standard Deviations by Instructional
Interaction Strategy and Executive Controls............................56
4.2 Analysis of Covariance Summary by Independent Variable.................57
xii


CHAPTER 1
STATEMENT OF THE PROBLEM
Introduction
A student pilot departs on a solo cross-country flight. The primary gyro
compass fails half way through the flight leaving the pilot with only the standby
magnetic compass for directional reference. The novice pilot remembers to correct
the readings for the deviation of the instrument, but does not correct for difference
between the earths magnetic and north poles and quickly becomes seriously
disoriented. Ultimately, disorientation and low fuel produce catastrophic
consequences.
In another incident, a motorist in a borrowed car on an icy road begins to
skid. Almost instinctively, the driver steers into the skid in an attempt to correct the
condition as he would have done in his own car. Unfortunately, the driver owns a
rear wheel drive vehicle, and the borrowed vehicle is a front wheel drive. The skid
becomes aggravated, and the car slides across the median totally out of control.
What do these two incidents have in common? Both cases are instances of
fragile knowledge. In the first case, the novice pilot is missing critical knowledge. In
the second case, the driver inappropriately applied knowledge from one situation to
an entirely different situation. What is fragile knowledge?
1


Definition of Fragile Knowledge
According to David Perkins (1992, p. 20) fragile knowledge is knowledge that a.
person does not remember, understand, or actively use much of what they have
supposedly learned. This type of knowledge has little substance or strength and is
flimsy or weak when demands are made on it. Research shows that fragile
knowledge can manifest itself in several forms. The six most common types of
fragile knowledge are:
inert knowledge (Whitehead, 1929)
partial or missing knowledge (Perkins & Martin, 1985, p. 5)
misplaced knowledge (Perkins & Martin, 1985, p. 5)
naive knowledge (Perkins & Martin, 1985, p. 5)
conglomerative knowledge (Perkins & Martin, 1985, p. 5)
ritual knowledge (Taba & Elzey, 1964, p. 524)
Perkins (1992) claims that these six types of fragile knowledge can combine in a
learner causing the display of a distinctive cluster of behaviors. This distinctive
cluster of behaviors has been labeled fragile knowledge syndrome. The following
example from Perkins (1992) illustrates this syndrome.
Brian is tackling fraction computation problems. For the simplest
problems, Brian proceeds nicely. Encountering a mixed number,
Brian has no idea what to do with it~a knowledge gap. On another
problem, Brian obtains an answer that needs reducing but forgets to
reduce it, even though he knows how. On an addition problem, Brian
cancels a 3 in the numerator of one term against a 3 in the
2


denominator of the other, mistakenly believing that cancellation
works for sums as well as products. However, on a similar problem,
he does not try canceling and solves the problem correctly, (p. 26)
The performance in this example is inconsistent. At times, the learner seems
competent. However, at other times, with similar problems the learner reaches an
impasse or elects to perform an inappropriate action. The cause of the impasses or
inappropriate actions can be traced to one of several common types of fragile
knowledge.
To provide a little more detail on the types of fragile knowledge, inert
knowledge refers to situations where the student fails to retrieve knowledge but in
fact possesses it, as revealed by a clinical probe (Perkins & Martin, 1985, p. 5).
Partial knowledge is the straightforward case of an impasse due to knowledge the
student has not retained or ever learned, as revealed by clinical probes failing to
reveal signs of the knowledge (Perkins & Martin, 1985, p. 5). Misplaced
knowledge is an instance of a student importing knowledge appropriate to one
context but inappropriate to the new context. Naive knowledge is knowledge that
takes the form of naive theories and stereotypes (Perkins, 1985, p. 5). A flat earth
is a common example of this type of knowledge. Conglomerative knowledge
signifies an attempt by the student to force together disparate elements of knowledge
to provide a solution to a problem, repair a bug, or resolve an impasse. Finally, ritual
knowledge is knowledge that is culturally defined and usually limited to a particular
3


context; for example, knowledge or tasks that are useful in school but not anywhere
else.
Educators have an ethical responsibility to minimize the construction of
fragile knowledge. Are there things that we can do as educational professionals to
minimize or inhibit the learners construction of fragile knowledge? Does how we
interact with learners in an instructional context influence what they learn? Do
instructional interaction strategies that we use with students influence what they
learn? If the answer is yes, then are some strategies more effective then others in
minimizing or inhibiting particular types of fragile knowledge construction? We
have a responsibility to our clients and students to seek answers to these questions.
Purpose of the Study
This study examined the effects of four common instructional interaction
strategies and their effects on the learners construction of fragile knowledge during
initial procedural skill acquisition. The study also examined the effect of providing
explicit instruction on executive control processes and strategies during initial
procedural skill acquisition. Executive control processes are those processes
involved in planning, monitoring, and evaluating performance on intellectual tasks
(Vygotsky, 1978).
This research effort addressed the following questions: Does the way we
interact with learners have an effect on the learners construction of fragile
4


knowledge? If so, are some interaction strategies more effective than others at
minimizing or inhibiting the construction of particular types of fragile knowledge?
Also, previous research has suggested that direct instruction on executive
control processes may reduce fragile knowledge construction in problem-based
learning environments (Perkins & Martin, 1985). Can direct instruction on executive
control processes reduce fragile knowledge construction during acquisition of
algorithmic-heuristic cognitive skills?
Finally, is there an interaction between the use of common instructional
interaction strategies and the explicit instruction of executive control processes? Are
there certain combinations of interaction strategies and executive control processes
that together produce a significant difference in the amount of fragile knowledge
construction?
Experimental Intervention
The design of this experiment was a 4 X 2 factorial design. The two
independent variables were instructional interaction strategies (with four levels) and
executive control strategies (with two levels).
The combination of levels of the two independent variables resulted in eight
treatments being administered in this experiment. The eight treatments were:
1. No instructional interaction with executive controls: Subjects in this
treatment group were given a coursebook which contained the information necessary
5


to complete the specific procedures addressed by each particular learning objective.
After receiving a coursebook, the subjects were sent to the simulator to complete the
simulation without any instructional assistance. Their coursebook contained explicit
instruction on executive control strategies applicable to each procedure they were
required to perform.
2. No instructional interaction without executive controls: These subjects
were given a coursebook and sent to the simulator to complete the lesson without
any instructional assistance. Their coursebook did not contain any instruction on
executive control strategies.
3. Directive/didactic instructional interaction with executive controls:
Instructors in this group directed subjects through each procedure. Each subject was
told in a direct step-by-step manner how to accomplish each of the required tasks.
The instructional interaction strategy focused on telling or making directive or
declarative statements to the learner to perform each step of the procedure.
Instructors in this group also provided explicit (but still directive) instruction on
executive control of the processes within the procedures.
4. Directive/didactic instructional interaction without executive controls:
Subjects in this group received the same type of interaction as the previous group,
but they did not receive any instruction on executive control of the processes within
the procedures.
6


5. Questioning/Reflective instructional interaction with executive controls:
Instructors in this group used questions to guide students through the same tasks as
the other groups. They also used questions to lead the student through executive
control strategies that could be used to resolve impasses or check the output of the
procedures being performed.
6. Questioning/Reflective instructional interaction without executive
controls: This group experienced the same interaction style as the previous group,
but it did not receive the additional questioning on executive control strategies.
7. Mixed Directive/Reflective instructional interaction with executive
controls: Instructors in this group presented the procedures in a directive style first,
and then they shifted to a questioning strategy as the subject performed the skill.
Subjects in the other instructor-assisted groups were simply directed through the
procedure a second time to ensure equal time-on-task for all treatment groups.
Subjects in this group also received explicit instruction (in a questioning type
interaction style) on executive control strategies.
8. Mixed Directive/Reflective instructional interaction without executive
controls: Subjects in this group received the same style of interaction as the previous
group, but they didnt receive any instruction on executive control strategies.
All eight treatments were conducted within the context of two actual
navigation simulation missions. These missions were part of a current course on
7


navigation taught at the United States Air Force Academy. The subjects were taking
their first air navigation course. Each subject received: (1) a pretest, (2) two
navigation simulation missions with instruction according to their treatment group,
(3) and an immediate posttest. The posttest measured how much fragile knowledge
on basic air navigation principles and procedures the subject still possessed after the
instructional treatment. The number of instances of fragile knowledge of basic air
navigation principles and procedures was the dependent variable in this study.
Research Hypotheses
The primary hypothesis stated that subjects who were provided instructional
interaction would perform better on the posttest than subjects who were not provided
instructional interaction. Better performance (higher scores) on the posttest indicated
less fragile knowledge construction. The reason for this is that expert guidance
during the knowledge compilation process in the associative stage of cognitive skill
development should have an effect on what the learner encodes. Including input
from experts should clarify what actually gets encoded during the compilation
process, thus strengthening the semantic connections the learner makes.
Strengthening the semantic connections should reduce the likelihood of constructing
a tenuous grasp on the knowledge.
A secondary hypothesis stated that subjects who were provided explicit
instruction on executive control processes would score better on the posttest than
8


subjects who werent provided this instruction. Once again, a better score on the
posttest indicated a lower amount of fragile knowledge present in the learner. The
reason for this hypothesis is that the possession and use of executive control
strategies should enable learners to solve their own impasses and gain a stronger
grasp on the knowledge to be gained. Subjects who were unable to resolve their own
impasses would not even have access to the additional knowledge opportunities that
existed beyond the initial point of their impasse.
Significance of the Study
This study provided evidence for educational professionals concerning the
issue of the influence of common instructional interaction strategies on their ability
to reduce, minimize, or inhibit the learners construction of fragile knowledge during
initial procedural skill acquisition. Knowing whether these strategies have an effect
on knowledge construction or which strategies are more effective than others in
reducing the construction of fragile knowledge will enable educational professionals
to make improvements to products and processes that will reduce the construction of
fragile knowledge and its consequences.
Definition of Terms
Cognitive Strategies: skills by which learners regulate their own internal processes
of attending, learning, remembering, and thinking (Gagne, 1985, p. 55).
9


