Genetic and Cultural Transmission of
Antisocial Behavior: An Extended Twin
Parent Model
Hermine H. Maes,1,2,3 Judy L. Silberg,1 Michael C. Neale,1,4 and Lindon J. Eaves1,4
1
Department of Human Genetics,Virginia Institute for Psychiatric and Behavioral Genetics,Virginia Commonwealth University, Richmond,Virginia,
United States of America
2
Massey Cancer Center,Virginia Commonwealth University, Richmond,Virginia, United States of America
3
Faculty of Kinesiology and Rehabilitation Sciences, Katholieke Universiteit Leuven, Leuven, Belgium
4
Department of Psychiatry,Virginia Institute for Psychiatric and Behavioral Genetics,Virginia Commonwealth University, Richmond,Virginia, United
States of America
C
onsiderable evidence from twin and adoption
studies indicates that both genetic and shared
environmental factors play a substantial role in the
liability to antisocial behavior. Although twin and
adoption designs can resolve genetic and environmental influences, they do not provide information
about assortative mating, parent–offspring transmission, or the contribution of these factors to trait
variation. We examined the role of genetic and environmental factors for conduct disorder (CD) using a
twin–parent design. This design allows the simultaneous estimation of additive genetic, shared and
individual-specific environmental effects, as well as
sex differences in the expression of genes and environment in the presence of assortative mating and
combined genetic and cultural transmission. A retrospective measure of CD was obtained from twins
and their parents or guardians in the Virginia Twin
Study of Adolescent Behavior Development and its
Young Adult Follow up sample. Both genetic and
environmental factors play a significant role in the
liability to CD. Major influences on individual differences appeared to be additive genetic (38%–40%)
and unique environmental (39%–42%) effects, with
smaller contributions from the shared environment
(18%–23%), assortative mating (~2%), cultural
transmission (~2%) and resulting genotype-environment covariance. This study showed significant
heritability, which is slightly increased by assortative
mating, and significant effects of primarily nonparental shared environment on CD.
In this article, we aim to accomplish four goals. First,
we highlight some of the main findings published so
far with respect to individual differences in internalizing and externalizing behavior from a range of
perspectives. Second, we briefly introduce the genetic
epidemiologic literature on antisocial behavior, with a
special focus on longitudinal applications. Third, we
describe the overall design, sample and measures in
136
the Virginia Twin Study of Adolescent Behavior
Development (VTSABD) and Young Adult Follow Up
(YAFU) in detail as well as the measures on antisocial
behavior used in the current analyses. Fourth, we
illustrate some of the unique features of its design by
extending the classical twin study with data collected
from biological and nonbiological (referred to as stepfor this article) parents to examine the role of genetic
and environmental factors in the variation of antisocial behavior. The added information from
parent–offspring correlations and spousal correlations
will allow us to (i) disentangle genetic and cultural
transmission, (ii) partition the environmental variance
into influences shared with parents, with twin siblings
and those specific to the individual, and (iii) test for
and quantify the effects of assortative mating. This
design thus provides an extended perspective to the
traditional longitudinal study by studying the sources
of continuity and change across generations.
Although we only illustrate this using retrospective
data in adolescence and one single measure, these
type of analyses can be readily applied to include
prospective longitudinal data from childhood to adolescence and young adulthood.
The VTSABD (Hewitt et al., 1997) and its YAFU
were designed specifically to focus on the developmental trajectories of behavior and disorders from
childhood to young adulthood and to study the interplay of genes and environment in shaping individual
differences in internalizing and externalizing behaviors. The study used a rich assessment battery
including dimensional and categorical measures, multiple raters, and environmental indices. Data were
collected at up to five occasions with varying intervals
Received 19 March, 2006; accepted 15 November, 2006.
Address for correspondence: Dr H. H. Maes, Virginia Institute for
Psychiatric and Behavioral Genetics, PO Box 980003, Richmond, VA,
23298-0003, USA. E-mail:
[email protected]
Twin Research and Human Genetics Volume 10 Number 1 pp. 136–150
Genetic and Cultural Transmission of Conduct Disorder
from genetically related and unrelated individuals,
aged 8 through 30.
