KURAM VE UYGULAMADA EĞİTİM BİLİMLERİ EDUCATIONAL SCIENCES: THEORY & PRACTICE

Received: January 20, 2016

Revision received: July 7, 2016 Copyright © 2017 EDAM
Accepted: October 26, 2016 www.estp.com.tr
OnlineFirst: February 10, 2017 DOI 10.12738/estp.2017.2.0051  April 2017  17(2)  515–528

Research Article

An Investigation of the Factor Structure of the

Torrance Tests of Creative Thinking*

William M. Bart1 Brad Hokanson2

University of Minnesota University of Minnesota
Iclal Can3, 4
Middle East Technical University Northern Cyprus Campus

Abstract
This study investigated the factorial structure of the Torrance Tests of Creative Thinking (TTCT) Figural
Form A with an aim to find out whether the two-factor structure of creative thinking established by Kim and
Kim, Cramond, and Bandalos holds true for the older participants. Data were gathered from 996 8th grade
students and 748 11th grade students from a suburban public school system in Minnesota. Based on previous
research and Kirton’s Adaption-Innovation (KAI) theory, one and two-factor models were tested in the study.
Upon checking the related assumptions, confirmatory factor analyses with Maximum Likelihood (ML)
estimation were conducted using LISREL 8.71. The results indicated that a two-factor model established
by Kim and her colleagues was a better fit than a one-factor model. This shows that fluency and originality
load onto the latent variable termed the innovative factor, elaboration and abstractness of titles load onto the
latent variable termed the adaptive factor, and resistance to premature closure loads on both latent variables.
Our results suggest that any efforts to assess and promote creativity should take into consideration the two
domains of creativity posited in KAI theory.

Keywords
TTCT • Creative thinking • Creativity • Factorial structure • Adaption-Innovation theory

*
This research was supported in part by a grant from the Metropolitan Research Grant Program of the
University Metropolitan Consortium and the University of Minnesota Center for Urban and Regional Affairs.
The authors express thanks to the School District personnel, who helped in the implementation of the research and in the
collection of the data.
1 Correspondence to: William M. Bart (PhD), Department of Educational Psychology, University of Minnesota; 162 Education
Sciences Building, 56 East River Road, Minneapolis, MN 55113, USA. Email: [email protected]
2 Department of Design, Housing, and Apparel, University of Minnesota; 32 McNeal Hall, 1985 Buford Ave., St. Paul, MN
55108-6136. Email: [email protected]
3 Guidance and Psychological Counseling Program, Middle East Technical University Northern Cyprus Campus; Academic
Buildings, R-138, Middle East Technical University Northern Cyprus Campus, 99738 Kalkanlı, Güzelyurt, Mersin 10,
Turkey. Email: [email protected]
4 Previously Iclal Sahin
Citation: Bart, W. M., Hokanson, B., & Can, I. (2017). An investigation of the factor structure of the Torrance Tests of
Creative Thinking. Educational Sciences: Theory & Practice, 17, 515–528. http://dx.doi.org/10.12738/estp.2017.2.0051
EDUCATIONAL SCIENCES: THEORY & PRACTICE

Measuring creative thinking abilities of individuals is an increasingly important

area in education. The Torrance Tests of Creative Thinking (TTCT) are “the most
commonly used test of divergent thinking” to serve this purpose (Plucker & Renzulli,
1999, p. 39). Devised by Torrance and his colleagues (Torrance, 2008), the TTCT has
more than 35 language translations available (Millar, 2002) and are “still enormously
popular” (Runco & Acar, 2012, p. 67). The TTCT has attracted the attention of many
researchers investigating its psychometric features and focusing on the structure
of creativity (e.g., Clapham, 1998; Heausler & Thompson, 1988; Hocevar, 1979;
Kim, 2006a; Kim, Cramond, & Bandalos, 2006). However, the latent structure of
creativity is still controversial. Although some researchers argue that creativity is
uni-dimensional, some other researchers suggest that it has more than a single factor
structure. This research served as an investigation of whether the two-factor structure
established by Kim (2006a) and Kim et al. (2006) holds true for seemingly two new
population groups: 8th and 11th graders.