Conglomerative Knowledge: signifies an attempt by the student to force together
disparate elements of knowledge to provide a solution to a problem, repair a bug, or
resolve an impasse (Perkins and Martin, 1985, p. 7).
Executive Control Processes: are those processes involved in planning,
monitoring, and evaluating performance on intellectual tasks (Sincoff & Sternberg,
1989, p. 33).
Executive Control Strategies: are a superset of cognitive strategies. According to
Robert M. Gagne (1985, p. 77) they are a part of the information-processing theory
of learning and memory that influences attention and selective perception which
determine what is encoded into short and ultimately long-term memory. The
executive control strategies used in this study included additional procedural tools
that were used to control and cross-check the execution of the navigation procedure.
These strategies included: close tracking of the procedure being executed, strategic
orientation, and self-prompting.
Fragile Knowledge: David Perkins (1985, p. 5) defines fragile knowledge as
knowledge that students do not remember, understand, or use actively much of
what they have supposedly learned; and poor thinking, which means that students do
10


not think very well with what they know. This type of knowledge has little
substance or strength and is flimsy or weak when demands are put on it.
Inert Knowledge: refers to situations where the student fails to retrieve knowledge
but in fact possesses it, as revealed by-a clinical probe (Perkins, 1985, p. 5).
Instructional Interaction Strategy: a superset of elemental methods or tactics for
conveying instruction to the learner and/or for receiving and responding to input
from the learner. Instructional interaction strategies are similar to instructional
delivery strategies, but they focus exclusively on the instructor/leamer exchange.
Some typical strategies include directive or didactic strategies, questioning or
reflective strategies, or combinations of the two.
Misplaced Knowledge: is an instance of a student importing knowledge
appropriate to another context but inappropriate to the context it is being imported
to.
Naive Knowledge: is knowledge that takes the form of naive theories and
stereotypes (Perkins, 1985, p. 5). A flat earth is a common example of this type of
knowledge.
11


Partial Knowledge: is the straightforward case of an impasse due to knowledge the
student has not retained or ever learned, as revealed by clinical probes failing to
reveal signs of the knowledge (Perkins, 1985, p. 5).
Ritual Knowledge: is knowledge that is culturally defined and usually limited to a
particular context; for example, knowledge or tasks that are useful in school but not
anywhere else (Taba & Elzey, 1964).
12


CHAPTER 2
REVIEW OF LITERATURE
Introduction
The concept and consequences of fragile knowledge are not new. However,
systematic research of fragile knowledge is limited in scope and dates back only ten
years. What insights into the nature of fragile knowledge do we currently possess?
We must assess the current state of our knowledge before finding and developing
new techniques to minimize construction of fragile knowledge and reduce our
exposure to its consequences.
This chapter surveys the literature on fragile knowledge and examines the
broader domain of procedural skill acquisition. It starts with the theoretical
limitations of the scope of the survey of the literature, and then it briefly delineates
the philosophical assumptions of the study. With the theoretical foundation of the
review set, it begins by examining the definition of fragile knowledge, and then it
seeks to define cognitive skill. Once the concept of cognitive skill is defined, the
focus then shifts to the acquisition of cognitive skill and to the origins and
development of procedural skill. The theoretical portion of the review concludes
with an examination of executive control processes and how they can effect the
procedural
13


skill acquisition process. Finally, literature on common instructional interaction
strategies concludes the chapter, establishes the theoretical rationale for the study,
and provides the connection to previous research on the topic of fragile knowledge
and its effects on learners.
Theoretical Limitations
Fragile knowledge is a broad concept with an infinite number of possible
contexts. This review is limited to literature pertaining to fragile knowledge
construction during initial algorithmic-heuristic procedural skill acquisition as
defined by Colley and Beech (1989). Algorithmic-heuristic procedural skills are a set
sequence of steps a performer uses to arrive at a solution (Colley & Beech, 1989).
Only the cognitive domain is addressed; the affective and psychomotor domains are
not reviewed. Furthermore, issues of transfer and retention of procedural skills are
also ignored. This is not to imply that the domains or issues not covered are not
important; they are simply not of interest to this particular study. This study focuses
on the construction of fragile knowledge at the point of initial cognitive procedural
skill acquisition.
Philosophical Assumptions
In another effort to stabilize the theoretical foundation, an information
processing model (Gagne, 1974, 1985) was assumed throughout this study. The
information processing model was chosen because of its broad representation in the
14


literature on procedural skill acquisition (Anderson, 1980). Furthermore, information
processing models address the process of skill acquisition which is the focus of this
study. Although the information processing model cannot completely explain the
complexities and intricacies associated with more holistic views on learning, the
model does provide a practical means to communicate observations of a complex
phenomenon such as cognitive procedural skill acquisition between researchers and
consumers of research.
Information-processing theory grew out of human factors work and
information theory; it gained prominence during the reemergence of cognitive
psychology after World War II (Anderson, 1980). Newell and Simons work during
the 1960s and 1970s on artificial intelligence at Canegie-Mellon has also strongly
influenced the information-processing model.
Information processing approaches to explaining human cognition center
around the analogy between how a computer is programmed to process information
and how a human processes information. Cognitive psychologists who hold this
view see the human mind as a computer (Andre & Phye, 1986). Information-
processing models suggest that people store concepts and rules in a place called
memory through a vast network or schema of interrelated symbols, propositions, and
knowledge components (Wilson & Teslow, 1992, p. 1). According to John
Anderson (1980, p. 13),
15


there is a clear sequence or serial ordering to mental operations,
and the important characteristic of an information-processing
analysis, then, is that it involves a tracing of a sequence of mental
operations and their products (information) in the performance of a
particular cognitive task
By this statement, Anderson asserts that there is a sequence of mental operations that
convert raw sensory information into knowledge. Do changes in the mental
operations in this sequence produce corresponding changes in the information that is
converted into knowledge? If so, can these variations in information processing
produce fragile knowledge?
Definition of Fragile Knowledge
The first and most obvious question is what is fragile knowledge? Fragile
knowledge is knowledge that students do not remember, understand, or use actively
much of what they have supposedly learned; and poor thinking, which means that
students do not think very well with what they know (Perkins, 1992, p. 20).
Research shows that fragile knowledge can manifest itself in several forms;
however, these various forms can also combine in a learner which results in the
display of a distinctive cluster of behaviors called fragile knowledge syndrome
(Perkins, 1992, p. 26). Several common forms of fragile knowledge are:
inert knowledge (Whitehead, 1929)
partial or missing knowledge (Perkins & Martin, 1985, p. 5)
misplaced knowledge (Perkins & Martin, 1985, p. 5)
16


naive knowledge (Perkins & Martin, 1985, p. 5)
conglomerative knowledge (Perkins & Martin, 1985, p. 5)
ritual knowledge (Taba & Elzey, 1964, p. 524)
To provide a little more detail, inert knowledge refers to situations where
the student fails to retrieve knowledge but in fact possesses it, as revealed by a
clinical probe (Perkins, 1985, p. 5). In aviation, an important ability that every pilot
must possess is the ability to determine when the aircraft will arrive at a specific
geographic point. Estimating the time of arrival accurately facilitates smooth
coordination with ground support personnel and enables accurate monitoring of fuel
requirements for the flight. Inability to recall the first step of how to compute an
estimated time of arrival (ETA) during the simulation is an example of inert
knowledge within the context of this study.
Partial knowledge is the straightforward case of an impasse due to
knowledge the student has not retained or ever learned, as revealed by clinical
probes failing to reveal signs of the knowledge (Perkins, 1985, p. 5). The symptoms
of partial knowledge are similar to the symptoms of inert knowledge. The key to
distinguishing between the two types is that in the case of inert knowledge probing
by the instructor reveals that the student does in fact remember the information. In
the case of partial knowledge, probing questions used by the instructor fail to reveal
that the student actually possessed the knowledge in the first place.
17


Misplaced knowledge is an instance of a student importing knowledge
appropriate to a particular context but inappropriate to the new context. Attempting
to compute a groundspeed for the ETA computation instead of reading it from the
Navigation Computer System is an example of misplaced knowledge. Computing a
groundspeed is appropriate for flight planning, but it is inappropriate for ETA
computation.
Naive knowledge is knowledge that takes the form of naive theories and
stereotypes (Perkins, 1985). A flat earth is a common example of this type of
knowledge. We know that the earth is not flat; it is round. You cannot fly off the end
of the earth. Using true airspeed instead of groundspeed for the ETA computation is
an indication of the presence of naive knowledge.
Conglomerative knowledge signifies an attempt by the student to force
together disparate elements of knowledge to provide a solution to a problem, repair a
bug, or resolve an impasse. Measuring a distance from a dead reckoning position
instead of measuring the distance from a navigational fix (as the ETA computation
procedure calls for) simply to be able to come up with a possible answer for the ETA
problem is indicative of conglomerative knowledge.
Finally, ritual knowledge is knowledge that is culturally defined and usually
limited to a particular context; for example, knowledge or tasks that are useful in
school but not anywhere else. Teaching students to use a slide rule to perform
18


mathematical calculations when the real world uses electronic calculators is an
example of ritual knowledge. All tasks performed in the study were real-world tasks.
None of the concepts or procedures were for training purposes only. Therefore, ritual
knowledge was not relevant to this study.
All of the concepts mentioned here are manifestations or instances of fragile
knowledge. Fragile knowledge, as mentioned previously, is a significant deterrent to
the successful performance of any cognitive skill. What is a cognitive skill? How is
it acquired? Answers to these questions will yield an insight into the nature of
' cognitive skills and how to minimize the effects of fragile knowledge during the
initial acquisition of a cognitive skill.
Definition of Cognitive Skill
A cognitive skill is the ability to perform an intellectual procedure
(Anderson, 1980). Cognitive skill encompasses two categories of knowledge. The
first category of knowledge in a cognitive skill is declarative knowledge. Declarative
knowledge is knowing that. The second category is procedural knowledge.
Procedural knowledge is knowing how (Gagne, 1985, p. 48). Possessing both
categories of knowledge is paramount to the ability to perform a cognitive skill.
Now that we know what a cognitive skill is, how is it acquired?
19


Acquisition of Cognitive Skill
Recent approaches to examining skill acquisition focus on the works of John
Anderson from Camegie-Mellon. According to Colley and Beech (1989), Anderson
provides a framework for understanding observations made earlier by Fitts (1964) on
the development of skill. Colley and Beech (1989) outline Fitts three stages of
cognitive skill development: _
the cognitive stage, in which the learner makes an initial approximation
to the skill, based upon background knowledge, observation or instruction;
the associative stage, in which performance is refined through the
elimination of errors; and the autonomous stage, in which skilled
performance is well-established but still continues to improve, albeit
very gradually, (p. 7)
Andersons (1982) adaptive control of thought (ACT) theory posits two
stages of skill development. His first stage is the declarative stage. In this stage,
declarative representations relevant to the skill guide behavior. The second stage is
the procedural stage where knowledge undergoes a process of conversion and
continual refinement of conditions and actions into procedures. He calls the process
of converting facts into procedures knowledge compilation (Colley and Beech,
1989). Knowledge compilation also results in an increase in the rate of production
application (condition-action pairs being applied to solve procedural problems) and
tuning processes. This becomes the basis for improving procedural skill performance
and facilitating the transition from novice to expert.
20