Previous analyses of the data in the VTSABD have
focused on baseline prevalence rates and univariate
genetic analyses. Several publications have addressed
the representativeness of the population-based
Adolescent Behavior Development (ABD) sample for
nosological and epidemiological studies. CBC-L externalizing behaviors were slightly but significantly
elevated in the ABD twin sample compared with norms
for nontwins and elevation for internalizing symptoms
in older twins (Gau et al., 1992). Meyer et al. (1996)
showed that families in the ABD sample demonstrate
only slight differences in the distribution of socioeconomic status (SES) from the census tracts in which the
targeted twins were located, there being a slight deficit
of the very highest SES respondents. The impact of this
bias on rates of psychopathology was slight. Simonoff
et al. (1997) found that age- and sex-specific prevalence
rates of the principal Diagnostic and Statistical Manual
of Mental Disorders (3rd ed., rev.; DSM-III-R;
American Psychiatric Association, 1987) diagnoses at
Wave 1 were close to those from other epidemiological
studies. Pickles et al. (2001) showed that prediction of
depressive symptomatology at Wave 2 is affected only
by Wave 1 symptoms and not impairment. For conduct
and oppositional defiant disorders, prior impairment
improved prediction of later symptomatology.
Eaves et al. (1997) presented cross-sectional modelfitting results, pooled across ages, for symptom counts
and factorially derived scales for ratings by mothers,
fathers and children on all aspects of behavior; then
showed that genetic contributions were generally small
to moderate, with some evidence of shared environmental influences on questionnaire ratings of conduct
problems. Sibling contrast effects were found for
parental ratings of attention-deficit/hyperactivity disorder (ADHD), but not for teacher ratings (Simonoff et
al., 1998). Manifest anxiety ratings by twins and
parents showed that children and parents rate genetically different aspects of the phenotype (Topolski et al.,
1999). Reynolds et al. (1996) analyzed Wave 1 oral
reading performance. Maes et al. (1999) reported crosssectional analysis of the first wave of data relating to
the lifetime and current use of alcohol, drugs and
tobacco. The findings demonstrate that the contribution of genetic and environmental factors depends
significantly on context and severity of use.
The VTSABD assesses a wide variety of putative
indices of the family and individual environment and is
therefore valuable for the study of nature-nurture interplay. Silberg at al. (1999) showed that the genetic
variance in depression symptoms increases significantly
during puberty in girls, as does the variance in life
events. Furthermore, (1) genetic variance in female postpubertal depression is greater in twins exposed to higher
‘doses’ of adverse environments (G × E interaction,
Silberg et al., 2001b); (2) depression is associated with
adverse life events (G–E correlation, Silberg et al.,
1999); and (3) prepubertal genetic differences in
anxiety predict postpubertal genetic differences in
depression (heterotypic continuity, Silberg et al.,
2001a). Subsequently, Eaves et al. (2003) used
Markov Chain Monte Carlo (MCMC) methods to
analyze the joint roles of G × E and rG–E in the developmental transition from early anxiety to later
depression, thus integrating Silberg et al.’s papers into
a single analysis of the simultaneous interaction and
correlation of random genetic and environmental
factors and identified three separate pathways by
which prepubertal differences in anxiety may produce
later differences in depression. Analyses focused on
the role of environment, G × E interaction and G–E
correlation for disruptive behavior have shown significant associations of conduct disorder (CD) with
shared environmental factors: parental alcoholism and
presence versus absence of a stepfather (Foley, Pickles,
et al., 2004); maternal smoking in pregnancy (Silberg
et al., 2003); marital discord and family adaptability
(Meyer et al., 2000); and differential treatment of
individual twins (Carbonneau et al., 2002).
Preliminary developmental analyses of antisocial
behavior measures suggested that the shared environment is more important prior to adolescence and that
genes play a greater role afterwards (Lyons et al.,
1995) and that puberty is an important threshold for
change in the expression of genetic factors on CD
(Eaves et al., 2000). Thus, for depression, genetic differences create differences in sensitivity to the
environment (G × E) while for anxiety and antisocial
behavior, genes influence exposure to the environment
(rG-E). Recent publications have explored the main
effects and interaction of childhood adversity
(parental neglect and exposure to violence between
parents in the home) and genotype at the (X- linked)
MAO-A locus in boys and girls from the VTSABD.
Foley, Eaves, et al. (2004) detected a significant G × E
interaction in boys that followed closely the pattern
reported by Caspi et al. (2002). A parallel analysis in
girls (Prom et al., 2006) shows a significant main
effect of the locus, with greater risk in the ‘low activity’ homozygotes with weak evidence of G × E.
A range of VTSABD analyses have focused on patterns of comorbidity and heterogeneity of disorders.
Multivariate analyses of mood disorders have shown a
common underlying genetic basis to prepubertal overanxious disorder (Silberg et al., 2001a) and
postpubertal juvenile depression in girls, while prepubertal depression may have a shared-environmental
component that is indexed by separation anxiety. Rowe
et al. (2002) explored the role of genes in comorbidity
between symptoms of bulimia and other disorders.