Being “the most widely used divergent thinking tests” (Clapham, 2004, p. 829),
the major purpose of the TTCT was to individualize instruction (Torrance, 1966,
1974). However, it is mostly used to identify gifted children (Kim, 2006b), although
it is also used for research and instructional planning (Kim, 2006b). The TTCT is
originally “based in part on Guilford’s Structure of Intellect model” (Krumm, Lemos,
& Filippetti, 2014, p. 72). Four dimensions of the TTCT, namely, fluency, originality,
elaboration, and flexibility, were adopted from Guilford’s divergent thinking factors
(Kim, 2006a). Although the TTCT is based on Guilford’s model, it also differs from
Guilford’s test batteries as “Guilford’s test tasks attempt to elicit as factorially pure
mental functioning as possible” (Torrance, 2008, p.47). In addition, it differs from the
test battery of Wallach and Kogan as “the Wallach and Kogan activities all attempt to
elicit associations” (Torrance, 2008, p. 47).

The TTCT has six versions. After being published in 1966, it was re-normed in
1974, 1984, 1990, 1998, and 2008. The re-norming process affected the scoring
procedures. However, the content of the TTCT did not go through any alteration
(Kim, 2006a, 2006b, 2011). The earliest two versions of the TTCT assessed fluency,
originality, elaboration, and flexibility.

In the 1984 version of the TTCT, flexibility was excluded from the analyses;
whereas, abstractness of titles and resistance to premature closure were included in
the TTCT (Hébert, Cramond, Neumeister, Millar, & Silvian, 2002; Torrance & Ball,
1984). The reason why Torrance excluded flexibility from the analysis was that there
was high correlation between fluency and flexibility (Hébert et al., 2002). A new
measure termed creative strengths was included in the scoring in 1984 as well. This
measure included thirteen criterion-referenced measures (indicators).

516
Bart, Hokanson, Can / An Investigation of the Factor Structure of the Torrance Tests of Creative Thinking

The TTCT battery has two different forms: the TTCT Figural and Verbal, each
having two parallel forms (Torrance, 1974, 1990, 1998, 2008). “With both measures,
the instructions are designed to motivate the respondents to give unusual, detailed
responses” (Cramond, Matthews-Morgan, Bandalos, & Zuo, 2005, p. 284). The
Figural TTCT has a wide range of application ranging from measuring creative
thinking skills among young children to identifying creative adults. It includes
three picture-based activities: picture construction, picture completion, and repeated
figures of lines or circles. This test lasts 30 minutes with each activity being allocated
10 minutes (Torrance, 1990, 1998, 2008).

Conceptual Framework
Many researchers have studied the structure of creativity using the TTCT (e.g.,
Clapham, 1998; Heausler & Thompson, 1988; Hocevar, 1979; Kim, 2006a; Kim
et al., 2006; Krumm et al., 2014). Many studies were also conducted using other
tests of creative thinking or test batteries from different creative thinking tests (e.g.,
Bachelor, 1986-1987; Michael & Bachelor, 1990; Richardson, 1986; Runco & Mraz,
1992). These studies have raised questions about the dimensionality of creativity.

Several researchers have contended that creativity is uni-dimensional and that

it has a single factor structure (e.g., Clapham, 1998; Heausler & Thompson, 1988;
Hocevar, 1979; Runco & Mraz, 1992). For example, Clapham (1998) investigated
the structure of TTCT Figural forms A and B among 334 university students. The
principal component analyses revealed that there was one general factor.

Similarly, Heausler and Thompson (1988) collected data from 69 kindergarten

and 63 second graders using the TTCT Figural form A. Their results suggested a
general creativity factor, too. In their study, Heausler and Thompson (1988) wrote
that “A fundamental problem with the derivation of the five separate scores is that the
scores are derived from the same response data; this introduces potentially spuriously
high subscale correlation” (p. 466), suggesting that the TTCT has one single factor
structure. However, Torrance (1974), himself, suggests that use of a composite score
should be avoided as each subscale in the TTCT measures a different area although
a composite score “provided a good measure of an individual’s creative energy”
(Torrance, as cited in Cramond et al., 2005, p. 284).

Many researchers have not accepted the idea that creativity is uni-dimensional.
Kim (2006a) argues that “Because Guilford (1959; 1962) viewed divergent thinking
as multidimensional, many researchers have come to the conclusion that creativity
consists of several psychological factors” (p. 252). The idea that creativity is not uni-
dimensional is also evident in Kirton’s Adaption-Innovation (KAI) theory (1976) in
which he introduced two cognitive styles to creative thinking, namely, adaptive and