In Andersons declarative stage, knowledge about how to perform a skill is
assembled from declarative memory, and from instruction or guidance, into working
memory. General problem-solving procedures then turn this declarative knowledge
into productions which are composed of condition-action pairs. Rum m el hart and
Norman (1981) discuss the utility of analogy to create or extend schematas ability
to incorporate task knowledge in procedures. How is declarative knowledge
converted into procedural knowledge? It happens through a process called
knowledge compilation. Knowledge of the compilation and tuning processes
provides a conceptual framework for understanding how and why instructional
interaction may have an effect on knowledge construction.
Knowledge compilation occurs between the declarative and procedural stages
of cognitive skill acquisition. Knowledge composition has two subprocesses:
composition and proceduralization. Composition (Lewis, 1978) collapses successive
productions (condition-action pairs) into a single production which has the same
effect. Proceduralization removes clauses in the condition of a production that
require matching from long-term memory via working memory. This removal of
clauses permits automatization of the procedural task. Compilation is a gradual
process which allows for errors in procedural information to be corrected over
practice.
21


In the procedural stage of acquisition, productions are tuned, that is, made
more appropriate and efficient for the task at hand. Tuning has three subprocesses:
generalization, discrimination, and strengthening. Generalization occurs when
common aspects of specific productions are used to create a more widely applicable
production which can be used in novel situations. Discrimination restricts the use of
productions to instances where they are successful. Strengthening involves repeating
applications so that the time taken to apply them diminishes (Colley & Beech, 1989).
Now that we understand the cognitive skill acquisition process, its time to look at
the procedural skill acquisition process.
Origins and Development of Procedural Skill
With the broader context of cognitive skills established, the focus now
narrows to procedural skills which are a subset of cognitive skill. Although some
controversy and a few alternative theories exist, Andersons (1982) adaptive control
of thought (ACT) theory remains the seminal work on procedural skill acquisition.
The controversy centers around Andersons claim that knowledge must be encoded
declaratively before proceduralization can occur and that proceduralization can only
occur as a by product of actually performing the procedure. Kieras and Bovair
(1985) disagree. Their study claims that learners can represent production rules (the
If (condition) THEN (action) statements) declaratively before they are
proceduralized without having to actually perform the procedure. In spite of the
22


disagreement on when a production can be proceduralized there is agreement on
what actually happens in the process.
According to ACT theory (Anderson, 1982) the procedural skill acquisition
process involves three distinct steps. These steps are encoding, proceduralization,
and composition (Neves & Anderson, 1981). All incoming knowledge is encoded
declaratively as a set of facts in a semantic network. This information is then
proceduralized by converting it into productions which are condition-action pairs.
These productions created during the proceduralization phase are then subject to the
composition process. During the composition process productions are combined into
larger productions to decrease the amount of time required to apply a procedure.
Anderson (1981, p. 67) provides the following illustration of composition:
Production 12: If
Then
you see a red light
assert danger.
Production 13: If
Then
there is danger and another person is near you
warn that person.
The simple composite is:
Production 14: If you see a red light
and there is danger
and another person is near you
Then assert danger
and warn that person
23


However, this process is not without its problems. Problems that result in
bugs or impasses can occur at any point during the acquisition process. Burton and
Brown (1982) conducted extensive research on this subject with their classic
program Debuggy. Bugs are systematic errors in the execution of procedural skills,
and these errors produce fragile knowledge. Burton and Brown (1982), VanLehn
(1982), Shaw (1982), and Sleeman (1984) observed and documented over 200
different kinds of bugs in the acquisition of simple mathematical skills alone. Bugs
indicate the presence of fragile knowledge in the learner.
VanLehn (1982) claims that mysteries abound in bug data and looks to
answers to the following questions as central to solving these mysteries.
Why are there so many different bugs? What causes them?
What causes bugs to migrate or disappear? Why do certain bugs
only migrate to certain other bugs? Often a student has more than
one bug at a timewhy do certain bugs almost always occur together?
Do co-occurring bugs have the same cause? Most important, how
is the educational process involved in the development of bugs? (p. 49)
VanLehn appears to believe that a correlation may exist between the educational
process and the development of bugs that produce fragile knowledge. Since bugs
indicate the presence of fragile knowledge in the learner, can the educational process
eliminate or reduce these bugs? Are there things educators can do to facilitate bug-
free procedural skill learning?
24


Executive Control Processes
Much of VanLehns work focuses on impasses caused by bugs and points to
executive control processes to solve these impasses and repair the bugs that caused
them. Executive control processes facilitate the planning, monitoring, and evaluating
of performance on intellectual tasks (Vygotsky, 1978).
Case (1984,1985) argues that the learners ability to assemble increasingly
sophisticated executive control structures underlies most of their cognitive change.
According to Case, executive control structures consist of three interrelated parts: a
representation of the problem situation, a representation of the problem objective,
and a representation of the problem strategy. This tells us what the structure of our
interaction with the learner must contain to be considered an executive control.
Sternberg (1984, 1985) further refines the executive control process. He
identifies nine types of metacomponential processing that fulfill executive
processing functions. Metacomponential processes are subprocesses of a procedural
skill much like cognitive strategies that help regulate the execution of the procedure.
Nine types of metacomponential processing account for many of the consistent
findings across the literature on cognitive development. He contends that a learner
accomplishing an intellectual task must:
Determine the nature of the problem under consideration
Select the appropriate performance components for solving the problem
Select a strategy for combining performance components in an optimal
way
25


Select one or more representations for the information crucial to problem
solution
Decide how to allocate processing resources
Monitor their performance
Interpret feedback about the effectiveness of their efforts
Decide how to act upon positive or negative feedback
Modify their performance in response to the feedback
Now that we know exactly what the elements of an effective executive control
strategy are. These elements must be incorporated into our instructional interaction
with the learner if our objective is to provide effective executive control strategies.
Since studies have shown executive control processes to be effective for
repairing bugs (Burton & Brown, 1982; VanLehn, 1982), can they also provide clues
to solving the problems associated with the construction of fragile knowledge?
Referring back to VanLehns most important issue earlier, how is the educational
process involved in the development of bugs? Research on teacher behavior sheds
some light on the issue.
Teacher Behavior and Student Achievement
Jere Brophy (1979) conducted extensive research on teacher behavior and its
effects on student achievement to show a significant correlation between teacher
behavior and student achievement. Brophy also exposes gaps in the literature and
research methodology. One of the significant gaps is research on the effectiveness of
questioning techniques on student achievement. A synthesis of findings over several
studies indicates that factual questions with single answers was correlated positively
26


and significantly with achievement, whereas the frequency of more complex,
difficult, or divergent questions had negative correlations (Brophy, 1979). The lack
of significant results for complex or higher level questions has puzzled researchers
and caused them to conclude that we need to rethink what is meant by types of
questions and their effects. Rosenshine (1976) summarized the findings of several
studies on teaching behaviors and student achievement conducted during the 1970s
by stating that we are as yet unaware of the optimal types and sequencing of
questions...optimal sequences may be different for different types of students and
outcomes, but we still do not know what these sequences are (p. 61). Meredith
Galls work at the Far West Laboratory for Educational Research and Development
supports Brophys claims. Gall (undated, p. 707) says that
Granting the importance of questions in teaching, researchers
still do not know much about them. What educational objectives
can questions help students achieve? What are the criteria of an
effective question and how can effective questions be identified?
Until researchers find answers to questions such as these, a viable
behavioral technology of teaching will remain unrealized.
Gall (undated, p. 707) goes on to say that very few researchers have explored the
relationship between teachers questions and student outcomes.
Rosenshine (1976, p. 61) confirms this by stating that Since 1940 fewer
than 25 studies have been conducted on any specific variable such as teacher praise
or teacher questions. More recent research is equally as sparse. The basic
27


conclusion here is that researchers agree that the issue is important, but little research
is actually being conducted in this area. Brophy and Gall support the need for more
research on teacher questions. While the need for this research is apparent, research
on the use of questioning as an instructional strategy is also needed.
Instructional Interaction Strategies
Instructional interaction strategies were previously defined in Chapter 1.
Common instructional interaction strategies are addressed in the Minnesota
Department of Educations publication entitled Classroom Instructional Design:
Options for Teacher/Student Interaction (1989). This publication details three
common teaching models for teacher/student interaction. These interaction strategies
are: directive, investigative, and interactive. The directive strategy involves telling
to present content. The investigative strategy involves a questioning approach to
achieving the lesson objective, and the interactive strategy is a mixture of both
telling and questioning.
Perkins (1992) hails Mortimer Adlers (1980) paieda proposal as a solution
to fragile knowledge syndrome. Adler proposes an interaction strategy that involves
didactic instruction followed by coaching and Socratic approaches to presenting
instruction. Here, the didactic approach corresponds to the directive approach from
the Minnesota group. Adlers coaching phase corresponds to the interactive stage,
28


and the Socratic approach corresponds to the investigative strategy mentioned
earlier.
In the directive approach to interaction, the instructor or intelligent tutoring
system is the possessor of knowledge. This knowledge is presented to the learner by
telling or making declarative statements to explicitly direct the learner through the
instructional activities to achieve the learning objectives and goals. These learning
objectives could be a complex process or procedure. The following example, which
is the actual procedure to compute an estimated time of arrival (ETA) at a
predetermined geographic point, illustrates the directive interaction strategy in a
typical instructional scenario:
Instructor: Its time to compute an ETA. The first thing you need
to do to compute an ETA is to get the aircrafts
groundspeed from the Navigation Computer System.
Then, take your plotter and measure from the fix you
resolved and plotted on your chart to the geographic
point that you would like to compute the ETA to.
Now, divide the distance you measured with your plotter
by the groundspeed. The result is how long it will take
to fly the distance you measured. Add this time to the
time of your fix and you now have your estimated
time of arrival.
At this point in the scenario the instructor would then repeat the same instruction
once again as the student performed the task.
In the questioning approach to interaction, questions are used to lead the
learner through the accomplishment of the process. Making declarative statements to
29


the learners is not part of this interaction strategy. The questioning approach should
not be confused with inquiry teaching, because in this case, what is to be learned is a
previously established process that has a correct sequence for accomplishment and
consequences for incorrect sequences. Learners are not encouraged to invent their
own sequences or cases and test their hypotheses. Again the same task is used to
illustrate the questioning approach:
Instructor: What do we need to do right now?
Student: We need to compute an ETA to our destination.
Instructor: How do we do that?
Student: First, we need to get the aircrafts groundspeed.
Instructor: Where do we get that information?
Student: We get it from the Navigation Computer System.
Instructor: What do we do with the groundspeed?
(The exchange then continues until the student has properly and correctly computed
an ETA).
In the mixed approach to instructional interaction, learners are first told how
to accomplish the process (as in the directive approach) and then questioned through
it as they perform it (as in the questioning approach). Remember, each of these
strategies involve various ways of interacting with the learner to present a previously
established procedure. The learners ability to perform the process correctly and
autonomously indicates successful acquisition of the procedural skill prescribed by
the learning objective.
30


