Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

This article was downloaded by: [Colorado College]

On: 30 October 2014, At: 15:29


Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

Creativity Research Journal


Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/hcrj20

Causal Analysis to Enhance Creative Problem-Solving:


Performance and Effects on Mental Models
a a a a
Kimberly S. Hester , Issac C. Robledo , Jamie D. Barrett , David R. Peterson , Dean P.
a a a
Hougen , Eric A. Day & Michael D. Mumford
a
Department of Psychology , University of Oklahoma
Published online: 08 Jun 2012.

To cite this article: Kimberly S. Hester , Issac C. Robledo , Jamie D. Barrett , David R. Peterson , Dean P. Hougen , Eric A.
Day & Michael D. Mumford (2012) Causal Analysis to Enhance Creative Problem-Solving: Performance and Effects on Mental
Models, Creativity Research Journal, 24:2-3, 115-133, DOI: 10.1080/10400419.2012.677249

To link to this article: http://dx.doi.org/10.1080/10400419.2012.677249

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
CREATIVITY RESEARCH JOURNAL, 24(2–3), 115–133, 2012
Copyright # Taylor & Francis Group, LLC
ISSN: 1040-0419 print=1532-6934 online
DOI: 10.1080/10400419.2012.677249

Causal Analysis to Enhance Creative Problem-Solving:


Performance and Effects on Mental Models
Kimberly S. Hester, Issac C. Robledo, Jamie D. Barrett, David R. Peterson, Dean P. Hougen,
Eric A. Day, and Michael D. Mumford
Department of Psychology, University of Oklahoma

In recent years, it has become apparent that knowledge is a critical component of


Downloaded by [Colorado College] at 15:29 30 October 2014

creative thought. One form of knowledge that might be particularly important to


creative thought relies on the mental models people employ to understand novel,
ill-defined problems. In this study, undergraduates were given training in the use of cau-
sal relationships in applying mental models in creative problem-solving. A pre-post
design was used to assess the effects of this training on mental models and creative
problem-solving. It was found that causal analysis training resulted in the acquisition
of better mental models (in terms of subjective and objective attributes) and better solu-
tions (in terms of quality, originality, and elegance) to problems calling for creative
thought among high-ability participants. The implications of these findings for under-
standing the role of mental models in creative problem-solving are discussed.

Many variables have been found to influence people’s In a more recent study, Hunter, Bedell-Avers, Ligon,
creative problem-solving. Divergent thinking (Basadur, Hunsicker, and Mumford (2008) have shown that many
Runco, & Vega, 2000), openness (McCrae & Ingraham, knowledge structures, for example associational, case-
1987), self-efficacy (Jaussi, Randel, & Dionne, 2007) and based, or experiential, and schematic, or conceptual
the strategies people employ in problem-solving (Scott, knowledge, might be employed in creative problem-
Lonergan, & Mumford, 2005) have been found to influ- solving. One kind of knowledge that might be parti-
ence the people’s ability to solve the kind of novel, com- cularly important to peoples’ creative problem-solving
plex, and ill-defined problems that call for creative efforts are the mental models people use to understand
thought (Mumford & Gustafson, 2007). Of particular and craft solutions to novel, complex, ill-defined prob-
importance in this regard is the knowledge people lems (Anderson, Barker, & Chen, 2006; Finke, Ward,
bring to creative problem-solving efforts (Ericsson & & Smith, 1992). Mental models represent a complex
Charness, 1994; Weisberg, 2006). For example, Rich form of domain specific knowledge (Hmelo-Silver &
and Weisberg (2004) found that development of one Pfeffer, 2004). One critical aspect of people’s mental
creative product, the television show All in the Family, models is the causal connections made among the
was based on the combination and reorganization of concepts embedded in their models (Goldvarg &
extant knowledge. Other work by Ward, Patterson, Johnson-Laird, 2001).
and Sifonis (2004) and Vincent, Decker, and Mumford The potential impact of mental models on people’s
(2002) have provided empirical evidence for the impor- creative thinking suggests that training interventions
tance of knowledge in creative problem-solving. intended to improve creative problem-solving might
focus on the strategies by which people work with the
We would like to thank Rich Marcy and Sam Hunter for their information embedded in mental models as they seek
contributions to this effort. Parts of this research were supported by to solve problems calling for creative thought (Scott,
a grant from the National Science Foundation (IIS – 0852589), Leritz, & Mumford, 2004a, 2004b). One key set of
Michael D. Mumford principal investigator.
Correspondence should be sent to Michael D. Mumford, Depart-
strategies that might be employed as people work with
ment of Psychology, The University of Oklahoma, Norman, OK mental models may be found in the procedures used
73019. E-mail: mmumford@ou.edu to identify critical causes (Marcy & Mumford, 2007).
116 HESTER ET AL.

Our intent in this study was to show that providing mental models might represent an important form of
people with strategies for working with causal opera- knowledge influencing creative thought.
tives contributed to production of the higher-quality, One line of evidence may be found in historic studies
more original, and more elegant solutions (Besemer & of eminent achievement. For example, Crouch (1992), in
O’Quin, 1999; Christaans, 2002) that are held to be a study of the development of powered flight, found that
the hallmark of creative thought. the success of the Wright brothers could be traced to
their use of a unique mental model—the bicycle. Simi-
larly, Carlson and Gorman (1992), in contrasting the
MENTAL MODELS AND CREATIVITY work of Bell and Edison on the telephone, traced Bell’s
success to the use of a more appropriate mental model—
Mental models are held to represent a domain-specific the ear, as opposed to the telegraph. Other work by
form of knowledge (Acton, Johnson, & Goldsmith, Anderson et al. (2006) has traced many radical innova-
1994; Al-Diban, 2008). Although mental models have tions in the sciences to the development of new mental
been defined in different ways by different investigations models for understanding the operation of phenomena
(Rouse & Morris, 1986), mental models are commonly within certain domains.
held to involve two critical components (Doyle & Ford, Experimental studies have also provided some evi-
1998; Goldvarg & Johnson-Laird, 2001). First, mental dence for the impact of mental models on creative
models are held to involve a set of concepts. And, thought. In one investigation along these lines, Finke
Downloaded by [Colorado College] at 15:29 30 October 2014

second, they are held to involve a set of assumptions et al. (1992) asked people to draw aliens. They found
about the causal relationships operating among these that people’s drawings of aliens often retained features
concepts. Assessment of the status of relevant concepts evident in people’s mental models of life on earth. Thus
vis-à-vis assumptions made about causation allow flying aliens were often depicted as having wings and
people to understand, predict, and act with respect to feathers. In another investigation along these lines,
problems bearing on this mental model. Ward et al. (2004) found that asking people to think
The available evidence indicates that mental models more abstractly resulted in the production of more
are formulated by people to provide solutions to the original drawings of aliens.
problems they are encountering within a given domain. In a field study examining the impact of mental
Thus Hmelo-Silver and Pfeffer (2004) found that the models on creative problem-solving, Shah, Vargas-
mental models of aquariums formulated by biologists Hernandez, Summers, and Kilkarni (2001) examined
differed from those formulated by expert hobbyists with the effects of mental models on engineering students’
respect to both concepts embedded in their models and creativity in product design. Students were either pre-
the causal connections held to link these concepts. Not sented, or not presented, with an initial design and were
only do individuals differ as a function of experience asked, through progressive refinements, to produce new
in the mental models they formulate, mental models designs. It was found that the quality and originality of
have been shown to have a number of influences on peo- the designs increased when students were presented with
ple’s thinking. It has been found that mental models an initial mental model. Similarly, Mumford, Feldman,
influence the rate of skill acquisition as people work Hein, and Nago (2001) found that when shared mental
on complex tasks (Day, Arthur, & Gettman, 2001; models were induced through training, the production
Kieras & Bovair, 1984). Mental models have also been of high quality and original problem solutions increased
found to influence the execution of critical cognitive pro- on a group creative problem-solving task.
cessing activities, such as information search and encod- Finally, Mumford et al. (in press) examined the
ing (He, Erdelez, Wang, & Shyu, 2008), idea generation impact of mental models in a psychometric study. In this
and evaluation (Weick, 1995), and solution monitoring study, participants were asked to illustrate their mental
(Ellis & David, 2005). More centrally, the nature of models of either marketing or education using a concept
the mental models people apply have been shown to mapping technique. Here, key concepts drawn from
influence spatial analysis (Jahn, Knauff, & Johnson- either the marketing or educational literatures were pre-
Laird, 2007), chemistry (Coll & Treagust, 2001), and sented. And, participants were asked to select relevant
knitwear design (Eckert & Stacey, 2003). concepts and indicate their causal relationships. Both
These effects of people’s mental models on learning objective, for example number of concepts and number
and problem-solving bring to the fore a new question. of causal connections, and subjective, for example
Is there reason to expect that people’s mental models coherence and workability, features of these conceptual
might also influence their ability to solve the novel, com- maps, or mental models, were assessed. It was found
plex, ill-defined problems that are commonly held to call that both objective and subjective features of the mental
for creative thought? Broadly speaking. the evidence models provided by undergraduates predicted the qual-
accrued in four lines of investigation all suggest that ity, originality, and elegance of solutions provided to the
MENTAL MODELS 117

marketing and education problems—producing multiple on social innovation problems. Undergraduates were
correlations in the .30 to .35 range and accounting for asked to provide solutions to six social innovation prob-
variance above and beyond that attributable to intelli- lems where three problems were drawn from the business
gence, divergent thinking, and need for cognition. domain (e.g., unionization) and three problems were
drawn from the educational domain (e.g., introduction
of authentic assessment techniques). Prior to beginning
CAUSAL ANALYSIS work on these problems, participants were asked to com-
plete a self-paced instructional program. This self-paced
Taken as a whole, the findings sketched out earlier indi- instructional program familiarized participants with
cate that the characteristics of people’s mental models each of seven strategies for analyzing causes: (a) work
do influence their ability to provide high quality, orig- with causes that can be manipulated, (b) work with
inal, and elegant solutions to problems calling for causes that influence multiple outcomes, (c) work with
creative thought. This observation, in turn, broaches causes that have large effects, (d) work with causes that
another question. Can training interventions be can be controlled, (e) work with causes that have syner-
designed that will result in the acquisition of stronger gistic effects, (f) work with causes that work together,
mental models or better procedures for applying mental and (g) work with causes that have direct effects. Follow-
models? Many approaches are available for training cre- ing introduction of each strategy, its role in creative
ative thinking skills. However, the evidence provided by problem solving was described and its application was
Downloaded by [Colorado College] at 15:29 30 October 2014

