JOURNAL OF RESEARCH IN PERSONALITY 14,224-234 (1980)

A Twin Study of Personality and Temperament during

Middle Childhood

ADAM P. MATHENY,JR. AND

ANNE BROWN DOLAN
University of Louisville

In a study of 105 same-sex twin pairs, ages 7 to 10 years, maternal ratings on 23

bipolar scales yielded six factors designated as compliant morality, applied cogni-
tive, sociability, emotionality, tough-mindedness, and activity-distractibility.
Analyses by twin pairs (68 identical, 37 fraternal) indicated that correlations for
identical pairs were significantly higher than fraternal pairs on all but one factor:
tough-mindedness. Profile analyses for the six factors indicated that the profiles of
scores across the factors were more concordant for the identical twin pairs.
Correlations obtained from the individual scales produced differences between the
identical and fraternal twin pairs as well. Overall, the data suggested that several
components and the total organization of those components of personality and
temperament are genetically influenced.

The vigor of growth in the number of studies examining possible genetic

contributions to individual variations in a child’s personality and temper-
ament attests to the lack of completeness found for explanations em-
phasizing exogenous factors alone. By and large, the studies have em-
ployed twin pairs, and with varying success, the results have indicated
that identical twins are more similar than same-sex fraternal twins for a
wide variety of behaviors broadly subsumed under trait names such as
sociability, anxiety, impulsivity, extraversion, and the like.
For practical reasons, the largest studies of twins before the adult years
have been made of adolescents given standard personality inventories
eliciting self-reports (e.g., Gottesman, 1966; Loehlin & Nichols, 1976;

This research has been supported in part by grants from the Grant Foundation, the
National Institute of Mental Health (MH 23884), the National Institute of Child Health and
Human Development (HD 07200), and Office of Child Development (90~C-922bR. S.
Wilson, Principal Investigator. The authors are indebted to R. S. Wilson for critical com-
ments regarding the analyses, and to M. Hinkle for help in data preparation.
Requests for reprints should be sent to Adam P. Matheny, Jr., Louisville Twin Study,
Room 111-H Medical-Dental Research Building, P.O. Box 35260, Louisville, KY 40232.

224
0092-6566/80/020224-11$02.00/O
Copyright @ 1980 by Academic Press, Inc.
All tights of reproduction in any form reserved.
 TWIN STUDY OF PERSONALITY/TEMPERAMENT 225

Vandenberg, 1962). Inventories requiring self-reports are less practical for

younger twins, however, and other methods, particularly ratings of twins
by parents or others (Lytton, 1974), have provided comparatively better
yields of information. Typically, the data from rating scales have shown
that identical twins are more similar than fraternal twins for several broad
dispositions: activity, sociability or extraversion, emotionality, and im-
pulsivity (Buss & Plomin, 1975; Starr, 1969). The results have provided a
bridge between those dispositions found for adolescent twins and those
found for infant twins (Freedman, 1965); however, the data have been
restricted to a small domain of behavioral characteristics construed in part
with adult status in mind. As Vandenberg has indicated (1962), the study
of hereditary factors in personality might profit more from examining a
host of rudimentary individual differences commonly observed among
children and not necessarily be limited by preexisting ideas of what the
crucial (i.e., clinically important) aspects of personality and temperament
might be.
As part of a series of ongoing investigations of twins’ development from
infancy to adolescence, the present research was undertaken to examine
hereditary influences on children’s behaviors during middle childhood.
Previously, parental reports of similarities and differences within younger
twin pairs (Wilson, Brown, & Matheny, 1971) have outlined common
developmental themes during infancy and early childhood. The data pre-
sented in this paper draw upon those previous reports and supplement
parental observations in order to take into account the increased diversity
of experiences available in middle childhood.
METHOD
Sample. The data were collected from rating scales filled out by mothers of 105 same-sex
pairs (58 female, 47 male) of white twins who participated in the Louisville Twin Study.
According to the index of fathers’ occupations as reported by Reiss (l%l), the families
represented the entire socioeconomic range found in the Louisville metropolitan area.
Distribution of families whose SES scores fall in the lowest to highest quintile of the index
was approximately IO,25 20,30, and 15%. The twins ranged in age from 7 to 10 years, the
median being 8 years. Zygosity had been determined by blood typing when the twins were
about 3 years old; there were 68 pairs of identical and 37 pairs of same-sex fraternal twins by
this criterion. There were no statistically significant differences between identical and
fraternal pairs or female and male pairs in age or in socioeconomic status; however, there
was a trend for identical pairs to represent slightly higher socioeconomic status.
Raring scales. Five-point rating scales were developed largely from the results garnered
from over 3000 interviews with mothers of twins 6 years of age and younger (Wilson et al.,
1971). To these scales, denoting, among other characteristics, sociability, emotionality,
sustained attentiveness, and level of activity, were added scales related to several categories
of temperament (Thomas & Chess, 1977), empirically derived source traits of personality
(Porter & Cattell, 1%8), and aspects of interpersonal and prosocial development (Piaget,
1948; Werner, 1948). Finally, scales pertaining to imagination, breadth of interests, and
226 MATHENY AND DOLAN

