4 0 - Y E A R F O L L O W- U P O F T H E T O R R A N C E T E S T S

A Report on the 40-Year Follow-Up

of the Torrance Tests of Creative Thinking:
Alive and Well in the New Millennium
Bonnie Cramond, Juanita Matthews-Morgan, Li Zuo
and Deborah Bandalos United Nations
University of Georgia

A B S T R A C T P U T T I N G T H E
R E S E A R C H T O U S E
This article updates information about the Torrance
Tests of Creative Thinking (TTCT) by reporting on This study should give individuals who are inter-
predictive validity data from the most recent data ested in using the Torrance Tests of Creative
collection point in Torrance’s longitudinal studies. Thinking some confidence that the tests can be good
First, we outline the background of the tests and predictors of adult creativity. So, why should educa-
changes in scoring over the years. Then, we detail tors consider using the TTCT? One reason is to find
the results of the analyses of the 40-year follow-up creative strengths that may be hidden in children
on the TTCT resulting in a structural equation who are seen as behavior problems. According to
model, which demonstrates the validity of the Torrance’s observations, discovering and nurturing
TTCT for predicting creative achievement 40 years creativity in children may redirect some children’s
after its administration. Finally, we provide a ration- misbehavior toward more positive pursuits and away
ale for the relevance of the test in schools today. from diagnoses of psychological disturbance.
Another reason that the TTCT are especially use-
ful to educators today is as one of the multiple crite-
Although their creator, E. Paul Torrance, passed away
ria to identify gifts and talents in children. The
in 2003, his eponymous tests of creative thinking are still
recent and growing emphasis on identifying and
very much in use around the world. Most educators in
nurturing talents in youth has created a need for dis-
the fields of gifted education and creativity are aware of
covering a broad array of talents using a variety of
these tests, but many people have outdated information measures. The TTCT are especially useful for iden-
about them. This article will update information about tifying more students from underserved populations,
the Torrance Tests of Creative Thinking (TTCT) by because they are not as culturally loaded as other
reporting on the background of the tests, changes in scor- standardized assessments that might be used
ing, and the existence of a new abbreviated form of the (Torrance, 1971, 1977). The TTCT figural, which
tests for adults, and by providing quantitative results of requires very little language, is especially helpful with
the latest data point in the 40+ years of study1, as well as a the increasing numbers of children with limited or
rationale and suggestions for their use in schools today. no proficiency in English.
Whether for identification, curriculum develop-
Background and Scoring ment, or both, the TTCT can provide some infor-
C h a n g e s i n t h e To r r a n c e mation about students that other assessments do not
Te s t s o f C r e a t i v e T h i n k i n g readily measure. Because the items are open-ended,
students are also able to express their interests, fears,
The TTCT battery is composed of verbal and figural hopes, knowledge about diverse topics, and emo-
tests. The verbal test is comprised of five activities, and tional states. The TTCT provide teachers with addi-
the figural is comprised of three activities. Responses to tional information about how their students think
the verbal and figural tests of the TTCT are not only through an activity that most children enjoy.
expressed in two different modalities—written or oral

G I F T E D C H I L D Q U A R T E R LY • F A L L 2 0 0 5 • V O L 4 9 N O 4 2 8 3
4 0 - Y E A R F O L L O W- U P O F T H E T O R R A N C E T E S T S