Osburn and Mumford (2006), Scott et al. (2005), and illustrated. Then participants were asked to answer
Scott et al. (2004a, 2004b) suggests that one approach questions bearing on this strategy where feedback was
that is likely to prove effective in helping people apply provided. Finally, participants were asked to apply this
knowledge in creative problem-solving is to provide strategy in solving a series of progressively more complex
them with more effective strategies for applying a given problems. It was found that training these causal analysis
form of knowledge in problem-solving. strategies contributed to the production of higher-
Of course, the strategies that contribute to more quality, more original, and more elegant problem solu-
effective application of knowledge in creative problem- tions, especially when people approached these problems
solving depend on the nature of the knowledge being form a greater psychological distance.
applied by people. Scott et al. (2005) asked undergradu- These findings are noteworthy in the context of this
ates to work on solving an educational leadership prob- investigation because they suggest one set of strategies
lem that called for creative thought. Prior to preparing that might be useful in enhancing the application of
their problem solutions, people were asked either to mental models in creative problem-solving. As noted
work with case-based or experiential knowledge or, alter- earlier, mental models include causal relationships
natively, schematic or conceptual knowledge through among concepts. Accordingly, providing people through
material provided as they began working on their prob- training with better strategies for working with these
lem solutions. When working with schematic knowledge, causes should contribute to the production of creative
they were asked to employ feature search and mapping problem solutions (Marcy & Mumford, 2010). These
strategies (e.g., Baughman & Mumford, 1995). When observations, in turn, led to the first hypothesis underly-
working with case-based knowledge, they were asked ing the present study.
to employ critical event identification and forecasting
strategies (e.g., Mumford, Friedrich, Caughron, & Hypothesis 1: Providing people with strategies for
Byrne, 2007). It was found that the strategies contribu- working with the causes embedded
ting to the production of original, high-quality, and in mental models will result in the
elegant problem solutions varied as a function of the type production of higher quality, more
of knowledge people were asked to work with. original, and more elegant solutions
when people are asked to solve cre-
As noted earlier, mental models involve two distinct
ative problems.
types of knowledge. First, mental models involve a set
of concepts. Second, mental models involve definition
of a set of causal relationships among these concepts. However, training in strategies for working with
Thus, Goldvarg and Johnson-Laird (2001) have shown causes may not only contribute to better application of
that people may formulate different mental models extant mental models in creative problem-solving.
and arrive at different problem solutions based on the Mental models are, in part, based on causal connections
assumptions they make about operative causes, such established among concepts (Hmelo-Silver & Pfeffer,
as A causes B, A allows B, A prevents B, not A causes B. 2004). Accordingly, when provided with training con-
Recently, Marcy and Mumford (2007) examined the cerning how causal connections should be established
value of training people to apply different types of causes within a domain, shifts may occur among the causal
118 HESTER ET AL.

relationships held to operate among concepts. These who agreed to participate in the study were recruited
shifts in causal connections may, in turn, give rise to from undergraduate psychology courses providing extra
acquisition of stronger mental models for understanding credit for involvement in experimental studies. Of these
problems arising in a given domain (Rouse & Morris, 232 participants, there were 70 men, 152 women, and
1986). Thus, our second hypothesis: 10 unidentified. Students seeking extra credit were
asked to review a departmental Web-site that provided
Hypothesis 2: Providing people with strategies for a brief, one-paragraph description of available studies
working with the causes embedded and then select the studies in which they wished to par-
in mental models will result in the ticipate. The demographic characteristics were typical of
acquisition of stronger mental mod- students taking introductory psychology classes. Sample
els for understanding events occur- members had an average age of 19. Their academic
ring within a domain.
ability, as indexed by scores on the academic achieve-
ment test, lay roughly a quarter of a standard deviation
Although it appears that training in strategies for above national norms for freshmen entering 4-year
working with causes may result in better creative institutions.
problem-solving and acquisition of stronger mental
models, a noteworthy caveat should be borne in
General Procedures
mind. Ceci and Papierno (2005) reviewed the available
Downloaded by [Colorado College] at 15:29 30 October 2014

literature on interactions between ability and evalu- Participants were recruited to take part in what was
ation interventions. Broadly speaking, the findings purported to be a study of complex problem-solving.
obtained in this effort indicated that many, if not During the first half-hour of this 4-hour study, parti-
all, educational interventions are subject to a Matthew cipants were asked to complete a background infor-
effect—those who are already strong, or already pos- mation form and a set of timed covariates. During the
sess greater ability, benefit more from educational last half-hour of this study, participants were asked to
interventions than those who are weaker or lack requi- complete a set of untimed covariates. After participants
site ability. In keeping with this observation, Osburn had completed the timed covariates, they began work on
and Mumford (2006) provided undergraduates with a general training task.
training in strategies contributing to the development All participants in this investigation were asked to
of plans for executing creative ideas such as forecast- work for 1 hour on a task intended to provide them with
ing (Byrne, Shipman, & Mumford, 2010) and pen- instruction for how to draw, or illustrate, their concep-
etration (Mumford, 2001). They found that the tual maps, or mental models. In this training program,
impact of this training on creative problem-solving participants were presented with a scenario where they
depended on students’ ability levels with regard to were to assume the role of general manager for a foot-
intelligence and divergent thinking. These findings, in ball team. Using this scenario, participants were asked
turn, suggest that the effects of training causal analysis to work through a self-paced instructional program that
strategies on creative problem-solving and relevant described the nature of mental models from a structural
mental models will depend on the person’s initial equations modeling framework. This training program
ability. Hence, our third and fourth hypotheses: illustrated both concept selection and the manner in
which different causal relationships should be specified.
Hypothesis 3: The effects of training causal After participants had completed this training pro-
analysis strategies on creative gram, they were asked to work on a pretraining task.
problem-solving will depend on On this task, participants were asked to assume the role
people’s initial performance of a senior manager developing a marketing campaign
capabilities. for a new product—a new line of athletic shoes for
Hypothesis 4: The effects of training causal analy- extreme sports. Prior to starting work or preparing their
sis strategies on people’s mental
marketing campaigns, participants were presented with a
models will depend on people’s
initial performance capabilities.
series of e-mails. The first e-mail described the company
and task at hand. The second e-mail presented a list of
key marketing concepts drawn from the relevant research
literature (Aspara, Olkkonen, Tikkanen, Moisander, &
METHOD
Parvinen, 2008; Bennett, 1995; Clemente, 1992; Meldrum
& Mcdonald, 2007) and asked participants to illustrate
Sample
their mental model of the problem. Subsequently, an
The sample used to test these hypotheses was drawn e-mail requested participants to provide a 2- to 3-page
from a large southwestern university. The 232 students written description of their marketing plan.
MENTAL MODELS 119

Once participants had completed the pretest, they and Mumford, Marks, Connelly, Zaccaro, and Johnson
were provided with training in strategies that might be (1998).
used in causal analysis. All training was provided in a To measure expertise in the marketing domain, a
self-paced instructional format following the procedures variation on the procedures developed by Scott et al.
applied by Marcy and Mumford (2007). Training was (2005) were employed. Here, participants were presented
provided in a crossed experimental design where parti- with six background data questions examining prior
cipants were asked to complete none or up to four of interest and involvement in marketing issues in their life
these self-paced instructional modules. (Mumford & Stokes, 1992). For example, questions
Following causal analysis training, participants were asked, ‘‘How often have you discussed advertisements
presented with the posttest marketing task. On this task, with friends?’’ or ‘‘How often have you thought of
participants were asked to assume the role of a senior going into advertising or marketing as your career?’’
marketing manager working for a beverage company Responses to these questions were scored on a 5-point
and design a marketing campaign for a new product— scale and yielded internal consistency coefficients above
an energy root beer drink. Prior to beginning work on .70. Scott et al. and Lonergan, Scott, and Mumford
their marketing campaigns, participants were provided (2004) have provided evidence for the construct validity
with a series of e-mails, the first of which described the of these life history questions as a basis for assessing
company and the product at hand. The second e-mail expertise.
presented participants with the same list of marketing Because the study at hand asked participants to draw,
Downloaded by [Colorado College] at 15:29 30 October 2014

concepts presented in the pretest and asked them to or illustrate, their mental models, participants were
draw their model of the marketing problem at hand. also asked to complete a measure of spatial ability.
The final e-mail asked participants to provide a two- The spatial ability measure employed was a 12-item
to three-page written description of their marketing block-rotation test drawn from the Redrawn Vanden-
campaign for this new product. berg & Kuse Mental Rotations Test (MRT-A; Peters
et al., 1995). On this test, participants were presented
with a target image and asked to indicate which of four
Covariate Controls
rotated images matched this target. The 12 items
Prior research has shown that performance on creative included in this measure of spatial ability yielded retest
problem-solving tasks is influenced by intelligence, reliabilities above .80. Evidence for the construct val-
divergent thinking, and expertise (Vincent et al., 2002). idity of this test as a measure of spatial ability has been
To measure intelligence, participants were asked to com- provided by Peters et al.
plete the Employee Aptitude Survey (EAS). The EAS In addition to these ability and expertise measures,
questions present a set of four to five factual statements. motivation for working on cognitive tasks was also
Based on these statements, participants are asked to assessed. Here, participants were asked to complete
indicate whether a conclusion is true, false, or they are Cacioppo and Petty’s (1982) need for cognition scale.
not sure. The 30 items measured in this test yielded ret- This 18-item scale presents a set of behavioral state-
est reliability coefficients above .80. Evidence for the ments reflecting engagement in cognitive activities; for
predictive validity of this test, and its construct validity example, ‘‘I prefer complex to simple problems’’ and
as a measure of intelligence, has been provided in studies ‘‘I really enjoy a task that involves coming up with a
by Grimsley, Ruch, Warren, and Ford (1985) and Ruch new solutions to a problem.’’ People are asked to rate,
and Ruch (1980). on a 5-point scale, the extent to which they agree with
To assess divergent thinking, participants were these behavioral statements. The resulting scale scores
asked to complete Guilford’s consequences measure yielded internal consistency coefficients above .80. Evi-
(Merrifield, Guilford, Christen, & Frick, 1962). The dence bearing on the validity of this measure has been
consequences measure presents five questions where provided by Cacioppo and Petty.
people are asked to list the outcomes of unlikely events, Finally, participants were asked to complete Gill and
such as ‘‘What could happen if gravity was cut in half?’’ Hodgkinson’s (2007) Five Factor Model Questionnaire
or ‘‘What would happen if everyone lost the ability to (FFMQ) measuring neuroticism, openness, conscien-
read and write?’’ Participants are given 2 minutes per tiousness, extraversion, and agreeableness to provide
question to write down as many consequences of these a global assessment of personality. On this measure,
events as they can think of. When scored for fluency, participants are presented with 100 adjectives such as
or the average number of consequences provided, the accurate, original, and tolerant. They are asked to indi-
measure yielded internal consistency coefficients in the cate, on a 7-point scale, how accurately the adjectives
.70s. Evidence pointing to the construct validity and describe them. The resulting scales for measuring neur-
predictive validity of this measure of divergent thinking oticism, openness, extraversion, conscientiousness, and
has been provided by Guilford (1967), Merrifield et al., agreeableness produced internal consistency coefficients
120 HESTER ET AL.