humor were included because such characteristics had been provided by some of the
mothers of twins as salient aspects of differences between twins.
Bipolar rating scales were composed so that the anchors were typically phrased in
everyday terms most easily understood by mothers from diverse backgrounds. Preliminary
analyses led to the deletion of rating scales so highly correlated with other scales as to be
redundant, Through successive efforts, a final set of 60 ratings was created; however, the
present report was based on only 23 of the 60 scales because those provided the most
complete data for the largest sample of twins.
The rating scales, one set for each twin, were completed by the mother during a routine
visit to the study. The mothers were given oral and written instructions for completing the
scales. The mothers were told to refrain from rating any scale felt to be ambiguous or not
applicable to their twins. Test-retest reliability data for the scales were obtained about 1
month later by sending duplicate sets of the rating scales to 20 of the mothers who then
completed the scales at home. The percentages of agreement (within 1 point) between the
mothers’ first and second ratings ranged from 80 to lOO%, the mode being 95% for the 23
rating scales reported in this study.
Critics of the twin method have been quick to point out that expectations for the greater
behavioral similarities of identical twin pairs than fraternal twin pairs have been due to
inequality of environmental effects. In the present study, the mothers’ ratings may have
been influenced or biased by inequalities in the mothers’ expectations of similarities;
mothers of identical pairs expected their twins to be more similar than mothers of fraternal
pairs. Although there is no obvious way to avoid this problem, evidence has steadily
accumulated that provides support for the twin method in general and the use of parental
observations of twins in particular (Cohen, Dibble, & Grawe, 1977; Loehlin & Nichols,
1976; Lytton, 1977; Matheny, Wilson, & Dolan, 1976; Plomin, Willerman, & Loehlin, 1976;
Starr, 1968).

RESULTS
Table 1 provides the means obtained for the ratings of the female and
male twins. Each scale is identified by the high score description. The few
significant sex differences found between the two groups of means indi-
cate that the males were rated as being more daring, active, impatient, and
socially bold. In combination, these behaviors depict the males as being
more assertive (aggressive?) than the females. The females’ higher ratings
were for prosocial or compliant behaviors in that they were rated higher
for the scales conscientious, self-disciplined, and takes others’ views. A
previous report (Cohen, et al., 1977) has shown similar sex differences for
twin children between the ages of one and six years.
In order to reduce the number of dimensions to be examined, principal
component factor analysis was applied to the scale by scale intercorrela-
tions obtained from the ratings. Unities were placed in the main diagonal,
followed by a varimax criterion of components, and factors with eigen-
values equal to or greater than 1.00 were retained for interpretation.
Application of the analysis by sex was considered; however, the relia-
bility of sex differences in the factor structures could not be assumed for
samples of this size. Therefore, the female and male ratings were com-
bined for the present study. As a step toward cross-validation of the factor
 TWIN STUDY OF PERSONALITY/TEMPERAMENT 227