responses for the verbal, and drawn responses on the fig- a person’s ability to stay open and tolerate ambiguity long
ural—but they are also measures of different creative abil- enough to come up with a creative response. When
ities. In fact, Torrance (1990) found very little correlation added to the other three norm-referenced scores—
(r = .06) between performance on the verbal and figural Fluency, Originality, and Elaboration—there are now
tests. five norm-referenced scores. The 13 criterion-refer-
Furthermore, there are two forms of each test, A and
enced measures of creative strength, derived from years of
B, and each activity on the tests is based on research link-
creativity testing and research, are: emotional expressive-
ing the required ability to creativity (Torrance, 1966,
1974). With both measures, the instructions are designed ness, storytelling articulateness, movement or action,
to motivate the respondents to give unusual, detailed expressiveness of titles, synthesis of incomplete figures,
responses. synthesis of lines or circles, unusual visualization, internal
The items on the test have been the same for many visualization, extending or breaking boundaries, humor,
years with the exception of the removal of one of the ver- richness of imagery, colorfulness of imagery, and fantasy.
bal tasks when it was not shown to provide additional Because of the difficulty of getting interrater reliabil-
meaningful information on the respondent’s creative ity for untrained scorers on Elaboration for the verbal
abilities. (Activities 1–5 and 7 of the original verbal tasks tests, the scoring was simplified to include only Fluency,
remain: Asking Questions, Guessing Causes, Guessing Flexibility, and Originality (Torrance, 1966, 1974).
Consequences, Product Improvement, Unusual Uses, Torrance intended to include a list of criterion-refer-
and Just Suppose. Activity 6, Unusual Questions, was enced measures of creative strength to the verbal scoring,
removed.) However, the scoring procedures have gone but did not get to do so before he died. He did develop a
through several iterations as Torrance attempted to find a list of creative strengths for the Demonstrator Form of
scoring procedure that could be taught easily to others the TTCT (Torrance, 1979, 1980), a shortened form of
and still produce valid, reliable results. For example, at the battery that combines elements of the figural and ver-
one point he adopted the U.S. Patent Office criteria for bal tests. He developed this Demonstrator Form, with
rating new inventions as a scoring scheme for the TTCT two verbal activities and two figural activities, as an
(Torrance, 1959). It proved to have very good predictive example of the kinds of activities that are found on the
validity, but it was too difficult and time consuming for actual TTCT. It was printed in the book, The Search for
the scorers, so it was eschewed (E. P. Torrance, personal Satori and Creativity (1979), so that readers could experi-
communication, September 26, 1991). For many years, ence the test activities without taking the actual test.
Torrance based his scoring of the TTCT on Guilford’s The combined form of verbal and figural activities was
(1956) four divergent thinking factors: (a) f luency—the later reconceptualized as the Abbreviated Torrance Test
number of relevant responses; (b) f lexibility—the num- for Adults (ATTA; Goff & Torrance, 2002). Although it
ber of different categories or shifts in responses; (c) orig- does not yet have the history and research trail of the
inality—the number of unusual yet relevant ideas as complete battery, it is suitable as a screening instrument
determined by statistical infrequency; and (d) elabora- and has great promise for increased use because it can be
tion—the number of details used to extend a response administered in 15 minutes. There is nothing about the
(Torrance, 1966, 1974, 1990). activities themselves that would preclude the ATTA’s use
Yet, he became dissatisfied with these scoring criteria with children; however, additional research must be done
for several reasons. On the figural test, he was concerned to establish its validity and reliability with that group.
about the high correlation between Flexibility and As a final note on scoring, Torrance maintained that
Fluency scores, and he determined that there were addi- the composite score was not the most useful way to look
tional creative attributes that individuals demonstrated on at a person’s creative functioning because he knew it
the tests that were not being measured (Torrance, 1979, could mask individual strengths; however, he conceded
1981b). So, in 1984, he designed the streamlined scoring that it provided a good measure of an individual’s creative
system, which includes five norm-referenced scores and energy (Torrance, 1966, 1974). Other ways of looking at
13 criterion-referenced scores (Ball & Torrance, 1984; the scores—according to individual dimensions, by inter-
Torrance, Ball, & Safter, 1992). He deleted Flexibility actions, as profiles, and through specific responses to cer-
from the scoring because of overlap with Fluency, and tain stimuli—can often provide much more information
added two new norm-referenced scores: Abstractness of about creative functioning than the composite scores.
Titles, as a verbal measure on the figural tests, and These methods are discussed in more detail elsewhere
Resistance to Premature Closure, as a gestalt measure of (Cramond, 1999).

2 8 4 G I F T E D C H I L D Q U A R T E R LY • F A L L 2 0 0 5 • V O L 4 9 N O 4
 4 0 - Y E A R F O L L O W- U P O F T H E T O R R A N C E T E S T S

O v e r 4 0 Ye a r s o f R e s e a r c h — With the second high school follow-up, 5 years later,