Analysis of the Wave 1 externalizing data (including
ADHD and oppositional disorder [ODD]) demonstrated the need to integrate the concepts of etiological
heterogeneity with those of comorbidity. Silberg,
Meyer, et al. (1996) identified three latent classes in
child and parent ratings of disruptive behaviors in male
Twin Research and Human Genetics February 2007
137
Hermine H. Maes, Judy L. Silberg, Michael C. Neale, and Lindon J. Eaves
twins: a ‘pure’ CD class, distinguished from ‘normals’
by additive genetic and shared environmental effects; a
comorbid CD/ADHD class, distinguished by additive
and nonadditive genetic effects; and a multiply disordered class characterized by additive genetic
differences. A cross-sectional analysis of the first wave
of data (Silberg, Rutter, et al., 1996) showed developmental change in the pattern of heterogeneity/
comorbidity in externalizing disorders, such that in
younger male twins there was virtually complete
genetic correlation between ADHD and conduct disturbance, while in older twins a further set of genetic
effects specific to CD are expressed (Nadder et al.,
2002). Eaves et al. (2000) showed significant genetic
and environmental correlation between the components of ADHD: impulsivity; inattention and
hyperactivity, but also detected significant genetic
effects specific to the components suggesting that
ADHD is genetically heterogeneous. There was
marked consistency of genetic effects on CD and ODD
across sexes, but the sexes were heterogeneous with
respect to genetic effects on ADHD. Nadder et al.
(2001) detected rater- and measure-specific genetic
effects on ADHD. Simonoff et al. (1998) fitted a multivariate ‘ACE’ model to four a priori subtypes of
disruptive behavior: property violations; status violations; oppositional behavior; and aggression. They
concluded that child self-reports showed high genetic
correlations between subtypes, whereas maternal
ratings showed a high degree of specificity across subtypes. Consistency between behavioral ratings by
children, parents and teachers was low and parents
attributed greater similarity to the symptoms of their
twin children than can be explained by similarity in
self-reports (Simonoff et al., 1995). Children’s selfreports were less highly correlated between twins than
parent or teacher reports (Eaves et al., 1997) but selfreports seem to show similar long-term consistency. In
the case of ADHD, parents appear to contrast twins
with one another, whereas teachers do not (Simonoff et
al., 1998). Eaves et al. (2000) noted significant ‘genetic
overlap’ between child and parent ratings of CD and
ODD, but also rater-specific genetic effects that imply
raters are sensitive to genetically distinct aspects of a
child’s behavior, or that they experience behavior in different contexts that interact with gene expression. The
rater’s own psychopathology may affect patters of rater
inconsistency (Foley et al., 2005).
The longitudinal nature of the VTSABD study has
permitted the analysis of continuity and change of the
relative roles of genes and the family environment during
development. The shared environment is more important
prior to adolescence and genes play a greater role afterwards for both depression (Silberg et al., 1999, 2001a,
2001b) and lifetime versus current use of alcohol (Maes
et al., 1999). In the short-term and mid-term (1–2 years),
there is little evidence to decide whether parent–offspring
interaction influences child psychopathology more than
child psychopathology influences parent–offspring inter-
138
action (Carbonneau et al., 2002, Foley, Pickles, et al.,
2004). Puberty is an important threshold for change in
the expression of genetic factors on anxiety/depression
(Eaves et al., 2003), ADHD/CD (Silberg, Rutter, et al.,
1996) and obesity (Meyer et al., 1997). Recently with
the availability of up to five waves of measurement necessary to track development into young adulthood,
Silberg et al. (submitted) examined the genetic and environmental causes of continuity or change in antisocial
behavior in males from childhood to young adulthood,
which revealed important developmental differences in
the etiology of antisocial behavior. Incorporating both
mother and child reported assessments of CD symptoms
on twins between ages 8 and 17 and antisocial behavior
of the twins as young adults revealed the following four
features: (1) a single genetic factor influencing antisocial
behavior beginning at age 10 through young adulthood
(‘life-course persistent’); (2) a shared environmental
effect beginning in adolescence (‘adolescent-onset’), (3) a
transient genetic effect at puberty, and (4) a specific
genetic influence on adult antisocial behavior. Overall,
these etiological findings coincide with Moffitt’s developmental theory of antisocial behavior (Moffit, 1993). The
genetic effect at puberty at age 12 to 15 is also consistent
with a genetically mediated influence on the timing of
puberty effecting the expression of genetic differences in
antisocial outcomes. Tracy et al. (submitted) focused on
sex differences in the genetic architecture of antisocial
behavior in young adults and its genetic and environmental relationship to retrospective assessment of
conduct disorder prior to age 18. They showed that both
genetic and shared environmental factors contribute to
variability of antisocial behavior in adolescence.