above .70. Gill and Hodgkinson have provided evidence participants with the key concepts needed to illustrate
for the validity of these scales as measures of personality. mental models in a structural form. Inspection of the
models produced in module four indicated that parti-
cipants had learned how to articulate mental models
Mental Model Training
using this instructional program. A postexperiment
Participants were asked to complete a training program questionnaire indicated that virtually all participants
intended to provide instruction in how to articulate found this training valuable and that it had provided
mental models prior to beginning work on the precausal them with the background information needed to illus-
training and postcausal training marketing tasks. The trate their mental models.
self-paced instructional program consisted of four
modules. In the first module, participants were asked
Pretraining Task
to assume the role of general manager of a new pro-
fessional football team. Four key concepts involved in Prior to being exposed to training in causal analysis stra-
sports management, such as sponsorship, selection of tegies, participants were asked to work on a pretraining
coaches, selection of team members, and profits, were marketing problem. This problem was presented in a
presented and operationally defined. Subsequently, an series of e-mails. In the first e-mail, a history of a com-
initial mental model was presented and three key learn- pany producing athletic shoes was provided, which
ing points were presented: (a) lines between concepts noted that the company had a history of success and
Downloaded by [Colorado College] at 15:29 30 October 2014

indicate causal relationships, (b) lines ending in other had developed a noteworthy set of television advertising
lines indicate contingencies, and (c) positive signs indi- campaigns. In this e-mail, participants were asked to
cate positive relationships. Subsequently, participants assume the role of director of marketing working for
were asked to answer, in a few lines, two questions the vice president of sales, who was requesting a new
intended to illustrate these principles. marketing campaign for extreme sports footwear
In the second module, four additional concepts were intended to position the company with younger consu-
introduced: size of industry, salary contracts, injuries, mers and increase brand awareness.
and public promotion. Participants were then presented In the next e-mail, the vice president for sales pre-
with an extension of the initial model intended to take sented a list of important issues they hoped would be
into account (a) use of curved lines to indicate correla- considered in developing the marketing campaign. An
tions, (b) use of multiple lines linking concepts to indi- attachment provided this list of concepts relevant to
cate reciprocal causation, (c) use of negative signs to marketing and operational definitions of each concept.
indicate a negative relationship, and (d) use of double All concepts presented were drawn from the literature
positive signs to indicate a strong positive relationship. on marketing (e.g., Aspara et al., 2008; Bennett, 1995;
A verbal description of how these principles applied to Clemente, 1992; Meldrum & Mcdonald, 2007). The 24
the extended model was provided. Subsequently, parti- concepts presented, along with operational definition
cipants were asked to provide two answers to open- of each concept, included: (a) organizational mission,
ended questions where they were to draw implications (b) organizational objectives, (c) public opinion, (d)
from this model. political environment, (e) economic environment, (f)
In the third module, two new concepts were intro- social environment, (g) demographic environment, (h)
duced, stadium quality and fan attendance, and the competitive environment, (i) product, (j) price, (k) distri-
model incorporating these two additional concepts bution, (l) promotion, (m) product positioning, (n) tar-
was illustrated. This extended model was used to illus- get market identification, (o) consumer motivation, (p)
trate feedback loops and variables operating to cause market segmentation, (q) concentrated market segmen-
multiple outcomes. Following a verbal summary of this tation strategy, (r) undifferentiated market segmentation
extended model, which stressed these new principles, strategy, (s) profit, (t) market share, (u) payback, (v)
participants were again asked to answer, in a few lines, market position, (w) customer satisfaction, and (x) word
two questions where they were asked to draw conclu- of mouth promotion. After reviewing this list of con-
sions bearing on these principles given the model pre- cepts, participants were asked to select those concepts
sented. In the fourth and final module, participants they believed to be important in formulating a market-
were presented with two additional concepts, compe- ing plan and then draw their model indicating the rela-
tition and win=loss ratio. No illustrative model was pre- tionships among these concepts. After participants had
sented. Instead, participants were asked to redraw the drawn their mental model, they were presented with a
models presented earlier to incorporate these two final e-mail, where the vice president for sales requested
additional concepts. their marketing plan for the new line of extreme sports
The self-paced instructional program described above footwear. This marketing plan was to be two to three
used a scaffolding instructional method to provide written pages in length.
MENTAL MODELS 121

Posttraining Task receiving training simply read the first page of the intro-
ductory module that contained no information about
Following training in causal analysis strategies, parti-
the causal analysis strategies. The full introductory
cipants were asked to provide a solution to a second
module described the general importance of the strate-
marketing problem, or the posttraining task, where they
gies people apply in problem-solving to the success of
were asked to develop a marketing campaign for a high
their efforts and the ways strategies help people solve
energy root beer. Development of the marketing cam-
problems. Subsequently, participants were asked to
paign again occurred through a series of e-mails. In
answer a series of five multiple-choice questions about
the first e-mail, a description of the company and its his-
this material, where feedback was provided. If parti-
tory was provided, which noted that IBC root beer was
cipants answered more than one question incorrectly,
a popular, if traditional, soda and that the company had
they were asked to review the introductory material.
been acquired recently by Cadbury Schweppes. They
After participants had completed the introductory
were to assume the role of director of marketing report-
material, they were asked to work through an instruc-
ing to the vice president for sales. In the next e-mail,
tional booklet that presented the training material for
participants were informed they were being asked to
each strategy applying in their experimental condition.
develop a marketing campaign for a highly caffeinated
Each training module consisted of three components.
root beer intended to appeal to the 15- to 25-year-old
In the first component, the strategy being trained was
demographic.
defined in concrete operational terms. Next, participants
Following the e-mail, the vice president for sales sent
Downloaded by [Colorado College] at 15:29 30 October 2014

were presented with three questions they should ask


another e-mail, in which he asked the marketing director
themselves when applying this strategy. Finally, three
to consider the critical concepts involved in marketing in
to four reasons why application of the strategy might
developing his or her campaign. Again, 24 concepts,
prove useful in problem-solving were provided. Follow-
along with their operational definitions, were presented
ing this material, participants were presented with three
in an attachment. After reading through these concepts,
to four multiple-choice questions intended to ensure
participants were asked to select those concepts that
active processing where feedback concerning answers
they believed to be relevant to the development of their
to these questions was provided.
marketing plan and draw a model illustrating the rela-
In the second component, a real-world illustration of
tionships among these concepts. After participants had
application of this strategy in solving a day-to-day life
drawn their mental models, they received a final e-mail
problem was provided. In this component, the problem
from the vice president for sales in which a marketing
at hand was described in a paragraph. Subsequently, a
plan for the new, high energy root beer was requested.
second paragraph provided an illustration of how this
These marketing plans were to be two to three written
strategy might be applied in solving the problem at hand
pages in length.
and how application of this strategy influenced the
resulting problem solution. After reading through this
material, participants were asked to answer four mul-
Causal Training
tiple choice questions where feedback was provided. If
After completing the premeasure but prior to complet- participants answered more than one question incor-
ing the postmeasure, participants were asked to work rectly, they were asked to review the material.
through a training program intended to illustrate the In the third component of each training module,
use of strategies in causal analysis. This self-paced, participants were presented with a one paragraph,
instructional program was based on earlier work by day-to-day life problem they were asked to solve. After
Marcy and Mumford (2007). Experimental manipula- generating a solution, they were asked to answer three
tions were based on the number of strategies for which multiple-choice questions bearing on application of the
training was provided. In all, up to four strategies might strategy being trained. The questions presented in the
be potentially trained, including (a) negative effects— third component were more complex than the second
can acting on a cause have other unanticipated negative component.
effects?, (b) criticality—work with causes that have large
effects, (c) contingencies—is a causal effect dependent
Mental Model Variables
on another variable?, and (d) multiple outcomes—work
with causes that effect multiple outcomes. These four The conceptual maps drawn by participants in respond-
training modules were administered in a crossed design, ing to the pretest and posttest problems provided the first
including a no-training condition. set of dependent variables used to assess the quality of
All participants receiving training began work of this the mental models people applied in problem-solving.
self-paced instructional program by completing an These dependent variables were drawn from earlier work
introductory module. Those participants who were not by Mumford et al. (in press). Two sets of dependent vari-
122 HESTER ET AL.

ables were examined: (a) objective attributes of parti- regression weight below .15 and otherwise the scale
cipants’ conceptual maps and (b) subjective attributes was given a weight of 0. These weighted scale scores
of participants’ conceptual maps. were used to construct a scale for appraising positive
Two judges, both undergraduate research assistants, objective features of participants’ mental models and
were asked to appraise the concept maps provided by positive subjective features of participants’ mental mod-
participants with respect to 11 features: (a) the number els. Scores on the objective and subjective scales reflect-
of critical causal concepts used, (b) the number of causal ing the quality of participants’ mental models were
linkages made, (c) the number of reciprocal causal lin- obtained on both the pretraining and posttraining tasks.
kages made, (d) the number of positive relationships The scores on the objective and subjective scales derived
indicated, (e) the number of feedback loops indicated, from performance in the posttraining task provided the
(f) the number of basic causal concepts used, (g) the first set of dependent variables examined in the present
number of moderator=mediator concepts used, (h) the study.
number of moderator=mediator concepts used as media-
tors, (i) the number of outcome concepts used, (j) the
Creative Problem-Solving Variables
number of important concepts used, and (k) the number
of concept links. It is of note that important, critical, The second set of dependent variables examined part-
basic, and moderator=mediator concepts were defined icipants’ creative problem-solving. Here, four judges,
based on the findings obtained in prior research. Judges’ all doctoral students in industrial and organizational
Downloaded by [Colorado College] at 15:29 30 October 2014

