Upload
stephen-m
View
215
Download
0
Embed Size (px)
Citation preview
Enhancing learning outcomes in computer-based trainingvia self-generated elaboration
Haydee M. Cuevas • Stephen M. Fiore
Received: 27 December 2012 / Accepted: 14 March 2014� Springer Science+Business Media Dordrecht 2014
Abstract The present study investigated the utility of an instructional strategy known as
the query method for enhancing learning outcomes in computer-based training. The query
method involves an embedded guided, sentence generation task requiring elaboration of
key concepts in the training material that encourages learners to ‘stop and think’ about the
information already presented before proceeding to new concepts. This study also inves-
tigated the effect of varying the level of elaboration (low or high) prompted by the queries.
Fifty-one undergraduate students from the general psychology department subject pool at a
major university in the southeastern United States received instruction on the basic prin-
ciples of flight via one of three versions of a computer-based tutorial (no query, low-level
elaboration query, or high-level elaboration query). Participants had no prior knowledge or
previous experience with the aviation domain. A one-way between-groups design was
employed, with the query method serving as the independent variable and a sample size of
17 per condition. Dependent variables included knowledge organization, knowledge
acquisition, and instructional efficiency. Overall, results showed that incorporating low-
level elaboration queries into the training resulted in improved organization, integration,
and application of task-relevant knowledge and higher instructional efficiency. High-level
elaboration queries consistently failed to produce significantly better post-training out-
comes, possibly due to the increased cognitive load imposed on learners during training.
The discussion centers on theoretical and practical implications for promoting and
assessing learning outcomes in computer-based training.
H. M. Cuevas (&)Department of Doctoral Studies, College of Aviation – Room 137E, Embry-Riddle AeronauticalUniversity, 600 South Clyde Morris Blvd, Daytona Beach, FL 32114, USAe-mail: [email protected]
S. M. FioreCognitive Sciences – Department of Philosophy, College of Arts and Humanities, University of CentralFlorida, 3100 Technology Parkway, Suite 140, Orlando, FL 32826, USAe-mail: [email protected]
123
Instr SciDOI 10.1007/s11251-014-9315-8
Keywords Cognitive load � Instructional efficiency � Knowledge acquisition �Knowledge organization � Self-generated elaboration
Introduction
Differences in learning outcomes in computer-based training environments can be clas-
sified as being due to factors internal and external to the learner (Fiore and Salas 2007).
Internal factors increasingly important due to the nature of training are those associated
with characteristics of the learner (e.g., individual aptitudes, knowledge structure devel-
opment). External factors correspond to the instructional techniques incorporated into such
environments (e.g., diagrams, animations). In this paper, we describe a study devised to
assess how the interaction of such factors influences learning in these environments. First,
we briefly describe the learning outcomes of interest to this study. We then describe how
embedded prompting within computer-based training may be a viable instructional strategy
to produce greater learning.
Knowledge acquisition and knowledge organization
Important learning outcomes include not only learners’ mastery of basic factual infor-
mation, but also the organization, integration, and application of task-relevant knowledge
to novel situations (Chipman et al. 2013). Thus, training should be targeted at facilitating
learners’ ability to effectively integrate differing knowledge components and apply this
newly acquired knowledge in a variety of dynamic task-relevant scenarios (Cuevas et al.
2002; Fiore et al. 2002, 2003). Such post-training outcomes would demonstrate if learners
have acquired a more flexible, higher level of understanding of the material. In addition,
training should also promote learners’ organization of their newly acquired knowledge,
that is, the degree to which elements of knowledge are interconnected and integrated
within meaningful networks in long-term memory, which has been shown to be critical to
the development of task expertise. In particular, acquiring task expertise requires one to
understand, not just the meaning and function of concepts associated with their domain, but
also how these fit together (Salas and Rosen 2010).
Cognitive load and instructional efficiency
Sweller’s (1994) Cognitive Load Theory proposes that training materials and activities
required of learners during training should be structured to minimize any avoidable load on
learners’ cognitive resources (e.g., working memory capacity) and maximize knowledge
structure development (knowledge organization). To achieve this objective, training design
needs to consider the cognitive load generated from the complex interaction between the
demands of the learning material, learning process, and presentation mode (Brunken et al.
2004, p. 131; for a discussion, see Sweller et al. 2011). Specifically, cognitive load may be
influenced by both the intrinsic cognitive load associated with the content of the training
material as well as extraneous cognitive load resulting from how the training material is
presented (Sweller et al. 1998, 2011). Intrinsic cognitive load arises from the inherent
complexity and level of integration of the concepts to be learned. Intrinsic cognitive load
(and the corresponding strain on learners’ limited cognitive resources) is highest when
H. M. Cuevas, S. M. Fiore
123
many information-rich elements need to integrated for knowledge structure development to
occur (Sweller 2010). Extraneous cognitive load is artificially produced due to inadequate
presentation modes that impose split-attention and redundancy of information (Kalyuga
et al. 1999, 2004; Mayer et al. 2001). Training programs that simultaneously impose high
levels of both intrinsic and extraneous cognitive load would be expected to lead to sub-
optimal post-training outcomes (Sweller et al. 2011).
Research has also investigated germane cognitive load, which stems from cognitive
effort purposely generated by instructional strategies designed to maximize elaboration of
the targeted material (Sweller 2010; van Merrienboer et al. 2002). In this case, training
design should promote germane cognitive load since deeper processing of the targeted
material may encourage learners to organize this knowledge into a coherent cognitive
representation, integrated with other representations and prior knowledge within mean-
ingful networks in long-term memory (Mayer et al. 2005). As such, germane cognitive load
is required for knowledge organization and knowledge integration of the to-be-learned
material in long-term memory (Kirschner 2002). Accordingly, one of the issues this study
explored was how instructional strategies might influence germane cognitive load. Theo-
retically, germane cognitive load produces optimal levels of processing of training content,
which, in turn, facilitates comprehension and retention of the to-be-learned material
(Kirschner 2002; Sweller et al. 2011). Training designers should, therefore, attempt to
identify what are the optimal levels of germane cognitive load in a given learning situation.
In order to understand the influence of cognitive load on the learner, research has
assessed the instructional efficiency of the learning experience. This is described as the
relative efficiency of the instructional program in terms of the demands imposed by the
training on learners’ cognitive resources. This measure of training effectiveness involves
examining the observed relation between subjective cognitive effort and task performance
in a particular learning condition (see Paas and van Merrienboer 1993). Cognitive (or
mental) effort, as indicated by subjective ratings, is the amount of resources allocated by
the learner to meet the cognitive load demands imposed by the task (Paas et al. 1994).
Training programs that reduce learners’ cognitive load while facilitating knowledge
acquisition have been shown to yield higher instructional efficiency (for a review, see
Clark et al. 2006).
Self-generated elaboration—query method
Explicitly prompting self-generated elaboration of concepts (via self-explanation) has been
shown to facilitate knowledge acquisition and result in a greater understanding of the
domain (e.g., Catrambone and Yuasa 2006; de Bruin et al. 2007; Hermanson et al. 1997;
O’Reilly et al. 1998; Wong et al. 2002). Our past research has also highlighted the ben-
eficial effects of self-generated elaboration on learning outcomes (e.g., Cuevas et al.
