Stimulation Seeking and Intelligence: A Prospective Longitudinal Study

Journal of Personality and Social Psychology
Copyright 2002 by the American Psychological Association, Inc.
2002, Vol. 82, No. 4, 663– 674
0022-3514/02/$5.00
DOI: 10.1037//0022-3514.82.4.663
Stimulation Seeking and Intelligence: A Prospective Longitudinal Study
Adrian Raine
Chandra Reynolds
University of Southern California
University of California, Riverside
Peter H. Venables
Sarnoff A. Mednick
University of Southern California and University of York
University of Southern California
The prediction that high stimulation seeking 3-year-olds would have higher IQs by 11 years old was
tested in 1,795 children on whom behavioral measures of stimulation seeking were taken at 3 years,
together with cognitive ability at 11 years. High 3-year-old stimulation seekers scored 12 points higher
on total IQ at age 11 compared with low stimulation seekers and also had superior scholastic and reading
ability. Results replicated across independent samples and were found for all gender and ethnic groups.
Effect sizes for the relationship between age 3 stimulation seeking and age 11 IQ ranged from 0.52
to 0.87. Findings appear to be the first to show a prospective link between stimulation seeking and
intelligence. It is hypothesized that young stimulation seekers create for themselves an enriched
environment that stimulates cognitive development.
Despite increasing interest in the ways personality can contrib-
research on intelligence, academic performance, and sensation
ute to intelligence in adults (Sternberg & Ruzgis, 1994), stimula-
seeking, reported findings from two unpublished studies of high
tion seeking in young children is greatly underresearched, and
school students. One of these studies (M. S. Buchsbaum & D. L.
almost nothing is known about its relationship to later intelligence.
Murphy, personal communication to M. Zuckerman, 1974; cited in
One study has shown that preschoolers who are low on shyness
Zuckerman, 1994) showed a significant, positive relationship (r
have higher creativity scores (Kemple, David, & Wang, 1996),
.22) between total sensation-seeking scores and the Wechsler
thus suggesting a possible link between increased sociability (a
Adult Intelligence Scale (Wechsler, 1955) in 138 high school
component of stimulation seeking) and a cognitive trait related to
students. Another unpublished study (Pemberton, 1971) indicated
intelligence. Nevertheless, no link was found in this study between
a significant, positive correlation (r
.19) between stimulation-
low shyness and increased scores on the Peabody Picture Vocab-
seeking scores and Scholastic Assessment Test scores in 200 male
ulary Test (Dunn, 1965), and there appear to be no published
graduates and exactly the same correlation in 200 female under-
studies on behavioral stimulation seeking (as opposed to other
graduates. These two unpublished findings are supported by pos-
traits of shyness or visual novelty preference in infants) and
itive correlations between stimulation seeking and arithmetic prob-
intelligence in early childhood.
lem solving (r
.21) and arithmetic concepts (r
.15) in a study
More is known about the link between stimulation seeking and
of 97 male and female junior high school students (Kish & Leahy,
intelligence in nonpathological adolescent and adult populations,
1970), although the latter finding failed to reach statistical
but the empirical data are sparse. Zuckerman (1994), in a review of
significance.
Findings from these studies on nonpathological groups are sup-
ported by similar findings of positive relationships between stim-
ulation seeking and IQ in three samples of drug abusers (Carrol &
Adrian Raine, Department of Psychology, University of Southern Cal-
Zuckerman, 1977), a sample of juvenile delinquents (L. Cohen,
ifornia; Chandra Reynolds, Department of Psychology, University of Cal-
Dingemans, Lesnik-Oberstein, & van der Vlugt, 1983), 9 –14-year-
ifornia, Riverside; Peter H. Venables, Department of Psychology, Univer-
old boys with psychiatric disorders (Russo, Stokes, Lahey, &
sity of Southern California and Department of Psychology, University of
Christ, 1993), and hospitalized alcoholics (Kish & Busse, 1968). It
York, York, England; Sarnoff A. Mednick, Department of Psychology and
Social Science Research Institute, University of Southern California.
is surprising that these wide-ranging studies of both pathological
This research was conducted with the support of Independent Scientist
and normal populations have been unusually consistent in observ-
Award K02 MH01114-01 and National Institute of Mental Health Grant
ing low-level but significant relationships between stimulation
RO1 MH 46435-02 to Adrian Raine; grants from the Medical Research
seeking and intelligence. Despite the potential significance of these
Council, Leverhulme Trust, and the Mental Health Foundation to Peter H.
findings, none of these studies has provided a conceptual or
Venables; National Institute of Mental Health Research Scientist Award 5
theoretical explanation for this relationship, and the empirical
K05MH00619-08 to Sarnoff A. Mednick; and a grant from the Ministry of
finding itself has virtually been ignored.
Health of the Mauritian government. We thank Marie-Clare Calambay,
To our knowledge, there have been no prospective studies of
Meena Calinghen, Athene Chiriaca, Cyril Dalais, and Devi Jaganathen for
any kind on the link between stimulation seeking in children—
their help in data collection.
whether measured through questionnaires, parental ratings, or lab-
Correspondence concerning this article should be addressed to Adrian
Raine, Department of Psychology, University of Southern California, Los
oratory observational assessments—and later cognitive ability.
Angeles, California 90089-1061. E-mail: raine@usc.edu
Indeed, there appears to have been very little research of any kind
663