TABLE 1
RATING SCALES: SEX DIFFERENCES

Means

Scale: high score description Females Males

1. Impatient 2.70 3.16*

2. Daring 2.38 2.95*
3. Overly active 2.02 2.34*
4. Outgoing 3.93 3.84
5. Grouchy 1.99 2.08
6. Hardhearted 1.69 1.84
7. Moody 2.38 2.36
8. A “loner” 1.64 1.75
9. Conscientious 3.70* 3.28
10. Socially bold 3.11 3.4s*
11. Self-disciplined 3.42* 2.84
12. Inattentive 1.60 1.80
13. Tense 2.52 2.66
14. Excitable 3.41 3.53
15. Emotional 2.03 2.19
16. Slow-to-discipline 2.54 2.72
17. Quick tempered 3.29 3.47
18. Dependable 3.20 3.02
19. Takes others’ views 3.22* 2.87
20. Imaginative 2.96 3.10
21. Artistic 2.66 2.72
22. Wide Interests 3.33 3.33
23. Witty 2.88 2.97

* Difference between means significant, by t test, p < .05.

structure, the total sample was split into two equivalent samples and
factor analysis was applied to the data from each. The split was obtained
by taking the first-named twin from each twin pair to make up one group
and the second group was made up of the co-twins from each pair. With
eigenvalues equal to or greater than 1.00, only six factors were retained
from each group. The order of extraction of the six factors was not the
same for both groups, but the six factors were equivalent between groups
according to items principally loading on each factor. As a further check,
the scale by scale intercorrelation matrices obtained from the groups’
ratings were examined to see if appreciable correspondence existed for
both sign and magnitude of the correlations. Except for minor differences
in the magnitude of the correlations from the two matrices, the two groups
provided comparable patterns of correlations as well. Consequently, the
two groups were recombined for a factor analysis based on the total
sample.
228 MATHENY AND DOLAN

Factors
Table 2 presents the loadings of the rating scales on the six factors with
eigenvalues equal to or greater than 1.00. Only those loadings of .30 or
greater are provided in the table for interpretation.
The first factor pertained to compliance with rules and expectations set
by others. At the opposite end, risk taking and disobedience appeared as
contrary indications of compliance. Descriptively, the factor resembled
Factor E outlined by Porter & Cattell (1968). In terms of the development
of morality, the factor represented a level of conventional morality
(Piaget, 1948) imposed by others. It is of interest to note that although
taking another’s view or perspective, a prosocial aspect of development,
contributed to this factor, no other socially oriented item did. For descrip-
tive purposes, the factor was called Compliant Morality.
Factor II was one representing all of the scales of a somewhat cognitive
nature-embodying imagination, breadth of interests, artistic inclinations,
and humor. Apparently, higher scoring children on these scales were
more placid-a link between readily applied cognitive skills and the more
tranquil aspects of emotionality depicted in Factor IV. This factor was
labeled Applied Cognitive.
Sociability, as represented by the scales of Factor III, incorporated the
usual notions associated with extraversion, a characteristic figuring prom-
inently in twin studies of temperament (Buss & Plomin, 1975; Starr, 1969;
Wilson et al., 1971). In contrast with Factor I this factor suggested a
quantitative dimension of interpersonal behaviors rather than a qualitative
one. That is, the children could have been judged as being more or less
gregarious without having been judged to have a correlative display of
interpersonal rules of behavior or taking others’ points of view. From
these data, therefore, prosocial behaviors or social perspectivism (Piaget,
1948; Werner, 1948) appear not to be directly related to the concept of
sociability. In part, negative mood added to the opposite dimension of the
factor, and a dark, gloomy, nay-saying disposition was somewhat anti-
thetical to characteristics of an extraverted nature.
Negative mood also doubled with attributes of Factor IV which was com-
prised of scales corresponding directly to a pattern of emotional behav-
iors emerging in the early childhood of twins (Wilson et al., 197 1). In one form
or another this characteristic has appeared frequently in the studies of
children’s personality and temperament (Buss & Plomin, 1975; Digman,
1964; Thomas & Chess, 1977). The negative to positive affect provided by
this factor appeared to relate, through one scale or another, to several
factors isolated in this study-suggesting that emotional lability or stabil-
ity makes at least a partial contribution to many domains of children’s
behaviors.
TWIN STUDY OF PERSONALITY/TEMPERAMENT 229