2004b; Fiorella et al. 2012). In particular, teaching students the cognitive strategy of
generating questions for new material has been shown to lead to significant gains in
comprehension, as measured by standardized and experimenter-generated tests following
the interventions (Rosenshine et al. 1996). For example, in a series of studies that
examined several learning and study strategies, King (1992) demonstrated how a guided
learner-generated questioning strategy (i.e., open-ended question stems; e.g., ‘How are
____ and ____ related?’), designed to prompt high-level elaboration of new material,
facilitated knowledge acquisition and led to superior performance on objective and essay
tests on the material, compared to learners not presented with such strategies.
Enhancing learning outcomes in computer-based training
123
The present study explored the effectiveness of the query method, a guided sentence-
generation task that prompts learners to elaborate on key concepts presented in the training
material. The query method can be described as an embedded content-dependent strategy
(Osman and Hannafin 1992). By embedding queries within the lessons, learners are
prompted to attend to and interact with the critical concepts in the presented material,
increasing processing of the information and facilitating knowledge acquisition. By
making the queries content-dependent, emphasis can be placed on the lesson’s concepts
and their unique interrelations, promoting learning of the target domain. Thus, the query
method may provide learners with active control of instruction (i.e., their knowledge
acquisition process) by enabling elaboration of the material (Fonseca and Chi 2011).
Relation to past research
Our past research efforts have explored the factors underlying knowledge acquisition and
integration within two distinct contexts: 1) the effect of diagrammatic presentations on the
acquisition of domain knowledge in aviation (Cuevas et al. 2002; Fiore et al. 2003), and 2)
knowledge integration and cognitively diagnostic assessment within distributed team
training environments (Cuevas et al. 2004b; Fiore et al. 2002; Scielzo et al. 2004, 2005).
Over the past successive iterations of our investigation, we have examined not only
knowledge acquisition and integration, but also the cognitive precursors to the develop-
ment of this knowledge, including constructs such as instructional efficiency (e.g., Paas and
van Merrienboer 1993) as well as the role of individual differences in learner aptitudes
(e.g., Gully and Chen 2010), to better converge on an understanding of technology-med-
iated learning. The overall goal of the present study was to further increase our under-
standing of the cognitive processes involved in learning within computer-based training
environments and to investigate how embedded content-dependent strategies can support
this learning process.
Hypotheses
The elaboration prompted by the query method was hypothesized to assist learners in
building internal associations between multiple concepts in the to-be-learned material and
as such, potentially foster a deeper, more integrative understanding of the information (as
indicated by greater similarity to an expert model), which, in turn, would facilitate suc-
cessful performance on questions requiring integration and application of knowledge
(Fonseca and Chi 2011). This effect would not be expected to be as strong for basic factual
knowledge assessment (cf. Cuevas et al. 2002, 2004b; Fiore et al. 2003). Furthermore, by
better enabling learners to build internal associations among the concepts presented, the
query method may reduce the demands imposed by the training on learners’ cognitive
resources (i.e., cognitive load), leading to higher performance on a knowledge assessment
task with less cognitive effort exerted. Thus, this study also evaluated the beneficial effect
of the query method on the training program’s instructional efficiency. Finally, by varying
the level of elaboration prompted by the queries, this study could determine what level of
processing of the to-be-learned material was necessary for this instructional strategy to be
most effective (cf. King 1992). Varying the level of elaboration would help us to under-
stand how embedded content-dependent strategies can be effectively used to promote
optimal levels of germane cognitive load, leading to productive learning outcomes. The
specific hypotheses investigated in this study were as follows:
H. M. Cuevas, S. M. Fiore
123
Hypothesis 1 Participants presented with the query method, particularly those in the
high-level elaboration training condition, were hypothesized to exhibit significantly greater
accuracy in their knowledge organization of the presented concepts (as measured via a card
sort task) than participants not presented with the query method.
Hypothesis 2 Participants presented with the query method, when compared to partici-
pants not presented with this instructional strategy, were hypothesized to exhibit signifi-
cantly greater performance on integrative knowledge questions.
Hypothesis 3 No significant difference in performance was hypothesized for declarative
knowledge or perceptual knowledge (concept recognition) questions.
Hypothesis 4 Incorporating the query method into its instructional design was hypoth-
esized to significantly improve the training program’s instructional efficiency in relation to
performance on the knowledge assessment task (declarative, perceptual, and integrative
questions).
Method
Participants
Fifty-four undergraduate students (14 males and 40 females, mean age = 21.39) from the
general psychology department subject pool at a major university in the southeastern
United States participated in this experiment for course credit. Participation in the
experiment was open to all students, regardless of age, race, gender, or nation of origin. A
demographic form was used to screen participants to ensure that data analyzed were only
from participants with no prior knowledge or previous experience in the aviation domain;
no participants indicated prior aviation knowledge or experience. Data from three partic-
ipants were excluded from the analysis due to procedural problems, resulting in a final N of
51 (13 males and 38 females, mean age = 21.39 years). Treatment of all participants was
in accordance with the ethical standards of the American Psychological Association.
Design
This study employed a one-way between-groups design, with the query method serving as
the independent variable, manipulated at three levels: no query (NQ), low-level elaboration
query (LLEQ), and high-level elaboration query (HLEQ). Participants were randomly
assigned to conditions to ensure that each participant had an equal chance of being
assigned to any one of the three experimental groups. Dependent variables included
knowledge organization, knowledge acquisition, and instructional efficiency.
Materials and apparatus
Aviation training tutorial
Three versions (NQ, LLEQ, HLEQ) of an interactive computer-based instructional tutorial
on the principles of flight were developed for this study (see example lesson in Fig. 1).
Material for the tutorial was adapted from the Jeppesen Sanderson Private Pilot Manual
(Jeppesen Sanderson Training Systems 1996c) and the Jeppesen Sanderson Private Pilot
Enhancing learning outcomes in computer-based training
123
Maneuvers Manual (Jeppesen Sanderson Training Systems 1996b), both standard training
products for the instruction of pilots in the private and public sector. The tutorial consisted
of the following three modules.
Module 1 (Airplane Parts) described a number of airplane parts critical for standard
flight operations. Participants were presented with an overview slide and two main slides
(wings, tail), with hyperlinks to four additional slides that provided more detailed expla-
nation of the concepts (e.g., ailerons, elevator).
Module 2 (Flight Movements) discussed the aerodynamics of flight, including infor-
mation about the axes around which an airplane moves and the movements possible in
standard airplane flight. Participants were presented with an overview slide and two main
slides (axes, movements), with hyperlinks to six additional slides that defined the various
axes and movements (e.g., vertical axis, yaw movement).
Module 3 (Flight Instruments) introduced the six primary flight instruments traditionally
used by pilots to navigate the airplane. Participants were presented with an overview slide
and two main slides (pitot-static instruments, gyroscopic instruments), with hyperlinks to
12 additional slides that described how to read the instruments and explained how changes
in the airplane’s movements affected the information displayed on the instruments (e.g.,
airspeed indicator, turn coordinator). Participants were also presented with a hyperlink to
an animated demonstration depicting each of the six flight instruments in motion.
Participants proceeded through the tutorial at their own pace, navigating the hyperlinks
embedded in the tutorial using a standard point-and-click mouse. Participants were free to
move backward and forward through the tutorial. After all the lessons in the respective
module had been viewed, participants in the query method conditions were presented with
the guided sentence-generation task, prompting them to engage in either low or high level
elaboration of the concepts (further described next). Participants in the no-query condition
were presented with the relevant information in the lessons only. At the end of each
module, all participants were given the opportunity to go back and review the lessons
before proceeding to the next module. Though no time limit was imposed, participants
took, on average, approximately 24 min to complete the tutorial.