---

664
RAINE, REYNOLDS, VENABLES, AND MEDNICK
on stimulation seeking in preschool children. Yet stimulation-
of the whole island. Informed consent was obtained from the mothers of the
seeking behaviors in which preschoolers physically explore their
participants.
environment and engage in verbal and nonverbal stimulation with
The total sample consisted of both boys (51%) and girls (49%). Ethnic
other children and adults are of significance in that they provide
distribution was as follows: 69% Indian, 26% General Population, and 6%
others (Chinese, English, French, and ethnically unidentified). The term
the basis for dimensions of stimulation seeking that have been
General Population is the official government term used predominantly to
found to characterize adults (Zuckerman, Eysenck, & Eysenck,
characterize descendants from slaves brought over by the French between
1978). Furthermore, physical exploration of the environment may
1670 and 1810 from Madagascar (45%), the East African coast (primarily
in part be driven by curiosity, a cognitive trait that has been found
Mozambique, 40%), India (13%), and West Africa (2%; Barker, 1996;
to characterize highly intelligent children (A. W. Gottfried, Gott-
Filliot, 1974). This ethnic group is predominantly identified by its constit-
fried, Bathurst, & Guerin, 1994), whereas the confident, outgoing
uents as Creole and was described by Bowman (1991) as being constituted
nature of stimulation seekers may help to provide such children
as follows: 85% African or Malagasy origin, 11% mixed origin, and 4%
with rich environmental experiences that stimulate cognitive de-
Franco-Mauritians. The Indian group are largely descendants of indentured
velopment. The neglect of the link between intelligence and stim-
laborers brought over by the English after slavery was abolished in 1835
ulation seeking in children is paralleled by its neglect in adults. For
and constitute Hindus (from Northern India, especially Bihar and Uttar
Pradesh), Tamils (from Madras), and Muslims (from the Gangetic plain of
example, Sternberg and Ruzgis (1994) argued that some of the
North India and West India; Bowman, 1991). Census data for the island as
most interesting insights on intelligence are now being gained
a whole indicated 66% Indian, 29% Creole, and 5% other, indicating that
from research on personality, yet none of the 11 chapters in their
the study largely achieved its goal of sampling a population that was
edited volume even mentioned stimulation seeking.
representative in gender and ethnicity.
The current study attempts to address this gap in the literature.
Four competing hypotheses on the relationship between early
Intelligence and Cognitive Ability at Ages 3 and 11 Years
stimulation seeking and later cognitive ability can be entertained.
First, the null hypothesis is that although there may be a concurrent
Age 3.
Measures of verbal and spatial cognitive ability were derived
relationship between early stimulation seeking and cognitive abil-
from subtests of the Boehm Test of Basic Concepts—Preschool Version
(Boehm, 1986). These tests were used to assess basic verbal and cognitive
ity, developmental changes in both temperament and cognitive
abilities at age 3 and took approximately 30 min (range 20 – 45) to admin-
ability result in no longitudinal relationship (zero effect size) over
ister. Although testing the children at age 11 posed few difficulties, pilot
an 8-year period from preschool to late childhood. Second, the
testing of the Boehm Test of Basic Concepts on 3-year olds indicated that
longitudinal relationship in children may mimic the concurrent
some changes in format were necessary for age 3 testing. In addition to
relationship observed in adult studies both in direction and size of
pilot tests in the laboratory, visits were made to the homes of pilot test
effect—that is, high stimulation seeking in preschoolers is weakly
children to observe them in a more natural context. This pilot testing led to
associated (i.e., positive correlation of approximately .20) with
minor modifications of the test for use with Mauritian children. For
higher intelligence in later childhood. Third, if stimulation seeking
example, sugar cane sticks were used for judgments of length (Mauritius
in childhood but not in later adolescence and adulthood is impor-
had a predominantly sugar cane economy in 1972), local rocks were used
tant in shaping intelligence, then effect sizes based on early (age 3
for judgments of size, pictures of Mauritian children were used for iden-
tification of body parts of children, and a tea set was used to assess ability
years) assessments of stimulation seeking may be stronger than
to follow directions (tea drinking is ubiquitous in Mauritius).
those observed in cross-sectional studies of adolescents and adults.
The modified test had six components: (a) block assembly (making
A fourth hypothesis is that a negative relationship may exist
constructions from blocks; e.g., bridge, circle, tower), (b) copying shapes
between stimulation seeking and IQ because stimulation seeking
(copying circle, triangle, and square), (c) information (identifying body
could be expected to be associated with disinhibited behaviors,
parts, pictures of boys and girls), (d) number, size, and length concepts
which could interfere with school performance and, hence, cogni-
(simple numeric ability, size and length discriminations), (e) color concepts
tive development (Zuckerman, 1994).
(naming and pointing to different colors), and (f) classification (making
These four competing hypotheses are tested in the context of a
discriminations between same and different objects).
longitudinal study of child health and human development on the
Many of these abilities parallel cognitive skills found in the Wechsler
island of Mauritius. A preschool analogue of stimulation seeking
Preschool and Primary Scale of Intelligence (WPPSI; Wechsler, 1967; e.g.,
labeling is similar to the WPPSI Information subscale, similar and different
was taken when children were 3 years old, and cognitive ability
discriminations are similar to the Similarities subscale, copying shapes is
was measured at ages 3 and 11 years. To assess whether relations
similar to Geometric Design, number, size, and length concepts are similar
between age 3 stimulation seeking and age 11 cognitive ability
to Arithmetic). Consequently, scale construction initially followed a face
extend beyond IQ, we also assessed prospective relationships
validity approach to form indices of verbal and spatial ability. The re-
between age 3 stimulation seeking and later school achievement,
searchers first normalized each scale by transforming the raw scores to
reading ability, and performance on neuropsychological tests.
percentiles and then finding the standard score for each percentile (Allen &
Yen, 1979). Scales were then standardized to have a mean of 10 and
Method
standard deviation of 3.
Two of the scales (block assembly and copying shapes) were spatial–
Participants
constructional in nature and were similar in nature to the Block Design and
Geometric Design spatial tests of the WPPSI. These two tests correlated
The larger population from which the participants were drawn consisted
significantly with age 11 spatial IQ (range
.22–.25, p
.0001). These
of 1,795 children from the island of Mauritius (a country lying in the Indian
tests were summated and further standardized to a mean of 100 and
Ocean between Africa and India). All children born in 1969 –1970 in two
standard deviation of 15 to form an index of age 3 spatial ability. Coeffi-
towns on the island were recruited into the study between September 1972
cient alpha for this spatial scale was .46.
and August 1973, when they were 3 years old. The two towns (Vacoas and
The remaining scales were verbal in nature. Some involved a verbal
Quatre Bornes) were chosen to be representative of the ethnic distribution
response (e.g., picture content, numbers), whereas others required verbal