However, shared environmental factors were no longer
significant in young adulthood, supporting the hypothesis that in addition to unique environmental factors,
‘adolescent-limited’ antisocial behaviors are at least
partly mediated by common environmental factors while
‘life-course persistent’ antisocial behaviors are influenced
primarily by genetic effects. While genetic effects were
stronger in males than in females, the shared environment had a greater impact on juvenile antisocial
behavior in females than in males.
The results from analyses of the VTSABD and
YAFU are remarkably consistent with considerable evidence, primarily from twin and adoption studies which
shows that both genetic, shared and specific environmental factors play a substantial role in antisocial
behavior. A meta-analysis of studies focused on antisocial behavior (Rhee & Waldman, 2002) concluded that
genetic factors accounted for about 32% of the variance, with smaller but significant proportions of
variance explained by dominance (9%) and shared
environmental factors (16%). The unique environmental variance, including measurement error, was
estimated at 43%. Their analysis included adolescent
and adult samples, males and females, prospective and
retrospective assessments of behavior, and both selfreport and reports by other raters. A few publications
Twin Research and Human Genetics February 2007
Genetic and Cultural Transmission of Conduct Disorder
Table 1
VTSABD/YAFU Sample Sizes by Age
Age (year)
8
9
10
11
12
13
14
15
119
203
158
167
165
79
126
154
105
42
87
16
17
18+
Total
165
128
137
139
31
1412
124
156
93
107
107
1051
133
102
106
158
628
26
50
69
48
193
443
382
421
344
3284
VTSABD
Wave 1
Wave 2
Wave 3
Wave 4
Total
119
282
284
321
312
376
YAFU
Age at Wave 1
% of Wave 1
85
165
134
133
147
144
112
125
129
28
0.71
0.81
0.85
0.80
0.89
0.87
0.88
0.91
0.93
0.90
have addressed the question of continuity in genetic and
environmental effects between childhood and adulthood. Lyons et al. (1995) identified genetic, common
and unique environmental factors that influenced both
juvenile (before age 15) and adult (before age 15)
behaviors, with additional unique environmental effects
specific to adults in 3326 twin pairs from the Vietnam
Era Twin Registry. Genetic factors had a larger impact
in adulthood, while shared environmental factors were
more important in childhood. Jacobson et al. (2002)
identified quantitative differences in genetic effects
between two retrospective measures of antisocial
behavior in childhood (before age 15) and adolescence
(age 15–18) with genetic factors having a larger impact
in adolescence and shared environmental factors being
more significant in childhood in 6806 male and female
twins from the Virginia Twin Registry. However, no
such difference was observed between adolescence and
adulthood. Focusing specifically on young adulthood in
males, Malone et al. (2004) found additive genetic and
unique environmental effects to be the only factors contributing to antisocial behavior at three time points
(ages 17, 20 and 24) in 289 twin pairs from the
Minnesota Twin Family Study.
Although these prior studies have answered several
important questions regarding the nature and nurture
of antisocial behavior and its developmental trajectory,
they did not address issues of transmission of behavior
from parents to children, or the role of genes and environment on the influences across generations. Such
questions require a longitudinal design that spans at
least two generations or a twin design augmented with
data from parents or other extended twin designs.
Materials and Methods
Virginia Twin Study of Adolescent Behavioral
Development (VTSABD) and Young Adult
Follow-Up (YAFU)
The VTSABD is the first population-based, multiwave, cohort–sequential twin study of adolescent
psychopathology and its risk factors. The design
includes Caucasian families of male and female
1202
1202
0.85
monozygotic (MZ) and dizygotic (DZ) twins and their
parents. Its goals are to assess genetic and environmental factors in developmental change, to identify
many of the major familial psychosocial risk factors,
and to characterize their correlation and interaction
with genetic risk. Twins were initially assessed at ages
8 to 16. The YAFU has assessed the sample at a
median follow-up age of 21 years. Details of sample
ascertainment and assessment of socioeconomic bias
were reported by Meyer et al. (1996).