All three interaction strategies mentioned previously represent the mean and
the extremes for common instructional interaction strategies within the context of
procedural skill acquisition. So, which of these strategies is most effective for
reducing fragile knowledge construction? Are some strategies more effective than
others in eliminating or reducing the construction of particular types of fragile
knowledge?
Brophy (1979) and Good (1979) conclude that direct instruction has
important advantages and is clearly superior to other approaches for producing
mastery of basic skills. Advocates of questioning approaches point to the ability to
encourage learners to reflect and thus stimulate thinking and learning of the process
or procedural skill being learned (Aschner, 1961). Yet, Adler (1980) and Perkins
(1992) argue that the best interaction strategy is to provide a didactic and
questioning approach together as a congruent instructional interaction strategy. Who
is right? Does the interaction strategy even matter at all? The purpose of this study is
to find an answer to these questions.
Hypotheses
This experiment examined the effect of common instructional interaction
strategies on the construction of fragile knowledge in learners. Furthermore, the
experiment assessed whether providing learners with executive control strategies can
inhibit the production of fragile knowledge during initial procedural skill
31


acquisition. Prominent educational theorists and researchers hold conflicting
positions on the efficacy of various instructional interaction strategies to produce
rich understanding (and less fragile knowledge) in the learner. This study examined
each of these positions and tested their effectiveness in eliminating or reducing the
construction of fragile knowledge.
Hypothesis 1 The Effect of Interaction Strategies on Fragile Knowledge
Construction
Learners who receive minimal or directive type strategies will construct more
fragile knowledge than learners who received more facilitative (questioning or
mixed) styles. With respect to the first independent variable, instructional interaction
strategies, which had four levels: no instructional interaction, directive, questioning,
and mixed (directive and questioning), the results of the experiment will show that
the interaction strategy makes a significant difference on the learners construction
of fragile knowledge. This will be indicated by the no instructional interaction and
directive groups presenting lower total scores on the posttest. The questioning and
mixed groups will present higher scores on the posttest.
Hypothesis 2 The Efficacy of Executive Control Strategies
Regarding the second independent variable, executive controls, which had
two levels: the explicit teaching of executive control strategies and no explicit
teaching of executive control strategies, the results of the experiment will show that
32


possessing and using executive control strategies will minimize the construction of
fragile knowledge. This will be indicated by higher scores on the posttest than the
corresponding group that did not receive explicit instruction on executive control
strategies. For example, the directive with executive controls group will achieve
higher scores on the posttest than the directive without executive controls group and
the same will follow for each of the remaining treatment groups.
Hypothesis 3 The Interaction of Instructional Interaction Strategy and Control
Processes
Concerning the issue of possible interaction between the two independent
variables, the results of this experiment will also show that there is no significant
interaction between instructional interaction strategies and the explicit teaching of
executive control processes. Thus, there will be no particular combinations of
instructional interaction strategies and the absence or presence of executive controls
that will produce statistically different results on the construction of fragile
knowledge.
The theory contained in the literature of this chapter indicates that the
interaction strategy should in fact matter. The theoretical argument is that encoding
of condition-action productions takes place during the associative stage of
procedural skill acquisition. In keeping with the information-processing model,
different knowledge compilers should produce different encoding of condition-
33


action productions. Input from domain experts to the compiler Or, ultimately, the
compilation process should produce encoded condition-action productions that are
more stable and less tenuous than unassisted knowledge compilation. Thus, assisted
compilation should be less fragile or contain less fragile knowledge. Furthermore,
encoding executive control processes along with the new procedural skill should
provide learners with more tools to resolve impasses or repair bugs. This ability to
autonomously resolve these impasses or bugs allows learners access to knowledge
not previously attainable and stabilizes what was previously fragile knowledge. The
research design, methodology, and material reflect the assertions posed in this
chapter and help to ensure meaningful and useful results.
34


CHAPTER 3
EXPERIMENTAL DESIGN, METHODOLOGY, AND MATERIALS
Methodology
This study examined the relative effects of four common instructional
interaction strategies: no instructional interaction, directive, inquiry or questioning,
and mixed (directive and questioning) strategies along with the presence or absence of
explicit instruction on executive control processes and their effects on the
construction of fragile knowledge during initial procedural skill acquisition. Subjects
were instructed in five basic navigation procedural skills according to the interaction
strategy prescribed by their respective treatment group. The five basic navigation
procedural skills are listed here in Table 3.1:
Table 3.1
Basic Navigation Procedural Skills
Determining an initial heading to fly to navigate to a prescribed geographic point
Determining the aircrafts actual position using a radio aid to navigation
Determining the aircrafts actual position using the RADAR
Computing a heading to alter the aircraft to when off the desired course
Computing an Estimated Time of Arrival (ETA)
Determining the aircrafts most probable position in the future
35


This chapter provides a detailed description of the participants, design,
experimental materials, instrumentation, experimental procedures, and evidence of
compliance with the assumptions for the statistical analyses conducted in this study.
Participants
One hundred and eighty-four freshmen undergraduate students enrolled in an
introductory aviation science course participated in this study. AVI00 (Introduction
to Aviation) is a required course for all first-year cadets at the United States Air
Force Academy. This study was conducted during the Fall 1995 semester. There were
approximately 159 males and 25 females, ranging in age from 17 to 20. The students
entering the Academy are selected from the applicants who typically score in the top
twenty-five percent on the SAT/ACT exams. The average SAT score was 1430. All
students, regardless of academic major, are required to complete the core curriculum
at the Academy. The core curriculum focuses on hard science and general
engineering. No subject selected for this study had any formal navigation training. All
subjects completed the Academys summer training and orientation programs. The
Academy environment for first-year cadets is very regimented, strict, and demanding.
Twenty-one credit hours with several concurrent extracurricular activities is a typical
load at the Academy. Consequently, first-year cadets are used to a highly
proceduralized learning and living environment and are compliance oriented toward
authority. Summer training and orientation programs acclimated the cadets to the
36


military instructional environment before the study began. They also completed the
introductory microcomputer orientation for first-year cadets. The orientation
provided the subjects with specific microcomputer training on hardware, start-up
procedures, keyboarding and use of the mouse, operating systems, operation of the
local area network, and word processing. At the time this study was conducted,
subjects had already completed assignments on their computers for other classes such
as English, history, physics or chemistry which was evidence of their proficiency with
the same microcomputer system that the simulator is based on. The students reading
skills were tested during inprocessing to the Academy and were at least an 11.5
reading level on the SRA reading scale for both speed and comprehension.
Student participation in this study was part of their normal core academic
schedule, specifically for this Aviation Science class. A pretest assessed their level of
prior knowledge of basic navigation. The pretest was a parallel form of the posttest.
The test covered the same basic air navigation principles and procedures as the
posttest. The purpose of the pretest was to assess the amount of prior knowledge the
subject possessed prior to the treatment. Any subject who scored above 80% was
considered to have a significant amount of prior knowledge and was not selected for
participation in the study. The eighty percent cut score on the pretest was established
at the recommendation of Academy institutional research personnel who have
conducted numerous studies on Academy cadets. Academy cadets are very high
37


ability subjects who are highly motivated in this particular subject area. However, no
subjects were actually eliminated from the study due to a significant amount of prior
knowledge because the highest score on the pretest was 76%. The participants were
randomly assigned to a class section and each class section was randomly assigned to
one of eight different treatment groups. Ten sections out of a possible sixty sections
were selected to participate in this study. Each section contained approximately 20
subjects who were randomly assigned by computer to that particular section. The
subjects were motivated by the typical pressures of a competitive academic
environment and special, additional, highly-desired incentive activities for those that
distinguished themselves by their performance on the pre and posttest instruments.
The tests used in the study were part of their regular academic program in AV-100
and counted toward their class standing. Class standing determined who received the
additional incentive activities.
Design
The design of the experiment was a 4 X 2 factorial design (see Table 3.2
below) in which the effects of the absence or presence of executive control strategies
within four different instructional interaction strategies were tested for their effects on
the amount of fragile knowledge constructed during initial procedural skill
acquisition.
38


Table 3.2
Experimental Design
Instructional Interaction Strategy
No Interaction Directive Questioning Mixed
Executive Controls Group 1 Group 3 Group 5 Group 7
No Executive Controls Group 2 Group 4 Group 6 Group 8
The experimental design followed Campbell and Stanleys (1963) pretest-posttest,
control group design. The graphical depiction of the experiment is:
ROXiO
R0X20
ROX3O
ROX4O
ROXjO
ROXfiO
ROX7O
RO 0
The first independent variable was the instructional interaction strategy. These
strategies included: no instructional interaction, directive, questioning, and Mixed
(directive and questioning) interaction strategies. The second independent variable
was absence or presence of explicit instruction on executive control processes to
double-check work or provide a broader and deeper understanding of the concepts
and principles underlying the procedures.
39


Eight treatments were used in this experiment. Each of these treatments was
composed of combinations of the two independent variables and was depicted earlier
in Table 3.2 entitled Experimental Design.
Variables Studied
The relative effects of two independent variables were studied. The first
independent variable was the interaction strategy used, with four levels. The second
independent variable was the presence versus absence of explicit instruction on
executive control processes. This independent variable had two levels.
Independent Variables
Instructional Interaction Strategy. The effect of varying the instructional
interaction strategy (one independent variable with four levels) was the first
independent variable studied (see Table 3.2). This experiment tested common
interaction strategies. These common instructional interaction strategies were no
instructional interaction (the control group), directive strategies, questioning
strategies, and mixed (directive and questioning) strategies.
Executive Control Processes. The second independent variable studied was
the effect of explicit instruction on executive control processes (one independent
variable with two levels). Perkins (1985) pointed to direct instruction on executive
processes as a significant inhibitor of fragile knowledge construction. Eight treatment
groups, as described previously, tested this posit.
40