---

STIMULATION SEEKING AND INTELLIGENCE
665
comprehension and knowledge of the names of objects (information). All
School Attainment and Reading Ability at Age 11
of these subtests correlated significantly with age 11 verbal IQ (range
.19 –.27, p
.0001). Several of the Boehm Test of Basic Concepts verbal
Reading ability was assessed at age 11 using the Holborn Reading Scale
tests had parallels with WPPSI verbal tests (e.g., information was similar
(Pumfrey, 1985). This word recognition test consists of 33 sentences of
to the Information subscale, classification was similar to Similarities,
increasing difficulty. Total scores (number of sentences correctly read)
number, size, and length were similar to Arithmetic). These subscales were
were standardized into reading quotients and were available for 1,264 of
the sample.
summated and standardized in the same way as were the spatial tests to
The measure of scholastic ability was based on scores on four standard-
form an index of age 3 verbal ability. Coefficient alpha for the verbal scale
ized academic tests (Certificates of Primary Education) that were given to
was .76. Age 3 verbal ability correlated significantly with age 3 spatial
all 11-year-old children throughout the country: English, French, Mathe-
ability (r
.41, n
1,387, p
.0001). Data were available on 1,387 of
matics, and Environmental Studies. Scores on these tests (graded 0 through
the sample.
5) were summated to form an overall index of school attainment and were
Age 11.
Estimates of verbal and spatial IQ were assessed at age 11
available for 1,415 of the sample. The correlation between school attain-
years using six subtests of the Wechsler Intelligence Scale for Children—
ment and reading ability was .69 (n
1,094, p
.0001).
Revised (WISC–R; Wechsler, 1974). Raw scores on the WISC–R sub-
scales were normalized and standardized in the same way as were the age 3
scales. The Similarities and Digit Span subscales formed an estimate of
Trails A and B at Age 11
verbal IQ, whereas the Block Design, Object Assembly, Coding, Mazes,
One neuropsychological measure, the Trail Making Test (Reitan, 1958),
and Picture Completion subscales formed an estimate of spatial IQ. Data
was administered to the children at age 11. The two components of this task
were available on 1,261 of the sample.
(Trails A and B) assess visuomotor tracking, motor speed, and attention,
and Trail B also contains a working memory component (Lezak, 1983).
The task requires the participant to draw lines to connect consecutively
Reliability and Validity of the Age 3 and Age 11
numbered circles (Trail A) and to connect consecutively numbered and
Cognitive Measures
lettered circles (Trail B). Participants were instructed to go as quickly as
they could without lifting the pencil off the paper, and time-to-completion
The two subcomponents of the age 3 spatial measure correlated signif-
scores for each component were corrected for age by residualization. Data
icantly with age 11 spatial IQ (range
.22–.25, p
.0001). Similarly, all
were available on 1,157 and 1,239 of the sample, respectively.
subcomponents of the age 3 verbal measure correlated significantly with
age 11 verbal IQ (range
.19 –.27, p
.0001). Age 3 verbal ability
Stimulation Seeking and Sociability
correlated significantly with age 3 spatial ability (r
.41, n
1,387, p
.0001). The age 3 spatial measure correlated .24 ( p
.0001) with the
Full details of the creation of the age 3 index of stimulation seeking and
age 11 spatial IQ measure, whereas the age 3 verbal measure correlated .25
sociability are given in Raine et al. (1998). Briefly, four putative indices of
( p
.0001) with the age 11 verbal IQ measure. Age 3 verbal ability
stimulation seeking and sociability were taken at age 3 as follows:
correlated significantly (r
.25, p
.0001) with age 11 reading ability.
1. The child’s physical exploration away from the mother toward new
Both verbal (r
.29, p
.0001) and spatial ability (r
.25, p
.0001)
toys was assessed in a laboratory room by a research assistant. Exploratory
measures taken at age 3 predicted scholastic ability at age 11. Age 3 verbal
behavior was rated on a 4-point scale as follows: 1
passive, clings to
ability correlated significantly (r
.31, p
.0001) with a rating of the
mother, withdrawn; 2
shows interest, examines toys but stays close to
amount of verbalizations the child made to the experimenter at age 3
mother; 3
leaves mother, mild independent exploration, comes and goes
(Raine, Reynolds, Venables, Mednick, & Farrington, 1998), whereas age 3
to mother; and 4
active independent exploration. This behavior was
spatial ability correlated significantly with a measure of motor ability (e.g.,
rated on four occasions during the entire testing session (soon after arrival,
jumping, hopping, balancing on one foot) at age 3 (r
.23, p
.0001).
before psychophysiological testing, between tests, and after completion of
tests). Scores for the four ratings were summated to obtain an overall index
Data from 73 participants who were given the Reynell Developmental
of physical exploration.
Language Scale (Reynell & Huntley, 1972) at age 6 showed a .36 corre-
2. Extent of verbalizations to the research assistant during cognitive
lation ( p
.002) with the age 3 verbal ability measure, compared with a
testing was rated on a 4-point scale ranging from 1 (very reluctant to speak)
.25 correlation ( p
.005) with the age 3 spatial measure ( p
.025);
to 4 (many spontaneous comments).
although the difference between these two correlations is not statistically
3. Friendliness with the research assistant during cognitive testing was
significant because of the modest sample size, the age 3 verbal measure
rated on a 4-point scale ranging from 1 (fearful) to 4 (immediately friendly).
predicted twice the amount of variance in age 6 language than did the age 3
4. Active social play with other children during free play in a sandbox
spatial measure. The intercorrelation between age 3 total cognitive score
was rated by a research assistant on a 5-point scale ranging from 1
(verbal
spatial) and age 11 estimated total IQ was .30 (n
969, p
(solitary) to 5 (cooperative relationship with role reciprocity).
.0001). This latter correlation is modest but in keeping with the facts that
Confirmatory factor analysis using LISREL 8 (Jo¨reskog & So¨rbom,
(a) IQ does not stabilize until later childhood, and correlations between
1993) confirmed that these items all loaded together on the same factor
early cognitive ability and later IQ are relatively small (A. W. Gottfried et
(Raine et al., 1998). Virtually identical findings were obtained for boys and
al., 1994), (b) different cognitive measures were used in the age 3 and
girls and for Indians and Creoles (see Raine et al., 1998, for full details).
age 11 test sessions, and (c) internal reliabilities for the age 3 measures are
The four items from this factor all intercorrelated from .25 to .68 (M
not high.
.43). Item–total correlations for this scale ranged from .48 to .59 (M
.53).
Confirmatory factor analysis was used to assess whether the two-factor
Coefficient alpha for the scale was .75. These items were z transformed and
(verbal–spatial) model was a significantly better fit than the one factor
summated into a scale of Stimulation Seeking, with high scores indicating
model. A one-factor solution resulted in a significant misfit,
2(20,
increased stimulation seeking. Data were available on 1,772 of the sample.
N
1,386)
817.94, p
.0001, root-mean-square error of approximation
(RMSEA)
.17. However, the two-factor model (spatial, verbal) resulted
Parental Education and Social Class
in a significant improvement in fit compared with the one-factor model,
2(1, N
1,386)
34.79, p
.0001. Coefficient alpha at the subtest
Demographic measures were recorded in an interview with the mother
level was .76 for the verbal scale and .46 for the spatial scale.
when the child was age 3. Parental occupation was rated on an 8-point