average counts of each of these variables provided the psychology, were asked to read through solutions to
measure of each of these variables employed. The the footwear and soft drink problems provided on
reliability of these counts for the objective features of the pretraining and posttraining tasks. In accordance
the conceptual maps provided lay in the mid-.90s. with observations of Besemer and O’Quin (1999) and
Evidence for the construct validity of these measures Christiaans (2002), judges were asked to appraise each
has been provided by Mumford et al. (in press). solution for: (a) quality—complete, coherent, and use-
Three judges, all doctoral students in industrial and ful solutions; (b) originality—novel, surprising solu-
organizational psychology, were asked to appraise the tions; and (c) elegance—refined solutions which
subjective attributes of the conceptual maps provided flowed well together in a clever way. These ratings of
by participants. Judges were asked to rate, on a 5-point problem solution quality, originality, and elegance
scale, the extent to which the concept maps provided were made on a set of 5-point benchmark rating scales
were (a1) coherent, (b) logical, (c) workable, (d) com- based on the findings of Redmond, Mumford, and
plete, (e) complex, (f) novel, (g) hierarchically organized, Teach (1993). Figure 1 illustrates the quality, orig-
(h) categorical where one concept subsumed others, and inality, and elegance benchmark rating scales
(i) causal=process based. Prior to making these evalua- developed for the pretraining problem (the footwear
tions, judges were asked to participate in a 20-hr train- problem); Figure 2 illustrates the quality, originality,
ing program. In this training program, operational and elegance rating scales developed for the posttrain-
definitions of each concept were provided and judges ing problem (the soft drink problem).
were provided with practice applying these rating scales To develop these benchmark rating scales, a sample
to a set of sample products. Judges met to discuss and of 40 solutions to the footwear marketing problem (pre-
resolve any discrepancies following training. The aver- training) and 40 solutions to the soft drink marketing
age interrater reliability obtained for these subjective problem (posttraining) were obtained. Subsequently,
ratings of participants’ mental models was .63. Again, four judges rated the quality, originality, and elegance
Mumford et al. (in press) have provided evidence for of these problem solutions. Anchors were selected that
the construct validity of these subjective evaluations of had high, medium, or low mean scores and low standard
participants’ conceptual maps. deviations across judges. Abstracts reflecting key attri-
To scale scores on the objective and subjective eva- butes of these solutions were then written to provide
luations of participants’ conceptual maps, the following scale anchors.
procedures were applied. Initially, the findings provided Prior to applying these rating scales, judges were
by Mumford et al. (in press) with regard to the predic- required to participate in a 40-hr training program. In
tion of creative problem solutions were reviewed. The this training, judges were familiarized with the oper-
median regression weight obtained in predicting the ational definitions of solution quality, originality, and
quality, originality, and elegance of creative problem elegance. Subsequently, how quality, originality, and
solutions was obtained. A given objective or a given sub- elegance were reflected in solutions to the footwear mar-
jective scale was given a weight of þ1 if it provided a keting problem and the soft drink marketing problem
median standardized regression weight above .15, a were described. Judges were then asked to apply these
weight of 1 if it provided a median standardized rating scales in evaluating a set of sample problems.
MENTAL MODELS 123

Judges then met to discuss and resolve any discrepancies


in their ratings.
Following training, the interrater agreement coeffi-
cients obtained for evaluations of quality, originality,
and elegance on the pretraining footwear marketing
problem were .72, .69, and .60. The interrater agreement
coefficients obtained for evaluations of quality, orig-
inality, and elegance on the posttraining soft drink mar-
keting problem were .76, .76, and .67. Examination of
the correlations observed among these evaluations of
quality, originality, and elegance also provided some
evidence for the construct validity of these ratings. Thus,
quality and elegance were found to be positively related
on both the footwear (r ¼ .74) and soft drink (r ¼ .78)
marketing problems. However, solution quality and
elegance were less strongly related to solution originality
on the footwear (r ¼ .59) and soft drink (r ¼ .52) market-
ing problems.
Downloaded by [Colorado College] at 15:29 30 October 2014

Analyses
Two sets of analysis of covariance tests were conducted
to appraise the effects of causal analysis training on
positive features of participants’ mental models and
FIGURE 1 Anchors selected for pretraining problem for solution the quality, originality, and elegance of creative problem
quality, originality, and elegance. solutions. In all analyses, a covariate control was
retained only if it produced a relationship significant
at the .05 level. Moreover, in all analyses, scores on
the pretest problem were treated as a blocking variable
where participants receiving above average scores were
coded high (2) and below average scores were coded
low (1). The dependent variables examined included
positive, objective features of participants’ mental mod-
els, positive, subjective features of participants’ mental
models, the quality of problem solutions, the originality
of problem solutions, and the elegance of problem
solutions. All dependent variables were derived from
measures obtained on the posttraining (soft drink)
marketing problem. Scores on the blocking variable
were derived from performance on the pre-training
(footwear) marketing problem.

RESULTS

Problem Solving
Analysis of covariance tests were conducted to appraise
the effects of causal analysis training on the quality,
originality, and elegance of creative problem solutions.
In all analyses, a covariate control was retained only if
it produced a relationship significant at the .05 level.
Moreover, in all analyses, scores on the pre-test problem
were treated as a blocking variable where participants
FIGURE 2 Anchors selected for post-training problem for solution receiving above average scores were coded high (2)
quality, originality, and elegance. and below average scores were coded low (1). The
124 HESTER ET AL.

dependent variables examined included the quality of As might be expected, the comparison of high and
problem solutions, the originality of problem solutions, low performing people on the pretest had a significant
and the elegance of problem solutions. These variables effect, F(1,189) ¼ 47.14, p  .001, of posttest quality for
were derived from measures obtained on the these marketing problems. It was found that people
post-training (soft drink) marketing problem. Scores who evidenced better performance on the pretest foot-
on the blocking variable were derived from performance wear problem evidenced better performance on the
on the pre-training (footwear) marketing problem. posttest soft drink problem (m ¼ 3.05, SE ¼ .083) than
people who evidenced poorer performance on the pret-
Quality. Table 1 presents the results obtained in the est (m ¼ 2.36, SE ¼ .057). Thus, initial performance
analysis of covariance conducted for solution quality. conditions later performance.
One covariate produced significant (p  .05) effects with A significant main effect, F(1,189) ¼ 10.98, p  .001,
regard to solution quality. More specifically, gender, was also obtained for instruction in the use of negative,
F(1,189) ¼ 8.61, p  .01, was found to be related to sol- causal relationships. It was found that when people were
ution quality with men producing higher quality solu- given instruction in the use of negative, causal relation-
tions than women. ships, poorer quality (m ¼ 2.54, SE ¼ .069) posttest

TABLE 1
Downloaded by [Colorado College] at 15:29 30 October 2014

Analysis of Covariance Results for Post-Training Solution Quality

Variable MS F(1, 189) p g2

Covariates
Gender 3.546 8.608 0.004 0.044
Main Effects
Pre Quality 19.412 47.138 0.000 0.200
Critical Causes 0.157 0.380 0.538 0.002
Multiple Outcomes 0.091 0.220 0.639 0.001
Negative Effects 4.522 10.977 0.001 0.055
Contingencies 0.233 0.565 0.453 0.003
2-way Interactions
Pre Quality  Critical Causes 0.383 0.929 0.336 0.005
Pre Quality  Multiple Outcomes 0.129 0.313 0.576 0.002
Critical Causes  Multiple Outcomes 0.468 1.136 0.288 0.006
Pre Quality  Negative Effects 1.235 2.998 0.085 0.016
Critical Causes  Negative Effects 0.793 1.924 0.167 0.010
Multiple Outcomes  Negative Effects 0.066 0.160 0.690 0.001
Pre Quality  Contingencies 0.204 0.496 0.482 0.003
Critical Causes  Contingencies 0.009 0.022 0.882 0.000
Multiple Outcomes  Contingencies 0.167 0.406 0.525 0.002
Negative Effects  Contingencies 0.355 0.861 0.355 0.005
3-way Interactions
Pre Quality  Critical Causes  Multiple Outcomes 1.867 4.532 0.035 0.023
Pre Quality  Critical Causes  Negative Effects 0.005 0.011 0.915 0.000
Pre Quality  Multiple Outcomes  Negative Effects 0.684 1.659 0.199 0.009
Critical Causes  Multiple Outcomes  Negative Effects 0.027 0.064 0.800 0.000
Pre Quality  Critical Causes  Contingencies 0.225 0.545 0.461 0.003
Pre Quality  Multiple Outcomes  Contingencies 1.691 4.105 0.044 0.021
Critical Causes  Multiple Outcomes  Contingencies 0.324 0.785 0.377 0.004
Pre Quality  Negative Effects  Contingencies 0.023 0.056 0.814 0.000
Critical Causes  Negative Effects  Contingencies 0.235 0.571 0.451 0.003
Multiple Outcomes  Negative Effects  Contingencies 2.436 5.912 0.016 0.030
4-way Interactions
Pre Quality  Critical Causes  Multiple Outcomes  Negative Effects 0.324 0.786 0.376 0.004
Pre Quality  Critical Causes  Multiple Outcomes  Contingencies 0.146 0.354 0.553 0.002
Pre Quality  Critical Causes  Negative Effects  Contingencies 0.293 0.711 0.400 0.004
Pre Quality  Multiple Outcomes  Negative Effects  Contingencies 0.287 0.698 0.405 0.004
Critical Causes  Multiple Outcomes  Negative Effects  Contingencies 0.013 0.032 0.859 0.000
5-way Interaction
Pre Quality  Critical Causes  Multiple Outcomes  Negative Effects  Contingencies 0.009 0.022 0.882 0.000

Note: 10 participants were excluded because they did not report their gender. MS ¼ mean square, F(df) ¼ F-ratio(degrees of freedom),
p ¼ significance level, g2 ¼ partial eta squared effect size.
MENTAL MODELS 125

solutions were obtained than when no such instruction with people producing more original pretest solutions
(m ¼ 2.87, SE ¼ .072) was provided. These effects may (m ¼ 2.76, SE ¼ .082) producing more original posttest
reflect people’s tendency to appraise causation in terms solutions in comparison to those who produced pretest
of positive influences (Hogarth, 1980). solutions of low originality (m ¼ 1.72, SE ¼ .075). Sig-
The quality pretest solutions were also found to pro- nificant main effects were also produced by instruction
vide a significant, F(1,189) ¼ 4.53, p  .05, interaction in the use of negative causes, F(1, 200) ¼ 4.22, p  .05,
with training focused on the use of critical causes and and contingencies, F(1, 200) ¼ 5.44, p  .05. Inspection
the use of causes affecting multiple outcomes. It was of the cell means indicated that more original posttest
found that when people produced high-quality solutions problem solutions were obtained when no training
on the pretest and received no instruction on use of (m ¼ 2.35, SE ¼ .079) as opposed to training (m ¼ 2.13,
causes, better-quality solutions were obtained on the SE ¼ .078) in the use of negative causes was provided
posttest (m ¼ 3.19, SE ¼ .147) than in all other con- and when no training (m ¼ 2.37, SE ¼ .079), as opposed
ditions (m ¼ 2.53, SE ¼ .134) except when people who to training (m ¼ 2.11, SE ¼ .078) in the use of contingen-
produced high-quality solutions were provided with cies was provided.
training in both the use of critical causes and causes In this regard, however, it is important to bear in
affecting multiple outcomes (m ¼ 3.23, SE ¼ .159). Thus, mind the significant interactions between training and
although training in the use of critical causes and causes the production of original solutions on the pretest. Pro-
affecting multiple outcomes might hurt the production duction of original pretest solutions was found to pro-
Downloaded by [Colorado College] at 15:29 30 October 2014