Longitudinal Studies Torrance sent the questionnaire to the entire 1959
University of Minnesota High School population. This
There have been four major points of data collection time there were enough respondents to break down the
in the longitudinal studies of the TTCT. There were information by gender. All of the creativity predictors
some preliminary studies done with smaller samples were found to be significant at the .01 level. By combin-
within a shorter interim of time (c.f. Cropley, 1971; ing the scores on the creativity test battery administered
Torrance, Tan, & Allman, 1970; Witt, 1971), but the in 1959 into a total creativity score to predict the com-
longitudinal studies conducted with the students who bined creativity criteria derived in 1971, a canonical cor-
had been in two elementary schools and a high school in relation of .51 was obtained for the full sample. When
Minneapolis in the late 1950s form the basis of the pre- looking at the genders separately, a canonical correlation
dictive validity evidence for the TTCT (Torrance 1969, of .59 was obtained for males and one of .46 for females.
1972a, 1972b, 1980, 1981a). This data collection point illustrated once again that cre-
From 1958 through 1964, all pupils enrolled in grades ativity measures administered in high school were able to
1–6 in two elementary schools were administered various accurately predict individuals’ adult creative accomplish-
batteries of the TTCT, standardized intelligence and ments on three dimensions: quantity of creative achieve-
achievement tests, and sociometric questionnaires each ments, quality of creative achievements, and creative
year. Beginning in September of 1959, all students aspirations or motivation (Torrance, 1972a, 1972b).
enrolled in grades 7–12 of the University of Minnesota Although these results were encouraging, Torrance
High School also completed the tests. The results of these faced the more daunting challenge of following up on the
tests became the predictor variables in the follow-up elementary school students. This task was more difficult
studies. Then, the high school respondents were followed for several reasons. First, because there was a much longer
up after 7 years (Torrance, 1969), and again after 12 years period of time between the testing and the questionnaire,
(Torrance, 1972a, 1972b), with a questionnaire. The he knew that the predictive ability of the variables would
questionnaire asked for biographical and demographic tend to decrease. Second, the additional time, as well as
information, as well as information about the subject’s Torrance’s move from Minneapolis to Georgia, exacer-
most creative achievement, aspirations, and a checklist of bated the problem of finding the participants. Third, the
creative accomplishments, which included type and follow-up of the elementary school students was more
degree of achievement in the arts, research, work innova- difficult because much more data had been collected
tions, inventions, and change in life philosophy from the elementary school students over several years, so
(Torrance, 1969). The two follow-ups of the elementary there was a more complex predictor set. Finally, there is
students were conducted 22 years after the initial testing, the complex nature of the construct itself and its unsub-
in 1980 (Torrance, 1980, 1981a), and again 40 years after stantiated developmental path.
the testing, in 1998. Detailed information on the first Torrance (1981a) realized that establishing adequate
three studies, the two high school follow-ups and the first predictive validity for tests of creative thinking ability
elementary school follow-up, has been presented else- presented particular difficulties. During the initial 6-year
where (Torrance, 1969, 1972a, 1972b, 1980, 1981a) and period of testing in Minnesota elementary schools, he
is also collectively reported (Cramond, 1994). had seen many of the children in his sample “surrender”
In summary, from the first follow-up of the high their creativity at an early age. He feared that a “fourth
school students who had been seniors in 1959, Torrance grade slump” (Torrance, 1966, 1968) was indicative of
found that three of the component scores of the Torrance this trend, which would destroy the best predictions.
Tests of Creative Thinking (Fluency, Flexibility, and Even reporting that he had found a “fourth grade slump”
Originality, but not Elaboration) were better predictors created an additional problem for him when it was noted
of creative achievement 7 years later than were intelli- by Senator William Proxmire as an example of wasteful
gence, high school achievement, or peer nominations. spending of taxpayer money; the resulting controversy
When the four TTCT scores were entered into a step- made it more difficult for Torrance to fund his longitudi-
wise regression equation, they accounted for 25% of the nal studies2 (Torrance, 1990).
variance noted in creative quality, 21% of the variance in Torrance realized that the best measure of creative
creative quantity, and 26% of the variance in creative abilities would not measure all creative abilities, much less
motivation (Torrance, 1969). such important considerations as motivation, opportu-

G I F T E D C H I L D Q U A R T E R LY • F A L L 2 0 0 5 • V O L 4 9 N O 4 2 8 5
4 0 - Y E A R F O L L O W- U P O F T H E T O R R A N C E T E S T S