517
EDUCATIONAL SCIENCES: THEORY & PRACTICE

innovative. Kirton (1976) introduced the view that creative thinking is a continuum
ranging from being adaptive to being innovative, and individuals may be classified as
adaptors or as innovators in terms of their preferred approaches to solving problems.
An individual is adaptive to the extent to which the individual prefers to engage
in activities in a better way. An individual is innovative to the extent to which the
individual prefers to engage in activities differently. Kirton (1976) posits that “... the
more the structure surrounding a problem is incorporated within and treated as part of
the problem, the more any solution is likely to be radical and innovative (i.e., “doing
things differently”). The less structure is challenged, the more any solution is likely
to be adaptive (i.e., “doing things better”)” (p. 622). Kirton (1976) developed an
inventory to locate the individuals on the adaptor and innovator continuum. Kirton’s
Adaption-Innovation (KAI) theory has informed subsequent research on creative
thinking (e.g., Kim, 2006a; Kim et al., 2006; Krumm et al., 2014).

Kim (2006a) analyzed creative thinking test score data from a sample of
approximately 500 6th grade students. The test whose data she analyzed was data for
the TTCT - Figural Form A. Using confirmatory factor analysis (CFA), Kim (2006a)
determined that the best-fitting model was a two-factor model compatible with
Kirton’s theory of cognitive styles to creativity. That model hypothesized that the
fluency and originality subscales loaded onto a factor that she termed the “innovative”
factor, that the elaboration and abstractness of titles subscales loaded onto a factor
that she termed the “adaptive” factor, and that the resistance to premature closure
subscale loaded on both factors.

Kim (2006a) offered the following statement as the basis for that third aspect of
her hypothesized two-factor model: “The double loading by resistance to premature
closure is consistent with Torrance’s (1984; 1990; 1998) theory that creative people
would be able to keep their mind open long enough to make mental leaps, whereas
less creative individuals tend to prematurely leap to conclusions.” (p. 253). The two-
factor model without the creative strengths subscale being included provided a better
fit than the two-factor model with the creative strengths subscale being included. That
result led Kim (2006a) to infer that the creative strengths subscale loads on a separate
third factor.

In another study conducted by Kim et al. (2006) based on KAI theory, TTCT-
Figural data from 1,000 kindergarten students, 1,000 third graders, and 1,000 sixth
graders were analyzed to test the factorial structure of the TTCT. Similar to what
Kim (2006a) found with 6th graders, the results revealed a two-factor solution for the
TTCT, fluency and originality loading on an innovative factor and elaboration and
abstractness of titles loading on an adaptive factor. Resistance to premature closure
had a double loading on both factors.

518
Bart, Hokanson, Can / An Investigation of the Factor Structure of the Torrance Tests of Creative Thinking

In a recent study, Krumm et al. (2014) used CFA to test four theoretical models
regarding the factorial structure of creativity. They used Figural form B data collected
from 577 Spanish-speaking children (331 girls and 246 boys) between 9 and 14 years
of age in Argentina. Krumm et al. (2014) found the existence of two general factors,
innovative and adaptive. Similarly, the CFA revealed that the model that best fit the
data was the one without creative strengths in which fluency and originality loaded on
the innovative factor and elaboration and abstractness of titles loaded on the adaptive
factor. However, the findings of Krumm et al.’s study were not entirely consistent
with the findings of Kim (2006a) and Kim et al. (2006), as Krumm et al.’s (2014)
study indicated that resistance to premature closure did not load on the innovative
factor, but only on the adaptive factor.

Overall, these studies highlight the need for investigating the factorial structure of
the TTCT in new population groups as the populations are changing and thus earlier
results may not apply to contemporary samples. The major purpose of this study is to
investigate the factor structure of creative thinking as measured by the TTCT with the
goal being to determine if the two-factor structure established by Kim (2006a) and
Kim et al. (2006) holds true for two completely different population groups (8th and
11th graders). Apart from this, as the status of premature closure seems unresolved
in previous studies, we believe that our study will provide new insights on the role
of premature closure in the structure of creativity as assessed by the TTCT in older
participants. The research question is as follows: Does the two-factor structure of
creative thinking as established by Kim (2006a) and Kim et al. (2006) hold true for
8th and 11th graders?

As a replication of the studies reported by Kim (2006a) and Kim et al. (2006), the
present study proposes a two-factor model for the latent structure of the TTCT. Figure
1 depicts the two-factor model for both grades. As demonstrated in the hypothesized
model, fluency and originality load onto the latent variable termed the innovative
factor; while, elaboration and abstractness of titles load onto the latent variable
termed the adaptive factor. The hypothesized model further proposes that resistance
to premature closure loads on both latent variables.