of high-quality solutions, if training in both strategies duce a significant, F(1, 200) ¼ 3.96, p  .05, interaction
is provided performance improves. with training in the use of critical causes and contingen-
A similar pattern of effects emerged in examining the cies. The cell means indicated that when people pro-
effects of training use of causes affecting multiple out- duced original solutions on the pretest the best
comes and use of contingencies. It was found that a sig- performance was observed when no training was pro-
nificant interaction, F(1,189) ¼ 4.11, p  .05, was vided (m ¼ 3.07, SE ¼ .183) or when training in both
obtained among these variables in accounting for postt- use of critical causes and contingencies (m ¼ 2.73,
est solution quality—the quality of solutions obtained SE ¼ .177) was provided in contrast to all other con-
for the soft drink campaign. Again, it was found that ditions (m ¼ 2.60, SE ¼ .144) for those producing orig-
when people produced high quality solutions on the inal solutions in the pretest. Among those producing
pretest, the footwear campaign, the highest-quality solu- unoriginal pretest solutions, however, particularly
tions were obtained when no training was provided unoriginal solutions (m ¼ 1.44, SE ¼ .142) were obtained
(m ¼ 3.25, SE ¼ .142) and when both training in use of when both critical causes and contingencies training was
multiple outcomes and use of contingent causes provided in contrast to all other conditions (m ¼ 1.81,
(m ¼ 3.11, SE ¼ .169) was provided in comparison to SE ¼ .153). Thus, training in the use of critical causes
other conditions for people producing high-quality, and contingencies helped people who had strong initial
pretest solutions (m ¼ 2.91, SE ¼ .175) and people, performance, but hurt people who had weak initial per-
regardless of condition, producing low-quality, pretest formance with regard to solution originality.
solutions (m ¼ 2.35, SE ¼ .112). A significant (F(1, 200) ¼ 4.17, p  .05) interaction
Finally, a significant, F(1,189) ¼ 5.91, p  .05, was also obtained between training in the use of critical
three-way interaction was observed among training causes, causes effecting multiple outcomes and contin-
examining use of multiple outcomes, negative causes, gencies. Inspection of the cell means indicated that the
and contingencies. Inspection of the cell means indicated especially unoriginal solutions were obtained when no
that the highest quality posttest solutions were obtained critical causes training was provided, but training in
when no training in these three strategies (m ¼ 3.15, multiple outcomes and contingencies was provided
SE ¼ .131) was provided in contrast to all other con- (m ¼ 1.89, SE ¼ .146) in comparison to all other con-
ditions (m ¼ 2.63, SE ¼ .142). Thus, causal analysis ditions (m ¼ 2.62, SE ¼ .150). Thus, incongruent train-
training proved useful with regard to the production ing patterns, for example, contingencies are often
of high quality solutions only when certain patterns of specified with respect to critical causes, appear to disrupt
strategy training were given to people already evidenc- the production of original problem solutions.
ing some skill as assessed by pretest solution quality. Training in negative causes, multiple outcomes, and
contingencies also produced a significant, F(1, 200) ¼
Originality. Table 2 presents the results obtained in 3.76, p  .05, interaction in accounting for solution orig-
the analysis of covariance when posttest solution orig- inality on the posttest. It was found that unoriginal solu-
inality was the dependent variable of concern. Again, tions were especially likely to be produced when all three
the production of original solutions on the pretest pro- training modules were presented (m ¼ 1.85, SE ¼ .144)
duced a significant effect, F(1, 200) ¼ 86.59, p  .001, in comparison to all other conditions (m ¼ 2.28,
126 HESTER ET AL.

TABLE 2
Analysis of Covariance Results for Post-Training Solution Originality

Variable MS F(1, 200) p g2

Main Effects
Pre Originality 55.499 86.586 0.000 0.302
Critical Causes 0.745 1.163 0.282 0.006
Multiple Outcomes 1.342 2.093 0.150 0.010
Negative Effects 2.702 4.216 0.041 0.021
Contingencies 3.485 5.438 0.021 0.026
2-way Interactions
Pre Originality  Critical Causes 0.071 0.111 0.739 0.001
Pre Originality  Multiple Outcomes 0.141 0.220 0.639 0.001
Critical Causes  Multiple Outcomes 0.003 0.005 0.941 0.000
Pre Originality  Negative Effects 0.326 0.509 0.476 0.003
Critical Causes  Negative Effects 1.245 1.943 0.165 0.010
Multiple Outcomes  Negative Effects 0.896 1.398 0.238 0.007
Pre Originality  Contingencies 0.268 0.418 0.519 0.002
Critical Causes  Contingencies 0.241 0.376 0.541 0.002
Multiple Outcomes  Contingencies 0.504 0.787 0.376 0.004
Negative Effects  Contingencies 0.749 1.169 0.281 0.006
Downloaded by [Colorado College] at 15:29 30 October 2014

3-way Interactions
Pre Originality  Critical Causes  Multiple Outcomes 0.385 0.601 0.439 0.003
Pre Originality  Critical Causes  Negative Effects 0.007 0.011 0.917 0.000
Pre Originality  Multiple Outcomes  Negative Effects 0.057 0.089 0.766 0.000
Critical Causes  Multiple Outcomes  Negative Effects 0.277 0.432 0.512 0.002
Pre Originality  Critical Causes  Contingencies 2.537 3.958 0.048 0.019
Pre Originality  Multiple Outcomes  Contingencies 0.165 0.257 0.612 0.001
Critical Causes  Multiple Outcomes  Contingencies 2.671 4.166 0.043 0.020
Pre Originality  Negative Effects  Contingencies 0.503 0.785 0.377 0.004
Critical Causes  Negative Effects  Contingencies 0.013 0.021 0.885 0.000
Multiple Outcomes  Negative Effects  Contingencies 2.412 3.763 0.054 0.018
4-way Interactions
Pre Originality  Critical Causes  Multiple Outcomes  Negative Effects 0.462 0.721 0.397 0.004
Pre Originality  Critical Causes  Multiple Outcomes  Contingencies 0.031 0.048 0.827 0.000
Pre Originality  Critical Causes  Negative Effects  Contingencies 0.850 1.326 0.251 0.007
Pre Originality  Multiple Outcomes  Negative Effects  Contingencies 0.030 0.046 0.830 0.000
Critical Causes  Multiple Outcomes  Negative Effects  Contingencies 1.683 2.626 0.107 0.013
5-way Interaction
Pre Originality  Critical Causes  Multiple Outcomes  Negative Effects 
Contingencies 0.133 0.207 0.650 0.001
2
Note: MS ¼ mean square, F(df) ¼ F-ratio(degrees of freedom), p ¼ significance level, g ¼ partial eta squared effect size.

SE ¼ .158). Again, if critical causes training was not pro- the pretest problem of developing a marketing campaign
vided, other forms of strategy training could apparently for a new athletic footwear. It was found that people
prove detrimental to the production of original solutions who produced more elegant solutions on the pretest
on the posttest. (m ¼ 3.09, SE ¼ .06) produced more elegant posttest
solutions than people who produced less elegant solu-
Elegance. Table 3 presents the results obtained in tions on the pretest (m ¼ 2.51, SE ¼ .064).
the analysis of covariance where posttest solution A significant, F(1, 198) ¼ 3.89, p  .05, interaction
elegance was the dependent variable. As may be seen, was obtained between training focused on the use of
both divergent thinking, F(1, 198) ¼ 2.42, p  .15, and critical causes, negative causes, and contingencies.
neuroticism, F(1, 198) ¼ 3.48, p  .10, produced margin- Inspection of the cell means indicated that when critical
ally significant relationships with the production of causes training was provided, but not negative causes
elegant solutions on the posttest, soft drink marketing training and contingency training, more elegant solu-
problem. It was found that divergent thinking was posi- tions (m ¼ 3.16, SE ¼ .120) were obtained on the
tively related to production of more elegant solutions, post-test in comparison to all other conditions
while neuroticism was negatively related to the pro- (m ¼ 2.75, SE ¼ .125). Thus, training in use of critical
duction of more elegant solutions. causes could contribute to production of more elegant
Again, a significant main effect, F(1, 198) ¼ 41.98, solutions when exposure to these alternative training
p  .001, was obtained for the elegance of solutions to conditions disrupted performance.
MENTAL MODELS 127

TABLE 3
Analysis of Covariance Results for Post-Training Solution Elegance

Variable MS F(1, 198) p g2

Covariates
Divergent Thinking 0.926 2.426 0.121 0.012
Neuroticism 1.330 3.485 0.063 0.017
Main Effects
Pre Elegance 16.017 41.982 0.000 0.175
Critical Causes 0.287 0.753 0.387 0.004
Multiple Outcomes 1.294 3.390 0.067 0.017
Negative Effects 3.360 8.806 0.003 0.043
Contingencies 0.413 1.081 0.300 0.005
2-way Interactions
Pre Elegance  Critical Causes 0.306 0.803 0.371 0.004
Pre Elegance  Multiple Outcomes 0.310 0.812 0.369 0.004
Critical Causes  Multiple Outcomes 0.763 1.999 0.159 0.010
Pre Elegance  Negative Effects 0.016 0.043 0.836 0.000
Critical Causes  Negative Effects 0.799 2.094 0.149 0.010
Multiple Outcomes  Negative Effects 0.001 0.002 0.965 0.000
Pre Elegance  Contingencies 0.191 0.501 0.480 0.003
Downloaded by [Colorado College] at 15:29 30 October 2014

Critical Causes  Contingencies 0.015 0.040 0.842 0.000


Multiple Outcomes  Contingencies 0.002 0.004 0.949 0.000
Negative Effects  Contingencies 0.859 2.251 0.135 0.011
3-way Interactions
Pre Elegance  Critical Causes  Multiple Outcomes 0.355 0.931 0.336 0.005
Pre Elegance  Critical Causes  Negative Effects 0.007 0.019 0.891 0.000
Pre Elegance  Multiple Outcomes  Negative Effects 0.241 0.631 0.428 0.003
Critical Causes  Multiple Outcomes  Negative Effects 0.006 0.014 0.905 0.000
Pre Elegance  Critical Causes  Contingencies 0.284 0.745 0.389 0.004
Pre Elegance  Multiple Outcomes  Contingencies 0.100 0.263 0.608 0.001
Critical Causes  Multiple Outcomes  Contingencies 0.425 1.115 0.292 0.006
Pre Elegance  Negative Effects  Contingencies 0.624 1.635 0.203 0.008
Critical Causes  Negative Effects  Contingencies 1.484 3.891 0.050 0.019
Multiple Outcomes  Negative Effects  Contingencies 3.071 8.050 0.005 0.039
4-way Interactions
Pre Elegance  Critical Causes  Multiple Outcomes  Negative Effects 0.317 0.830 0.363 0.004
Pre Elegance  Critical Causes  Multiple Outcomes  Contingencies 0.256 0.671 0.414 0.003
Pre Elegance  Critical Causes  Negative Effects  Contingencies 0.168 0.441 0.507 0.002
Pre Elegance  Multiple Outcomes  Negative Effects  Contingencies 0.325 0.853 0.357 0.004
Critical Causes  Multiple Outcomes  Negative Effects  Contingencies 0.381 0.998 0.319 0.005
5-way Interaction
Pre Elegance  Critical Causes  Multiple Outcomes  Negative Effects  Contingencies 1.158 3.036 0.083 0.015

Note: MS ¼ mean square, F(df) ¼ F-ratio(degrees of freedom), p ¼ significance level, g2 ¼ partial eta squared effect size.