nity, and social inf luences, as well as physical, emotional, The 40-Year Follow-Up
and experiential factors that can impact the expression of
creativity. For example, several of Torrance’s initial par- Participants
ticipants were later conscripted into service during the
Vietnam War. For those who survived, the impact on the In 1998, 40 years after the initial testing, Torrance ini-
nature and timeline of their creative development cannot tiated another data collection point based on the same
be measured. Also, limited opportunities for creative design as the last one with a few modifications. This
expression may explain the lower predictive validity of time, 99 of the 170 respondents who were located
the TTCT for females and for children from the housing returned the questionnaire: 45 males and 55 females.
projects. This is approximately one fourth of the original eligible
group of participants, and approximately one half of the
In both schools many of the children lived in federal number that responded in 1980.
housing developments and were from definitely
disadvantaged backgrounds. They were difficult to Analyses
locate 22 years later. We did locate enough of them
to get some notion of how poverty inf luenced the There were eight predictor variables in the analyses
chances of creative achievement. They just don’t this time. The predictors included: (1) IQ, (2) f luency,
have a chance. (E. P. Torrance, personal communi- (3) f lexibility, (4) originality, (5) elaboration, (6) the same
cation, October 29, 1990) index of creativity that was derived for the elementary
sample based on 3 years of TTCT testing, (7) whether
In order to calculate the predictive validity of the cre-
respondents had a mentor in 1980, and (8) whether
ativity tests, a creativity index was calculated for the
respondents had a mentor in 1998. The predictor vari-
respondents using data collected over 3 years of creativity
ables were from IQ tests (#1), the TTCT (#2, #3, #4, and
testing. Then, questionnaires were sent to all of the par-
#5), 3 years of TTCT testing as in the earlier elementary
ticipants to elicit biographical and demographic informa-
studies (#6), and from the questionnaire (#7 and #8). The
tion similar to that requested in the follow-up of the high
criterion variables were the quantity and quality of cre-
school students (Torrance, 1972a, 1972b), as well as
ative achievements reported on the follow-up question-
“Creative Style of Life Achievements,” achievements that
naire.
are not ordinarily publicly recognized and acknowledged,
Two criteria of creative achievement were derived
such as organizing an action-oriented group (Torrance,
from the questionnaire responses: (1) quantity of publicly
1981a).
recognized creative achievements, and (2) quality of pub-
For this follow-up, five indices of creative achieve- lic achievements. Because the respondents were now
ment were derived by the expert judges from the ques- mature adults, it didn’t seem suitable to measure their
tionnaire responses and used as criteria: (1) number of aspirations, so the third criterion from the earlier studies
high school creative achievements, (2) number of post- was deleted. As in the 1980 study, three expert judges
high-school creative achievements, (3) number of “cre- rated the quantity and quality of creative achievements.
ative style of life” achievements, (4) quality of highest The 1998 judges’ interrater reliabilities were obtained
creative achievements, and (5) creativeness of future using Pearson product-moment correlations resulting in
career image. coefficients ranging from .78 to .88.
To determine how well the creativity measures pre- A Pearson product-moment correlation was used to
dicted future achievements, Pearson product-moment examine the relationship between the predictors obtained
correlation coefficients were calculated between the cre- in childhood (IQ, Fluency, Flexibility, Originality,
ativity index derived from elementary school assessments Elaboration, and the creative index), two predictors
and each of the five indices of creative achievement derived from the follow-up questionnaires (having a
obtained 22 years later. All correlation coefficients were mentor in 1980 and having a mentor in 1998), and the
significant at the .001 level (see Torrance, 1981a, p. 60). criteria of adult creative achievement obtained 40 years
A multiple correlation coefficient of .63 was obtained for later (quantity and quality). Another Pearson product-
the five criteria entered into a stepwise multiple regres- moment correlation was used to analyze the relationships
sion equation (Torrance). of the predictors to the criterion variables by gender.

2 8 6 G I F T E D C H I L D Q U A R T E R LY • F A L L 2 0 0 5 • V O L 4 9 N O 4
 4 0 - Y E A R F O L L O W- U P O F T H E T O R R A N C E T E S T S

In order to examine the differences between the most Ta b l e 1

and least creative of the respondents, individuals were
Product-Moment Correlation Coefficients
assigned to the high creative group if their number of Between Predictors of Creativity Obtained in
public achievements was greater than or equal to 61 (n = 1958–64 and Criteria of Creative Achievement
27, 31% of the sample), and to the low creative group if Obtained in 1998
their number of public achievements was less than or
equal to 22 (n = 26, 31% of the sample). Additional analy- Criteria
ses were then conducted on this sample of 53 respon-
dents. A chi-square analysis was used to determine if Predictors Quantity Quality
there were any differences in the number of males and
females in the two groups. Then, the grouping variable IQ .30* .32**
was used in a two-way ANOVA to determine if there was Fluency .23* .30
a difference in quality ratings for the genders in the two Flexibility .17 .35**
quantity groups. Originality .24* .39**
Finally, in order to examine the relationships of the Elaboration .09 .14
various variables more clearly, structural equation model- CIQ .27 .43**
ing was employed. Plucker (1999) had analyzed
Mentor ’80 .29** .39**
Torrance’s 1980 follow-up data using structural equation
Mentor ’98 .26* .26*
modeling and found a significant effect of the verbal
TTCT scores and IQ on adult creative achievement. The Quality .85**
intention to replicate the Plucker model on the 1998 fol- *
p ≤ .05, **p ≤ .01
low-up had to undergo some modifications due to the
following difficulties. Results
First, the 1998 data had a smaller sample size (N = 80
for this analysis) than the 1980 data because of the attri- The results of a Pearson product-moment correlation
tion typical in longitudinal studies. The small sample size between the predictors obtained in childhood and the
created a problem for a stable estimation of parameters in criteria of creative achievement are illustrated in Table 1.
a structural equation model. In order to ameliorate the IQ, Fluency, and Originality scores obtained in child-
sample size problem, the Plucker model had to be simpli- hood were the best test predictors of quantity of creative
fied by reducing the number of variables entered. One achievement 40 years later. The best test predictors of
way to reduce the number of variables was to concentrate quality of creative achievements were IQ, Flexibility,
on only one form of the TTCT, Verbal or Figural. As Originality, and the creativity index. From the question-
Plucker had already derived a model using the scores naires, the predictors of having had a mentor, in 1980 and
from the Verbal TTCT, it was decided to use only the 1998 respectively, were correlated with quantity and
Figural scores. Therefore, the Verbal TTCT scores of quality of creative achievements. Quality and quantity
Fluency, Flexibility, and Originality, which composed were highly correlated with each other.
the latent factor of Verbal TTCT in Plucker’s model, As indicated in Table 2, when the data were analyzed
were left out. by gender, IQ was still a better predictor of adult creative
Second, the Figural TTCT in its older form yielded achievements for females than for males, as was found in
four scores: Fluency, Flexibility, Originality, and 1980. Flexibility, Originality, and the creativity index
Elaboration. In fitting the model to the 1980 follow-up were better predictors of quality of creative achievements
data, Flexibility and Elaboration were pruned out (Plucker, than of quantity for both genders.
1999). Nevertheless, we included all four scores in the Having had a mentor in 1980 was significantly related
model fitting in order to preserve the integrity of the to females’ quantity and quality of creative production,
TTCT and to have more indicators for the latent factor. but not to males’. Having had a mentor in 1998 was
Third, considering that the latent factor of IQ had related to both males’ and females’ quality of creative
only a single indicator and that IQ is known to have a achievement, but not to the quantity. As might be
high reliability but is not free from measurement error, expected, quantity and quality of creative production
we specified a 10% measurement error in the IQ vari- were highly correlated for both males and females. This
able. provides further evidence of the link between quantity of