In the VTSABD, adolescent male and female twins
aged 8 through 16 were ascertained by statewide
recruitment through Virginia schools. Of the 1894
Virginia families that were initially eligible for study,
1412 Caucasian families (74.5%) participated in the
first wave of data collection (2775 individuals twins
comprising 1384 complete pairs). Twins under age 18
and currently enrolled in high school were followed
every 18 months up to 3 times. One thousand and
fifty-one out of the 1302 families that continued to
meet the age and Virginia residence requirements of
the study completed a second home interview (80%
participation rate), and 628 of the 777 eligible twins
families (81%) participated in a third wave of assessment; 193 families also completed Wave 4. Table 1
summarizes the numbers of pairs assessed in each year
of life and each assessment wave. The aggregate data
provide the equivalent of 4486 pairs of twins assessed
at ages distributed between 8 and young adulthood.
All assessments were done by face-to-face interview in
the twins’ home. A pair of interviewers each assessed
one twin and one parent in each family.
At age 18 or older, all twins who participated in the
first wave of the VTSABD were targeted for a young
adult assessment. To date (8/06), 1185/1412 (84%) of
pairs have been followed up in the YAFU. Of the 2,692
individuals eligible for participation, 2291 have completed the telephone interview and 186 remain to be
interviewed. We have been unable to contact 97, and
118 have refused to participate. Twenty-four per cent
of the YAFU subjects participated in only the first wave
of the VTSABD, 32% participated in two waves, 31%
in three waves, and 13% in all four waves. Ages of the
Twin Research and Human Genetics February 2007
139
Hermine H. Maes, Judy L. Silberg, Michael C. Neale, and Lindon J. Eaves
Table 2
VTSABD/YAFU Sample Size by Zygosity and Sex of Twins and by Type of Relative/Informant
Zygosity and sex of twins
Sample
MZ male MZ female DZ male DZ female
Type of relative/informant
DZ MF
Unknown Total pairs
Indiv twins Mother
Father
Teacher
VTSABD adolescents
313
408
183
189
295
24
1412
2824
1365
1105
1369
YAFU young adults
272
349
155
169
240
17
1185
2289
—
—
—
Note: Indiv = Individual
twins during participation in the YAFU ranged from 18
to 30 with a mean age of 21.4 years. Greater than 75%
have had some level of college education and about
50% of the subjects were currently in school at the time
of interview.
Table 2 gives the structure of the samples broken
down by zygosity and sex of twin pairs, and the
number of individuals twins, relatives and/or informants. VTSABD data comprise 6282 face-to-face
assessments of juveniles across four waves and 2289
assessments of young adult twins. The twins’ parents
(N = 2470) also completed psychiatric assessments
during the home interview, comprising 97% of
mothers (N = 1365) and 78% of fathers (N = 1105).
Of the mothers, 1119 were the biological mothers of
the twins, and 59 were adoptive mothers, stepmothers
or female guardians. 946 biological fathers and 224
nonbiological fathers participated. Parents as well as
teachers also acted as informants about the adolescent
twins’ behavior. Teacher assessments of both twins on
at least one occasion are available for 97% of the families (N = 1369). Zygosity assignment of twins was
based on DNA when available, and otherwise on an
algorithm combining standard questions gathered
from the parents about the twins’ similarity, and photographs of the twins rated independently by multiple
raters. A detailed description can be found elsewhere
(Eaves et al., 1997).
Measures
The VTSABD data comprise psychiatric assessment of
parents and offspring as children and young adults,
together with the principal psychosocial and environmental risk factors. Table 3 details the instruments
used, their mode and wave of assessment, and rater(s).
Prospective ratings were secured about each child and
home by direct assessment of the child and by reports
of parents and teachers. The core juvenile assessment
comprised face-to-face interviews with each twin and
both parents using the Child and Adolescent
Psychiatric Assessment (CAPA, Angold et al., 1995)
adapted for use with twins and their parents. The
CAPA has forms for interviewing parents about their
children (CAPA-P) and for the children themselves
(CAPA-C); it yields over 300 coded pages of symptom
data relevant to the common areas of childhood and
adolescent disorder nuanced with onset, frequency,
duration, incapacity, treatment, and context for each
clinical domain. The core assessment of psychopathol-
140
ogy in the twin pairs as young adults and their parents
is in the M and S-sections that are based on the
Structured Clinical Interview of DSM (SCID, Spitzer
et al., 1990). Twins and parents also filled out selfreport questionnaires (SRQ) which contained the
following scales: What I Think and Feel (Manifest
Anxiety, MA; Reynolds & Richmond, 1978), Mood
and Feelings Questionnaire (MFQ, Costello &
Angold, 1988), Behavior and Activities Checklist
(BAQ, adapted from Olweus, 1989), Fears and
Phobias (FSSC, adapted from Ollandick et al. 1989),
Life Events Checklist (LEQ, adapted from Johnson and
McCutcheon 1980), Twin Index of Rearing
Environment (TIRE, adapted from Sibling Inventory
for Differential Experience, SIDE, Daniels & Plomin,
1985), Family Adaptability and Togetherness Scale
(FAT, Olsson et al., 1979), EASI Temperament Scale
(EASI, Buss & Plomin, 1975). The Junior Eysenck
Personality Questionnaire (Juvenile, JEPQ, Eysenck &
Eysenck, 1975) was added in Wave 3; young adult
twins filled out the Eysenck Personality Questionnaire
(Adult Short form, Eysenck & Eysenck, 1975). Parents
filled out questions about Health, Habits and Behavior
(HHB), the Rutter ‘A’ Scale (RA, Rutter et al., 1970),
and the Dyadic Adjustment Scale (DAS, Spanier, 1976).