Training in the use of negative causes, multiple out- effects, on the quality, originality, and elegance of peo-
comes, and contingencies also produced a significant, ple’s creative problem solutions, the question remains
F(1, 198) ¼ 8.05, p  .01, three-way interaction. Examin- as to whether these training interventions also affected
ation of the cell means indicated that the effect could be attributes of people’s mental models as reflected in their
traced to production of especially inelegant posttest conceptual maps. To address this issue, composite
solutions when all three forms of training were provided scores reflecting desirable objective and subjective attri-
(m ¼ 2.48, SE ¼ .113), as opposed to all other conditions butes of people’s mental models were formulated based
(m ¼ 2.84, SE ¼ .126). Thus, extensive training, where on the results obtained by Mumford et al. (in press).
critical causes training was not provided, could disrupt Then analysis of covariance tests were conducted to
the production of more elegant solutions to the posttest appraise the effects of causal analysis training on posi-
soft drink, marketing problem. tive features of participants’ mental models. The same
procedures used in the problem solving analyses for cov-
ariates and blocking variables were used in the mental
Mental Models
model analyses. The dependent variables examined
Although our foregoing findings indicate that causal included positive, objective features of participants’
analysis training could exert effects, albeit complex mental models and positive, subjective features of
128 HESTER ET AL.

TABLE 4
Analysis of Covariance Results for Post-Training Objective Mental Model Attributes

Variable MS F(1, 200) p g2

Main Effects
Critical Causes 4.436 1.519 0.219 0.008
Multiple Outcomes 4.529 1.551 0.214 0.008
Negative Effects 2.071 0.709 0.401 0.004
Contingencies 10.964 3.755 0.054 0.018
Pre Objective Model Quality 439.450 150.488 0.000 0.429
2-way Interactions
Critical Causes  Multiple Outcomes 4.908 1.681 0.196 0.008
Critical Causes  Negative Effects 0.635 0.218 0.641 0.001
Multiple Outcomes  Negative Effects 0.337 0.115 0.735 0.001
Critical Causes  Contingencies 6.234 2.135 0.146 0.011
Multiple Outcomes  Contingencies 0.002 0.001 0.978 0.000
Negative Effects  Contingencies 9.817 3.362 0.068 0.017
Critical Causes  Pre Objective Model Quality 1.634 0.560 0.455 0.003
Multiple Outcomes  Pre Objective Model Quality 0.351 0.120 0.729 0.001
Negative Effects  Pre Objective Model Quality 0.001 0.000 0.988 0.000
Contingencies  Pre Objective Model Quality 0.456 0.156 0.693 0.001
Downloaded by [Colorado College] at 15:29 30 October 2014

3-way Interactions
Critical Causes  Multiple Outcomes  Negative Effects 2.464 0.844 0.359 0.004
Critical Causes  Multiple Outcomes  Contingencies 0.020 0.007 0.933 0.000
Critical Causes  Negative Effects  Contingencies 0.256 0.088 0.768 0.000
Multiple Outcomes  Negative Effects  Contingencies 1.579 0.541 0.463 0.003
Critical Causes  Multiple Outcomes  Pre Objective Model Quality 2.012 0.689 0.408 0.003
Critical Causes  Negative Effects  Pre Objective Model Quality 0.653 0.224 0.637 0.001
Multiple Outcomes  Negative Effects  Pre Objective Model Quality 12.946 4.433 0.036 0.022
Critical Causes  Contingencies  Pre Objective Model Quality 4.271 1.463 0.228 0.007
Multiple Outcomes  Contingencies  Pre Objective Model Quality 0.743 0.255 0.614 0.001
Negative Effects  Contingencies  Pre Objective Model Quality 2.824 0.967 0.327 0.005
4-way Interactions
Critical Causes  Multiple Outcomes  Negative Effects  Contingencies 0.040 0.014 0.907 0.000
Critical Causes  Multiple Outcomes  Negative Effects  Pre Objective Model Quality 15.695 5.375 0.021 0.026
Critical Causes  Multiple Outcomes  Contingencies  Pre Objective Model Quality 0.236 0.081 0.776 0.000
Critical Causes  Negative Effects  Contingencies  Pre Objective Model Quality 11.040 3.781 0.053 0.019
Multiple Outcomes  Negative Effects  Contingencies  Pre Objective Model Quality 0.190 0.065 0.799 0.000
5-way Interaction
Critical Causes  Multiple Outcomes  Negative Effects  Contingencies  Pre Objective Model Quality 0.652 0.223 0.637 0.001
2
Note: MS ¼ mean square, F(df) ¼ F-ratio(degrees of freedom), p ¼ significance level, g ¼ partial eta squared effect size.

participants’ mental models. These variables were compared to people who evidenced weaker pretest men-
derived from measures obtained on the posttraining tal models (m ¼ 3.91, SE ¼ .170). It was also found that
(soft drink) marketing problem. Scores on the blocking contingency training produced a significant, F(1,
variable were derived from performance on the pretrain- 200) ¼ 3.75, p  .05, main effect with people receiving
ing (footwear) marketing problem. contingency training evidencing weaker mental models
(m ¼ 5.14, SE ¼ .172) than those who received no train-
Objective attributes. Table 4 presents the results ing (m ¼ 5.61, SE ¼ .166).
obtained when objective attributes of the conceptual However, a significant, F(1, 200) ¼ 4.43, p  .05,
maps provided for the posttest, soft drink, marketing three-way interaction emerged between pretraining men-
campaign were treated as the dependent variable. tal models and training in the use of multiple outcomes
Although no significant covariates emerged in this and negative causes. The cell means indicated that when
analysis, scores on the inclusion of desirable objective people had poor pretest mental models, as defined by
attributes in people’s pretest mental models was signifi- their objective features, providing training in use of both
cantly, F(1, 200) ¼ 150.48, p  .001, related to inclusion multiple outcomes and negative causes (m ¼ 4.32,
of desirable objective attributes on peoples posttest men- SE ¼ .390) resulted in better performance than all other
tal models. Inspection of the cell means indicated that conditions (m ¼ 3.76, SE ¼ .320). In contrast, when
stronger pretest mental models resulted in stronger people evidence stronger mental models on the pretest,
posttest mental models (m ¼ 6.84, SE ¼ .168), as all training conditions (m ¼ 7.03, SE ¼ .317) resulted in
MENTAL MODELS 129

stronger mental models than no training (m ¼ 6.26, mental models on the pretest problem, particularly
SE ¼ .279). Thus, the causal analysis training for use strong posttest mental models were obtained when
of multiple outcomes and negative causes did appear training was provided articulating use of negative causes
beneficial with respect to post-training mental models. (m ¼ 7.92, SE ¼ .442) or when training was provided
A significant interaction, F(1, 200) ¼ 5.38, p  .05, indicating use of multiple outcomes (m ¼ 7.60,
was also obtained between pretest mental models and SE ¼ .403) relative to all other conditions (m ¼ 6.51,
training in both critical and negative causes along with SE ¼ .486). Apparently, different patterns of causal
use of multiple outcomes. The cell means indicated that training interventions are beneficial for those who have
when people evidence weak mental models on the pret- stronger as opposed to weaker initial mental models.
est, as defined by objective attributes, particularly strong Finally, a significant interaction, F(1, 200) ¼ 3.78,
posttest mental models were obtained when training was p  .05, was obtained between the strength of initial
provided to stress the use of negative causes and mul- mental models, as defined by their objective features,
tiple outcomes (m ¼ 4.61, SE ¼ .493) relative to all other and training in the use of critical causes, negative causes
conditions (m ¼ 3.80, SE ¼ .472). In contrast, when and contingencies. Inspection of the cell means indicated
people evidenced stronger objective attributes of their that when people had strong initial mental models,

TABLE 5
Downloaded by [Colorado College] at 15:29 30 October 2014

Analysis of Covariance Results for Post-Training Subjective Mental Model Attributes

Variable MS F(1, 195) p g2

Covariates
Domain Experience 1.834 10.240 0.002 0.050
Main Effects
Critical Causes 0.019 0.107 0.744 0.001
Multiple Outcomes 0.249 1.389 0.240 0.007
Negative Effects 0.008 0.044 0.835 0.000
Contingencies 0.154 0.862 0.354 0.004
Pre Subjective Model Quality 9.737 54.354 0.000 0.218
2-way Interactions
Critical Causes  Multiple Outcomes 0.096 0.538 0.464 0.003
Critical Causes  Negative Effects 0.032 0.176 0.675 0.001
Multiple Outcomes  Negative Effects 0.007 0.038 0.846 0.000
Critical Causes  Contingencies 0.009 0.049 0.825 0.000
Multiple Outcomes  Contingencies 0.004 0.025 0.875 0.000
Negative Effects  Contingencies 0.143 0.801 0.372 0.004
Critical Causes  Pre Subjective Model Quality 0.329 1.839 0.177 0.009
Multiple Outcomes  Pre Subjective Model Quality 0.062 0.347 0.556 0.002
Negative Effects  Pre Subjective Model Quality 0.340 1.898 0.170 0.010
Contingencies  Pre Subjective Model Quality 0.000 0.001 0.976 0.000
3-way Interactions
Critical Causes  Multiple Outcomes  Negative Effects 0.694 3.877 0.050 0.019
Critical Causes  Multiple Outcomes  Contingencies 0.085 0.475 0.492 0.002
Critical Causes  Negative Effects  Contingencies 0.338 1.886 0.171 0.010
Multiple Outcomes  Negative Effects  Contingencies 0.057 0.320 0.572 0.002
Critical Causes  Multiple Outcomes  Pre Subjective Model Quality 0.075 0.420 0.517 0.002
Critical Causes  Negative Effects  Pre Subjective Model Quality 0.006 0.034 0.853 0.000
Multiple Outcomes  Negative Effects  Pre Subjective Model Quality 0.061 0.339 0.561 0.002
Critical Causes  Contingencies  Pre Subjective Model Quality 0.141 0.787 0.376 0.004
Multiple Outcomes  Contingencies  Pre Subjective Model Quality 0.025 0.140 0.708 0.001
Negative Effects  Contingencies  Pre Subjective Model Quality 0.001 0.005 0.945 0.000
4-way Interactions
Critical Causes  Multiple Outcomes  Negative Effects  Contingencies 0.034 0.190 0.664 0.001
Critical Causes  Multiple Outcomes  Negative Effects  Pre Subjective Model Quality 0.076 0.427 0.514 0.002
Critical Causes  Multiple Outcomes  Contingencies  Pre Subjective Model Quality 0.053 0.294 0.588 0.002
Critical Causes  Negative Effects  Contingencies  Pre Subjective Model Quality 0.425 2.371 0.125 0.012
Multiple Outcomes  Negative Effects  Contingencies  Pre Subjective Model Quality 0.019 0.108 0.743 0.001
5-way Interaction
Critical Causes  Multiple Outcomes  Negative Effects  Contingencies  Pre Subjective Model Quality 0.167 0.935 0.335 0.005