Dependent Variables
The dependent variables were the amount and types of fragile knowledge that
were constructed as a result the instructional interaction strategy the subject received.
The dependent measure was a multiple-choice immediate posttest given at the
conclusion of the treatment to assess the amount of fragile knowledge the subject
possessed. The test represented a solid grasp of navigational concepts, principles, and
procedures. A low score on the measure indicated the presence and degree of fragile
knowledge present in the subject. A detailed description of the posttest is provided
under the discussion of the individual instruments.
Treatments
There were eight treatment groups (see Table 3.2 earlier in this chapter).
Groups 1 and 2, groups 3 and 4, groups 5 and 6, and groups 7 and 8 differed by the
instructional interaction style or strategy they received. Each odd numbered group
additionally differed by the presence or absence of explicit instruction on executive
control processes that could be used to cross-check or provide a better understanding
of the air navigation process.
Treatment groups 1 and 2 did not receive any instructional interaction. Group
1 was provided information on executive control processes by additional text
covering these subjects. Group 2 was the control group. This treatment group was
41


not provided any instructional interaction or the additional coverage on executive
control processes in its text.
Treatment groups 3 and 4 were given directive or didactic instruction on
navigation principles and procedures. Subjects in this group were told what to do and
how to do it from the beginning until the end of the simulator mission. Group 3 was
given additional instruction on how to cross-check their work, identity and correct
their own errors, and orient themselves strategically. Group 4 was not provided with
that instruction.
Groups 5 and 6 experienced a more facilitative instructional strategy.
Instructors in this group used questions instead of making statements to lead subjects
through the procedures. Instructors provided subjects in the 5th group with the key
questions they needed to ask themselves to prevent or correct errors and orient
themselves. Group 6 did not receive this instruction.
Groups 7 and 8 received a combination of interaction strategies from the two
previous groups. Instructors used statements and questions to instruct subjects in this
group. Once again, subjects in Group 7 were given explicit instruction on strategies
and executive control processes to facilitate autonomous performance.
Materials
The primary device used for the experiment was a microcomputer-based
simulation of the navigator station of an Air Force T-43 A. The T-43 is a military
42


version of the Boeing 737-200 commercial airliner. The NAVSIM (navigation
simulator) graphically reproduces the panel in the aircraft and is supported by the
actual flight model for the Boeing 737 resulting in simulation that is realistic enough
to reduce the number of inflight training missions.
NAVSIM runs on an IBM PC based microcomputer and requires at least an
80386 microprocessor, 1 MB RAM, 1 floppy disk drive, and an EGA monitor. The
actual system used for the experiment was an 80486DX/66 MHZ PC with 8MB
RAM, 120B hard drive, and an SVGA monitor. The NAVSIM has been used to train
students and instructors in the T-43 for the past five years. The software is updated
regularly to accurately reflect changes to the aircrafts systems and performance.
The other materials used were standard Air Force issue items. These items
included a current aeronautical navigation chart for the route to be flown, a
navigation plotter, a Dead Reckoning Computer, a current IFR Supplement, and a
navigation worksheet.
Topic Selection
Three fundamental concerns prompted the selection of the subject. The first
factor in the topic selection was the concern to address a fundamental, authentic,
problem. Fragile knowledge can have dire consequences. Educational professionals
have an ethical responsibility to address the issue of fragile knowledge and any
influence they might have on the construction of it in the products they produce or the
43


people they instruct. The second factor was the need to teach a new procedural
subject to motivate the learners. Teaching a new subject reduced possible
contamination from prior general knowledge or experience. The need for a large
number of subjects was the final factor in the selection of this particular topic for the
study. A sufficiently large sample was required to cover the range of common
interaction strategies, examine the effect of strategic controls within these groups, and
retain statistical power. Navigation is a good candidate because all first-year students
at the Air Force Academy are required to complete the course to graduate.
Once the academic subject area was established, ten subject matter experts
reviewed the course content and objectives to establish the content validity of the
instructional materials. The Subject Matter Experts identified five primary procedural
skills used in basic air navigation (see Table 3.1). These procedures and their
underlying concepts and principles were taught lecture style in the first four lessons of
the course (eight contact hours). This experimental treatment used the actual training
methods and materials the Air Force Academy uses to teach flight training students to
navigate the T-43. Only the instructional interaction strategy was changed for this
study; the goals, objectives, content, methods, or materials were not altered for the
experiment.
44


Instruments
Pretest and Posttest
These two tests were instructor developed multiple choice examinations
designed to measure the subjects ability to perform the five basic navigational
procedures (see Table 3.1) and test their understanding of the concepts and principles
underlying the procedures.
Each test contained 21 multiple choice questions with an additional 29 fill-in-
the-blank substeps (see Appendices D and E). Each question required the subject to
demonstrate his or her ability to perform an entire process or a part of one of the five
basic navigation procedures mentioned earlier. To answer these questions subjects
had to refer to print screens from the simulator which were included as supplemental
material in the test. This provided graphical depictions of real-world conditions which
the subject then had to interpret just as they would inflight. Each question had three
or four possible answers, one correct answer and two or three distracters. The
distracters were based on fragile knowledge of navigational concepts, principles, or
procedures. The lower the subjects score on the instrument the higher degree of
fragile knowledge the subject possessed.
The pretest and posttest were initially developed by a highly experienced
instructor. Ten subject matter experts then reviewed the instruments for their content
validity, comprehensiveness, and accuracy. The ten subject matter experts also
45


reviewed the distracters in each question to ensure that these distracters represented
valid examples of the common types of fragile knowledge as defined in the definitions
section of chapter 1. The tests were changed based on input from the subject matter
experts until a consensus was reached among the ten subject matter experts. Parallel
forms were used to reduce testing effects and to ensure the security and integrity of
the instruments. Recent graduates of the course were used to conduct a formative
evaluation on both instruments. The formative evaluation group took both the pretest
and the posttest during the same testing session. Results of the formative evaluation
were used to improve the form and the function of the pre and posttest instruments.
An item analysis showed the parallel forms reliability to be .88.
Procedure
Pilot Study
Once the instrumentation was finished, I conducted a pilot study. I randomly
selected eight additional recent graduates to participate in the pilot study. These
participants were randomly assigned to one of four possible treatment groups in the
pilot study. The materials and procedures mirrored the main study. Results from the
pilot study were generally in the hypothesized directions. Only minor changes were
made as a result of the pilot study. These changes generally focused on clarifying
verbiage in the instructions given to the subjects during the simulator mission.
46


The experimental procedure followed a pretest, intervention, and posttest
sequence. The pretest was administered immediately after the regular academic
portion of the course. The treatment was conducted as two instructional practice
simulator missions. The immediate posttest was a mid-course examination.
Pre-Treatment Training of Instructors
Variability among the sixty instructors who participated in this study was a
significant concern. Several steps were taken to minimize this possible effect. First,
instructors were sent an overview of the study with sufficient details explaining the
purpose, scope, methodology, and procedures of the research effort. I then held a
meeting with the instructors to provide direct interaction and answer any questions
before instructor training commenced.
Instructor training consisted of a one hour and forty minute practice teaching
simulator mission. I gave each instructor a job aid to remind them of how to conduct
their interaction with the subject according to their treatment group and a checksheet
to remind the instructor of the content that needed to be covered during the lesson.
During the practice mission on-the-spot corrections and constructive feedback were
given when instructors deviated from the prescribed interaction strategy.
During the actual treatment, instructors used the same job aids from the
practice mission to remind them of their interaction strategy roles. I continued to
provide on-the-spot corrections to the offending instructors when they deviated from
47


their prescribed roles. If I had to provide more than one correction to an instructor I
rejected his or her data.
Classroom Pretest
I administered the pretest in the classroom prior to the start of the fifth
classroom lesson on navigation principles. By this point, subjects had attended eight
contact hours of classroom. The pretest was a parallel form of the posttest. The
pretest took twenty-two minutes for the quickest student and thirty-one minutes for
the slowest student. The classroom instructor told the subjects that the test would
count toward their class standing and opportunity for incentive flights. No one gave
the subjects any feedback on how they performed on the pretest. Furthermore, they
were restricted from discussing anything about the test with anyone else or going
back and reviewing any of the content areas on the test until after completing the
second simulator mission (which included the posttest).
Assignment to Treatment
The participants were randomly selected and randomly assigned to one of
eight treatment groups. Heres how this was accomplished. First, the Registrar
segmented the course offering into six modules. Each module contained ten sections
(one section contained approximately 20 students). Three modules were offered per
semester, and each module occurred one after the other. Then, the scheduling
computer randomly assigned each subject to one of the sixty possible sections of AV-
48


100 within these modules. The computer did this by alphabetically assigning each
participant to the first available section until the section was full. I conducted this
study during the fall semester; so, that left three modules to choose from. Due to
coordination problems with the department, I wasnt able to conduct the study until
late in the semester. The coordination-problem limited my access, and effectively
made the choice of which module to study for me. I got together with the Course
Director and we each picked random numbers from one to ten to assign each of the
ten sections within the module to one of the eight treatment groups.
Instructional Intervention
I gave the subjects the pretest when they arrived for their first simulator
mission. The instructors started the simulator mission once all the subjects had
finished the pretest. Each group had approximately twenty students. Additional
instructors were used to reduce the student to instructor ratio to 4:1. All simulator
missions started at the same prescribed geographical point, flew a prescribed route,
and ended at an established termination point. Only the instructional interaction style
and instruction on executive control processes varied by treatment group.
Subjects started the second simulator mission immediately after they arrived.
The second simulator mission was identical to the previous one. Instructors
administered the posttest when the subjects either completed the mission or could no
49


longer continue the mission. The instructors thanked and released the participants
once they completed the posttest.
Data Collection
I handscored the multiple-choice pre and posttests to prevent errors due to
machine misfeeds or the mismarking of answer sheets. I entered student responses to
each question along with pretest and posttest scores directly into the SPSS Data
Editor. This served as the master database for all statistical analyses. This database
also contained the pre-study data collected earlier. All simulator missions were
recorded on 3.5 floppy diskettes. Every action the subject did during the entire
mission was recorded for future reference and could be played back for analysis.
Data Analysis
SPSS for Windows 6.1, Statistical Package for the Social Sciences
(SPSS)(1994), a microcomputer based statistics package was used on an IBM based
80486 microcomputer, to conduct all statistical analyses. All data were entered
directly into the programs data editor.
I performed an Analysis of Covariance (ANCOVA) on the total score
obtained on the posttest, with the total score obtained on the pretest as the covariate.
The categories of the 4X2 factorial design were instructional interaction style by
executive control process. Also, I conducted a Multivariate Analysis of Variance
(MANCOVA) to determine whether specific instructional interaction strategies
50


produced particular types of fragile knowledge. I did this by recoding each student
response to every question on the posttest into one of the five types of fragile
knowledge that the response represented. For example, if a subject selected answer B
to question number one and the subject matter experts had previously determined that
answer B was a valid example of inert-knowledge, then the subjects response was
recoded into one instance of a display of inert knowledge. For the MANCOVA, the
instructional interaction strategy was the factor variable and the five types of fragile
knowledge (inert, partial, etc.) were the dependent variables. As a result of the
recoding, these dependent variables contained the number of instances each subject
displayed that particular type of fragile knowledge on the posttest. An alpha level of
.05 was used for all statistical analyses. A discussion and evidence of compliance with
the assumptions necessary for conducting a valid analysis of variance and multivariate
analysis of variance follows.
Assumptions
The assumptions for analysis of covariance are : (1) random and independent
sampling, (2) samples from a normal distribution, (3) true variances of each group are
equal, (4) independence of means and variances, (5) at least interval scaling (Gabriel
& Hopkins, 1974, p. 378), (6) a correlation between the covariate and the dependent
variable, and (7) homogeneity of regression (Keppel & Zedeck, 1989). The students
were randomly assigned to groups according to the plan presented earlier in this
51