519
EDUCATIONAL SCIENCES: THEORY & PRACTICE

Fluency    

 
Originality   Innovative  
 

Resistance  to  
Premature  Closure  

 
Elaboration   Adaptive  
 

Abstractness  of  
Titles  

Figure 1. Hypothesized two-factor model in the study.

Method
This study was conducted as part of a comprehensive project on creativity and
scholastic achievement in 8th and 11th graders, and funded by the University
of Minnesota. Using the same data, in a previous study, we explored the gender
differences among student creative thinking skills (Bart, Hokanson, & Sahin, 2015).
In the present study, however, we focused on the factorial structure of the TTCT and
primarily tested the tenability of the two-factor structure offered by Kim (2006a) and
Kim et al. (2006). We primarily aimed to find out if the two-factor structure would
appear in the same form among 8th and 11th graders.

Participants
996 8th graders (503 boys and 493 girls) and 748 11th graders (407 boys and 341
girls) from a suburban public school district in Minnesota participated in the study.
The 8th graders had a Mage = 14.11 years with SDage= .34 years and the 11th graders
had a Mage = 17.32 years with SDage= .35 years. Descriptive information for boys and
girls separately can be found in our previous study (Bart et al., 2015).

Instrument
The instrument used to assess creative thinking is the Torrance Tests of Creative
Thinking (TTCT), Figural Form A. “The figural form, especially, has equity benefits in
terms of gender and race and for persons who have various language, socioeconomic
status, and cultural backgrounds” (Cramond; Torrance, as cited in Kim, 2006a, p.

520
Bart, Hokanson, Can / An Investigation of the Factor Structure of the Torrance Tests of Creative Thinking

251). With regard to the reliability of the TTCT, Kuder-Richardson 21 “reliability

coefficients involving the creativity index are centered at .90 for the various grades,
and at .89 for various ages. Coefficients for the average standard score, primarily an
intermediate measure involved in developing the creativity index, are only slightly
lower” (Torrance, 2008, p. 44).

With regard to the inter-rater reliability of the TTCT, the results of five rater-
reliability studies conducted with 2nd, 5th, and 8th graders as well as college level
students and a random group of students ranged between .90 and .99 for five subscales
except the college level student group which had a coefficient of .78 for the resistance
to premature closure subscale (Torrance, 2008). As for the validity of the TTCT, the
test has relatively strong predictive validity as can be seen in Torrance’s follow up
study which was conducted 40 years after the TTCT was first administered to the
participants (Cramond et al., 2005).

The Figural TTCT provides norm-referenced scores for five subtests: fluency (“the
total number of relevant responses”), originality (“the statistical infrequency and
unusualness of the response”), abstractness of titles (“the subject’s synthesizing and
organizing processes of thinking”), elaboration (“the imagination and exposition of
detail”), and resistance to premature closure (“ability to keep open and delay closure
long enough to make the mental leap that makes possible original ideas”) (Torrance,
2008, p. 39). The TTCT also provides criterion-referenced scores for the creative
strengths subscale, scores that were not included in our analyses for the present study.
The participants in this study completed the TCTT-Figural, Form A with five creative
thinking subtest scores being recorded for each participant.

Procedure
The researchers worked in collaboration with the cooperating school district
during the data collection process. The district office used a purposive sampling
method in the selection of participants so that the sample of participants would reflect
the diversity of the school district student population in terms of social class, ethnic
group, and gender. After receiving human subjects board review approval for the
study, the parents were notified that the district office collected the data for research
purposes. Upon receiving parental assent, the teachers who were trained by one of the
researchers administered the data collection tools: TTCT and the general demographic
questionnaire to the students in groups. The test responses of the students were sent
to the publisher of the TTCT for scoring. After the scoring process, the district office
matched the student creativity scores with student information, and deleted personal
data to ensure the confidentiality of the data.

521
EDUCATIONAL SCIENCES: THEORY & PRACTICE

Data Analysis
The raw subscale scores were used in this study as the use of the standard scores
produced quite similar results. Table 1 presents the means and standard deviation
scores of the creativity subscales for the 8th and 11th grade students.
Table 1
Descriptive Statistics
Subscale Grade M SD
8th 24.02 9.41
Fluency
11th 20.73 9.19
8th 16.66 7.03
Originality
11th 14.82 6.94
8th 7.75 2.72
Elaboration
11th 7.16 2.46
8th 6.18 4.02
Abstractness of Titles
11th 6.31 3.95
8th 13.41 4.02
Resistance to Premature Closure
11th 12.89 4.33

Data analysis was conducted in two successive steps: (a) preliminary analyses and
(b) confirmatory factor analyses. Preliminary analyses entailed conducting missing
data analysis, screening for outliers, and checking univariate and multivariate
normality, as maximum likelihood (ML) estimation assumes adequate sample size
and univariate and multivariate normality (Brown, 2006). These assumptions were
tested using SPSS 20.