Note: 4 participants were excluded because they did not successfully complete the domain experience measure. MS ¼ mean square,
F(df) ¼ F-ratio(degrees of freedom), p ¼ significance level, g2 ¼ partial eta squared effect size.
130 HESTER ET AL.

training in use of negative causes (m ¼ 8.38, SE ¼ .457) true that both problems were drawn from the marketing
was beneficial relevant to all other conditions domain. Because prior research has indicated some
(m ¼ 6.48, SE ¼ .477). When, however, people had weak domain specificity in the strategies people employ in cre-
initial mental models providing all three forms of ative problem-solving (Baer, 2003; Mumford, Antes,
instruction resulted in particularly weak post-test mental Caughron, Connelly, & Beeler, 2010), some caution is
models (m ¼ 3.38, SE ¼ .513) relative to all other con- called for extending the findings obtained in this study
ditions (m ¼ 3.98, SE ¼ .474). Apparently, extensive to other domains where creative problem-solving is
training in causal analysis may not prove beneficial for required.
those who begin with weak mental models. It should also be noted that we have in this study pro-
vided training only with respect to four strategies
Subjective attributes. Table 5 presents the results involved in causal analysis—working with (a) critical
obtained when positive, subjective attributes of posttest causes, (b) negative effects, (c) multiple outcomes, and
mental models were treated as the dependent variable. (d) contingencies. Marcy and Mumford (2007), how-
As might be expected, domain exposure, or expertise, ever, identified some seven strategies that might contrib-
proved to be a significant covariate, F(1, 195) ¼ 10.24, ute to causal analysis and creative thought. As a result,
p  .01, with those having more expertise evidencing it is not clear whether the findings obtained in this effort
stronger posttest mental models. Moreover, a significant would necessarily hold if other strategies had been
main effect, F(1, 195) ¼ 54.35, p  .001, was obtained for examined.
Downloaded by [Colorado College] at 15:29 30 October 2014

evaluations of the quality of pretest mental models. Finally, it should be noted that the instructional mod-
Those people who evidenced stronger (m ¼ 3.08, ules were presented in a fixed sequence. Thus, instruc-
SE ¼ .043) pretest mental models produced stronger tional modules were dropped across conditions, but
posttest mental models than people evidencing weaker the order in which they appeared in a given condition
(m ¼ 2.61, SE ¼ .047) pretest mental models. was fixed. Fixing the order of presentation of the
Finally, a significant interaction, F(1, 195) ¼ 3.88, self-paced instructional modules is, of course, a useful
p  .05, was observed between training in the use of critical control. By the same token, however, this study does
causes, training in the use of negative causes and training not allow examination of different orderings of instruc-
in the use multiple outcomes. In keeping with our earlier tion. This limitation may be noteworthy if the effects of
observations, it was found that the quality of posttest training one strategy are contingent on earlier training
mental models, as assessed by their subjective attributes, of other strategies.
declined when use of multiple outcomes and negative Even bearing these limitations in mind, we believe
causes was trained with no training in critical causes that this study has some noteworthy implications both
(m ¼ 2.70, SE ¼ .086) relative to all other conditions for creative thinking in general and the use of mental
(m ¼ 2.86, SE ¼ .089). Thus, certain sets of strategy train- models in creative thought. Turning first to creative
ing, and potentially certain sequences of training for these thought as a general phenomenon, it was clear that per-
strategies, may be needed to result in mental models formance in producing high-quality, originality, and
evidencing strong subjective characteristics. elegant solutions on the posttest depended, and
depended strongly, on pretest performance. Thus, cre-
ative problem-solving is not inherently flexible but,
DISCUSSION instead, depends on prior development (Mumford,
2002). One implication of this finding is that creative
Before turning to the broader conclusions flowing from potential will depend on the long-term, systematic devel-
this study, certain limitations should be noted. To begin, opment of individuals (Feldhusen, 1998).
this study was based on a classic experimental paradigm One key to the development of creative potential is to
where the sample examined was university undergradu- provide people with greater knowledge or more expert-
ates. Although undergraduates are not unfamiliar with ise (Rich & Weisberg, 2004; Weisberg, 2006). However,
the demands of marketing tasks (Lonergan et al., it should be recognized that multiple, different types of
2004; Redmond et al., 1993), they cannot be assumed knowledge may be applied in creative problem-solving
to have the same level of expertise as professionals (Hunter et al., 2008). One of the knowledge structures
(Ericsson, 2004). As a result, the question remains as that might be critical to creative problem-solving are
to whether similar findings would be obtained in more the mental models people apply to understand the kind
experienced populations. of novel, ill-defined complex problems that call for cre-
Along related lines, it should be recognized that per- ative thought (Mumford & Gustafson, 2007). In fact,
formance was examined in only one domain. Although prior research (e.g., Carlson & Gorman, 1992; Mumford
multiple marketing problems, the footwear and soft et al., in press; Ward et al. 2004) indicated that mental
drink problems, were presented in this effort, it is also models may, in fact, represent one form of knowledge
MENTAL MODELS 131

influencing people’s prior performance on creative performance improved with these individuals producing
problem-solving tasks. And the results obtained in this solutions of comparable, or greater quality, originality,
study provide some further support for this proposition. and elegance to the untrained talented individuals.
By the same token, the findings obtained in this study These findings are noteworthy for three reasons.
indicate that the mental models people apply in solving First, they provide evidence supporting the third and
problems within a domain are relatively stable. Thus, fourth hypotheses underlying this study. Second, they
the quality of the pretest mental models, as assessed suggest that training in causal analysis strategies will
by either objective or subjective model attributes, was transfer to performance only for individuals who display
found to be a powerful influence on posttest mental some talent—although less talented individuals’ perfor-
models, and, thus, given the findings of Mumford et al. mance is not disrupted by such instruction. Third, rela-
(in press), subsequent generation of high quality, more tively extensive instruction in strategies, training a larger
original, and more elegant problem solutions. The stab- number of strategies for more time, may be required
ility of people’s mental models, in turn, suggests that before large gains in solutions quality, originality, and
training intended to enhance application of mental elegance are observed.
models in creative problem-solving will prove difficult. These observations about training bring us to the
Our intent in this study was, broadly speaking, to implications of this study for how such training should
show that training interventions intended to provide be conducted. The results obtained in the study indicate
people with more effective strategies for working with that solution quality improved when talented indivi-
Downloaded by [Colorado College] at 15:29 30 October 2014

mental models would contribute to both creative duals were given training in strategies for the use of criti-
problem-solving and the development of stronger men- cal causes and multiple outcomes; solution originality
tal models as a result of training. In our effort, we improve when talented individuals were given training
focused on strategies people might employ in causal in the use of critical causes and contingencies, and
analysis because prior studies have indicated that the elegance improved when training was provided in criti-
structure (Goldvarg & Johnson-Laird, 2001) and appli- cal causes vis-à-vis negative causes and contingencies.
cation (Rouse & Morris, 1986) of mental models This pattern of findings suggests that training intended
depends on the nature of the causal linkages formulated to improve people’s mental models for use in creative
among concepts. problem-solving should focus on critical causes and at
The results obtained in this study indicate that least one other strategy—recognizing that the other,
instruction in more effective strategies for the analysis additional, strategy, or strategies provided in training
of causes contributes to improvements in the production will differ if the goal is to increase the quality as opposed
of high quality, more original, and more elegant solu- to the originality as opposed to the elegance of people’s
tions to the kind of problems that call for creative creative problem solutions. In fact, the results obtained
thought. Moreover, at least in the case of positive, in examining the acquisition of positive objective and
objective features of the mental models people applied subjective attributes in people’s mental models also
in problem-solving, it appears that the instruction con- point to the value of this particular instructional
tributed to the acquisition of stronger mental models. strategy.
Thus, the findings obtained in this study provide some At a general level it is, perhaps, not especially surpris-
support for our first two hypotheses. ing that providing talented people with training in the
In this regard, however, it is important to bear in use of critical causes along with an additional strategy,
mind an important caveat on these general conclusions depending on the performance outcome being sought,
emerging from this study. Providing training in causal would prove beneficial in enhancing creative perfor-
analysis strategies had little effect on these individuals mance. Prior studies by Coll and Treagust (2001) and
who evidenced little potential for creative thought, as Jahn et al. (2007) indicated that when working with
indicated by their scores on the pretest. Thus, strategy mental models in problem-solving, people prefer to
training appeared to have little effect on the creative work with a limited number of critical causes. Creative
problem-solving of less talented individuals. problem-solving, however, may require that they work
When strategy training proved effective in enhancing with critical causes along with one, or more, other stra-
creative problem-solving, and, to some extent, people’s tegies in producing creative problem solutions—
mental models, was when the individual evidenced solutions of high quality, originality, and elegance.
adequate talent as indexed by scores on the pretest. Although this study has provided some initial clues as
Here, it was found that talented people who received to how one might go about developing people’s capacity
no training in causal analysis generally evidenced better for creative thought through the mental models they
performance than trained individuals, however, as tal- apply in problem-solving, it cannot be said that this
ented people were provided with more strategies, and study has provided a complete answer to this question.
a select combination of causal analysis strategies, their The effects of training other strategies must be
132 HESTER ET AL.

investigated and the parameters shaping effective edu- Day, E., Arthur, W., & Gettman, D. (2001). Knowledge structures and
cational interventions, such as time, instructional the acquisition of a complex skill. Journal of Applied Psychology, 86,
1022–1033.
method, and instructional technique, must be investi- Doyle, J. K., & Ford, D. N. (1998). Mental models concepts for system
gated. Nonetheless, we hope that the findings obtained dynamics research. System Dynamics Review, 14, 3–29.
in this effort provide an impetus for further research Eckert, C., & Stacey, M. (2003). Adaptation of sources of inspiration
along these lines. in knitwear design. Creativity Research Journal, 15, 355–384.
Ellis, S., & David, I. (2005). After-event reviews: Drawing lessons from
successful and failed experience. Journal of Applied Psychology, 90,
857–871.
REFERENCES Ericsson, K. A. (2004). Deliberate practice and the acquisition and
maintenance of expert performance in medicine and related
Acton, W. H., Johnson, P. J., & Goldsmith, T. E. (1994). Structural domains. Academic Medicine, 79, S70–S81.
knowledge assessment: Comparison of referent structures. Journal Ericsson, K. A., & Charness, N. (1994). Expert performance: Its struc-
of Educational Psychology, 86, 303–311. ture and acquisition. American Psychologist, 49, 725–747.
Al-Diban, S. (2008). Progress in the diagnostics of mental models. In Feldhusen, J. F. (1998). Programs for the gifted few or talent develop-
D. Ifenthaler, P. Pirnay-Dummer & J. M. Spector (Eds.), Under- ment for the many? Phi Delta Kappan, 79, 735–738.
standing models for learning and instruction. Essays in honor of Nor- Finke, R. A., Ward, T. B., & Smith, S. M. (1992). Creative cognition:
bert M. Seel (pp. 81–101). New York, NY: Springer. Theory, research, and applications. Cambridge, MA: MIT Press.
Anderson, H., Barker, P., & Chen, X. (2006). The cognitive structure of Gill, C. M., & Hodgkinson, G. P. (2007). Development and validation
scientific revolutions. New York, NY: Cambridge University Press. of the five-factor model questionnaire (FFMQ): An adjectival-based
Aspara, J., Olkkonen, R., Tikkanen, H., Moisander, J., & Parvinen, P. personality inventory for use in occupational settings. Personnel
Downloaded by [Colorado College] at 15:29 30 October 2014