---

666
RAINE, REYNOLDS, VENABLES, AND MEDNICK
rating scale that ranged from 1 ( part-time laborer) to 8 (academic or head
Extremes in Stimulation Seeking and Cognitive Ability
of business with 50 or more employees), whereas parental education was
scored as the number of years in full-time education. Data were available
Extremes of stimulation seeking have been associated in the
on 1,795 and 1,785 of the sample, respectively.
adult and child literatures with drug and alcohol abuse and anti-
social behavior (Cloninger, Sigvardsson, & Bohman, 1988; Raine
et al., 1998; Zuckerman, 1994), behaviors that, in turn, have been
Results
associated with lower, not higher, cognitive functioning (Lynam,
Stimulation Seeking and Cognitive Ability at Ages 3
Moffitt, & Stouthamer-Loeber, 1993; Moffitt & Silva, 1988;
and 11
Tarter, Jacob, & Laird, 1993). It is possible that individuals at the
end of the stimulation-seeking continuum do not show enhanced
IQ.
Means and standard deviations of all variables are given in
cognitive ability. To test this possibility, we compared the top 15%
Table 1. The sample was randomly divided into two subsamples to
of scorers on stimulation seeking (n
193–270) at age 3 years
assess replicability of findings. Intercorrelations between stimula-
with those falling as close as possible to the bottom 15% cut off
tion seeking at age 3 and cognitive ability at ages 3 and 11 for
(n
80 –204)1 on age 3 and age 11 cognitive, school, and
these two samples, together with the total sample, are given in
neuropsychological measures. This same 15% cut off was used in
Table 2. All correlations for the two replication samples were
the one other previous study of stimulation seeking with this
statistically significant and in a positive direction (i.e., high stim-
population (Raine et al., 1998). Comparisons for these groups are
ulation seeking was associated with high cognitive ability), with
shown in Table 4.
relatively little variability in the absolute magnitude of the corre-
High stimulation seekers had significantly higher scores on
lations from one sample to another. The effect size for the corre-
all 10 measures at both ages. The effect size for age 3 total
lation of stimulation seeking and the age 3 total cognitive score
cognitive ability was 1.41, and there was an effect size of 0.87 for
(r
.36) was calculated as d
0.77 (J. Cohen, 1988), whereas for
age 11 IQ. Consequently, the positive association between stimu-
the correlation of age 3 stimulation seeking and age 11 total IQ
lation seeking and cognitive ability that was observed for the entire
(r
.25), the effect size was 0.52.
population also held for extreme stimulation seekers.
As a comparison, we calculated intercorrelations between age 3
and age 11 measures of cognitive ability as follows: verbal, r
Gender and Ethnicity
.25, n
971, p
.0001; spatial, r
.24, n
1,025, p
.0001;
and total, r
.30, n
969, p
.0001. Consequently, the relation
To assess whether the stimulation seeking– cognitive ability
between age 3 stimulation seeking and age 11 spatial ability (r
relationship was the same in both boys and girls, we computed
.24) was as high as the correlation between age 3 spatial ability and
intercorrelations separately for the two sexes. Results are given in
age 11 spatial ability (r
.24).
Table 2. Intercorrelations in boys were again positive and statis-
Age 11 reading, school achievement, and performance on Trail
tically significant, ranging from .18 to .34. These relationships
Making Test.
Intercorrelations between age 3 stimulation seeking
were replicated in girls, with correlations ranging from .23 to .38.
and age 11 scholastic ability (reading and school achievement
On average, there was a difference of .033 between boys and girls
scores) are given in Table 2. Again, findings replicated from one
in the size of the sensation seeking– cognitive ability correlation,
sample to the other, with correlations statistically significant and in
favoring girls. Similar findings were obtained for scholastic ability
the predicted direction but somewhat lower in magnitude than for
and neuropsychological test performance.
cognitive ability and IQ. For the total sample, correlations of .15
The sample was then divided into Indians and Creoles to test
(reading) and .17 (school achievement) were observed. Effect sizes
cross-ethnic replicability of findings. Correlations for Indians were
were calculated as 0.30 for reading and 0.35 for school
positive and statistically significant, ranging from .20 to .33.
achievement.
Correlations for Creoles similarly ranged from .18 to .42. On
Similarly sized and replicable correlations were obtained for
average, there was a difference of .032 between Indians and
neuropsychological test performance (Trails A and B), with neg-
Creoles, favoring Creoles. Similar findings were obtained for
ative correlations indicating faster completion times for high
scholastic ability and neuropsychological test performance.
stimulation-seeking scorers. Effect size for both measures were
calculated as 0.30.
Components of Stimulation Seeking and Cognitive Ability
It is possible that the link between stimulation seeking at age 3 and
Stimulation Seeking in the Context of g
IQ at age 11 is driven only by the verbal component of stimulation
To obtain the purest estimate of the size of the relationship
seeking, with no contribution made by nonverbal, behavioral aspects.
between stimulation seeking and general cognitive ability, we
For example, it is conceivable that a child who makes more sponta-
entered stimulation seeking at age 3 into the principal-components
neous verbalizations at age 3 is more verbally skilled and that it is this
analysis with the six age 3 cognitive subtest measures and also
specific component of age 3 stimulation seeking that relates to later IQ
with the six age 11 IQ subtest measures. The loadings of cognitive
and stimulation-seeking measures on the first principal component
1 The smaller sample of low relative to high sensation seekers at age 3
for these two analyses are given in Table 3. It can be seen that for
reflects the fact that 19% of the sample was untestable on cognitive tasks
age 3 cognitive measures, stimulation seeking loaded .55
at this age, with a disproportionate number of low stimulation seekers
(d
1.32) on the first principal component and that on age 11
falling into this category. In contrast, very few high stimulation seekers
cognitive measures it loaded .35 (d
0.75).
were untestable. All children were testable at age 11.