s
230 MATHENY AND DOLAN

Factor V had contributions from four scales, all of which provided

higher mean scores for males than females. The factor appeared to tap a
no-nonsense, bold style of interaction. This factor seemed like the tough-
versus tender-minded distinction outlined by Cattell, Eber, and Tatsuka
(1970). High scorers would be seen as forceful, demanding, unsentimen-
tal, and impatient with or complaining about obstacles.
The final factor involved activity and inattentiveness or distractibility.
The behavioral aspects of this factor have figured so prominently in
studies of childrens’ personality and temperament that no further explana-
tion is needed.
Twin Concordance
In the preceding section, the data from the ratings of the twin children
were presented in the form of summarized behavioral dimensions ob-
tained from factor analysis. The next step was to examine the degree of
concordance for those behavioral dimensions shared by twins belonging
to pairs. Factor scores, obtained from the factor analysis, were used for
the comparisons. These scores were in standard score form, distributed
with a mean of zero, variance of one, and the vectors of the scores were
orthogonal. In this application, concordance was demonstrated by within-
pair (intraclass) correlations based on the six factor scores empirically
determined for each twin. The expectation was that concordance found
within pairs of identical twins would be more pronounced than concor-
dance found within same-sex fraternal pairs.
Table 3 provides the within-pair correlations for the identical and fra-
ternal pairs. The general feature of the correlations was that all factors had
preponderantly higher correlations for the identical pairs than for the

TABLE 3
WITHIN-PAIR CORRELATIONS FOR IDENTICAL AND FRATERNAL TWINS’
BEHAVIORAL FACTOR SCORES

Identical pairs Fraternal pairs

Factor (N = 68) (N = 37)

I Compliant Morality .60* .2.5

II Applied Cognitive .67* .I8
III Sociability .56* .06
IV Emotionality .45* -.ll
V Tough-minded .39 .15
VI Activity-Distractibility .66* .19
Profile correlation
for six factors .61* .13

* p < .05, one tailed for identical > fraternal

 TWIN STUDY OF PERSONALITY/TEMPERAMENT 231

fraternal pairs, but the differences were not always significant. Some of
the fraternal correlations were not appreciably different from zero-order
correlations, and in one instance, the fraternal correlation was even
negative. Loehlin and Nichols (1976) have noted that twin studies of
personality and temperament often produce low or sometimes negative
correlations for fraternal pairs, but the magnitude of the correlations
fluctuates considerably among characteristics examined and there is no
appreciable consistency among results from different studies.
Turning from the concordance of the twin subgroups for each factor, it
was of interest to determine if the identical twins were more concordant
than fraternal twins for the pattern of factor scores across the six factors.
That is, did the identical twin pairs display a more congruent pattern in the
total organization of personality and temperament as provided by the
factors? The analysis determined how closely a profile of scores for each
twin was matched by that of his co-twin over factors, and was obtained
from an analysis of variance design (Haggard, 1958) adapted for twin data.
Table 3 also provides the profile correlations. It is quite apparent that the
identical pairs were more concordant than the fraternal pairs for the
pattern of factor scores, irrespective of the degree of concordance for
each factor considered in isolation.
Because specific factor structures often have been difficult to replicate,
particularly when sample sizes have been small, Table 4 provides the
within-pair correlations derived from each of the 23 behavioral scales.
Thereby, the differences in concordances of the identical and fraternal
twin pairs can be evaluated without regard to factor organization. For
convenience, the scales are listed according to their primary position in
the factor structure.
The correlations found in Table 4 were based on the raw scores from
each rating scale. Despite the fact that the scores were unweighted and
were taken from each individual scale, the general trend among the
correlations was similar to that found for the correlations derived from a
weighted combination of scales.
DISCUSSION
The results from the factor analyses of the rating scales revealed several
characteristics of personality and temperament that have figured largely in
the literature on childhood. Among the children, the prominent features
or orientations of individual differences pertained to the degree of emo-
tional lability, sociability, activity, and application of commonplace, cog-
nitively based skills. Moreover, compliance with rules of interpersonal
conduct and a dimension of tough- versus tender-mindedness appeared as
isolable components of the children’s behaviors. In large part, the results
corresponded to previous reports on similar aspects of children’s per-
232 MATHENY AND DOLAN