Teachers filled out the Rutter ‘B’ Scale (Teacher, (RB,
Rutter et al., 1970), the Conners Teacher Rating Scale
(CO, Conners) and the (Child Behavior Checklist
(CBC, Achenbach). Prior to Wave 1 of the VTSABD,
parents filled out the CBC, which was repeated in Wave
4 by parents and twins. These questionnaires cover personality measures, internalizing and externalizing
constructs. The behavioral assessment was augmented
with cognitive measures (Slosson Oral Reading Test
(Slosson, 1990) and Raven’s Standard Progressive
Matrices (Raven, 1956) in Wave 3) and physical measures (height, weight) and questions about the twins’
pregnancy and health problems. Environmental assessment includes the Home Environment Interview (based
on Robins et al., 1985). Census block/tract data from
1990 US Census are also available.
In this article we use data on antisocial behavior
(AB) to illustrate features of the data set and the types
of analyses that may be conducted. AB was measured
both prospectively in the twins between the ages of 8
and 16, and retrospectively in the young adult twins
after they reached the age of 18 and in the twins’
parents. Here we use the Conduct Disorder (CD) data
obtained as part of the semi-structured interview for the
Twin Research and Human Genetics February 2007
Genetic and Cultural Transmission of Conduct Disorder
Table 3
Measures Used in the VTSABD/YAFU with their Mode and Wave of Assessment and Rater(s)
VTSABD
Abbreviation
Content
Mode
Wave 1
Wave 2
Wave 3
Wave 4
YAFU
Behavior and psychopathology (family, peers, school,
anxiety, depression, suicide, eating, puberty,
hyperactivity, oppositional and conduct, tobacco,
alcohol, drugs, incapacity)
I
TMoFa
TMoFa
T(MoFa)
T(MoFa)
Adult psychopathology (depression, anxiety, panic,
phobia, alcohol, drugs)
I
MoFa$ [M]
MoFa$ [M]
MoFa$ [L] (M2)*
MoFa$ [L2]
T^
I
MoFa [S]
MoFa [S]
MoFa [S2]
(ASP.A)*
MoFa$ [S3]
T
Behavioral
CAPA
M, L, L2, M2
S, S2, S3, ASP Adult and childhood antisocial behavior
$
ADDR
Retrospective attention-deficit
SRQT
MA, MFQ, BAQ, FSSC, LEQ, SIDE, FAT , EASI, JEPQ
SRQP
HHB, RA, MA, MFQ, BAQ, FSSC, LEQ, SIDE, FAT$, EASI,
DAS$
Q
SRQS
RB, CO, CBC
Q
CBC
Internalizing and externalizing behavior
Q
P#
EPQ
Personality
$
I
$
&
Q
$
MoFa$
T
T
T
T
MoFa
MoFa
MoFa
MoFa
Te
Te
Te
Te
TP
T
Cognitive
SORT
Reading
I
SPM
Nonverbal reasoning
I
A
Twins’ pregnancy, birth, similarity
I
AA, AA.1
Medical/health problems
I
Measures
Height, weight
T
T
T
T
T
Physical
MoFa
MoFa*
MoFa*
MoFa*
MoFa [AA]
MoFa [AA.1]
T
T
T
T
MoFa
MoFa
MoFa*
MoFa*
P
P*
P*
MoFa [S]
MoFa [S2]
MoFa [S3]
DNA
Environmental
CL/CL3
Religion, attendance, RELY
I
P/Q
Household socioeconomic status
I
S, S2, S3
Employment, relationships
I
E
Home environment
I
TP
R
Peer relations
I
T
Census
Demographic indices
MoFa [S]
T
T
Note: I: interview (face to face); Q: questionnaire; T: twins self-report, Mo: mother, Fa: father, Te: Teacher, P: Parent, (MoFa) mother or father rating, $ assessment of parents of twins,
[] section used, * if not done in Wave 1 or 2, ^telephone interview, &Wave 3 and 4 only, #mailed prior to Wave 1, YYAFU only
CAPA: Child and Adolescent Psychiatric Assessment (Angold et al., 1995); M, L, M2, L2, S, S2, S3, ASP, ADDR, A, AA, AA.1, CL, CL3, P, Q, E, R: sections of interview with
twins/parents; SRQ: Self-Report Questionnaire of T: twin, P: parent, S: school teacher; CBC: Child Behavior Checklist (Achenbach, 1988); JEPQ: Junior Eysenck Personality