G I F T E D C H I L D Q U A R T E R LY • F A L L 2 0 0 5 • V O L 4 9 N O 4 2 8 7
 4 0 - Y E A R F O L L O W- U P O F T H E T O R R A N C E T E S T S

Ta b l e 2
Product-Moment Correlation Coefficients by Gender Between Predictors of Creativity Obtained in 1958–64
and Criteria of Creative Achievement Obtained in 1998

Criteria

Males Females

Predictors Quantity Quality Quantity Quality

IQ .15 .22 .38* .43**

Fluency .16 .30 .30 .35*
Flexibility .20 .37* .16 .35*
Originality .20 .36* .29 .45**
Elaboration -.02 .13 .22 .20
CIQ .33 .51** .26 .43**
Mentor ’80 .17 .30 .41** .50**
Mentor ’98 .27 .36* .25 .40**
Quality .90** .81**
*
p ≤ .05, **p ≤ .01

Figure 1. Structural equation model

ideas and the production of quality ideas (c.f. Runco, 1. This model specified that intelligence as indicated by IQ
1990). scores should have an effect on the TTCT, indicated by
The chi-square analysis of gender by high vs. low fluency, flexibility, originality, and elaboration, and also
quantity of public achievements was not significant. have an effect on creative achievement, indicated by expert
There were no differences in the proportion of males and judges’ mean ratings on the creativeness and importance of
females in the two quantity groups. A two-way ANOVA participants’ creative products. Moreover, the model speci-
of quality, by gender by group (high vs. low creative), was fied that performance on the TTCT should have an effect
statistically significant (F [1,49] = 9.53, p < .01). The on creative achievement, too. This path from TTCT to
quality ratings for both males and females were signifi- creative achievement is considered essential because the
cantly higher for the high quantity group. purpose of Torrance’s longitudinal studies is to test the
The structural equation model is diagrammed in Figure predictive ability of the TTCT.

2 8 8 G I F T E D C H I L D Q U A R T E R LY • F A L L 2 0 0 5 • V O L 4 9 N O 4
 4 0 - Y E A R F O L L O W- U P O F T H E T O R R A N C E T E S T S