Fig. 1 Illustrative content of Aviation Training Tutorial (Module 1—Airplane Parts)
H. M. Cuevas, S. M. Fiore
123
Query method
Upon completion of the lessons in each module, participants in the query conditions were
presented with a ‘Stop and Think’ exercise in an open-ended question format that asked
them to generate one sentence from a list of key concepts presented in the training.
Participants in the LLEQ condition were prompted to generate a sentence using only one of
the terms from this list (simple sentence). Participants in the HLEQ condition were
prompted to generate a sentence that connected three or more concepts from the list that
best described the relation among those concepts (complex sentence). For example, the list
for Module 1 included the terms: Wings, Tail, Ailerons, Flaps, Vertical Stabilizer, Hori-
zontal Stabilizer, Rudder, Elevator. Choosing among these concepts, a participant in the
LLEQ condition could generate the following simple sentence: ‘‘Pilots use the rudder to
move the airplane’s nose left and right.’’ Conversely, a participant in the HLEQ condition
could generate a complex sentence linking three or more of these concepts in the following
manner: ‘‘During flight, pilots initiate a turn by using the ailerons on the wings in com-
bination with the rudder, which is attached to the back of the vertical stabilizer.’’ Once
submitted, participants could not change their responses to the sentence generation task.
Tutorial survey (instructional efficiency)
Following the tutorial, all participants were presented with a brief Tutorial Survey, which
included an item designed to assess their perceived cognitive load associated with learning
the training material. Specifically, participants were asked to report how easy or difficult
they found it to understand the concepts presented in the training, with responses recorded
on a 7-point anchored scale, ranging from 1 (very easy) to 7 (very difficult). These
responses were used to calculate the training’s instructional efficiency (described later in
the analysis of the results). The use of a single-item measure to assess learner’s perceived
cognitive load has been successfully demonstrated in our previous work (e.g., Cuevas et al.
2002; Scielzo et al. 2005) as well as in other studies (e.g., Kirwan et al. 1997).
Card sort task (knowledge organization)
To evaluate participants’ knowledge organization of task-relevant concepts, the present study
employed the TPL-KATS-card sort software (Harper et al. 2003), a computer-based knowledge
structure assessment product developed by the Team Performance Laboratory at the University
of Central Florida (Copyright 2001). Participants were instructed to group 26 key concepts
extracted from the Aviation Training Tutorial into as many categories as they desired and then
name or describe the categories they created. Though no time limit was imposed, participants
took, on average, approximately 20 min to complete this task, including receiving instructions
on the software program and performing the card sort task itself.
Knowledge assessment task (knowledge acquisition)
Participants completed a self-paced computer-based knowledge assessment task (overall
a = .859) consisting of 48 multiple-choice questions, divided into three sections (see
example question in Fig. 2). Only one question was presented at a time on the screen and
participants proceeded from one question to the next using a standard point-and-click
mouse. Participants were not able to go back and review or change their responses once
they had proceeded to the next question and no feedback was provided as to the accuracy
Enhancing learning outcomes in computer-based training
123
of their responses. Participants, on average, completed this task in about 20 min. The three
sets of questions were presented in the following order:
Factual knowledge assessment (declarative knowledge questions) Twenty questions
(a = .741), adopted from a standard introductory flight manual (Jeppesen Sanderson
Private Pilot Exercise Book; Jeppesen Sanderson Training Systems 1996a), assessed
participants’ mastery of basic factual information associated with the training (e.g., defi-
nitions of the various parts of the plane). Participants were presented with text-based
definitions taken from the Aviation Training Tutorial and were asked to identify the
concept being described.
Concept recognition assessment (perceptual knowledge questions) Eighteen questions
(a = .675) tested participants’ perceptual knowledge with regard to their ability to rec-
ognize key concepts from the Aviation Training Tutorial. Participants were presented with
static illustrations of the principle concepts (airplane parts, axes, movements, and instru-
ments) from the tutorial and were asked to identify the concept depicted.
Airplane function assessment (integrative knowledge questions) Ten questions
(a = .634) assessed participants’ ability to integrate and apply task knowledge. Partici-
pants were presented with a dynamic animated scenario (using audio–video interleaved file
format) illustrating an airplane performing a maneuver and were asked to determine, for
example, which airplane parts and flight instruments were being utilized in this maneuver.
Verbal comprehension ability (covariate)
Since the nature of the material presented in the Aviation Training Tutorial required the
understanding and integration of complex concepts and relations, Part 1 (Verbal
Fig. 2 Illustrative content of knowledge assessment task (integrative knowledge question)
H. M. Cuevas, S. M. Fiore
123
Comprehension) of the Guilford-Zimmerman Aptitude Survey (Guilford and Zimmerman
1981) was administered to covary out any effects due to individual differences in this
learner aptitude. Part 1 has a computed odd–even estimate of reliability of .96 and factorial
validity is demonstrated from the results of three factor analyses, with factor loadings
ranging from .70 to .86 (Guilford and Zimmerman 1981). For this paper-and-pencil task,
participants were given 10 min to respond to 72 multiple-choice questions assessing
knowledge of semantic meanings. For each item, participants were presented with a word
(e.g., earth) and were asked to select from among five other words (e.g., sugar, farm, sun,
soil, horse) which one had a meaning like the first word (e.g., soil).
Apparatus
The software program for the Aviation Training Tutorial, knowledge assessment task, and
card sort program were hosted on an IBM compatible Pentium 586 computer with a 15-inch
color monitor, run on Windows XP� operating system. The Aviation Training Tutorial and
knowledge assessment task were presented utilizing Microsoft PowerPoint XP�. Participants
navigated through the tutorial and test using a standard point-and-click mouse. Both keyboard
inputs and the use of a standard point-and-click mouse were required to perform the card sort
task. Multimedia presentation using audio–video-interleaved files were incorporated into the
Airplane Function Assessment (integrative knowledge questions). A paper-and-pencil for-
mat was used to record participants’ responses to the queries embedded in the training, the
Tutorial Survey, and the knowledge assessment task.
Procedure
Upon arrival, participants were randomly assigned to one of the three experimental groups.
Participants first completed an informed consent form, a biographical data form (age,
gender, prior knowledge of aviation), and the measure of verbal comprehension ability.
Participants then proceeded with self-paced computer-based instruction on the principles of
flight using the Aviation Training Tutorial. For the query conditions, the sentence task was
embedded immediately following the last lesson presented in each module. Following the
tutorial, participants were asked to complete the Tutorial Survey. Participants were then
presented with the card sort task, followed by the knowledge assessment task. Finally,
participants were debriefed and extra credit was assigned. On average, the total length of
the experiment, including training and assessment, was approximately 95 min.
Results
Analyses
The experimental results were analyzed separately in terms of post-training learning
outcomes (knowledge organization, knowledge acquisition) and instructional efficiency.
Multivariate analysis statistics are reported using Roy’s largest root. Pairwise comparison
statistics are reported using Fisher’s Least Significant Difference (LSD). Correlations were
calculated, as appropriate. An alpha level of .05 was used for all statistical analyses. Verbal
comprehension ability was significantly correlated with several of the measures and thus
was treated as a covariate in the analysis of these variables. Table 1 lists the
Enhancing learning outcomes in computer-based training
123
intercorrelations, means, and standard deviations for the dependent measures and verbal
comprehension ability.
Check of random assignment
To check the effectiveness of the random assignment procedure used in this study, a
univariate one-way ANOVA was performed on participants’ scores on the verbal com-
prehension ability measure, as this variable should not have been influenced by the
manipulation. Analysis revealed no significant differences among the three experimental
groups on this variable, F (2, 48) \ 1.