First of all, the data were screened for missing data, and no missing data was
identified. Next, the univariate outliers were detected through the stem-and-leaf plots
and box plots (Parke, 2013) of SPSS. The stem and leaf plots were used to calculate
the extreme values that will be used to mark outliers in each subscale data and the
boxplots were used to detect the outliers that have these extreme values with their
case numbers. The outliers detected in this way were eliminated from the data sets
as even a small number of outliers in a large data can affect statistical results for a
data set (Thompson, 2004). The skewness and kurtosis values for each creativity
subscale were assessed and they were all close to zero for the 8th and 11th grades
data. This indicated that the univariate normality assumption was met. To test
multivariate normality, Mahalanobis distance (D2) for each individual in the data sets
was calculated (Thompson, 2004), and no value in the probability distribution of
Mahalanobis distance exceeded .0001. Thus, no multivariate outliers were detected.

Results
Confirmatory factor analyses with ML estimation were performed using LISREL
8.71 (Jöreskog & Sörbom, 2004) to test the fit of the one-factor and two-factor models
for the factor structure of the TTCT. Consistent with the preliminary studies of Kim

522
Bart, Hokanson, Can / An Investigation of the Factor Structure of the Torrance Tests of Creative Thinking

(2006a) and Kim et al. (2006) and with Kirton’s theory of two cognitive styles to
creativity, confirmatory factor analyses with two-factors were conducted with five
observed indicators. In the hypothesized two-factor model, two indicators, fluency
and originality, load onto the latent variable termed the innovative factor; two other
indicators, elaboration and abstractness of titles load onto the latent variable termed
the adaptive factor, and the last indicator, resistance to premature closure, loads on
both latent variables. Confirmatory factor analysis with one factor was also performed
based on the previous findings suggesting that creativity is uni-dimensional (e.g.,
Clapham, 1998; Hocevar, 1979). Apart from this, with an aim to resolve the status of
premature closure in the previous studies (e.g., Krumm et al. 2014), confirmatory factor
analyses with two-factors in which premature closure loading onto either on adaptive
or innovative latent variable were also conducted. To enable the readers to compare
the findings of the present study with those of Kim (2006a) and Kim et al. (2006), the
results, tables, and figures were presented in a similar fashion with these two studies.

Table 2 presents the results of Pearson product-moment correlation coefficients

between the variables. The correlation coefficients between the subscales of creativity
were significant with p < .01. The highest correlation was between originality and
fluency for both grades (8th grades = .78; 11th grades = .79).

Table 2
Correlations between Creativity Subscales for 8th and 11th Grades
Resistance
Abstractness
Subscale Grade Fluency Originality Elaboration to Premature
of Titles
Closure
8th
Fluency
11th -
8th .78**
Originality
11th .79** -
8th .31** .42**
Elaboration
11th .34** .44** -
Abstractness of 8th .09** .22** .44**
Titles 11th .15** .21** .44** -
Resistance to 8th .42** .39** .31** .29**
Premature Closure 11th .50** .43** .29** .31** -
** p < .01.

The model fit was assessed based on the two-index strategy of Hu and Bentler (1999).
In order to discriminate good models from poor ones, Hu and Bentler (1999) suggest
using the ML-based standardized root mean square residual (SRMR) and supporting it
with an index such as the incremental fit index (IFI), the comparative fit index (CFI),
the non-normed fit index (NNFI), and the root mean square error of approximation
(RMSEA). For this purpose, the goodness of fit was evaluated using root SRMR, IFI,
CFI, NNFI, and RMSEA using the cut off scores of SRMR (≤ 0.8), IFI (≥ .95), CFI (≥
.95), NNFI (≥ .95), and RMSEA (≤ .06), as suggested by Hu and Bentler (1999). We
used SRMR as the major index and supplemented it with IFI and CFI.