chapter. No subjects were assigned to more than one treatment group. The number of
participants in each treatment group varied from 11 to 31 in the final statistical
analysis. Stem and Leaf Plot analysis of posttest scores revealed that students fulfilled
the requirements for a normal distribution. Additionally, posttest scores further
indicated that each group followed a normal curve centered around a 58.8% mean
with an 11.8% standard deviation. Cochrans C and Bartlett-Box homogeneity of
variance tests conducted on the posttest data. Cochrans C(17,8)=. 14569 p=1.000
and Bartlett-Box(7,14,688)=.06243 p=1.000 supported the null hypothesis that all
cell variances were equal. Data in Tables 3.2 and 4.1 support the independence of
means and variances. Interval scaling was assumed by constructing the posttest so
that scores ranged from 0 to 100. Correlation between the covariate and the
dependent variable is established by the fact that the pretest (covariate) was a parallel
form of the posttest. Independence was achieved by taking the pretest measure before
the treatment occurred. Homogeneity of regression was verified by examining the
residual plots for each treatment group. All plots showed only minor deviations from
linearity. The deviations between the pretest residuals and posttest residuals were in
the same direction and nearly the same amounts.
The assumptions for multivariate analysis of variance are an extension of the
assumptions for univariate analysis of variance. As mentioned previously, the
assumptions for analysis of covariance are : (1) random and independent sampling, (2)
52


samples from a normal distribution, (3) true variances of each group are equal, (4)
independence of means and variances, (5) at least interval scaling (Gabriel & Hopkins,
1974), (6) a correlation between the covariate and the dependent variable, and (7)
homogeneity of regression (Keppel & Zedeck, 1989). These assumptions still hold for
multivariate analysis with the added assumption that the composite observations are
normally distributed, equally variable in the populations sampled, and independent
(Gabriel & Hopkins, 1974). The data used in this analysis were the same data used in
the previous univariate analysis. Posttest responses were recoded to represent the
type of fragile knowledge indicated by the answer the subject selected during the
posttest measurement. Since these assumptions have already been met, they will not
be addressed again. However, compliance with the additional assumption posed by
the multivariate analysis has not yet been demonstrated, and a brief examination
follows.
The assumption of normal distribution, equal variance, and independence of
composite observations for the multivariate analysis was not met for the recoded data.
Analysis of stem and leaf plots for the five types of fragile knowledge
(conglomerative, inert, misplaced, naive, and partial) showed a normal distribution for
all types except inert knowledge. The inert category also violated the homogeneity of
variance assumption with a Cochrans C(36,4)=.62344 and p<.0001. Multivariate

53


tests for homogeneity of dispersion were also violated. These tests revealed a Boxs
Mof F=T57.3, df(45,30425), p<0001.
Although multivariate assumptions for all five categories were not met due to
the effect of the inert category, assumptions for the remaining four categories were
met. Since the assumptions for the remaining categories were met, a MANCOVA was
conducted with the inert category removed from the_analysis. The PRETEST was
used as a covariate to account for aptitude effects and keep this analysis consistent
with the ANCOVA for the main effects of the first research question.
With the assumptions for the analyses met, the analyses were completed, and
the results of the study follow.
54


CHAPTER 4
RESULTS
This chapter presents the results of the statistical analysis of the
collected data. The primary focus of the study was the impact of common
instructional interaction strategies on the construction of fragile knowledge
during initial procedural skill acquisition. It also examined whether certain
interaction strategies produced particular types of fragile knowledge. The
secondary focus of the study was whether explicit instruction on executive
control strategies can influence fragile knowledge construction.
55


Table 4.1
Summary of Means and Standard Deviations bv Instructional Interaction
Strategy and Executive Controls
Instructional Interaction Strategy Pretest M
None SD
(Control Group) Posttest M SD n Pretest M
Directive SD Posttest M SD n Pretest M
Questioning SD Posttest M SD n Pretest M
Mixed SD Posttest M SD n Pretest
Executive M
Controls SD
Totals Posttest M SD n
Totals for
Executive Controls Strategy
Yes No
46.3 34.2 39.1
9.6 13.0 13.1
59.7 55.1 60.0
11.4 11.4 11.6
11 16 27
38.6 46.2 42.4
10.9 10.2 11.0
55.8 55.0 55.4
10.6 10.7 10.5
13 13 26
44.8 45.1 44.9
10.3 11.4 10.6
57.4 55.7 56.8
11.4 11.8 11.5
31 17 48
46.0 43.7 45.1
11.5 16.1 13.3
63.3 65.2 64.0
11.5 12.3 11.7
29 17 46
Yes No
44.4 42.2
10.8 13.6
59.5 58.0
11.5 12.2
84 63
56


Effect of Instructional Interaction Strategy: Research Question #1
Does the instructional interaction strategy used influence the
construction of fragile knowledge during initial procedural skill acquisition
in the learner? Yes, a main effect for instructional interaction strategy was
found in the ANCOVA (F(3,138)=5.43, p=.001) see Table 4.2 and Figure 4.1
below. Post-hoc Scheffe comparison indicatedthat the mixed strategy group
(the group that received the combined directive and questioning interaction
strategy) scored significantly higher on the posttest than the other groups.
Table 4.2
Analysis of Covariance Summary bv Independent Variable
DEP VAR: POSTTEST COVARIATE: PRETEST N: 147
SOURCE SUM SQUARES DF MEAN-SQUARE F-RATIO P
INST INTER STRAT 1590.18 3 530.06 5.43
.001* EXEC CONTROLS 7.33 1 7.33 .08
.784 INST INTER STRAT by EXEC CONTROLS 274.11 3 91.37 .94 .425
ERROR 13466.57 39 97.58
* p < .05
Note. The abbreviations used in this table are as follows:
INST INTER STRAT-Instructional Interaction Strategy. This is one of four
common elemental methods or tactics for conveying instruction to the learner
EXEC CONTROLS-Executive Control Strategies. These are additional
procedural tools used to control and cross-check the execution of the
57


procedure. Effect size and observed power at the .05 level were .264 and
1.00 respectively.
Figure 4.1
Posttest Scores bv Instructional Interaction Strategy
100
PL,
20
0J--------.----------------------=-----------5----------
N= 27 26 48 46
None Directive Questioning Mixed
Interaction Strategy
Types of Fragile Knowledge Construction: Research Question la
Do certain instructional interaction strategies result in a particular
type of fragile knowledge construction over another type? No significant
differences were found among the five types of fragile knowledge for any
specific instructional interaction strategy (multivariate F(12,423) = 1.69,
p=.067). See Figure 4.2 for a graphical summary of the results. The
multivariate effect size and power at the .05 level were .046 and .86
58


Instances of Fragile Knowledge
respectively. No further analyses were conducted on types of fragile
knowledge construction.
Figure 4.2
Instances of Specific Types of Fragile Knowledge bv Interaction Strategy
CONGLOM
INERT
MISPLACE
NAIVE
PARTIAL
Effect of Executive Controls: Research Question #2
Does the explicit teaching of executive control processes (cross-
checks and double-checks) influence the construction of fragile knowledge
during initial procedural skill acquisition? No, there was no main effect for
executive controls (F(l,138)=.08, p=.784). See Figure 4.3 for a graphical
presentation of the results of this analysis. Furthermore, there was no
None Directive Questioning Mixed
Interaction Strategy
59


interaction of instructional interaction strategies and executive controls
(F(3,138)=.94, p=.425).
Figure 4.3
Posttest Scores bv Executive Controls
N= 63 &
None Yes
Executive Controls
Now that weve seen the numerical results of the statistical analysis,
what does all this mean in terms of the study. Its now necessary to look at
the results and interpret them in light of the instructional theory that was
presented in Chapters 1 and 2. What implications do these results have for
educational professionals who are concerned about fragile knowledge
construction in learners?
60


CHAPTER 5
SUMMARY, DISCUSSION, AND RECOMMENDATIONS
Summary of the Study
This study focused on the effects of four common instructional interaction
strategies on the construction of fragile knowledge. It also examined whether the
explicit teaching of executive control strategies had an effect on fragile knowledge
construction. This chapter provides a summary of the research effort to answer these
questions, a discussion of the results with implications for the improvement of
instruction involving procedural skill acquisition, and recommendations for future
studies in this area.
The central research question was whether the strategy used to interact with
the learner effects the stability of the knowledge the learner acquires? If so, does a
particular interaction strategy tend to create a specific type of fragile knowledge?
Furthermore, does the explicit teaching of executive control strategies also influence
the construction of fragile knowledge in the learner?
The treatment consisted of common variations in the way instructors
interacted with the students. Forty-three instructors were trained and participated in
the administration of the treatment. The instructional interactions included no int
61


eraction with the student, directive interactions, questioning interactions, and a mix
of directive and questioning interactions. Explicit instruction on executive control
strategies was also provided to only one of the two sub-groups within each
instructional interaction strategy treatment group. The executive control strategies
were cross-checks and double-checks of the computations from each of the
navigational procedures.
Subjects in the No Instructional Interaction without Executive Controls group
were informed of the lesson objective, given the coursebook, lesson materials, and
sent to the simulator. Subjects in the No Instructional Interaction with Executive
Controls groups were given the same instructions and materials except that their
coursebook provided instruction on executive controls.
Subjects in the Directive without Executive Controls group were explicitly
told how to accomplish each step of the required procedures. Those in the Directive
with Executive Controls group received the same style of instruction with the
addition of explicit teaching of specific strategies to control the execution of the
required procedures.
Participants in the Questioning without Executive Controls group were
questioned all the way through each step of the required procedures. The questions
focused on how to accomplish each step. In the Questioning with Executive Controls
62


group subjects were also questioned through strategies to control the execution of the
required procedures.
The Mixed without Executive Controls group received a mixture of both the
directive and the questioning approaches. The Mixed with Executive Controls group
received the same approach with the addition of instruction on how to control
process execution.
An ANCOVA was conducted to examine the data for main effects and any
interaction between the independent variables. Pretest results were used as a
covariate to account for aptitude effects. Results of the analysis showed that the
mixed group performed significantly better than the other three groups on the
posttest. The posttest was a measure of the amount of fragile knowledge a subject
possessed. A MANCOVA for types of fragile knowledge constructed showed no
significant differences between treatment groups. An ANCOVA for the effect of
executive controls on fragile knowledge construction showed no significant
differences between treatment groups.
Discussion
Are certain instructional interaction strategies more effective than others in
minimizing fragile knowledge construction in procedural skill acquisition? If they
are, do certain interaction strategies have a tendency to produce specific types of
63


fragile knowledge? Does explicit instruction on executive control strategies also
help minimize fragile knowledge construction?
Epistemological differences have fueled the debate about which approaches
to instructional interaction are the most effective. Each competing educational
philosophy would answer these questions differently. A proponent of direct
instruction would support telling learners what they need to know. A proponent of
discovery learning would want to put them in an environment that is conducive to
learning and let them learn. Proponents of the Socratic method might want to
challenge learners to find answers by using questions to allow them to probe their
understanding. Adler (1980) and Perkins (1992) suggest a mixed approach. They
would provide didactic instruction, coach them by observing and guiding, and follow
up with the Socratic method. With all these apparently competing approaches to
achieving the same objective, whos right? Which method is most likely to reduce
fragile knowledge construction during procedural skill acquisition?
The results of this study support the importance of using mixed instructional
interaction strategies to minimize the construction of fragile knowledge in learners.
Subjects in this study that were initially provided a directive approach to
instructional delivery supported by a shift to a questioning style during learner
performance performed significantly better than subjects who received no
instructional interaction or only the directive or questioning approach.
64