Effect of query method on knowledge organization and knowledge acquisition
The cognitive measures were analyzed using a one-way between-groups MANCOVA, with
query method serving as the independent variable and verbal comprehension ability treated
as a covariate. The dependent measures included knowledge organization, as measured via
the card sort task, and knowledge acquisition, as measured using the knowledge assessment
task. Multivariate analysis revealed a significant effect for the query method on these
cognitive measures, F (4, 45) = 3.202, p = .021, gp2 = .222. Verbal comprehension ability
was a significant covariate, F (4, 44) = 2.733, p = .041, gp2 = .199. Univariate analysis
for each of the measures will be presented next. Adjusted means and standard errors are
reported in Table 2.
Knowledge organization (card sort task) (Hypothesis 1)
The card sort task was used to assess the degree to which presentation of the query method
affected similarity to an expert model. A quantitative measure was derived from the card
sort data to determine the connectedness among concepts. First, a list of all possible
pairings of the 26 concepts was generated (N = 325). A value of 1 was assigned to pairings
of concepts falling within the same group (i.e., if the participant grouped the pair of
concepts together in the same category) and a value of 0 was assigned for the remaining
concept pairs (i.e., for pairings where the participants did not group the two concepts
together in the same category). Each participant’s card sort data (i.e., the generated list of
Table 1 Intercorrelations, means, and standard deviations of verbal comprehension ability and cognitivemeasures a
Dependent variable 1 2 3 4 5 6 M SD
(1) Verbal comprehension – .05 .38** .32* .41** .23 .359 .123
(2) Knowledgeorganization
– .30* .27 .39** .06 .519 .219
Knowledge acquisition
(3) Total – .89** .90** .77** .571 .173
(4) Declarative – .68** .52** .582 .195
(5) Perceptual – .61** .611 .185
(6) Integrative – .478 .235
a N = 51
* p \ .05 (two-tailed). ** p \ .01 (two-tailed)
H. M. Cuevas, S. M. Fiore
123
the participant’s pairings of all the concepts) was compared to the card sort data generated
by the subject matter expert. This expert had approximately 7,000 h as a pilot and
approximately 2,700 h as an instructor and participated in the creation and evaluation of
the tutorial. By calculating the Pearson r correlation coefficient between the participant’s
and subject matter expert’s card sort data, a participant’s sensitivity to identifying the
critical relations among the concepts can be evaluated. Hence, the similarity of their
pairings to the expert model would indicate the accuracy of the participant’s knowledge
organization of the task.
Univariate tests revealed a significant effect of the query method on participants’
knowledge organization, as indicated by the mean correlation between the participants’ and
the expert model card sort pairings (see Table 2). However, pairwise comparisons showed
an unexpected pattern of results. Specifically, the LLEQ participants exhibited a signifi-
cantly greater mean correlation with the expert model than the HLEQ participants,
p = .016. Although the LLEQ participants’ card sort mean correlation to the expert model
was also greater than the NQ participants, this difference was not significant, p = .098. No
significant difference was found between the NQ and the HLEQ participants, p = .413.
Knowledge acquisition (Hypotheses 2 and 3)
Univariate tests revealed a significant effect of the query method on participants’ overall
performance on the knowledge assessment task (see Table 2). Pairwise comparisons
showed that the LLEQ participants significantly outperformed the NQ participants on the
knowledge assessment task overall, p = .006. Although the LLEQ participants’ perfor-
mance was also greater than the HLEQ participants, this difference was not significant,
Table 2 Adjusted means and standard errors for knowledge organization and knowledge acquisitionmeasures by training condition
Dependent variable Training conditiona ANOVA results
NQ LLEQ HLEQ F (2, 47) p, gp2obs. power
Knowledge organization .499 (.051)[.395, .602]
.621 (.051)a
[.518, .725].438 (.052)b
[.334, .542]3.269 .047, .122, .594
Knowledge acquisition
Total .500 (.037)c
[.426, .574].650 (.037)d
[.576, .725].564 (.037)[.489, .638]
4.177 .021, .151, .708
Declarative .526 (.045)[.436, .616]
.646 (.045)[.556, .736]
.575 (0.45)[.485, .666]
1.840 .170, .073, .365
Perceptual .551 (.040)a
[.471, .631].691 (.040)b
[.612, .771].591 (.040)[.511, .671]
3.343 .044, .125, .604
Integrative .358 (.052)c
[.253, .464].585 (.052)d
[.480, .690]492 (.052)[.387, .598]
4.742 .013, .168, .765
Knowledge Organization values represent mean correlation with expert model. Values for KnowledgeAcquisition represent mean percent correct. Numbers in brackets represent 95 % confidence interval.gp2 = partial eta squared; obs. power = observed power computed using alpha = .05. Means in the samerow with different superscripts ‘a’ and ‘b’ differ significantly at p \ .05, two-tailed, by the Fisher LeastSignificant Difference (LSD) comparison; different superscripts ‘c’ and ‘d’ indicate significant difference atp \ .01, two-taileda n = 17 for each condition
Enhancing learning outcomes in computer-based training
123
p = .106. No significant difference was found between the NQ and the HLEQ participants,
p = .230.
Univariate tests revealed no significant difference in performance on the declarative
knowledge (basic factual knowledge) questions (see Table 2). Although the LLEQ par-
ticipants did perform better than the NQ participants and HLEQ participants, these dif-
ferences were not significant, p = .062 and p = .270, respectively.
Univariate tests revealed a significant difference in performance on the perceptual
knowledge (concept recognition) questions (see Table 2). Pairwise comparisons showed
that the LLEQ participants significantly outperformed the NQ participants, p = .016.
Although the LLEQ participants’ performance was also greater than the HLEQ partici-
pants, this difference was not significant, p = .081. No significant difference was found
between the NQ and the HLEQ participants, p = .476.
Univariate tests revealed a significant difference in performance on the integrative
knowledge (airplane function) questions (see Table 2). Pairwise comparisons showed that
the LLEQ participants significantly outperformed the NQ participants, p = .004. Although
the LLEQ participants’ performance was also greater than the HLEQ participants, this
difference was not significant, p = .219. No significant difference was found between the
NQ and the HLEQ participants, p = .078.
Effect of query method on instructional efficiency (Hypothesis 4)
The means and standard errors for perceived cognitive load and the instructional efficiency
scores associated with the different training conditions are reported in Table 3. A one-way
between-groups ANOVA was conducted, comparing participants’ self-reported cognitive
load in the three training conditions. Univariate tests revealed no significant differences in
perceived cognitive load during training.
Next, the instructional efficiency (E) of the training program was calculated by plotting
the standardized scores on measures of cognitive effort (R) (i.e., self-report of task diffi-
culty as indicated by responses on the Tutorial Survey cognitive load item) against the
standardized scores on measures of performance (P) (declarative, perceptual, and inte-
grative questions), displayed as a cross of axes (for a detailed description of this procedure,
see Paas and van Merrienboer 1993). Instructional efficiency was calculated using the
following equation (adapted from Kalyuga et al. 1999): E = (P – R)/SQRT (2). The values
of P and R determine the sign of E. If P [ R, then E will be positive, indicating higher
efficiency (i.e., cognitive effort exerted is less, relative to the standard effort required to
achieve that level of performance). If P \ R, then E will be negative, indicating lower
efficiency (i.e., cognitive effort exerted is greater, relative to the standard effort required to
achieve that level of performance). Baseline (or standard level of efficiency) is represented
by E = 0.