---

STIMULATION SEEKING AND INTELLIGENCE
667
Table 1
Means, Standard Deviations, and Sample Sizes for Stimulation Seeking, Age 3 and Age 11
Cognitive Measures, and Parental Education and Occupation for the Total Sample and Gender
and Ethnic Subgroups
Variable and statistic
Total sample
Boys
Girls
Indians
Creoles
Age 3
Sensation seeking
M
0.00
0.03
0.06
0.06
0.07
SD
0.71
0.67
0.69
0.66
0.68
n
1,772
919
852
490
1,213
Verbal
M
100.00
99.52
100.56
100.96
99.53
SD
15.00
15.25
14.83
14.53
15.28
n
1,387
724
664
400
937
Spatial
M
100.00
99.00
100.90
100.40
99.81
SD
15.00
14.97
15.13
15.00
15.12
n
1,452
752
701
419
980
Total
M
100.00
99.24
100.87
100.84
99.63
SD
15.00
15.00
14.94
14.61
15.29
n
1,384
722
663
399
935
g
M
0.03
0.05
0.06
0.06
0.03
SD
1.00
1.01
1.03
0.98
1.00
n
1,384
722
663
399
935
Age 11
Verbal
M
100.00
100.57
99.18
101.42
99.44
SD
15.00
15.52
14.58
15.02
15.03
n
1,261
636
622
331
875
Spatial
M
100.00
103.33
96.48
101.75
99.36
SD
15.00
14.69
14.56
14.94
15.06
n
1,261
636
622
331
875
Total
M
100.00
102.81
97.00
101.82
99.39
SD
15.00
14.93
14.57
15.11
15.00
n
1,261
636
622
331
875
g
M
0.02
0.17
0.17
0.12
0.05
SD
1.00
0.99
0.98
1.00
0.99
n
1,261
636
622
331
875
Reading
M
91.13
86.45
95.63
86.96
91.45
SD
55.92
57.66
53.78
58.24
54.93
n
1,264
637
624
331
878
Achievement
M
9.81
9.66
9.97
9.73
9.70
SD
7.07
7.18
7.07
7.18
7.06
n
1,415
720
691
366
991
Trails A
M
0.00
0.13
0.13
0.03
0.02
SD
1.00
0.89
1.14
0.94
1.02
n
1,239
627
609
323
863
Trails B
M
0.00
0.07
0.08
0.06
0.03
SD
1.00
0.98
1.01
1.02
1.01
n
1,157
586
568
303
804
Parent
Occupation
M
3.87
3.90
3.82
4.03
3.82
SD
1.45
1.44
1.44
1.54
1.40
n
1,785
929
856
491
1,227
Education
M
4.55
4.67
4.41
4.79
4.45
SD
3.58
3.65
3.48
3.65
3.54
n
1,795
930
865
495
1,232
Note. g
general intelligence.

---

668
RAINE, REYNOLDS, VENABLES, AND MEDNICK
Table 2
Intercorrelations Between Stimulation Seeking and Cognitive Ability at Age 3 and IQ, Reading,
School Achievement, and Neuropsychological Performance at Age 11
Variable and statistic
Total
Sample 1
Sample 2
Boys
Girls
Indian
Creole
Age 3
Verbal
.36
.36
.36
.34
.38
.33
.42
n
1,366
718
669
714
652
919
396
Spatial
.22
.24
.20
.21
.23
.20
.25
n
1,429
750
702
741
658
961
414
Total
.36
.36
.35
.35
.38
.33
.43
n
1,363
716
668
712
651
917
395
g
.37
.37
.36
.36
.38
.33
.44
n
1,363
716
668
712
651
917
395
Age 11
Verbal
.20
.18
.21
.18
.23
.21
.18
n
1,255
651
610
629
615
861
331
Spatial
.24
.22
.26
.22
.26
.24
.24
n
1,244
651
610
629
615
861
331
Total
.25
.23
.27
.23
.27
.26
.24
n
1,244
651
610
629
615
861
331
g
.25
.24
.27
.23
.26
.24
.26
n
1,244
651
610
629
615
861
331
Scholastic ability
Reading
.15
.17
.13
.13
.18
.14
.19
n
1,247
651
613
630
617
864
331
Achievement
.17
.18
.15
.15
.19
.17
.19
n
1,391
726
689
711
680
775
362
Neuropsychological functioning
Trails A
.15
.14
.15
.14
.16
.15
.16
n
1,222
633
606
620
602
849
323
Trails B
.15
.19
.11
.15
.15
.15
.17
n
1,153
595
562
580
561
791
303
Note.
All correlations are significant at p
.05. g
general intelligence.
rather than the nonverbal components such as exploration of the
coefficients using the structural equation modeling program Mx 1.5
environment. We tested this possibility by relating the individual
(Neale, Boker, Xie, & Maes, 1999).
subcomponents of age 3 stimulation seeking with age 3 and age 11
Results are shown in Table 5. It can be seen that all correlations
cognitive ability and calculating confidence intervals for correlation
were statistically significant and in the direction of high sensation
seeking being associated with better performance. Exploration was
the component of stimulation seeking most highly associated with
Table 3
cognitive ability, with a correlation with age 11 total IQ of .24
Factor Loadings of Stimulation Seeking on First Principal
(d
0.50), compared with a correlation of .15 (d
0.30) for the
Component of Cognitive Measures at Age 3 and Age 11
verbalization component. Although confidence intervals over-
lapped (see Table 5), it is clear that the age 11 stimulation seeking–
Measure/scale
Factor loading
intelligence relationship cannot be attributed solely to the verbal
component of stimulation seeking. For the age 3 sensation-seeking
Age 3
components, gregariousness was the component most strongly
Stimulation seeking
.55
linked to cognition. The 95% confidence interval for the correla-
Information
.76
tions associated with gregariousness did not overlap with those for
Similarities
.62
the other three components, indicating that the cognition relations
Color
.62
Arithmetic
.60
with gregariousness were significantly higher than were all other
Geometry
.60
sensation-seeking components (see Table 5).
Block
.50
Age 11
Role of Parental Education and Occupation
Stimulation seeking
.35
It is conceivable that the link between stimulation seeking and
Block Design
.79
later IQ is mediated by parental education and occupation. For
Object Assembly
.74
Mazes
.70
example, parents of higher educational and occupational status
Coding
.68
may provide both genetic and environmental contributions to their
Digit Span
.68
child’s intellectual ability and also may encourage exploratory,
Similarities
.67
socially stimulating behaviors. In support of this possibility, we