523
EDUCATIONAL SCIENCES: THEORY & PRACTICE

The chi-square difference between the one-factor and two-factor models was also
evaluated. The chi-square differences between the one-factor model and two-factor
model with premature closure loading onto both latent variables were significant for
both 8th and 11th grades. This indicated that the two-factor model hypothesized in the
present study fit the data better than the one-factor model (See Table 3). In addition,
the goodness-of-fit indices except for NNFI and RMSEA were within the cut-off
values for both the 8th grade (χ2(3) = 76.03, SRMR = .04, IFI = .95, CFI = .95, NNFI
= .82, RMSEA = .16) and the 11th grade (χ2(3) = 69.84, SRMR = .04, IFI = .95, CFI
= .95, NNFI = .82, RMSEA = .18). As Table 3 indicates, SRMR is supplemented by
two incremental fit indices (IFI and CFI). This shows that the 8th and 11th grade data
fit the two-factor model hypothesized in the study, but the one-factor model failed to
fit the 8th and 11th grade data, as all model fit indices indicated poor fit.

As for the two-factor models with resistance to premature closure loading onto
either adaptive or innovative factor, resistance to premature closure loading onto
innovative factor did not converge in both 8th and 11th grade data. In addition,
resistance to premature closure loading onto adaptive factor did not converge in 11th
grade data while it converged in 8th grade data. As Table 3 shows, the chi-square
differences between (a) the one-factor model and the two-factor model with premature
closure loading onto adaptive factor and (b) the two-factor model hypothesized in the
study and two-factor model with premature closure loading onto adaptive factor in
8th grade data were significant. Although the chi-square differences may indicate that
two-factor model in which premature closure loading onto adaptive factor seemed
to fit the 8th grade data better than the one-factor model, it failed to fit the 8th grade
data when compared to the two-factor model hypothesized in the present study. This
is further supported by all goodness-of-fit indices except for SRMR, which were not
within the cut-off values for the 8th grade data.

Thus, the results suggest that the two-factor model hypothesized in the study fits
both 8th and 11th grade data and that fluency and originality loaded on the latent
variable termed the innovative factor and elaboration and abstractness of titles
loaded on the latent variable termed the adaptive factor. The results also indicate that
resistance to premature closure loaded on both latent variables.

Parameter estimates were also investigated to examine the two-factor model

hypothesized in the study. The parameter estimates indicated that the indicators were
highly related to their latent variables (range of R2 = .68 to .90 for 8th grades and R2 =
.69 to .79 for 11th grades), except for abstractness of titles and resistance to premature
closure, which had low R2 values (R2 = .28 and .23 for the 8th grade students and
R2 = .28 and .29 for the 11th grade students respectively). These values suggest that
fluency, originality, and elaboration are better indicators of their latent factors than
abstractness of titles and resistance to premature closure, which had low R2 values.

524
Bart, Hokanson, Can / An Investigation of the Factor Structure of the Torrance Tests of Creative Thinking

Table 3
Results of Model Comparison with One and Two Factors
Grades Number of Factors c2 df NNFI IFI CFI RMSEA SRMR ∆χ2
One 264.82 5 .66 .83 .83 .24 .11
Twoa 121.743 4 .81 .92 .92 .178 .074 143.077c**
8
188.79c**
Two b
76.03 3 .82 .95 .95 .16 .04
45.713d**
One 188.53 5 .70 .85 .85 .23 .10
11
Two 69.84 3 .82 .95 .95 .18 .04 118.69c**
Note. NNFI = Non-normed fit index; IFI = incremental fit index; CFI = comparative fit index;
RMSEA = root mean square error of approximation; SRMR = standardized root mean square
residual
a
Two-factor model with resistance to premature closure loading onto adaptive factor
b
Two-factor model with resistance to premature closure loading onto both factors (hypothesized
model)
c
Comparison to one-factor model
d
Comparison to two-factor model with resistance to premature closure loading onto adaptive factor
** p < .01.

Discussion
This study offers support for the idea that creativity is two-dimensional and
contributes to our understanding of the latent structure of creativity as assessed by
the TTCT, suggesting that the two-factor structure exists in the same structure from
elementary to high school. Although the results of the present study differ from some
published studies which suggest that creativity is uni-dimensional (e.g., Clapham,
1998; Heausler & Thompson, 1988; Hocevar, 1979; Runco & Mraz, 1992), they are
consistent with those of Kim (2006a) and Kim et al. (2006), who obtained a two-
factor solution for creativity as measured by the TTCT. In accordance with the present
results, Kim (2006a) and Kim et al. (2006) specified a two-factor model in which
fluency and originality loaded on an innovative factor, elaboration and abstractness
of titles loaded on an adaptive factor, and resistance to premature closure loaded on
both factors.