Ta b l e 3
Parameter estimate values, standard errors, and values
Parameter Estimate, Standard Error, and R2 Values
for Model A of R2 are shown in Table 3. All hypothesized paths were
significant and in the expected direction. No out of range
Parameter Parameter Standard R2 parameter estimates or inf lated standard errors were
Estimate Error noted. As can be seen in the table, the scales measuring
f luency and f lexibility resulted in the highest values of R2
Fluency ‘ TTCT 1.00a —— .63 for the TTCT, while quality of creative achievements
Flexibility ‘ TTCT 1.05 .16 .58 yielded the highest value for the creative achievement
Originality ‘ TTCT 1.00 .17 .46 scale. Although intelligence was a significant predictor of
Elaboration ‘ TTCT 0.79 .17 .30 TTCT scores, it explained very little of the variance in
Quality ‘ Creative Ach 1.00 —— .87 these scores (R2 = .085). However, the combination of
Quantity ‘ Creative Ach 1.70 .44 .31
TTCT and intelligence explained 54% of the variance in
Intelligence ‘ TTCT 0.10 .043 .085
creative achievement.
TTCT ‘ Creative Ach 0.71 .14
Intelligence ‘ Creative Ach 0.06 .04 .54b
a
Value fixed for identification purposes. S u m m a r y o f t h e 4 0 - Ye a r
b
This value represents the amount of variance explained by both the TTCT and
intelligence scales.
Study
A difference between the 1980 and 1998 follow-ups The predictive validity of the Torrance Tests of Creative
was the evaluation of creative achievement. In 1980, it Thinking, Verbal and Figural, is relatively strong consider-
was evaluated by the participants themselves, who ing the span of 40 years during which time data were col-
selected three of their most creative products, which were lected. The various studies demonstrated that the creativity
then rated by expert judges on their creative quality. In index, a combined TTCT score, was predictive of future
1998, it was the expert judges who rated the creativeness creative production, as were the dimension scores when
and importance of participants’ creative achievement analyzed separately. The fact that TTCT scores could
based on the information participants provided on the explain 23% of the variance in creative production is
questionnaires. The univariate distributions of mean rat- important in view of the length of the study and the fact
ings from three expert judges on the creativeness and that other important variables such as motivation and
importance of participants’ creative products (labeled as opportunity mediate creative production.
Overall, the data support the value of intelligence and
M Creativeness and M Importance) deviated slightly
creativity measures taken in childhood for predicting adult
from normality (kurtosis = -1.095, skewness = .10 for M
productive success. However, it is very clear that there are
Creativeness; and kurtosis = .912, skewness = .093 for M
many additional factors that can help or hinder this success.
Importance), but transformation was not considered to
As in Terman’s (Terman & Oden, 1959) and Bloom’s
be necessary (Wang, Fan, & Willson, 1996).
(1985) studies with highly intelligent and talented individ-
The model presented provided a good fit to the data:
uals, Torrance’s longitudinal studies point out the critical
χ212 = 17.07 (p = .15; RMSEA [Root Mean Square Error of
role that personalogical factors, chance experiences, and
Approximation] = .088; CFI [Comparative Fit Index] =
other environmental situations play in the development of
.98; SRMR [Standardized Root Mean squared Residual] = the productive individual. This further validates the sys-
.05). Although the value of the RMSEA was slightly tems emphasis for the study of creativity (Feldman,
higher than that recommended by Hu and Bentler (1999), Csikszentmihalyi, & Gardner, 1994) of which creative
all other fit indices indicated an adequate fit. Therefore, aptitude is but one, albeit critical, part of the formula.
although modification indices indicated that the model
could be improved by allowing a correlation between the
residuals of f luency and flexibility, this parameter was not Relevance to the Schools
added. However, this suggested correlation does indicate of This Millennium
that the two scales share more variance than can be
accounted for by the fact that they are part of the same fac- The work on measuring creative potential that was
tor. This finding lends some support to Torrance’s argu- begun in 1958 (Torrance, 1990; Torrance & DeYoung,
ment that these two scales should be highly correlated. 1958) has direct ramifications for the classrooms of today.

G I F T E D C H I L D Q U A R T E R LY • F A L L 2 0 0 5 • V O L 4 9 N O 4 2 8 9
4 0 - Y E A R F O L L O W- U P O F T H E T O R R A N C E T E S T S