TABLE 4
WITHIN-PAIR CORRELATIONS FOR IDENTICAL AND FRATERNAL TWINS’
BEHAVIORAL RATING SCALES

Identical pairs Fraternal pairs

Rating scales (N = 68) (N = 37)

I Conscientious .60 .lO

Dependable .52 .27
Self-disciplined .48 .20
Daring .37 -.20
Slow-to-discipline .58 .I1
Takes others’ views .58 .26
II Imaginative .69 .29
Artistic .63 .05
Interests .75 .57
Witty .62 .I3
Excitable .37 .02
III Socially bold .45 -.ll
Outgoing .23 - .20
Loner .31 .15
Grouchy .41 .09
Moody .35 .03
IV Tense .58 -.Ol
Emotionality .25 -.28
Quick tempered .60 .14
V Hardhearted .28 .29
Impatient .40 -.I1
VI Active .49 .08
Inattentive .60 .20

sonalities and temperaments (e.g., Bronson, 1966; Escalona, 1968; Dig-

man, 1965; Schaefer & Bayley, 1963; Thomas & Chess, 1977; Walker,
1967). Although the results were determined empirically, rational cluster-
ings of the scales would have anticipated somewhat similar findings.
Additional significance was added to the analyses because the data, exam-
ined for twin pairs, provided evidence for genetic influences on the be-
havioral features considered singly or in total combination as organized
patterns of personality and temperament.
According to method, behaviors, and the age of the twins, previous
efforts by Buss and Plomin (1975) were most comparable to the present
study. Buss and Plomin isolated emotionality, activity, sociability, and, to
a lesser extent impulsivity, as the primary factors of a temperament
theory of personality development, and presented evidence from highly
homogeneous scales that there was a pronounced genetic influence on
these four temperaments. The first three of their factors were identified in
the present study, and both lines of investigation indicate that emotional-
 TWIN STUDY OF PERSONALITY/TEMPERAMENT 233

ity, sociability, and activity are isolable, and genetically influenced, as-
pects of children’s behaviors. The present study, however, indicated that
the three factors were not as prominent as previous data indicated.
Perhaps the fact that the present scales were obtained more empirically
from a wider range of behaviors would account for the differences be-
tween these results and those reported by.Buss and Plomin. In any case,
the three factors combined accounted for less than 30% of the total
variance. Other characteristics-compliance with rules or principles, or
some combination of putatively cognitive behaviors-seemed to have had
equal or greater importance than any one of the “primary” temperament
factors considered alone.
The greater similarity of the identical pairs than fraternal pairs for the
profile correlations suggests that there is a genetic regulation at a more
complex level-the organization of personality and temperament. In es-
sence, each identical twin within a pair created a pattern of scores across
the eight factors, and that pattern was rather closely matched by the
identical co-twin. It should be stressed that the similarity for pattern is not
dependent on the congruence for level of scores for the twin pair on each
factor. The twins may differ in the level of factor scores, yet they may be
quite similar in the direction and degree of change of scores from one
factor to the next.
As a final note, it is apparent that the relative contribution of genetic
influences on factors of personality and temperament depends on the
specificity of the behavioral factors examined. For example, the sociabil-
ity factor in the present study was composed of items denoting sociability
in the usual sense: gregarious, needing to be with people, and the like.
Described in this way, sociability was found to be an isolable and ge-
netically influenced factor, but its relations with other factors of a social
nature did not produce any evidence for a strong link between actions
toward others and actions taken to be with others, a distinction between
the quality and quantity of social interactions. Therefore, one might
expect to find genetic influences on some types of social behaviors and not
others. In this regard, Horn, Plomin, & Rosenman (1976) provided data
that indicated a pronounced genetic influence on some aspects of sociabil-
ity (conversational poise with strangers and social ease) but not on others
(leadership activities and social exhibitionism). If such nuances exist in
the varieties of ways children are socially oriented, specificities of genetic
influences should also be hypothesized. It is equally likely that such
specificities may be found among other molar characteristics of per-
sonality and temperament as well.
REFERENCES
Bronson, W. C. Central orientations: A study of behavior organization from childhood to
adolescence. Child Development, 1966, 37, 125-155.
 234 MATHENY AND DOLAN