Furthermore, the group that received no instructional interaction performed
equally as well, if not better in some cases, than the directive group or the
questioning group. Another way of looking at this is that the directive or questioning
groups performed no better than the group that had no instructional interaction at all.
Based on post treatment interviews one possible explanation for this is that the
directive approach did not accommodate questions from the students. The students
may have had questions, but the instructors addressed them. The questioning
approach created ambiguity and insecurity because this interaction strategy created
more questions in the learner than it answered. Subjects frequently stated that they
understood the material but the questioning made them unsure of their
understanding.
Although certain groups produced more instances of fragile knowledge than
others, no treatment group produced a statistically significant amount of any one
specific type of fragile knowledge than any other group. This was expected.
Furthermore, subjects who also received explicit instruction on executive control
strategies performed better than other subjects in most instances; however, the
difference in performance was not statistically significant.
Limitations
The answer to the first question of whether there are certain interaction
strategies that minimize fragile knowledge construction is soundly supported by the
65


results of this study. The answers to the other two questions are less firmly supported
due to methodological and theoretical problems with the study. Consequently, these
results should be interpreted with caution.
The reason for caution, in regard to the question of whether certain
interaction strategies produce specific types of fragile knowledge, is that the analysis
done to examine this question was accomplished post-hoc. The primary purpose of
the instruments used in the research was to detect the presence of fragile knowledge
not to discriminate between the different types. Although the pre and posttests were
able to discriminate between the different types of fragile knowledge, they did not
present all five types equal opportunity in every instance throughout the entire
measurement. For example, some questions contained the correct answer and
instances of only one type of fragile knowledge as distracters. The effect of this
would be to provide more opportunity to find significant difference with this type of
fragile knowledge simply because it received more exposure than the other types.
The reason for caution with regard to the second main research question
(whether explicit instruction on executive control strategies has an effect on fragile
knowledge construction) is the theoretical interaction between executive control
strategies themselves and the resultant effect of questioning or mixed directive and
questioning interaction strategies. If subjects simply replicated the questions posed
by the instructor during the initial delivery, then they in effect unintentionally
66


constructed an executive control strategy. For example, if an instructor questioned a
subject on where to get a certain piece of data to solve a navigation equation
required in a procedure, and the subject remembered this interaction, then the subject
could use this to repair a procedural impasse that would have ordinarily resulted in
inert fragile knowledge. The end result of this would have been that even though
subjects were not given explicit instruction on executive control strategies, they
could have constructed different, but effective, executive control strategies by
themselves. Consequently, data would have shown no significant differences
between groups that did or did not receive explicit instruction on executive control
strategies. The results of this study support this interpretation.
Finally, although this study is theoretically generalizable to similar learning
environments that involve complex procedural skill acquistion, the reader must be
the one to make the generalization. This study examined high ability, highly
motivated subjects in a rigid learning environment; other environments or conditions
may vary significantly.
Recom m endations
Based upon the results of this study, the most effective way to minimize the
construction of fragile knowledge in learners during initial procedural skill
acquisition is to open the delivery with a directive or didactic approach; then, shift to
a questioning strategy as the learner performs the procedure. Other strategies may
67


work, but according to this study, they will have a tendency to produce more fragile
knowledge.
Initial delivery should clearly highlight the critical attributes of the procedure
and use an input-process-output approach to teaching each step in the procedure. The
instruction should explicitly cover where to get input data from, what to do with it,
and where put the result. This then becomes the input for the next step of the process
or sub-process. Once this has been accomplished, instructional interaction should
shift to a questioning style while the learner is performing the procedure. The
questions used during this phase should center on where one gets data for process
input, what to do with it, what to do with the results, and the learners evaluation of
the performance of the process.
Instruction involving procedural skill acquisition should avoid interaction
strategies with the learner that are entirely directive or declarative in nature or
interaction that is entirely question-based. A mix of the two strategies in the
sequence presented above will produce less fragile knowledge in the learner.
Future Studies
Future studies on fragile knowledge need to address the issues still
unresolved in this study and further examine why mixed strategies are better than the
others. Finding answers to these questions would clarify our understanding of fragile
knowledge and help to minimize its consequences.
68


The predisposition of certain instructional interaction strategies to produce
specific types of fragile knowledge and the effect of executive control strategies are
issues that remain unresolved in this study. Instrumentation designed specifically to
detect the types of fragile knowledge constructed would resolve the first issue. Also,
studies should be conducted on subjects of lower ability in an environment that is
less competitive. Competitive subjects were more willing and better able to produce
conglomerative fragile knowledge (combining disparate elements of knowledge to
be able to produce an answer) than inert fragile knowledge (simply not answering
the question). Conducting these studies would also help establish a range for the
magnitude of the effect of interaction strategies on fragile knowledge construction.
With regard to further examination of the effect of executive control strategies,
changing the nature of the questions used during the study would eliminate the
confounding nature of the interaction between the questions used to minimize the
construction of fragile knowledge and the executive control strategies themselves.
For study purposes, instructors should not use questions that learners can construct
control strategies from by mere recall. When attempting to measure the effect of
executive control strategies, avoiding questions like where do you get this
information? or what do you do with this data? should eliminate the confounding
effects in the study. However, it may be necessary to answer this question in a
completely separate study. This should provide a more definitive answer as to
69


whether the explicit instruction of executive control strategies influences the
construction of fragile knowledge.
Conclusion
Fragile knowledge can have significant consequences. Educational
professionals have an ethical responsibility to minimize or eliminate the construction
of fragile knowledge in learners. This study has shown that the instructional
interaction strategies we use with learners has an effect on the amount of fragile
knowledge they construct. Therefore, it is incumbent upon us as professional
educators to prescribe, and properly implement, interaction strategies that have been
shown to be effective against fragile knowledge construction.
Using a mixed strategy of declarative delivery initially and question-based
support during learner performance has been shown to be an effective strategy.
Avoidance of entirely declarative or question-based approaches is also effective in
reducing the construction of fragile knowledge. While questions still remain as to the
nature of fragile knowledge and the effect of executive control strategies, the
question of whether or not how instructors or instructional systems interact with the
learner effects their construction of fragile knowledge has been answered. The type
or style of interaction with a learner effects the stability of the knowledge they
construct. The tragic consequences that may be a result of fragile knowledge can be
avoided by the use of effective instructional interaction strategies. Can we afford to
70


continue to use strategies that are economical or efficient but less effective in
minimizing fragile knowledge construction and its consequences?
71


bje<
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
62
43
57
43
62
71
48
43
43
52
38
67
62
62
52
76
48
76
62
62
43
57
67
57
43
71
71
43
APPENDIX A
DATA SUMMARY FOR ANCOVA
TreatGP InstStrat ExecCtrls Pretest
1 None
1 None
1 None
1 None
1 None
1 None
1 None
1 None
1 None
1 None
1 None
1 None
1 None
1 None
1 None
1 None
2 None
2 None
2 None
2 None
2 None
2 None
2 None
2 None
2 None
2 None
2 None
3 Directive
3 Directive
3 Directive
3 Directive
None 10
None 33
None 33
None 29
None 29
None 48
None 33
None 38
None 43
None 33
None 14
None 62
None 52
None 24
None 29
None 38
Yes 43
Yes 57
Yes 57
Yes 52
Yes 33
Yes 43
Yes 38
Yes 48
Yes 38
Yes 62
Yes 38
None 29
None 62
None 52
None 48
72


62
67
57
29
62
52
62
57
43
67
38
48
43
38
62
62
57
57
57
62
67
67
57
48
57
71
52
43
62
24
52
62
76
67
52
48
57
62
57
67
3 Directive
3 Directive
3 Directive
3 Directive
3 Directive
3 Directive
3 Directive
3 Directive
3 Directive
4 Directive
4 Directive
4 Directive
4 Directive
4 Directive
4 Directive
4 Directive
4 Directive
4 Directive
4 Directive
4 Directive
4 Directive
4 Directive
5 Questioning
5 Questioning
5 Questioning
5 Questioning
5 Questioning
5 Questioning
5 Questioning
5 Questioning
5 Questioning
5 Questioning
5 Questioning
5 Questioning
5 Questioning
5 Questioning
5 Questioning
5 Questioning
5 Questioning
6 Questioning
None 52
None 57
None 48
None 29
None 52
None 38
None 52
None 43
None 38
Yes 43
Yes 29
Yes 48
Yes 43
Yes 33
Yes 48
Yes 24
Yes 38
Yes 38
Yes 14
Yes 48
Yes 48
Yes 48
None 43
None 43
None 48
None 52
None 52
None 38
None 52
None 14
None 48
None 57
None 38
None 57
None 48
None 43
None 43
None 62
None 29
Yes 62
73


48
52
62
71
57
71
38
62
48
76
67
48
48
67
43
76
38
29
48
57
62
52
62
67
57
57
67
62
62
57
71
43
62
76
67
57
57
81
81
76
6 Questioning Yes 38
6 Questioning Yes 38
6 Questioning Yes 52
6 Questioning Yes 52
6 Questioning Yes 31
6 Questioning Yes 57
6 Questioning Yes 38
6 Questioning Yes 52
6 Questioning Yes 43
6 Questioning Yes 57
6 Questioning Yes 57
6 Questioning Yes 33
6 Questioning Yes 38
6 Questioning Yes 52
6 Questioning Yes 38
6 Questioning Yes 62
6 Questioning Yes 38
6 Questioning Yes 29
6 Questioning Yes 43
6 Questioning Yes 57
6 Questioning Yes 33
6 Questioning Yes 38
6 Questioning Yes 57
6 Questioning Yes 43
6 Questioning Yes 33
6 Questioning Yes 33
6 Questioning Yes 38
6 Questioning Yes 57
6 Questioning Yes 38
6 Questioning Yes 52
7 Mixed None 57
7 Mixed None 38
7 Mixed None 43
7 Mixed None 52
7 Mixed None 5
7 Mixed None 43
7 Mixed None 19
7 Mixed None 38
7 Mixed None 62
7 Mixed None 43
74