Questionnaire (Juvenile), EPQ: Eysenck Personality Questionnaire (Adult Short form; Eysenck & Eysenck, 1975); MA: Manifest Anxiety/What I Think and Feel (Reynolds &
Richmond, 1978), MFQ: Mood and Feelings Questionnaire (Costello & Angold, 1988), BAQ: Behavior and Activities Questionnaire (adapted from Olweus, 1989), FSSC: Fears and
Phobias (adapted from Ollandick et al., 1989), LEQ: Life Events Checklist (adapted from Johnson & McCutcheon, 1980), TIRE: Twin Index of Rearing Environment (adapted from
SIDE; Daniels & Plomin, 1985), FAT: Family Adaptability and Togetherness Scale (Olsson et al., 1979), EASI: EASI Temperament Scale (Buss & Plomin, 1975), HHB: Health,
Habits & Opinions, RA: Rutter ‘A’ Scale (Rutter et al., 1970), DAS: Dyadic Adjustment Scale (Spanier, 1976), RB: Rutter B Scale (Teacher, see Rutter et al., 1970), CO: Conners
Teacher Ratings Scale (Conners CK); SORT: Slosson Oral Reading Test (Slosson, 1990), SPM: Raven’s Standard Progressive Matrices (Raven, 1956); REL: Young Adult Religious
Practices Interview, Home Environment Interview (based on Robins et al., 1985).
evaluation of adult Antisocial Personality Disorder
(ASPD). The answers were incorporated into a scoring
algorithm designed to correspond to DSM-III-R criteria
for CD, and were used to generate a sum score of the
number of symptoms present. DSM-III-R criteria were
chosen for the basis of these measurement algorithms
so that informative comparisons could be made with
results from previously reported analyses using the
VTSABD data. Parental data were obtained during any
of the waves of the home interview, as part of the
section about employment, relationships, adult and
childhood behavior. Section S (including 11 out of 12
items, and excluding ‘stolen with confrontation of
victim’) was used in Wave 1 or Wave 2 if not done in
Wave 1. In Waves 3 and 4, parents were given a short
version of the section (S2, 10 out of 12 items, excluding
‘stolen with confrontation of victim’ and ‘forced sexual
activity’) if they had completed section S in Wave 1 or
2. Otherwise, they were given a full antisocial personality (ASP) section including all 12 CD items. This latter
version was also administered to the young adult
assessment of the twins. In all cases, the presence of the
Twin Research and Human Genetics February 2007
141
Hermine H. Maes, Judy L. Silberg, Michael C. Neale, and Lindon J. Eaves
VTSABD
Wave 1
N = 1412 pairs
F = 1104 M = 1394
Section S
VTSABD
Wave2
N = 1051 pairs
F = 798 M = 1012
YAFU
VTSABD
Wave 4
N = 193 pairs
F = 136 M = 174
Section S2
YES
NO
VTSABD
Wave 3
N = 628 pairs
F = 474 M = 598
Section S
Section ASP
NO
NO
N = 1185 pairs
Section S2
Section ASP
NO
Section ASP
Figure 1
Design of the Virginia Twin Study of Adolescent Behavioral Development (VTSABD) and the Young Adult Follow Up (YAFU), with specific reference
to the assessment of antisocial behavior.
Note: F: number of fathers; M: number of mothers; S: S-section in Waves 1 or 2 for parents or guardians; S2: short S-section in Waves 3 or 4 for parents or guardians;
ASP: full S-section in Waves 3 or 4 for parents or guardians if not done in Waves 1 or 2, and for YAFU young adults.
items prior to age 18 was assessed and a diagnosis of
CD was assigned if three or more symptoms were
present. Missing values were imputed if less than 25%
of the items were not missing. Figure 1 provides a
graphical representation of the design of the study and
the measures of antisocial behavior in parents and
young adult twins used at each of the assessments.