Instructional efficiency (E) scores were analyzed using a one-way between-groups
MANOVA, with query method serving as the independent variable. Verbal comprehension
ability was not a significant covariate, and thus was not included in the analysis. The
dependent measures reflected the instructional efficiency of the training program in relation
to performance on the declarative, perceptual, and integrative questions. Multivariate
analysis revealed a significant effect for the query method on the training program’s
instructional efficiency, F (3, 47) = 3.648, p = .019, gp2 = .189. Univariate tests revealed
a significant effect of the query method on the training program’s instructional efficiency in
relation to performance on all three sets of knowledge questions (see Table 3).
H. M. Cuevas, S. M. Fiore
123
The LLEQ training condition consistently yielded positive instructional efficiency
scores (i.e., greater performance was achieved with less perceived cognitive effort),
whereas the NQ training condition consistently yielded negative instructional efficiency
scores (i.e., poorer performance with greater perceived cognitive effort) (see Fig. 3).
Instructional efficiency for the HLEQ training condition was typically at baseline (near
zero) (i.e., standard level of performance was achieved relative to perceived cognitive
effort). Pairwise comparisons showed that instructional efficiency scores between the
LLEQ and NQ training conditions significantly differed for the declarative (p = .005),
perceptual (p = .003), and integrative (p = .002) questions. Significant differences
between the LLEQ and HLEQ training conditions were found only for the perceptual
questions, p = .043. No significant differences were found between the NQ and HLEQ
training conditions.
Table 3 Means and standard errors for cognitive load and instructional efficiency scores by trainingcondition
Dependent variable Training conditionsa ANOVA results
NQ LLEQ HLEQ F (2, 48) p, gp2obs. power
Cognitive load 3.412 (0.283)[2.843, 3.981]
2.471 (0.283)[1.902, 3.039]
2.941 (0.283)[2.372, 3.510]
2.768 .073, .103, .520
Instructional efficiency
Declarative -0.457 (0.236)c
[-0.932, 0.017]0.520 (0.236)d
[0.046, 0.994]-0.065 (0.236)[-0.539, 0.410]
4.346 .018, .153, .727
Perceptual -0.476 (0.247)c
[-0.972, 0.200]0.599 (0.247)ad
[0.103, 1.096]-0.125 (0.247)b
[-0.622, 0.371]4.937 .011, .171, .784
Integrative -0.618 (0.266)c
[-1.154, -0.83]0.605 (0.266)d
[0.069, 1.141]0.011 (0.266)[-0.525, 0.547]
5.272 .009, .180, .811
Instructional Efficiency values represent mean E score. Numbers in brackets represent 95 % confidenceinterval. gp2 = partial eta squared; obs. power = observed power computed using alpha = .05. Means inthe same row with different superscripts ‘a’ and ‘b’ differ significantly at p \ .05, two-tailed, by the FisherLeast Significant Difference (LSD) comparison; different superscripts ‘c’ and ‘d’ indicate significant dif-ference at p \ .01, two-taileda n = 17 for each condition
Fig. 3 Effect of query method on training program’s instructional efficiency
Enhancing learning outcomes in computer-based training
123
Time-on-task
To evaluate any potential differences among the query conditions in the time to complete the
training and performance tasks, a one-way between-groups MANOVA was conducted, with
query method serving as the independent variable. Dependent variables included time-on-
task (recorded in minutes) for completing the Aviation Training Tutorial (training) and the
card sort and knowledge assessment tasks (performance). Multivariate analysis revealed a
significant effect for the query method on time-on-task, F (3,47) = 14.899, p \ .0005,
gp2 = .487. However, univariate tests revealed a significant difference in time-on-task only
for completion of the Aviation Training Tutorial (refer to Table 4). As would be expected,
pairwise comparisons showed that both the LLEQ and the HLEQ participants invested sig-
nificantly greater time-on-task for their training than the NQ participants, p \ .0005. No
significant difference for training time-on-task was found between the LLEQ and HLEQ
participants, p = .190. Univariate tests revealed no significant difference for time-on-task in
performing either the card sort task or the knowledge assessment task. Overall, these findings
indicate that participants in the query method condition invested more time in their training;
however, this was likely due to the extra time required to complete the sentence generation
task. Yet, the LLEQ and HLEQ participants did not necessarily need to take any longer than
the NQ participants in performing the post-training tasks.
Content analysis of query responses
The sentences generated by the participants in the LLEQ and HLEQ training conditions
were evaluated with regard to the number of concepts used in each sentence as well as the
accuracy of the sentences. Recall that the HLEQ training condition prompted participants
to generate a sentence using three or more concepts from the list presented whereas the
LLEQ training condition prompted participants to generate a sentence using only one of
these concepts. The number of concepts used in the sentence generated for each of the
three modules was summed to calculate the total number of concepts used during training.
The expected minimum number of concepts possible would be three for the LLEQ training
condition (one concept per module) and nine for the HLEQ training condition (at least
three concepts per module). Accuracy of the sentence generated for each module was rated
by a subject matter expert as either ‘0’ for inaccurate or ‘1’ for accurate, resulting in a
range from 0 (none of the sentences were accurate) to 3 (all three sentences were accurate).
Table 4 Means and standard errors for time-on-task by training condition
Dependent variable Training conditionsa ANOVA results
NQ LLEQ HLEQ F (2, 48) p, gp2
Time on task
Tutorial 17.824 (1.378)a 25.529 (1.378)b 28.118 (1.378)b 15.111 \.005, .386
Card sort 18.765 (1.225) 20.529 (1.225) 18.588 (1.225) \ 1 .470, .031
Knowledge assessment 18.706 (1.231) 20.235 (1.231) 21.000 (1.231) \ 1 .413, .036
Values for Time on task represent mean time in minutes. gp2 = partial eta squared. Means in the same rowwith different superscripts ‘a’ and ‘b’ differ significantly at p \ .01, two-tailed, by the Fisher Least Sig-nificant Difference (LSD) comparisona n = 17 for each condition
H. M. Cuevas, S. M. Fiore
123
An independent samples t test was conducted to analyze the total number of concepts
used and sentence accuracy, with training condition (LLEQ or HLEQ) serving as the
between-groups factor. As would be expected, analysis showed that participants in the
HLEQ training condition (M = 10.940; SD = 2.384) used a significantly greater total
number of concepts in generating their sentences than participants in the LLEQ training
condition (M = 5.060; SD = 2.164), t (32) = 7.532, p \ .0005, two-tailed. These results
suggest that participants were able to complete the sentence generation task as instructed.
With regard to the accuracy of the sentences, participants in the LLEQ training condition
(M = 2.820; SD = 0.529) generated significantly more accurate sentences than partici-
pants in the HLEQ training condition (M = 2.350; SD = 0.786), t (32) = 2.049, p = .049,
two-tailed.
Discussion
Training effectiveness can be evaluated in terms of how well instruction enhances learning
outcomes while minimizing the cognitive load imposed on learners during training. The
present study explored the effectiveness of a guided, learner-generated instructional
strategy (query method) in achieving this objective within computer-based training. This
study also examined the effect of varying the level of elaboration prompted by the queries,
asking participants to generate either simple (low-level elaboration) or complex (high-level
elaboration) sentences. Overall, results consistently highlighted the beneficial effect of
presenting participants with low-level elaboration queries, as compared to no queries or
high-level elaboration queries. Further, incorporating the high-level elaboration queries
into the training consistently failed to produce significantly better post-training outcomes
than the no-query training condition. These findings are next discussed in the context of
instructional strategies.