The present findings seem to be consistent with Krumm et al.’s (2014) research,
which found that creativity as assessed by the TTCT had a two-factor structure
(innovative and adaptive). However, the results of the present study also differ from
those reported by Krumm et al. (2014), as the present findings indicated that resistance
to premature closure loads onto both factors; whereas, Krumm et al (2014) found that
only the adaptive factor is loaded by resistance to premature closure. This study also
contributes to our understanding of creativity by providing evidence that there is a
high correlation between fluency and originality. This finding is in agreement with
what Chase (1985), Dixon (1979), Heausler and Thompson (1988), Kim (2006a), and
Kim et al., (2006) found. Torrance himself reported a high correlation among these

525
EDUCATIONAL SCIENCES: THEORY & PRACTICE

subscales. Although the high correlation between fluency and originality is considered
an indication of uni-dimensionality by some researchers (e.g., Chase, 1985; Heausler
& Thompson, 1988), it seems that these two measures assess two different abilities
and should be interpreted accordingly as two major indices of the innovative factor.

This study has some limitations. First, creativity subscale scores of the 8th and 11th
graders in a large public school district in Minnesota were used in the study. As a result,
this study may not reflect the structure of the TTCT in all grade levels or similar grades
in different educational contexts. Secondly, R2 values pertaining to the abstractness of
titles and resistance to premature closure in both the 8th and 11th grade data indicated
that they are weak indicators of the adaptive factor on which they load compared to
fluency, originality, and elaboration. Thus, the results pertaining to these subscales
should be interpreted accordingly. Next, some extraneous variables like student
motivation (Torrance, 1974) may have affected student responses to the test items.
Apart from the aforementioned limitations of the study, one should also note that the
scoring of the TTCT could be regarded as a limitation as well as “Reliable and valid
scoring requires extensive experience” and “This frequently necessitates expensive and
time-consuming professional scoring” (Clapham, 2004, p. 829).

In conclusion, this study suggests that any efforts to assess and promote creativity
should take into consideration two domains of creativity, that is, adaption and
innovation, to obtain a realistic assessment and effective promotion of creative
thinking abilities of individuals. Our results further suggest that two-factor structure
hypothesized in the study may be incorporated into TTCT’s scoring and norms.

References
Bachelor, P. (1986-1987). Higher-order factors in Guilford’s SOI model: Divergent productions.
Educational Research Quarterly, 11(2), 29–40.
Bart, W. M., Hokanson, B., & Sahin, I. (2015). An investigation of the gender differences in creative
thinking abilities among 8th and 11th grade students. Thinking Skills and Creativity, 17, 17–24.
http://dx.doi.org/10.1016/j.tsc.2015.03.003
Brown, T. A. (2006). Confirmatory factor analysis for applied research. New York, NY: Guilford
Press.
Chase, C. I. (1985). Review of the Torrance Tests of Creative Thinking. In J. V. Mitchell, Jr.
(Ed.), The ninth mental measurements yearbook (pp. 1631–1632). Lincoln, NE: University of
Nebraska, Buros Institute of Mental Measurements.
Clapham, M. M. (1998). Structure of Figural Forms A and B of the Torrance Tests of Creative
Thinking. Educational & Psychological Measurement, 58, 275–283. http://dx.doi.org/10.1177/
0013164498058002010
Clapham, M. M. (2004). The convergent validity of the Torrance Tests of Creative Thinking and
creativity interest inventories. Educational & Psychological Measurement, 64, 828–841. http://
dx.doi.org/10.1177/0013164404263883