---

STIMULATION SEEKING AND INTELLIGENCE
669
Table 4
Extremes of Sensation Seeking at Age 3 and Cognitive Functioning at Ages 3 and 11
Stimulation seeking
High
Low
Variable and statistic
M
SD
n
M
SD
n
t
df
p
d
Age 3
Verbal
105.9
13.7
256
85.4
13.6
80
11.7
334
.0001
1.50
Spatial
103.2
14.7
270
90.7
13.3
83
7.0
351
.0001
0.91
Total
105.7
13.4
256
85.1
12.9
80
12.1
334
.0001
1.57
g
0.39
0.89
256
1.01
0.86
80
12.4
334
.0001
1.61
Age 11
Verbal
104.3
13.9
197
93.9
15.7
177
6.8
372
.0001
0.71
Spatial
104.0
14.0
197
93.1
15.5
177
7.2
372
.0001
0.74
Total
104.5
13.8
197
92.7
15.6
177
7.8
372
.0001
0.81
g
0.30
0.94
197
0.49
1.01
177
7.8
372
.0001
0.81
Scholastic ability
Reading
104.6
55.4
198
75.9
60.9
177
4.8
373
.0001
0.50
Achievement
11.4
7.0
212
8.6
7.0
204
4.1
414
.0001
0.40
Neuropsychological functioning
Trails A
0.25
0.74
195
0.28
1.20
169
5.1
362
.0001
0.56
Trails B
0.20
0.83
193
0.29
0.97
148
5.0
339
.0001
0.55
Note. g
general intelligence.
observed significant correlations between parental education and
was as high as the correlation between age 3 spatial ability and
stimulation seeking (r
.14, p
.0001), age 3 cognitive ability,
age 11 spatial ability (r
.24). Furthermore, the relationship was
r(1385)
.21, p
.0001, and age 11 total IQ, r(1258)
.31, p
not lost in more extreme stimulation-seeking groups who have
.0001, and between parental occupation and stimulation seeking
been viewed as pathological, with an effect size of 0.87 being
(r
.14, p
.0001), age 3 cognitive ability, r(1378)
.18, p
observed. Consequently, results are strongly inconsistent with
.0001, and age 11 IQ, r(1251)
.30, p
.0001. Parental education
Hypothesis 4 (negative relationship), inconsistent with Hypo-
and occupation intercorrelated at r(1785)
.45, p
.0001. We
thesis 1 (zero relationship), moderately consistent with Hypothe-
tested the potential mediating role of parental education and oc-
sis 2 (small positive effect size), and most consistent with Hypoth-
cupation by computing partial correlations for the total sample
esis 3 (medium positive effect). To our knowledge, this is the first
between stimulation seeking and age 11 total IQ, controlling for
report of a prospective relationship between stimulation seeking in
these two factors individually and also in combination. Results are
preschool children and enhanced cognitive ability in later
given in Table 6. The sensation seeking–age 11 IQ relationship
childhood.
remained significant after we partialed out these effects, with little
Previous studies in adults have not provided a theoretical ac-
change in the absolute size of the correlations, as compared with
count of why increased stimulation seeking might be associated
Table 2.
with better cognitive ability later in life. We hypothesize an envi-
ronmental enrichment explanation of the stimulation seeking–IQ
Discussion
relationship, which argues that young children who physically
explore their environment, engage socially with other children, and
The key finding from this study is that increased stimulation
verbally interact with adults create for themselves an enriched,
seeking at age 3 years is associated with increased cognitive,
stimulating, varied, and challenging environment. This environ-
scholastic, and neuropsychological test performance at age 11
mental enrichment in turn is hypothesized to result in enhanced
years. It is important to note that the results replicated across
cognitive ability and better school performance. This environmen-
independent samples, were found in both boys and girls, applied to
tal enrichment hypothesis is broadly consistent with the seminal
both Indians and Creoles, and were not mediated by parental
research of Bell (1968), which showed that children affect their
education and occupation. The effect size for the relationship
own environment, and also with the triarchic theory of experience
between age 3 stimulation seeking and age 11 IQ was not small,
developed by Scarr (1992), which argues that individuals both
with d medium to large in size (J. Cohen, 1988) and ranging
create their own environments and select environments that are
from 0.52 (correlational analysis) through 0.75 (principal-
correlated with their interests and personality characteristics. Pro-
components analysis) to 0.87 (extreme group analysis). These
vision of an educationally enriched environment is known to result
longitudinal effect sizes are, if anything, higher than the cross-
in short-term but generally not long-term increases in IQ (Neisser
sectional effect sizes found for normal adults; pooling the four
et al., 1996), and it is conceivable that early stimulation seeking is
studies on school children reported above results in an average
associated with more substantive, longer term increases in IQ,
effect size of 0.40 based on a combined sample of 732. The
because stimulation seeking provides a continuous (not short-term)
strength of the relation between age 3 stimulation seeking and
enrichment of the environment. Nevertheless, this is only one of a
age 11 spatial ability (r
.24) can be gauged by the fact that it
number of possible explanations for the significant longitudinal

---

670
RAINE, REYNOLDS, VENABLES, AND MEDNICK
Table 5
Correlations and Their 95% Confidence Intervals for Relationships Between the Four
Subcomponents of Sensation Seeking and Cognitive Ability
Measure and CI
n
Verbalizations
Play
Gregarious
Explore
Age 3
Verbal
1,366
.31
.08
.42
.24
Lower CI
.27
.03
.37
.19
Upper CI
.36
.14
.46
.29
Spatial
1,429
.17
.10
.24
.12
Lower CI
.12
.05
.19
.07
Upper CI
.22
.16
.29
.17
Total
1,363
.30
.10
.42
.23
Lower CI
.25
.05
.37
.18
Upper CI
.35
.16
.46
.28
g
1,363
.31
.10
.43
.24
Lower CI
.26
.05
.38
.18
Upper CI
.36
.16
.47
.28
Age 11
Verbal
1,243
.13
.12
.17
.19
Lower CI
.08
.06
.11
.14
Upper CI
.19
.17
.22
.24
Spatial
1,243
.15
.14
.19
.24
Lower CI
.09
.09
.13
.18
Upper CI
.20
.20
.24
.29
Total
1,243
.16
.15
.20
.24
Lower CI
.10
.09
.14
.19
Upper CI
.21
.20
.25
.30
g
1,243
.15
.15
.19
.24
Lower CI
.10
.09
.14
.19
Upper CI
.21
.20
.24
.29
Reading
1,246
.09
.08
.14
.14
Lower CI
.03
.03
.08
.09
Upper CI
.14
.14
.19
.20
Achievement
1,390
.10
.08
.15
.16
Lower CI
.05
.03
.10
.11
Upper CI
.15
.14
.20
.21
Trails A
1,221
.06
.10
.11
.17
Lower CI
.11
.15
.16
.22
Upper CI
.00
.04
.05
.11
Trails B
1,140
.10
.09
.11
.15
Lower CI
.15
.15
.17
.21
Upper CI
.04
.03
.05
.10
Note.
All correlations in boldface are statistically significant at p
.05, two-tailed. CI
confidence interval;
g
general intelligence.
relationship that exists between early stimulation seeking and later
a brain area of importance in both attention and memory (Bur-
increased intelligence.
gess, Jeffery, & O’Keefe, 1999; Newman & Grace, 1999). In
A competing explanation is suggested by the fact that the
humans, there is increasing evidence that exercise can be ben-
component of age 3 stimulation seeking that was most predic-
eficial to both cognitive functioning (Thomas, Landers, Salazar,
tive in absolute terms of later intelligence at age 11 was
& Etnier, 1994) and psychophysiological measures of informa-
behavioral exploration. Specifically, the correlation between
tion processing and arousal (Lardon & Polich, 1996; Raine et
age 3 behavioral exploration and age 11 total IQ was nonsig-
al., 2001). It is also known that environmental enrichment
nificantly higher than the other three components of stimulation
increases novelty seeking in rats (Fernandez-Teruel, Escori-
seeking (see Table 5). It is conceivable that sensation-seeking
huela, Casetellano, Gonzalez, & Tobena, 1997). As such, the
children are more intelligent because they are more physically
link between stimulation seeking and intelligence could be
active and that it is physical activity per se rather than the
mediated at least in part by the increase in exercise that is a
creation of an enriched environment that results in increased
byproduct of stimulation-seeking activities.
cognitive ability. Studies that experimentally manipulated phys-
The physical activity explanation of the intelligence–stimulation
ical activity in rats have shown that physical exercise and
seeking relationship is seemingly inconsistent with some studies
environmental enrichment result in the growth of new neurons
on activity levels in children. For example, Halverson and Wal-
in the hippocampus (van Praag, Kempermann, & Gage, 1999),
drop (1976) found that vigorous, high-activity behavior in pre-