Learning as a constructive cognitive activity
The results of the present study support the utility of embedded content-dependent strat-
egies within computer-based training. Embedding the low-level elaboration queries within
the training may have prompted participants to attend to and interact with the critical
concepts in the presented material, increasing the efficiency of their information processing
and facilitating their knowledge acquisition. The content-dependent nature of the queries
may have promoted learning of the target domain by emphasizing the key concepts and
their unique interrelations, resulting in more accurate knowledge organization and better
integration of task-relevant concepts.
Query method—low versus high level elaboration
Incorporating the low-level elaboration queries into the training program significantly
enhanced post-training learning outcomes. Participants presented with the low-level
elaboration queries exhibited significantly more accurate knowledge organization (as
indicated by greater similarity to an expert model), better concept recognition (perceptual
knowledge questions), and superior performance on integrative knowledge questions
involving the integration and application of task-relevant concepts. Consistent with our
previous studies (Cuevas et al. 2002, 2004b; Fiore et al. 2003), no significant differences in
performance were found on basic factual knowledge assessment. Presentation of the low-
Enhancing learning outcomes in computer-based training
123
level elaboration queries also significantly improved the training program’s instructional
efficiency; that is, greater performance was achieved with less perceived cognitive effort.
Elaboration, learner control, and cognitive load
Prompting a high level of elaboration did not result in a significant difference in post-
training outcomes, when compared to no elaboration. Two potential explanations may
account for this somewhat counterintuitive finding. First, participants proceeded through
the aviation training tutorial at their own pace and could freely review the material,
allowing for a high level of learner control during initial training. However, one could
argue that learner control for reinforcing their newly acquired knowledge may have varied
across the two query method conditions. Specifically, arbitrarily forcing learners to gen-
erate complex sentences may have limited the level of control they could exercise as they
attempted to complete tasks designed to further develop their understanding of the training
concepts. In contrast, giving learners the option of generating simple sentences may have
allowed them to freely choose the appropriate level of complexity required in their elab-
oration of the material (Hasler et al. 2007). Learner control is essential if learners are to be
encouraged to take an active approach to learning, that is, make a deliberate and systematic
cognitive effort to engage the material during the learning process (Fonseca and Chi 2011).
A second explanation for this lack of effect for the high-level elaboration query focuses
on the cognitive load imposed on learners during their training. The findings in the present
study suggest that there may exist an optimal level of elaboration necessary to achieve the
desired learning gains. Indeed, the most optimal performance outcomes were yielded by
prompting participants to engage in a low level of elaboration of the training material. No
elaboration of the training material, as associated with the no-query training condition,
yielded significantly lower levels of post-training performance. Yet, requiring participants
to generate a high level of elaboration of the training material did not yield a corresponding
gain in post-training outcomes. Moreover, performance was not significantly greater than
the no-query training condition.
This lack of significant effect for the high-level elaboration query training condition
may have been due to the increased cognitive load associated with generating complex
sentences. According to Marcus et al. (1996, p. 50), ‘‘If multiple elements must be con-
sidered simultaneously because of high element interactivity, cognitive load may be high
and understanding difficult.’’ Similarly, forcing learners to attend to multiple sources of
information simultaneously may overburden their working memory capacity, reducing the
cognitive resources available for successful learning to occur (Sweller et al. 2011).
Interestingly, no significant differences were found on ratings of perceived cognitive load
during training between the two query method conditions. One possible explanation for
this result may be because participants did not consider the sentence generation task in
their ratings and focused on rating the training material itself. Further research is warranted
to investigate this unexpected finding.
By prompting participants to build internal associations between three or more concepts
in the to-be-learned material, the high-level elaboration queries were expected to promote a
deeper, more integrated understanding of the information (Fonseca and Chi 2011), leading
to more accurate knowledge organization. Yet, participants presented with the low-level
elaboration queries exhibited significantly more accurate knowledge organization and
generated more accurate sentences than those presented with the high-level elaboration
queries. Additionally, no significant differences were found between the no-query and
high-level elaboration query training conditions on knowledge organization, knowledge
H. M. Cuevas, S. M. Fiore
123
acquisition, or instructional efficiency. These findings suggest that generating complex
sentences was no more effective than not generating any sentences at all.
The beneficial effects of the query method on learners’ cognitive processes may have
been diminished by the increased cognitive load associated with the complexity of com-
pleting the high-level elaboration sentence generation task. Although this method was
designed to purposely enhance processing of the training material, the high-level elabo-
ration may have increased cognitive processing over a theoretical optimal level, thus
possibly imposing too great a germane cognitive load on participants during their training
or inadvertently introducing extraneous cognitive load. Specifically, when coupled with the
high intrinsic cognitive load of the training material, the increased germane or extraneous
cognitive load associated with the high-level elaboration may have minimized the cog-
nitive resources available for enhanced learning, as compared to the low-level elaboration
training condition. However, these findings are tentative pending further research to more
precisely investigate the effect of the query method on different types of cognitive load
(intrinsic, germane, and/or extraneous) and evaluate the resulting impact on learning
outcomes.
Instructional efficiency
The present study also demonstrated that evaluating a training program’s instructional
efficiency may serve as a potentially diagnostic measure of training effectiveness. Com-
bining subjective ratings of cognitive effort with performance scores may reveal useful
information about the effectiveness of training programs in terms of the cognitive costs
associated with complex task training over and above what would be found by using
measures of cognitive effort or performance alone (Paas and van Merrienboer 1993). As
discussed, results indicated that requiring participants to generate complex sentences (high-
level elaboration query) may have inadvertently increased the cognitive load associated
with the training, limiting the cognitive resources available for successful learning to
occur. Although, overall, this training condition yielded standard levels of performance
relative to the perceived cognitive effort exerted during training, such baseline (near zero)
instructional efficiency scores are not as ideal as the significantly higher instructional
efficiency scores yielded for the low-level elaboration query training condition.
Furthermore, an important distinction should be made between time and effort. On the
one hand, efficiency can be discerned from performance and amount of time spent on
learning. On the other hand, efficiency can be discerned from performance and amount of
effort invested in learning. Although time and effort are associated, these constructs are not
always equal. In this study, we focused on the effort/performance relationship for inves-
tigating the instructional efficiency of training programs.
Limitations and implications for future research
Key limitations to the present study should be noted. First, the training material used in this
study was based on introductory concepts related to the principles of flight. Although this
domain is more complex relative to the training material explored in prior studies (e.g.,
how a bicycle pump works, how lightning forms; Mayer, 2001; Moreno and Mayer, 2002),
to increase the external validity of this study’s findings, the query method needs to be
investigated with concepts of increasing complexity that are more relevant to advanced
training programs. Also, because participants used in this study were undergraduate stu-
dents enrolled in psychology courses, the generalizability of these findings to training in
Enhancing learning outcomes in computer-based training
123
complex operational environments is limited. Thus, it is necessary to examine how the
query method might affect learners from different populations, such as students in tech-
nical training courses.
In addition, the utility of the query method could be evaluated with learners with greater
prior knowledge in the target domain. In this way, future research can investigate the
degree to which the query method may have a differential effect on more experienced
populations (cf. de Bruin et al. 2007). Researchers in the study of expertise have argued
that instructional strategies need to be better tailored to where the learner falls on the
continuum of expertise (Fiore et al. 2008). As noted earlier, participants used in this study
had no prior knowledge or previous experience in the aviation domain. Yet, it is possible
that the high-level elaboration queries may be more beneficial for learners who have
already acquired some prior knowledge of this domain (e.g., student pilots in basic aviation
courses).