526
Bart, Hokanson, Can / An Investigation of the Factor Structure of the Torrance Tests of Creative Thinking

Cramond, B., Matthews-Morgan, J., Bandalos, D., & Zuo, L. (2005). A report on the 40-year follow-
up of the Torrance tests of creative thinking: Alive and well in the new millennium. Gifted Child
Quarterly, 49, 283–291. http://dx.doi.org/10.1080/10400419.2011.627805
Dixon, J. (1979). Quality versus quantity: The need to control for the fluency factor in originality
scores from the Torrance tests. Journal for the Education of the Gifted, 2, 70–79.
Heausler, N. L., & Thompson, B. (1988). Structure of the Torrance Tests of Creative Thinking.
Educational and Psychological Measurement, 48, 463–468. http://dx.doi.org/10.1177/
0013164488482021
Hébert, T. P., Cramond, B., Neumeister, K. L. S., Millar, G., & Silvian, A. F. (2002). E. Paul
Torrance: His life, accomplishments, and legacy. Storrs: The University of Connecticut, The
National Research Center on the Gifted and Talented (NRC/GT).
Hocevar, D. (1979). The unidimensional nature of creative thinking in fifth grade children. Child
Study Journal, 9, 273–278.
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indices in covariance structure analysis:
Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55. http://
dx.doi.org/10.1080/10705519909540118
Jöreskog, K. G., & Sörbom, D. (2004). LISREL 8.71 for Windows [Computer software].
Lincolnwood, IL: Scientific Software International, Inc.
Kim, K. H. (2006a). Is creativity unidimensional or multidimensional? Analyses of the Torrance
Tests of Creative Thinking. Creativity Research Journal, 18, 251–259. http://dx.doi.org/10.1207/
s15326934crj1803_2
Kim, K. H. (2006b). Can we trust creativity tests? A review of the Torrance Tests of Creative Thinking
(TTCT). Creativity Research Journal, 18, 3–14. http://dx.doi.org/10.1207/s15326934crj1801_2
Kim, K. H., Cramond, B., & Bandalos, D. L. (2006). The latent structure and measurement
invariance of scores on the Torrance Tests of Creative Thinking-Figural. Educational and
Psychological Measurement, 66, 459–477. http://dx.doi.org/10.1177/0013164405282456
Kim, K. H. (2011). The creativity crisis: The decrease in creative thinking scores on the Torrance
Tests of Creative Thinking. Creativity Research Journal, 23, 285–295. http://dx.doi.org/10.108
0/10400419.2011.627805
Kirton, M. J. (1976). Adaptors and innovators: A description and measure. Journal of Applied
Psychology, 61, 622–629.
Krumm, G., Lemos, V., & Filippetti, V. A. (2014). Factor structure of the Torrance tests of creative
thinking figural Form B in Spanish-speaking children: Measurement invariance across gender.
Creativity Research Journal, 26, 72–81. http://dx.doi.org/10.1080/10400419.2013.843908
Michael, W. B., & Bachelor, P. (1990). Higher-order structure-of-intellect creativity factors in
divergent production tests: A re-analysis of a Guilford data base. Creativity Research Journal, 3,
58–74. http://dx.doi.org/10.1080/10400419009534333
Millar, G. W. (2002). The Torrance kids at mid-life: Selected case studies of creative behavior.
Westport, CT: Ablex.
Parke, C. (2013). Essential first steps to data analysis: Scenario-based examples using SPSS.
California, CA: Sage.
Plucker, J. A., & Renzulli, J. S. (1999). Psychometric approaches to the study of human creativity.
In R. J. Sternberg (Ed.), Handbook of creativity (pp. 35–61). Cambridge, UK: Cambridge
University Press.

527
EDUCATIONAL SCIENCES: THEORY & PRACTICE

Richardson, A. G. (1986). Two factors of creativity. Perceptual and Motor Skills, 63, 379–384.
Runco, M. A., & Acar, S. (2012). Divergent thinking as an indicator of creative potential. Creativity
Research Journal, 24, 66–75. http://dx.doi.org/10.1080/10400419.2012.652929
Runco, M. A., & Mraz, W. (1992). Scoring divergent thinking tests using total ideational output
and a creativity index. Educational and Psychological Measurement, 52, 213–221. http://dx.doi.
org/10.1177/001316449205200126
Thompson, B. (2004). Exploratory and confirmatory factor analysis: Understanding concepts and
applications. Washington, DC: American Psychological Association.
Torrance, E. P. (1966). The Torrance tests of creative thinking: Norms-technical manual (Research
ed.). Princeton, NJ: Personnel Press.
Torrance, E. P. (1974). The Torrance tests of creative thinking: Norms-technical manual (Research
ed.). Princeton, NJ: Personnel Press.
Torrance, E. P. (1990). The Torrance tests of creative thinking: Norms-technical manual figural
(streamlined) forms A & B. Bensenville, IL: Scholastic Testing Service.
Torrance, E. P. (1998). The Torrance tests of creative thinking: Norms-technical manual figural
(streamlined) forms A & B. Bensenville, IL: Scholastic Testing Service.
Torrance, E. P. (2008). The Torrance tests of creative thinking: Norms-technical manual figural
(streamlined) forms A & B. Bensenville, IL: Scholastic Testing Service.
Torrance, E. P., & Ball, O. E. (1984). The Torrance tests of creative thinking: Streamlined scoring
guide Figural A and B. Bensenville, IL: Scholastic Testing Service.

528
Copyright of Educational Sciences: Theory & Practice is the property of EDAM- Education
Consultancy Limited and its content may not be copied or emailed to multiple sites or posted
to a listserv without the copyright holder's express written permission. However, users may
print, download, or email articles for individual use.