---

STIMULATION SEEKING AND INTELLIGENCE
671
Table 6
portance of the social components of sensation seeking in facili-
Partial Correlations Between Age 3 Stimulation Seeking and
tating superior cognitive abilities.
Cognitive Ability at Age 3 and IQ, Reading, School
Behavioral exploration may also be a marker for curiosity,
Achievement, and Neuropsychological Performance at Age 11
which in turn motivates learning and task persistence. A. W.
Gottfried et al. (1994) have reviewed evidence showing that gifted
Controlled for
children and adolescents have higher levels of curiosity than do
controls, and A. E. Gottfried (1990) argued that “academic intrin-
Education and
sic motivation” (p. 525) may be an important factor that contrib-
Variable and statistic
Education Occupation
occupation
utes to intellectual giftedness. Similarly, novelty preference in
Age 3
children (which may parallel the stimulation-seeking component
Verbal
.34
.35
.34
of boredom susceptibility in adults) has been related to higher
n
1,363
1,356
1,355
achievement scores in second- and fifth-grade children (Cahill-
Spatial
.21
.21
.21
n
1,426
1,418
1,417
Solis & Witryol, 1994). Visual novelty preference and habituation
Total
.34
.35
.34
in infants has also been found to relate to increased intellectual
n
1,360
1,351
1,352
functioning in later childhood (Fagan, 1984; McCall & Garriger,
g
.35
.35
.34
1993), although it is clear that measures of infant novelty prefer-
n
1,360
1,351
1,352
Age 11
ence may be better construed as measures of information process-
Verbal
.18
.18
.17
ing rather than of temperament and personality (Rose & Tamis-
n
1,241
1,234
1,233
LeMonda, 1999). Furthermore, positive associations have been
Spatial
.21
.21
.20
found in 6 –10-year-old children between increased stimulation
n
1,241
1,234
1,233
seeking and increased preference for complex puzzles and pictures
Total
.22
.22
.21
n
1,241
1,234
1,236
(Kafry, 1982). Consequently, curiosity may result in both
g
.21
.22
.20
stimulation-seeking behavior and increased intelligence rather than
n
1,241
1,234
1,236
stimulation seeking itself causing increased intelligence through
Scholastic ability
environmental enrichment.
Reading
.12
.12
.11
n
1,244
1,237
1,236
Genetic explanations of the stimulation seeking–intelligence
Achievement
.13
.14
.12
link are as plausible as the above environmental explanations and
n
1,388
1,381
1,380
represent an uncontrolled third factor. On the basis of 1,591 twin
Neuropsychological functioning
pairs ranging from 12 to 24 years old, 48% to 63% of the vari-
Trails A
.15
.13
.12
ability in stimulation seeking is accounted for by genetic factors,
n
1,219
1,212
1,211
Trails B
.14
.13
.13
with heritability levels being equal in both males and females
n
1,138
1,131
1,130
(Koopmans, Boomsma, Heath, van Doornen, & Lorenz, 1995).
Childhood intelligence is also known to show significant herita-
Note.
All correlations are statistically significant at p
.001. g
general
bility, though not as strong as adult intelligence (Neisser et al.,
intelligence.
1996), and it is conceivable that the same set of genes may
influence both intelligence and stimulation seeking. At this level,
stimulation seeking could be viewed as an integral component of
schoolers was associated with poorer, not better, intellectual ability
the wider construct of intelligence rather than being a separate
at age 7.5 years. It is likely that the precise nature of physical
construct. In addition, it is possible that inherent brain differences
activity is of central importance with respect to intellectual out-
may contribute to both stimulation seeking and cognitive ability
come. In the current study, a high level of active social play was
rather than stimulation-seeking experiences themselves causing
defined in terms of expressing cooperative social relationships
increased cognitive ability. Although human studies are lacking,
with other children with full role reciprocity. Similarly, Halverson
studies of rats have shown that normal rats with an increase in
and Waldrop (1976) found that social participation in preschoolers
volume of the cerebellar molecular layer show increased explor-
that was equally as vigorous and intense as high-activity behaviors
atory behavior (Anderson, 1994), and there is increasing evidence
was indeed associated with higher intelligence in later childhood.
that the cerebellum plays a significant role in complex cognitive
High-activity behaviors that are unstructured and independent of
functions (Schmahmann & Sherman, 1998). Furthermore, rat
social reciprocity are more likely to be markers of hyperactivity
strains that are almost totally lacking in Purkinje and granule cells
and inattention, whereas highly active social play and goal-
in the cerebellar cortex both show low levels of exploration of
directed exploration of the environment (both of which contributed
novel environments and have spatial deficits, compared with other
to the stimulation-seeking measure in this study) may better facil-
rat strains (Caston, Chianale, Delhaye-Bouchaud, & Mariani,
itate or reflect superior cognitive functioning. Consequently, de-
1998). Consequently, the cerebellar cortex may represent one brain
spite recent evidence from animal studies, high levels of physical
area that mediates both cognitive ability and stimulus-seeking
activity per se may not necessarily facilitate or reflect superior
behavior. Future twin studies that assess stimulation seeking, in-
intellectual performance in young children, and social involvement
telligence, and cerebellar structure and function are needed to
with others may be particularly important. In support of this
further test this possibility of a genetic correlation between stim-
possibility, the gregarious component of sensation seeking was
ulation seeking and intelligence.
significantly more highly related to age 3 cognition than were all
Stimulation seeking as measured behaviorally in this study in
other sensation-seeking components, indicating the potential im-
children and sensation seeking as assessed through self-report