That is because Torrance’s initial interest in creativity test- schools, but could only hope that the schools would
ing derived from his observation that children seen as someday use measures of creativity as another way of
behavior problems were often also very creative. This finding gifts in children who did not necessarily score
observation was strengthened when he studied the survival highly on IQ tests. The demographics of the U.S. schools
training of air force pilots; he observed that the risk-taking and society were so different in the 1950s, he would not
fighter pilots were also very creative (Torrance, 1954). have anticipated the numbers of ESOL children in the
However, in the late 1950s, there were fewer psychological schools of today. However, he knew that children of all
classifications for children whose behaviors did not fit the colors and nationalities can demonstrate creative think-
classroom norm; some would argue that there were fewer ing, and he created a test to measure that ability in them.
children who misbehaved. In any case, when Torrance The reasons for using the TTCT as stated in 1966 are
began work on his tests of creativity, the now common clas- as relevant today, or more so, as they were then: namely,
sification of Attention Deficit Hyperactivity Disorder did
not exist. Until about 1960, the syndrome, when diag- (a) to promote understanding of the human mind,
nosed, was known as minimal brain dysfunction (Barkley, its functioning and development; (b) to assist in the
1990). By 1991, ADHD was listed as the most common development of individualized instruction; (c) to
reason for referral and diagnosis in children seen in psycho- provide additional information for remedial and
logical clinics (Frick & Lahey, 1991). Currently, Asperger’s psychotherapeutic programs; (d) to assess the differ-
Syndrome, an autism-type syndrome, is gaining as a diag- ential effects of educational materials, programs,
nosis for bright children who don’t fit in socially (Boyle, curricula, procedures, and so on; and (e) to point
2003; Gallagher & Gallagher, 2002; Neihart, 2000). out potentialities that might otherwise go unno-
When Torrance began his work on creativity testing in ticed—especially in children from culturally diverse
1958, he could have no way of knowing that the schools and lower socioeconomic backgrounds. (Torrance,
of the next millennium would be beset with multiple 1966, p. 6)
diagnoses for the “wild colts” children (Torrance, 1990,
p. 2) who interested him. He knew they were creative,
and he set about to design an instrument to measure that. References
Perhaps, now, with the proliferation of psycho-educa-
tional syndromes in the schools, it is especially important Ball, O. E., & Torrance, E. P. (1984). Torrance Tests of Creative
to be able to discern aberrant behavior that is indicative of Thinking: Streamlined scoring guide Figural A and B.
high creativity rather than psychological disturbance. Bensenville, IL: Scholastic Testing Service.
Another reason that the TTCT are especially relevant in Barkley, R. A. (Ed.). (1990). Attention-Deficit Hyperactivity
education today is the increased emphasis on using multiple Disorder: A handbook for diagnosis and treatment. New York:
criteria to identify gifts and talents in children, and a widen- Guilford Press.
ing belief in the existence of multiple intelligences. When Bloom, B. S. (Ed.). (1985). Developing talent in young people. New
intelligence tests alone were an accepted way of identifying York: Ballantine Books.
the gifted, measures of creativity seemed less critical. Boyle, D. (2003, October 6). The syndrome that became an
However, the change in emphasis in recent years on identi- epidemic. New Statesman, 132, 27.
Brown v. Board of Education of Topeka, Kansas, 347 U.S. 483
fying and nurturing talents in youth has necessitated various
(1954).
methods for assessing potential in a wide variety of ways.
Cramond, B. (1994). The Torrance Tests of Creative Thinking:
The TTCT is one way to assess creative potential.
From creation through establishment of predictive validity.
Also, with the emphasis shifting to identifying more
In R. F. Subotnik & K. D. Arnold (Eds.), Beyond Terman:
students from underserved populations, the use of a cul-
Longitudinal studies in contemporary education (pp. 229–254).
ture-fair instrument like the TTCT-Figural (Torrance, Norwood, NJ: Ablex.
1971, 1977) is indicated. With the increased enrollment Cramond, B. (1999). The Torrance Tests of Creative Thinking:
of children with limited or no proficiency in English, the Going beyond the scores. In A. S. Fishkin, B. Cramond, &
use of a test that requires minimal language, like the P. Olszewski-Kubilius (Eds.), Investigating creativity in youth
TTCT-Figural, is helpful. (pp. 307–328). Cresskill, NJ: Hampton Press.
In the wake of the Brown decision (Brown v. Board of Cropley, A. J. (1971). Some Canadian creativity research.
Education, 1954), Torrance may have had some idea about Journal of Research and Development in Education, 4(3),
the effects of the developing civil rights movement on the 113–115.

2 9 0 G I F T E D C H I L D Q U A R T E R LY • F A L L 2 0 0 5 • V O L 4 9 N O 4
 4 0 - Y E A R F O L L O W- U P O F T H E T O R R A N C E T E S T S