(2008). A theory of affective self-affinity: Definitions and appli- Psychology, 60, 731–766.
cation to a company and its business. Academy of Marketing Science Goldvarg, E., & Johnson-Laird, P. N. (2001). Naı̈ve causality: A
Review, 12, 1–37. mental model theory of causal meaning and reasoning. Cognitive
Baer, J. (2003). Evaluative thinking, creativity, and task specificity: Science, 25, 565–610.
Separating wheat from chaff is not the same as finding needle in Grimsley, G., Ruch, F. L., Warren, N. P., & Ford, J. S. (1985). Manual
haystacks. In M. A. Runco (Ed.), Critical creative processes (pp. for the employee attitude survey, test of verbal reasoning. Glendale,
129–152). Cresskill, NJ: Hampton. CA: Psychological Services.
Basadur, M., Runco, M. A., & Vega, L. A. (2000). Understanding how Guilford, J. P. (1967). The nature of human intelligence. New York,
creative thinking skills, attitudes, and behaviors work together: A NY: McGraw-Hill.
causal process model. Journal of Creative Behavior, 34, 77–100. He, W., Erdelez, S., Wang, F. K., & Shyu, C. R. (2008). The effects of
Baughman, W. A., & Mumford, M. D. (1995). Process analytic models conceptual description and search practice on users’ mental models
of creative capacities: Operations involved in the combination and and information seeking in a case-based reasoning retrieval system.
reorganization process. Creativity Research Journal, 8, 37–62. Information Processing and Management, 44, 294–301.
Bennett, P. D. (1995). Dictionary of marketing terms (2nd ed.). Chi- Hmelo-Silver, C. E., & Pfeffer, M. G. (2004). Comparing expert
cago, IL: American Marketing Association. and novice understanding of a complex system from the perspec-
Besemer, S. P., & O’Quin, K. (1999). Confirming the three-factor tive of structures, behaviors, and functions. Cognitive Science,
creative product analysis matrix model in an American sample. 28, 127–138.
Creativity Research Journal, 12, 287–296. Hogarth, R. M. (1980). Judgment and choice. New York, NY: Wiley.
Byrne, C. L., Shipman, A. S., & Mumford, M. D. (2010). The effects of Hunter, S. T., Bedell-Avers, K. E., Ligon, G. S., Hunsicker, C. M., &
forecasting on creative problem-solving: An experimental study. Mumford, M. D. (2008). Applying multiple knowledge structures in
Creativity Research Journal, 22, 119–138. creative thought: Effects on idea generation and problem-solving.
Cacioppo, J. T., & Petty, R. E. (1982). The need for cognition. Journal Creativity Research Journal, 20, 137–154.
of Personality and Social Psychology, 42, 116–131. Jahn, G., Knauff, M., & Johnson-Laird, P. (2007). Preferred mental
Carlson, W. B., & Gorman, M. E. (1992). A cognitive framework to models in reasoning about spatial relations. Memory and Cognition,
understand technological creativity: Bell, Edison, and the Tele- 35, 2075–2087.
phone. In R. J. Weber & D. N. Perkins (Eds.), Inventive minds: Jaussi, K., Randel, A., & Dionne, S. (2007). I am, I think I can, and I
Creativity in technology (pp. 48–79). New York, NY: Oxford do: The role of personal identity, self-efficacy, and cross-application
University Press. of experiences in creativity at work. Creativity Research Journal, 19,
Ceci, S. J., & Papierno, P. B. (2005). The rhetoric and reality of gaps 247–258.
closing—When the ‘‘have-nots’’ gain but the ‘‘haves’’ gain even Kieras, D. E., & Bovair, S. (1984). The role of a mental model in learn-
more. American Psychologist, 60, 149–160. ing to operate a device. Cognitive Science, 8, 255–273.
Christiaans, H. H. C. M. (2002). Creativity as a design criterion. Lonergan, D. C., Scott, G. M., & Mumford, M. D. (2004). Evaluative
Creativity Research Journal, 14, 41–54. aspects of creative thought: Effects of idea appraisal and revision
Clemente, M. N. (1992). The marketing glossary. New York, NY: standards. Creativity Research Journal, 16, 231–246.
AMACOM. Marcy, R. T., & Mumford, M. D. (2007). Social innovation: Enhanc-
Coll, R. K., & Treagust, D. F. (2001). Learners’ mental models of ing creative performance through casual analysis. Creativity
chemical bonding. Humanities, Social Sciences and Law, 31, Research Journal, 19, 123–140.
357–382. Marcy, R. T., & Mumford, M. D. (2010). Leader cognition: Improving
Crouch, T. D. (1992). Why Wilbur and Orville? Some thoughts on the leader performance through causal analysis. Leadership Quarterly,
Wright brothers and the process of invention. In R. J. Weber & D. 21, 1–19.
N. Perkins (Eds.), Inventive minds: Creativity in technology: McCrae, R. R., & Ingraham, L. J. (1987). Creativity, divergent think-
Creativity in technology (pp. 80–92). New York, NY: Oxford ing, and openness to experience. Journal of Personality and Social
University Press. Psychology, 52, 1258–1265.
MENTAL MODELS 133

Meldrum, M., & McDonald, M. (2007). Marketing in a nutshell: Key Peters, M., Laeng, B., Latham, K., Jackson, M., Zaiyouna, R., &
concepts for non-specialists. Burlington, MA: Elsevier. Richardson, C. (1995). A redrawn Vandenberg & Kuse mental rota-
Merrifield, P. R., Guilford, J. P., Christensen, P. R., & Frick, J. W. tions test: Different versions and factors that affect performance.
(1962). The role of intellectual factors in problem solving. Brain and Cognition, 28, 39–58.
Psychological Monographs, 76, 1–21. Redmond, M. R., Mumford, M. D., & Teach, R. J. (1993). Putting crea-
Mumford, M. D. (2001). Something old, something new: Revisiting tivity to work: Leader influences on subordinate creativity. Organiza-
Guilford’s conception of creative problem solving. Creativity tional Behavior and Human Decision Processes, 55, 120–151.
Research Journal, 13, 267–276. Rich, J. D., & Weisberg, R. W. (2004). Creating all in the family: A
Mumford, M. D. (2002). Social innovation: Ten cases from Benjamin case study in creative thinking. Creativity Research Journal, 16,
Franklin. Creativity Research Journal, 14, 253–266. 247–260.
Mumford, M. D., Antes, A. L., Caughron, J. J., Connelly, M. S., & Rouse, W. B., & Morris, N. M. (1986). On looking into the black box:
Beeler, C. (2010). Cross-field differences in creative problem-solving Prospects and limits in the search for mental models. Psychological
skills: A comparison of health, biological, and social services. Crea- Bulletin, 100, 349–363.
tivity Research Journal, 22, 14–26. Ruch, F. L., & Ruch, W. W. (1980). Employee aptitude survey.
Mumford, M. D., Feldman, J. M., Hein, M. B., & Nago, D. J. (2001). Los Angeles, CA: Psychological Services.
Tradeoffs between ideas and structure: Individual versus group Scott, G. M., Leritz, L. E., & Mumford, M. D. (2004a). The effective-
performance in creative-problem solving. Journal of Creative Beha- ness of creativity training: A meta-analysis. Creativity Research
vior, 35, 1–23. Journal, 16, 361–388.
Mumford, M. D., Friedrich, T. L., Caughron, J. J., & Byrne, C. E. Scott, G. M., Leritz, L. E., & Mumford, M. D. (2004b). Types of
(2007). Leader cognition in real-world settings: How do leaders creativity training: Approaches and their effectiveness. Journal of
think about crises? Leadership Quarterly, 18, 515–543. Creative Behavior, 38, 311–342.
Mumford, M. D., & Gustafson, S. B. (2007). Creative thought: Scott, G. M., Lonergan, D. C., & Mumford, M. D. (2005). Conceptual
Downloaded by [Colorado College] at 15:29 30 October 2014

Cognition and problem solving in a dynamic system. In M. A. combination: Alternative knowledge structures, alternative heuris-
Runco (Ed.), Creativity Research Handbook: Vol. 2 (pp. 33–77). tics. Creativity Research Journal, 16, 121–148.
Cresskill, NJ: Hampton. Shah, J. J., Vargas-Hernandez, N., Summers, J. D., & Kulkarni,
Mumford, M. D., Hester, K. S., Robledo, I. C., Peterson, D. R., Day, S. (2001). Collaborative sketching (C-Sketch)—An idea generation
E. A., Hougen, D. P., & Barrett, J. D. (in press). Mental models and technique for engineering design. Journal of Creative Behavior, 35,
creative problem-solving: The relationship of objective and subjec- 168–198.
tive model attributes. Creativity Research Journal. Vincent, A. S., Decker, B. D., & Mumford, M. D. (2002). Divergent
Mumford, M. D., Marks, M. A., Connelly, M. S., Zaccaro, S. J., & thinking, intelligence, and expertise: A test of alternative models.
Johnson, J. F. (1998). Domain based scoring of divergent thinking Creativity Research Journal, 14, 163–178.
tests: Validation evidence in an occupational sample. Creativity Ward, T. B., Patterson, M. J., & Sifonis, C. M. (2004). The role of
Research Journal, 11, 151–164. specificity and abstraction in creative idea generation. Creativity
Mumford, M. D., & Stokes, G. S. (1992). Development determinants Research Journal, 16, 1–9.
of individual action: Theory and practice in applying background Weick, K. E. (1995). Sensemaking in organizations. Thousand Oaks,
data measures. In M. D. Dunnette & L. M. Hough (Eds.), Hand- CA: Sage.
book of Industrial and Organizational Psychology (2nd ed., pp. Weisberg, R. W. (2006). Modes of expertise in creative thinking: Evi-
61–138). Palo Alto, CA: Consulting Psychologists Press. dence from case studies. In K. A. Ericsson, N. Charness,
Osburn, H. K., & Mumford, M. D. (2006). Creativity and planning: P. J. Feltovich & R. R. Hoffman (Eds.), The Cambridge handbook
Training interventions to develop creative problem-solving skills. of expertise and expert performance (pp. 761–787). New York,
Creativity Research Journal, 18, 173–190. NY: Cambridge University Press.

You might also like