112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
81
76
67
62
52
52
48
76
43
57
76
71
90
67
62
67
57
48
62
67
76
57
76
71
52
62
57
38
76
52
71
52
67
57
71
57
7 Mixed
7 Mixed
7 Mixed
7 Mixed
7 Mixed
7 Mixed
7 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
8 Mixed
None 62
None 67
None 62
None 33
None 38
None 43
None 38
Yes 57
Yes 43
Yes 38
Yes 52
Yes 38
Yes 67
Yes 62
Yes 43
Yes 38
Yes 43
Yes 48
Yes 52
Yes 33
Yes 62
Yes 38
Yes 43
Yes 43
Yes 33
Yes 38
Yes 24
Yes 38
Yes 76
Yes 52
Yes 33
Yes 48
Yes 48
Yes 48
Yes 38
Yes 57
75


APPENDIX B
DATA SUMMARY FOR MANCOVA
Subject# Partial Conglom Misplaced Naive Inert
1 2 2 1 2 1
2 5 2 4 1 0
3 4 2 1 2 0
4 5 2 0 1 0
5 4 5 2 1 0
6 3 2 1 0 0
7 3 2 3 3 0
8 4 4 2 2 0
9 4 6 1 0 1
10 4 2 1 2 1
11 3 5 3 2 0
12 2 4 1 0 0
13 1 5 1 1 0
14 6 1 0 1 0
15 4 2 2 1 1
16 1 3 0 1 0
17 3 4 2 1 1
18 2 3 0 0 0
19 3 2 1 2 0
20 5 0 2 1 0
21 6 5 1 1 1
22 4 1 3 1 0
23 3 2 1 1 0
24 4 2 1 2 0
25 4 4 3 1 0
26 5 0 0 1 0
27 3 2 1 1 0
28 2 5 2 2 1
29 6 1 0 1 0
30 3 2 1 1 1
76


0
0
1
0
5
0
2
0
0
0
0
0
0
1
4
0
0
0
2
0
0
0
1
0
0
1
1
4
1
0
11
0
0
0
0
0
0
4 3 1 1
4 2 1 1
3 4 0 0
2 4 2 1
2 4 1 3
3 2 2 1
1 4 2 1
4 1 1 2
4 3 1 1
2 6 3 1
1 2 1 3
6 3 1 3
2 3 1 3
3 4 2 2
4 2 2 2
2 1 3 2
5 2 0 1
3 2 2 2
2 1 2 2
5 3 0 1
4 1 1 2
4 0 3 0
3 2 1 0
4 3 1 1
6 0 3 2
2 2 2 2
3 1 1 0
3 2 1 0
2 6 2 1
3 2 1 2
5 0 0 0
6 1 2 1
4 2 1 1
1 1 1 2
3 1 2 1
2 6 1 1
4 3 2 2
77


68 4 0 3 2 0
69 2 4 1 1 0
70 5 1 2 1 0
71 3 2 2 0 0
72 6 1 1 2 1
73 4 1 3 2 0
74 4 2 0 2 0
75 3 0 1 2 0
76 4 2 1 2 0
77 3 1 1 1 0
78 6 1 3 3 0
79 5 0 1 2 0
80 3 3 2 3 0
81 2 1 1 1 0
82 1 2 2 2 0
83 6 1 2 1 1
84 4 3 1 3 0
85 2 3 1 0 1
86 5 3 2 0 2
87 4 0 0 1 0
88 2 3 2 1 5
89 9 2 1 3 0
90 5 4 2 2 0
91 3 4 2 0 0
92 3 3 2 0 0
93 3 2 2 2 1
94 3 3 1 1 0
95 4 2 1 0 0
96 5 2 2 0 0
97 3 2 2 0 2
98 2 4 1 0 0
99 5 2 0 1 0
100 6 0 1 1 0
101 4 2 3 0 0
102 2 2 2 0 0
103 5 2 4 2 0
104 3 3 0 2 0
78


105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
2 0 2 0 1
3 1 1 2 0
3 1 3 2 0
5 0 2 2 0
2 0 1 2 0
3 1 0 0 0
2 0 1 2 0
2 0 0 2 0
2 1 1 1 0
1 3 1 1 1
2 2 2 1 1
3 4 1 0 2
3 4 2 1 0
5 3 0 3 0
2 2 0 1 0
3 5 2 1 1
2 4 2 1 0
3 1 0 1 0
4 1 1 0 0
2 0 0 0 0
4 1 1 1 0
1 5 1 1 0
1 4 1 1 0
3 2 2 2 0
5 3 1 2 0
3 1 2 2 0
5 0 1 1 0
3 0 2 0 0
4 2 2 1 0
3 1 0 1 0
3 3 0 0 0
2 5 1 2 0
5 1 1 1 0
4 2 2 1 0
6 4 0 3 0
3 0 0 2 0
5 2 1 2 0
79


142 1 3 1 1 0
143 5 4 0 1 0
144 2 3 1 1 0
145 2 3 2 2 0
146 1 4 0 0 1
147 3 3 2 1 0
80


APPENDIX C
Pre-Study Questionnaire
Note
This data will remain strictly confidential. Under no circumstances will this be divulged to
anyone other than you without your knowledge and prior written consent.
'lame:
^ge:
jender: M F
Nationality: U.S. Other___________________________
SAT Composite Score:
figh School GPA:
s English your native language? Yes No
*revious Pilot Flight Experience (ex: Private Pilot, 40 hours)
devious Navigation Experience (ex: Navy Sea Cadets, 2 yrs; NJROTC, Orienteering, etc.):
ire you Pilot or Nav Qualified? Yes No Dont Know
f you couldnt fly for the Air Force would you leave the Academy? Yes No
Vere your favorite courses in High School Math and Science type courses? Yes
)o you intend to pursue a technical/engineering major at the Academy? Yes
)id you attend a Preparatory School? Yes No
ire you Prior Service? Yes No
If yes, were you an Air Crewmember? Yes No
or an Instructor/Trainer of Air Crewmembers? Yes
rocedural Knowledge Memory Test Score_________
No
No
No
81


APPENDIX D
Name:______________________________________________
Aeronautical Knowledge Test #1
Choose the best answer. More than one answer per question may be correct.
1. Given the information shown in Figures 1 & 2., what heading should the pilot fly at the
Departure Point?
A. 254 degrees
B. 256 degrees
C. 266 degrees
D. 268 degrees
2. Given the indications shown in Figure 2., which of the following most accurately depicts the
aircrafts position? (Show your work on Figure 1.)
3. Based on the results of the fix resolved and plotted on Figure 1. and the indications in Figure
2., which magnetic heading should be flown to correct to course?
A. 265 degrees
B. 263 degrees
C. 254 degrees
D. 252 degrees
4. Using the information provided in Figure 3., which of the following is the most accurate ETA
for Point B?
A. 1216Z
B. 1217Z
C. 1234Z
D. 1235Z
82


5. Using the indications shown in Figure 3., which of the following depictions most accurately
represents where the aircraft will be at 1230Z?
6. Given the indications shown in Figure 4., which of the following most accurately depicts the
aircrafts position at 1230Z?
7. The distance between each tick mark on a line of longitude on a JNC chart
A. equals 2 NM
B. equals 5 NM
C. equals 60 NM
D. cannot be determined because they are not equidistant due to convergence
8. Variation is the difference between
A. True North and True Heading
B. True North and Magnetic North
C. True Course and Magnetic Course
D. subsequent readings of the same instrument
83


9. The difference between Desired Course and Actual Course is called
A. Variation
B. Deviation
C. Drift
D. Slant Error
10. What effect will a 25 knot headwind instead of a 50 knot headwind have on your Proposed
ETA to your final destination?
A. It will take only half as much time to get there
B. It will take twice as much time to get there
C. Arrive later than originally planned
D. Arrive earlier than originally planned
11. To maintain a planned/desired ground track on a JNC chart, a pilot must correct for
A. crosstrack error and precession
B. drift and precession
C. variation and precession
D. variation and drift
12. If a pilot must fly a heading of260 degrees to maintain a track of270 degrees and now is 10
minutes earlier to the arrival gate than originally planned which general direction is the wind
coming from?
A. Northwest
B. Northeast
C. Southwest
D. Southeast
13. AFixisa(n)
A. estimate of the aircrafts current position
B. measurement of the aircrafts current position
C. estimate of the aircrafts future position
D. measurement of the aircrafts future position
14. A Dead Reckoning (DR) position is a(n)
A. estimate of the aircrafts current position
B. measurement of the aircrafts current position
C. estimate of the aircrafts future position
D. measurement of the aircrafts future position
15. When taking a TACAN fix, where do we get the variation from?
A. the Navigation Computer System (NCS)
B. the line of variation halfway between the 2 points on the chart
C. the line of variation closest to the TACAN station on the chart
D. the IFR Supplement
84


16. What is the distance between N 21-18 W 105-15 and N 22-18 W 105-15?
A. 60 NM
B. 6 NM
C. 1 NM
D. .6 NM
17. How far will the aircraft have traveled if its been maintaining 400 GS for 9 minutes
A. 59.9 NM
B. approximately 63 NM
C. a distance equal to about one degree of latitude
D. about 66 NM
18. If your Desired Course is 180 degrees and you experience 1 degree more right drift than you
computed in your DR, where would you expect your fix to fall in relation to your DR assuming
240 NM between your last fix and current DR?
A. 1 NM due east of the DR
B. . 4 NM due east of the DR
C. 1 NM due west of the DR
D. 4 NM due west of the DR
19. The angular difference measured in degrees between two points on a JNC chart is called
A. True Course
B. True Heading
C. Magnetic Course
D. Magnetic Heading
20. For fixing purposes, the RADAR is oriented to
A. True Course
B. True Heading
C. True North
D. Magnetic Heading
21. Which of the following navigational facilities does not provide enough information to
establish a fix?
A. VOR
B. VORTAC
C. VOR/DME
D. TAC
85


Figure 1. Chare Extract
86


Figure 2. TACAN Fix Figure 3. ETA and DR Information
87


Figure 4. RADAR Fix
88