Feldman, D. H., Csikszentmihalyi, M., & Gardner, H. (1994). Torrance, E. P. (1974). Norms-technical manual: Torrance Tests of
Changing the world: A framework for the study of creativity. Creative Thinking. Lexington, MA: Ginn & Co.
Westport, CT: Praeger Publishers. Torrance, E. P. (1977). Discovery and nurturance of giftedness in the
Frick, P. J., & Lahey, B. B. (1991). The nature and characteris- culturally different. Reston, VA: ERIC Clearinghouse on
tics of Attention-Deficit Hyperactivity Disorder. School Handicapped and Gifted Children. (ERIC Document
Psychology Review, 20, 163–173. Reproduction Service No. ED145621)
Gallagher, S., & Gallagher, J. J. (2002). Giftedness and Torrance, E. P. (1979). The search for satori and creativity. Buffalo,
Asperger’s Syndrome: A new agenda for education. NY: Creative Education Foundation.
Understanding Our Gifted, 14(2), 7–12. Torrance, E. P. (1980). Norms-technical manual: Demonstrator form
Goff, K., & Torrance, E. P. (2002). The Abbreviated Torrance Test for the Torrance Tests of Creative Thinking. Unpublished
for Adults (ATTA). Bensenville, IL: Scholastic Testing Manuscript, Georgia Studies of Creative Behavior, Athens,
Service. GA.
Guilford, J. P. (1956). Structure of intellect. Psychological Review, Torrance, E. P. (1981a). Predicting the creativity of elementary
53, 267–293. school children (1958–80)—and the teacher who “made a
Hu, I., & Bentler, P. (1999). Cutoff criteria for fit indexes in difference.” Gifted Child Quarterly, 25, 55–62.
covariance structure analysis: Conventional criteria versus Torrance, E. P. (1981b). Empirical validation of criterion-refer-
new alternatives. Structural Equation Modeling, 6, l–55. enced indicators of creative ability through a longitudinal
Millar, G. W., & Torrance, E. P. (2001). The Torrance kids at study. Creative Child and Adult Quarterly, 6, 136–140.
midlife: Selected case studies of creative behavior. Westport, CT: Torrance, E. P. (1990). Experiences in developing creativity measures:
Greenwood. Insights, discoveries, decisions. Unpublished Manuscript,
Neihart, M. (2000). Gifted children with Asperger’s Syndrome. Torrance Center for Creative Studies and Talent
Gifted Child Quarterly, 44, 222–230. Development, University of Georgia, Athens, GA.
Plucker, J. A. (1999). Is the proof really in the pudding? Torrance, E. P., Ball, E. O., & Safter, H. T. (1992). Torrance Tests
Reanalysis of Torrance’s longitudinal data. Creativity of Creative Thinking: Streamlined scoring guide Figural A and B.
Research Journal, 12, 103–114. Bensenville, IL: Scholastic Testing Service.
Runco, M. A. (1990) Implicit theories and ideational creativity. Torrance, E. P., & DeYoung, K. (1958). Life Experience
In M. A. Runco & R. S. Albert (Eds.), Theories of creativity
Inventory: Creativity Form. Minneapolis, MN: Bureau of
(pp. 234–252). Newbury Park, CA: Sage.
Educational Research, University of Minnesota.
Terman, L. M., & Oden, M. H. (1959). Genetic studies of genius:
Torrance, E. P., Tan, C. A., & Allman, T. (1970). Verbal origi-
Vol. 5. The gifted group at midlife: Thirty-five years’ follow-up of
nality and teacher behavior: A predictive validity study.
a superior group. Stanford, CA: Stanford University Press.
Journal of Teacher Education, 21, 335–341.
Torrance, E. P. (1954). The development of a preliminary Life
Wang, L., Fan, X., & Willson, V. L. (1996). Effects of nonnor-
Experience Inventory for the study of fighter interceptor pilot com-
mal data on parameter estimates and fit indices for a model
bat effectiveness. San Antonio, TX: Air Force Personnel and
with latent and manifest variables: An empirical study.
Training Research Center, Lackland AFB.
Structural Equation Modeling, 3, 228–246.
Torrance, E. P. (1959). Current research on the nature of cre-
Witt, G. (1971). The Life Enrichment Activity Program, Inc.:
ative talent. Journal of Counseling Psychology, 6, 309–316.
A continuing program for creative, disadvantaged children.
Torrance, E. P. (1966). Torrance tests of creative thinking: Norms-
Journal of Research and Development in Education, 4(3), 14–22.
technical manual (Research ed.). Princeton, NJ: Personnel
Press.
Torrance, E. P. (1968). A longitudinal examination of the
fourth grade slump in creativity. Gifted Child Quarterly, 12, End Notes
195–199.
Torrance, E. P. (1969). Prediction of adult creative achievement 1. Although qualitative data were also collected and
among high school seniors. Gifted Child Quarterly, 13, analyzed, the report of that information is too lengthy to
223–229. be included here. Some of the qualitative data will be
Torrance, E. P. (1971). Are the Torrance Tests of Creative reviewed in a future article. Also, readers are referred to
Thinking biased against or in favor of “disadvantaged”
Millar, G. W., & Torrance, E. P. (2001). The Torrance kids
groups? Gifted Child Quarterly, 15, 75–80.
at midlife: Selected case studies of creative behavior. Westport,
Torrance, E. P. (1972a). Career patterns and peak creative
achievements of creative high school students twelve years
CT: Greenwood.
later. Gifted Child Quarterly, 16, 75–88. 2. For the rest of his life, Torrance used his own
Torrance, E. P. (1972b). Predictive validity of the Torrance money, usually from his royalties, to pay for most of his
Tests of Creative Thinking. Journal of Creative Behavior, 6, research. He bought his own copy machine, paid for
236–252. graduate student help, purchased his own postage, etc.

G I F T E D C H I L D Q U A R T E R LY • F A L L 2 0 0 5 • V O L 4 9 N O 4 2 9 1