Buss, A. H., & Plomin, R. A temperament theory of personality development. New York:
Wiley, 1975.
Cattell, R. B., Eber, H. W., & Tatsuoka, M. M. Handbook for the Sixteen Personality
Factor Questionnaire (16PF). Champaign, Ill.: Institute for Personality and Ability
Testing, 1970.
Cohen, D. J., Dibble, E., & Grawe, J. M. Fathers’ and mothers’ perceptions of children’s
personalities. Archives of General Psychiatry, 1977, 34, 480-487.
Digman, J. M. Child behavior ratings: Further evidence of a multiple factor model of child
personality. Educational and Psychological Measurement, 1965, 25, 787-799.
Escalona, S. The roots of individuafity. Chicago: Aldine, 1968.
Freedman, D. G. An ethological approach to the genetic study of human behavior. In S. G.
Vandenberg (Ed.), Methods and goals in human behavior genetics. New York:
Academic Press, 1%5.
Gottesman, I. Genetic variance in adaptive personality traits. Journal of Child Psychology
and Psychiatry, 1%6, 7, 199-208.
Horn, J. M., Plomin, R., & Rosenman, R. Heritability of personality traits in adult male
twins. Behavior Genetics, 1976, 6, 17-30.
Loehlin, J. C., & Nichols, R. C. Heredity, environment and personality. Austin, Texas:
Univ. Texas Press, 1976.
Lytton, H. Comparative yield of three data sources in the study of parent-child interaction.
Merrill-Palmer Quarterly, 1974, 20, 53-64.
Lytton, H. Do parents create, or respond to, differences in twins? Developmental Psychol-
ogy, 1977, 13, 456-459.
Matheny, A. P., Jr., Wilson, R. S., & Dolan, A. Brown. Relations between twins’ similarity
of appearance and behavioral similarity: Testing an assumption. Behavior Genetics,
1976, 6, 343-351.
P&et, J. The moral judgement of the child. New York: Free Press, 1948.
Plomin, R., Willerman, L., & Loehlin, J. C. Resemblance in appearance and the equal
environments assumption in twin studies of personality traits. Behavior Genetics, 1976,
6, 43-52.
Porter, R. B., & Cattell, R. B. Children’s personality questionnaire. Champaign, Illinois.:
Institute for Personality and Ability Testing, 1968.
Reiss, A. J., Jr. Occupations and social status. New York: Free Press of Glencoe, l%l.
Starr, S. Environmental bias in twin studies. Eugenics Quarterly, 1%8, 15, 34-40.
Starr, S. Social introversion-extraversion as a heritable response. Child Development,
1969, 40, 823-832.
Schaefer, E. S., & Bayley, N. Maternal behavior, child behavior, and their intercorrelations
from infancy through adolescence. Monographs of the Society for Research in Child
Development, 1%3, 28 (3, Serial No. 87).
Thomas, A., & Chess, S. Temperament and development. New York: Brunner/Mazel, 1977.
Vandenberg, S. G. The hereditary abilities study: Hereditary components in a psychological
test battery. American Journal of Human Genetics, 1%2, 14, 220-237.
Walker, R. N. Some temperament traits in children as viewed by their peers, their teachers,
and themselves. Monographs of the Society for Research in Child Development, 1967,
32 (6, Serial No. 114).
Werner, H. Comparative psychology of mental development. New York: Intern. Univ.
Press, 1948.
Wilson, R. S., Brown, A. M., & Matheny, A. P., Jr. Emergence and persistence of
behavioral differences in twins. Child Development, 1971, 42, 1381-1398.