Second, although the reliability of the overall knowledge assessment task was accept-
able, the internal consistency of the individual sections was not as high as would be desired
to draw definitive conclusions from these results. This decrease may be due to the limited
number of questions in each section. This issue can be addressed with further testing using
a larger set of items. Finally, although the findings of this study are promising, conclusions
drawn from these results are limited by the small sample size (n = 17 per condition). Thus,
further research is required to validate the effectiveness of the query method.
Theoretical and practical implications
The results reported here build upon the findings of our past research investigating the use
of interactive computer-based training technology to facilitate knowledge acquisition and
integration for complex task training ( Cuevas et al. 2002, 2004b; Fiore et al. 2002, 2003;
Scielzo et al. 2004, 2005; for a review, see Cuevas et al. 2004a). This final section attempts
to integrate the findings in the present study with our prior work in order to highlight both
theoretical and practical implications for enhancing learning outcomes in computer-based
training.
Theoretical implications
Our earlier work (Cuevas et al. 2002; Fiore et al. 2003) demonstrated how diagrammatic
presentation can be effectively used in computer-based training to facilitate knowledge
acquisition of complex concepts. Illustrative diagrams may provide learners with a scaf-
folding framework for knowledge construction, enabling them to integrate the concepts
presented in the training more effectively. In particular, diagrams supported the devel-
opment of mental models more similar to an expert, leading to increased performance on
tests requiring the integration of concepts from across the training. The present study
continued with this line of research by demonstrating how learner-generated elaboration
may also support knowledge integration of complex concepts.
In addition, the findings in the present study suggest that when attempting to master an
already inherently complex domain (high intrinsic cognitive load), instructional strategies
that force the integration of these concepts (e.g., requiring generation of complex sen-
tences) may create too great a germane cognitive load or inadvertently introduce extra-
neous cognitive load, negatively interfering with learners’ knowledge construction by
overloading their already limited cognitive resources. As such, instructional strategies
H. M. Cuevas, S. M. Fiore
123
should be designed to lead learners to levels of cognitive effort appropriate to the training
content and not be overly burdensome if the content itself is already complex (i.e., has a
high intrinsic load).
Practical implications
Our programmatic research efforts have consistently demonstrated the value of adopting a
multi-faceted approach for evaluating post-training outcomes within computer-based
training environments (Cuevas et al. 2002; Fiore et al. 2002, 2003; Scielzo et al. 2004). As
in our previous work, this study employed a computer-based training and performance
assessment system that incorporated instructional strategies (query method) with tests
designed with different types of knowledge assessment questions (declarative, perceptual,
integrative). These current findings provide further validation for utilizing several distinct
yet related approaches designed to evaluate the impact of instructional strategies, such as
the query method, on learning outcomes.
In addition to mastery of both basic factual knowledge and concept recognition, suc-
cessful training programs must also prepare learners to effectively apply their newly
acquired knowledge to more complex situations than were experienced during their
training (Chipman et al. 2013). As such, post-training assessment needs to include tasks,
utilizing dynamic task-relevant scenarios, that provide opportunities for evaluating how
well learners can integrate and apply different knowledge concepts (Fiore et al. 2002).
Post-training assessment of learning outcomes should also evaluate the level of accuracy of
learners’ knowledge organization, that is, the degree to which learners’ knowledge
structures exhibit similarity to an expert model. The beneficial effect of the low-level
elaboration queries on learners’ cognitive processes was better diagnosed via such a multi-
faceted approach to knowledge assessment. Finally, the present study also highlighted the
importance of evaluating a training program’s instructional efficiency, as this measure may
be more diagnostic in determining why seemingly useful instructional design features may
not lead to the most optimal post-training outcomes.
Conclusion
In sum, the results of this study suggest important implications with regard to the design
and evaluation of computer-based training programs. Prompting learners to generate low
level elaborations of the training material led to improved post-training learning outcomes,
including more accurate knowledge organization, better concept recognition, superior
performance on tasks involving integration and application of concepts, and higher
instructional efficiency. These findings also implicate the importance of utilizing a multi-
faceted approach to assessing training effectiveness. Equally important is gauging the
training program’s instructional efficiency; that is, the evaluation of a training program’s
design must also consider the cognitive costs associated with the training relative to the
performance achieved. Advances in technology and instructional design will continue to
drive organizations to rely more on technology-mediated learning approaches. As such, it
is essential that training program designers incorporate useful instructional strategies,
guided by theory and research, to prompt learners to actively engage the material during
the learning process (Fonseca and Chi 2011).
Enhancing learning outcomes in computer-based training
123
Acknowledgments The views herein are those of the authors and do not necessarily reflect those of theorganizations with which the authors are affiliated. The research reported in this paper is based upon thedoctoral dissertation of Haydee M. Cuevas, University of Central Florida. Portions of this paper werepresented at the Human Factors and Ergonomics Society 50th Annual Meeting. This research was partiallysupported by funding through Grant Number F49620-01-1-0214 from the Air Force Office of ScientificResearch to Eduardo Salas, Stephen M. Fiore, and Clint A. Bowers.
References
Brunken, R., Plass, J. L., & Leutner, D. (2004). Assessment of cognitive load in multimedia learning withdual-task methodology: Auditory load and modality effects. Instructional Science, 32, 115–132.
Catrambone, R., & Yuasa, M. (2006). Acquisition of procedures: The effects of example elaborations andactive learning exercises. Learning and Instruction, 16(2), 139–153.
Chipman, S. F., Segal, J. W., & Glaser, R. (Eds.) (2013). Thinking and learning skills: Vol. 2: Research andopen questions. New York, NY: Routledge–Taylor and Francis Group.
Clark, R. C., Nguyen, F., & Sweller, J. (2006). Efficiency in learning: Evidence-based guidelines to managecognitive load. San Francisco: Jossey-Bass.
Cuevas, H. M., Fiore, S. M., Bowers, C. A., & Salas, E. (2004a). Fostering constructive cognitive andmetacognitive activity in computer-based complex task training environments. Computers in HumanBehavior, 20, 225–241.
Cuevas, H. M., Fiore, S. M., Bowers, C. A., & Salas, E. (2004b). Using guided learner-generatedinstructional strategies to transform learning into a constructive cognitive and metacognitive activity.Proceedings of the 48th Annual Meeting of the Human Factors and Ergonomics Society, (pp.1049–1053). Santa Monica, CA: Human Factors and Ergonomics Society.
Cuevas, H. M., Fiore, S. M., & Oser, R. L. (2002). Scaffolding cognitive and metacognitive processes in lowverbal ability learners: Use of diagrams in computer-based training environments. Instructional Sci-ence, 30, 433–464.
de Bruin, A., Rikers, R., & Schmidt, H. (2007). The effect of self-explanation and prediction on thedevelopment of principled understanding of chess in novices. Contemporary Educational Psychology,32(2), 188–205.
Fiore, S. M., Cuevas, H. M., & Oser, R. L. (2003). A picture is worth a thousand connections: Thefacilitative effects of diagrams on task performance and mental model development. Computers inHuman Behavior, 19, 185–199.
Fiore, S. M., Cuevas, H. M., Scielzo, S., & Salas, E. (2002). Training individuals for distributed teams:Problem solving assessment for distributed mission research. Computers in Human Behavior, 18,729–744.