---

672
RAINE, REYNOLDS, VENABLES, AND MEDNICK
questionnaires in adults are conceptualized as related although not
preferences for four levels of novelty in toy construction. Genetic,
entirely identical constructs. Sensation seeking in adults is partic-
Social, and General Psychology Monographs, 120, 393– 408.
ularly related to the impulsivity component of extraversion (Zuck-
Carrol, E. N., & Zuckerman, M. (1977). Psychopathology and sensation
erman, Kuhlman, Joireman, Teta, & Kraft, 1993), whereas sensa-
seeking in “downers,” “speeders,” and “trippers”: A study of the rela-
tion seeking and neuroticism have been found to load on separate
tionships between personality and drug choice. International Journal of
Addictions, 12, 591– 601.
factors (Glickshon & Abulafia, 1998; Zuckerman, 1991). In the
Caston, J., Chianale, C., Delhaye-Bouchaud, N., & Mariani, J. (1998). Role
context of the Big Five, sensation seeking has been most clearly
of the cerebellum in exploration behavior. Brain Research, 808, 232–
related to Openness to Experience (Rawlings, Twomey, Burns, &
237.
Morris, 1998; Zuckerman, 1972). Stimulation seeking as measured
Cloninger, C. R., Sigvardsson, S., & Bohman, M. (1988). Childhood
behaviorally in this study is conceptualized as lying at the interface
personality predicts alcohol abuse in young adults. Alcoholism: Clinical
of exploration, curiosity, experience seeking, and sociability and is
and Experimental Research, 12, 494 –505.
anticipated as predisposing one to have a sensation-seeking per-
Cohen, J. (1988). Statistical power analysis for the behavioral sciences
sonality as an adult. It remains to be seen, therefore, whether
(2nd ed.). Hillsdale, NJ: Erlbaum.
stimulation seekers at age 3 are more likely to develop sensation-
Cohen, L., Dingemans, P. M. A. J., Lesnik-Oberstein, M., & van der Vlugt,
seeking personalities in adulthood.
H. (1983, August). Report of the findings with the Sensation Seeking
The prospective, longitudinal nature of the findings showing
Scale in Holland. Poster session presented at the 91st Annual Conven-
associations between stimulation seeking at age 3 and cognitive
tion of the American Psychological Association, Anaheim, CA.
Detterman, D. K., & Sternberg, R. J. (Eds.). (1982). How and how much
ability at age 11 is consistent with but does not prove the hypoth-
can intelligence be increased? Norwood, NJ: Ablex.
esis that stimulation-seeking behavior leads to enhanced cognitive
Dunn, L. M. (1965). Expanded manual for the Peabody Picture Vocabu-
ability. Future studies need to replicate the nature of this longitu-
lary Test. Minneapolis, MN: American Guidance Service.
dinal relationship in Western societies to assess cross-cultural
Fagan, J. F. (1984). The relationship of novelty preferences during infancy
generalizability, although the consistency of the current findings
to later intelligence and later recognition memory. Intelligence, 8, 339 –
with cross-sectional findings from the United States suggests that
346.
the findings should generalize. More critically, experimental ma-
Fernandez-Teruel, A., Escorihuela, R. M., Casetellano, B., Gonzalez, B., &
nipulations that encourage appropriate stimulation seeking in in-
Tobena, A. (1997). Neonatal handling and environmental enrichment
fants and preschoolers compared with controls are needed to
effects on emotionality, novelty/reward seeking, and age-related cogni-
demonstrate a casual relationship between stimulation seeking and
tive and hippocampal impairments: Focus on the Roman rat lines.
cognitive ability. If such causality can be demonstrated in future
Behavior Genetics, 27, 513–526.
Filliot, J. M. (1974). La traite des esclaves vers les mascareignes au XVIIIe
studies, results could have significant theoretical implications for
siecle [The slave trade to the Mauritius, Rodriguez, and reunion in the
the development of intelligence, practical implications on how
18th century]. Paris: ORSTOM.
intelligence can be increased (Detterman & Sternberg, 1982), and
Glicksohn, J., & Abulafia, J. (1998). Embedding sensation seeking within
clinical implications for behavioral conditions such as antisocial
the Big Three. Personality and Individual Differences, 25, 1085–1099.
behavior that are linked to low intelligence (Moffitt & Silva,
Gottfried, A. E. (1990). Academic intrinsic motivation in young elemen-
1988). Provision of an educationally enriched head start to children
tary school children. Journal of Educational Psychology, 82, 525–538.
is known to result in short-term increases in IQ that disappear by
Gottfried, A. W., Gottfried, A. E., Bathurst, K., & Guerin, D. W. (1994).
age 11 (Neisser et al., 1996). Findings of this study suggest that
Gifted IQ: Early developmental aspects. New York: Plenum.
stimulation-seeking children may provide for themselves a more
Halverson, C. F., & Waldrop, M. F. (1976). Relations between preschool
potent and continuous environmental enrichment than traditional
activity and aspects of intellectual and social behavior at age 7 1?2.
educational enrichment can provide, and in contrast to such en-
Developmental Psychology, 12, 107–112.
richment programs, can produce long-term IQ changes that last
Jo¨reskog, K. G., & So¨rbom, D. (1993). LISREL 8 (Version 8.03). Chicago:
Scientific Software International.
throughout childhood.
Kafry, D. (1982). Sensation seeking of young adults. Personality and
Individual Differences, 3, 161–166.
References
Kemple, K. M., David, G. M., & Wang, Y. (1996). Preschoolers’ creativ-
ity, shyness, and self-esteem. Creativity Research Journal, 9, 317–326.
Allen, M. J., & Yen, W. M. (1979). Introduction to measurement theory.
Kish, G. B., & Busse, W. (1968). Corr