Fiore, S. M., Hoffman, R. R., & Salas, E. (2008). Learning and performance across disciplines: An epiloguefor moving multidisciplinary research towards an interdisciplinary science of expertise. MilitaryPsychology, 20(S1), S155–S170.
Fiore, S. M., & Salas, E. (Eds.). (2007). Toward a science of distributed learning. Washington, DC:American Psychological Association.
Fiorella, L., Vogel-Walcutt, J. J., & Fiore, S. (2012). Differential impact of two types of metacognitiveprompting provided during simulation-based training. Computers in Human Behavior, 28(2), 696–702.
Fonseca, B., & Chi, M. T. H. (2011). The self-explanation effect: A constructive learning activity. In R.E. Mayer & P. A. Alexander (Eds.), The handbook of research on learning and instruction (pp.296–321). New York: Routledge—Taylor and Frances Group.
Guilford, J. P., & Zimmerman, W. S. (1981). Manual of instructions and interpretations for the Guilford-Zimmerman Aptitude Survey (revised ed.). Palo Alto, CA: Consulting Psychological Press.
Gully, S., & Chen, G. (2010). Individual differences, attribute-treatment interactions, and training outcomes.In S. W. J. Kozlowski & E. Salas (Eds.), Learning, training, and development in organizations (pp.3–64). New York: Routledge–Taylor and Francis Group.
Harper, M. E., Jentsch, F., Berry, D., Lau, H. C., Bowers, C., & Salas, E. (2003). TPL-KATS—Card Sort: Atool for assessing structural knowledge. Behavior Research Methods, Instruments, and Computers,35(4), 577–584.
Hasler, B. S., Kersten, B., & Sweller, J. (2007). Learner control, cognitive load and instructional animation.Applied Cognitive Psychology, 21, 713–729.
Hermanson, D. R., Hermanson, H. M., & Tompkins, J. G, I. V. (1997). The impact of self-generatedelaboration on students’ recall of finance concepts. Journal of Financial Education (Fall), 23, 27–34.
H. M. Cuevas, S. M. Fiore
123
Jeppesen Sanderson Training Systems. (1996a). Jeppesen Sanderson Private Pilot Exercises Book. Engle-wood, CO: Jeppesen Sanderson Inc.
Jeppesen Sanderson Training Systems. (1996b). Jeppesen Sanderson Private Pilot Maneuvers Manual (6thed.). Englewood, CO: Jeppesen Sanderson Inc.
Jeppesen Sanderson Training Systems. (1996c). Jeppesen Sanderson Private Pilot Manual (15th ed.).Englewood, CO: Jeppesen Sanderson Inc.
Kalyuga, S., Chandler, P., & Sweller, P. (1999). Managing split-attention and redundancy in multimediainstruction. Applied Cognitive Psychology, 13, 351–371.
Kalyuga, S., Chandler, P., & Sweller, J. (2004). When redundant on-screen text in multimedia technicalinstruction can interfere with learning. Human Factors, 46, 567–581.
King, A. (1992). Facilitating elaborative learning through guided student-generated questioning. Educa-tional Psychologist, 27, 111–126.
Kirschner, P. A. (2002). Cognitive load theory: Implications of cognitive load theory on the design oflearning. Learning and Instruction, 12, 1–10.
Kirwan, B., Evans, A., Donohoe, L., Kilner, A., Lamoureux, Atkinson, T. & MacKendrick, H. (1997).Human factors in the ATM system design life cycle. In Paper presented at the FAA/Eurocontrol ATMR&D Seminar, Paris, France, 16–20 June 1997. Retrieved from http://www.atmseminar.org/seminarContent/seminar1/papers/p_007_CDR.pdf. Accessed 19 Jan 2012.
Marcus, N., Cooper, M., & Sweller, J. (1996). Understanding instructions. Journal of Educational Psy-chology, 88, 49–63.
Mayer, R. E. (2001). Multimedia learning. Cambridge, England: Cambridge University Press.Mayer, R. E., Hegarty, M., Mayer, S., & Campbell, J. (2005). When static media promote active learning:
Annotated illustrations versus narrated animations in multimedia instruction. Journal of ExperimentalPsychology: Applied, 11(4), 256–265.
Mayer, R. E., Heiser, J., & Lonn, S. (2001). Cognitive constraints on multimedia learning: When presentingmore material results in less understanding. Journal of Educational Psychology, 93, 187–198.
Moreno, R., & Mayer, R. E. (2002). Verbal redundancy in multimedia learning: When reading helpslistening. Journal of Educational Psychology, 94(1), 156–163.
O’Reilly, T., Symons, S., & MacLatchy-Gaudet, H. (1998). A comparison of self-explanation and elabo-rative interrogation. Contemporary Educational Psychology, 23(4), 434–445.
Osman, M. E., & Hannafin, M. J. (1992). Metacognition research and theory: Analysis and implications forinstructional design. Educational Technology Research and Development, 40, 83–99.
Paas, F. G. W. C., & van Merrienboer, J. J. G. (1993). The efficiency of instructional conditions: Anapproach to combine mental effort and performance measures. Human Factors, 35, 737–743.
Paas, F. G. W. C., van Merrienboer, J. J. G., & Adam, J. J. (1994). Measurement of cognitive load ininstructional research. Perceptual and Motor Skills, 79, 419–430.
Rosenshine, B., Meister, C., & Chapman, S. (1996). Teaching students to generate questions: A review ofthe intervention studies. Review of Educational Research, 66, 181–221.
Salas, E., & Rosen, M. A. (2010). Experts at work: Principles for developing expertise in organizations. In S.W. J. Kozlowski & E. Salas (Eds.), Learning, training, and development in organizations (pp. 99–134).New York: Routledge–Taylor and Francis Group.
Scielzo, S., Cuevas, H. M, & Fiore, S. M. (2005). Investigating individual differences and instructional efficiency incomputer-based training environments. Proceedings of the Human Factors and Ergonomics Society 49thAnnual Meeting (pp. 1251–1255). Santa Monica, CA: Human Factors and Ergonomics Society.
Scielzo, S., Fiore, S. M., Cuevas, H. M., & Salas, E. (2004). Diagnosticity of mental models in cognitive andmetacognitive processes: Implications for synthetic task environment training. In S. G. Schiflett, L.R. Elliott, E. Salas, & M. D. Coovert (Eds.), Scaled worlds: Development, validation, and applications(pp. 181–199). Aldershot, UK: Ashgate.
Sweller, J. (1994). Cognitive load theory, learning difficulty and instructional design. Learning andInstruction, 4, 295–312.
Sweller, J. (2010). Element interactivity and intrinsic, extraneous, and germane cognitive load. EducationalPsychology Review, 22, 123–138.
Sweller, J., Ayers, P., & Kalyuga, S. (2011). Cognitive load theory. New York: Springer.Sweller, J., van Merrienboer, J. J. G., & Paas, F. (1998). Cognitive architecture and instructional design.
Educational Psychology Review, 10, 251–296.van Merrienboer, J. J. G., Schuurman, J. G., de Croock, M. B. M., & Paas, F. G. W. C. (2002). Redirecting
learners’ attention during training: Effects on cognitive load, transfer test performance and trainingefficiency. Learning and Instruction, 12, 11–37.
Wong, R. M. F., Lawson, M. J., & Keeves, J. (2002). The effects of self-explanation training on students’problem solving in high-school mathematics. Learning and Instruction, 12, 233–262.
Enhancing learning outcomes in computer-based training
123