Statistical Methods
Structural equation modeling of the data was undertaken using methods described in Eaves et al. (1999),
which assesses the contributions of genetic effects in
the presence of effects including vertical cultural inheritance, phenotypic assortative mating, shared and
within-family environment. While this model was
described for the extended twin kinship design (ET
model), the core parameters can be estimated with the
twin–parent design. Data from twins and their parents
permit quantification of the contributions of additive
genetic, dominance or shared environmental, and specific environmental factors as is the case for the
classical twin study. In addition, it is possible (i) to
estimate the degree of assortment (or to what extent
the data deviate from the assumption of random
mating), and (ii) the degree to which parents phenotypes influence their offspring (termed vertical cultural
transmission). This second component effectively partitions the shared environment into parental versus
nonparental sources. Although alternative models
exist, the most often used model includes phenotypic
assortment and phenotypic cultural transmission
(Fulker, 1988; Heath & Eaves, 1985; Neale & Fulker,
1984). Phenotypic assortment occurs when mate selection is based at least partly on the trait being studied,
and generates a correlation between the observed phenotypes of spouses. Also, the impact of assortment on
the contribution of genetic and shared environmental
factors can be estimated (Neale & Cardon, 1992).
Furthermore, the contribution of the genetic and envi-
142
ronmental factors may depend on sex, both in their
magnitude and nature.
To date, this model has been applied to data from
nuclear twin families, and also from adoption designs
(Baker et al., 1983). In the VTSABD, data were
obtained from the parents/guardians who were most
familiar with the adolescent twins. Therefore, a relatively large number of nonbiological fathers, and a
smaller number of nonbiological mothers participated
in the study. We therefore extended the traditional
twin–parent model to allow for both biological and
nonbiological parents, resulting in four different types
of families. Given the relatively small percentage of
families with a biological father and nonbiological
mother (1.5%) or with two nonbiological parents
(3.5%), we limited the current analyses to families
with either (i) both biological parents; or (ii) a biological mother and nonbiological father. In addition to the
assumption of equilibrium of variances across generations, three additional assumptions are made: (1)
equal variance for biological and nonbiological
parents (while allowing for different means); (2) equal
assortment between two biological parents and
between a biological and nonbiological parent; and (3)
equal cultural transmission from a biological or nonbiological parent. When genetic factors are operating,
the expected parent–offspring covariance will be
smaller between children with their nonbiological
parent than with their biological parents. Although
the genetic and shared environmental covariance will
be the same, genotype–environment covariance
(resulting from the combined presence of genetic and
cultural transmission) will be greater in twins with
both biological parents versus those with one biological and one nonbiological parent. Figure 2a presents a
path diagram of the extended twin parent model (ETP
model), which was implemented in the statistical modeling package Mx (Neale et al., 2006, see Maes et al.,
Twin Research and Human Genetics February 2007
Genetic and Cultural Transmission of Conduct Disorder
1
q
E
sm
r
u
A
vm
xm
B
1
C
q
E
em am bm cm
sm
r
u
A
vm
xm
B
xf
C
C
em am bm cm
nPF
.5
bPF
sf
vf
u
r
q
B
A
cf
E
af ef
i
.5
.5
1
.5
bPM
.5
i'
p'
1
o'
.5
p
.5
E
A
o
m
rc
1
B
1
C
n
.5
.5
C
em am bm cm
B
-.06
1
1.06
E
A
B
1
C
0
1
1.07
E
.48 .50 .00 .34
A
af ef
-.15
0
1.11
B
0
1
1.10
C
C
.48 .50 .00 .34
nPF
.5
0
1.07
B
.28
bPF
.36
.5
.5
.26
.01 .01
E
Pf
-.15
0
1.02
A
cf
Pm
0
1
1
-.34 -.25
1
A
E
.41 .44
bPM
.5
.5
-.24 -.32
1
1
.5
E
.5
A
1
B
.48 .50 .00 .34
1
C
.5
C
.5
B
.28
Pm
1
A
E
.41 .44
Pf
Figure 2a and 2b
Full twin–parent resemblance model for opposite-sex DZ twins (Pm, male twin; Pf, female twin) and their parents (either both biological parents,
bPF, biological father and bPM, biological mother; or biological mother and nPF nonbiological father).
Note: Path coefficients are the same in both generations, and gene–gene and gene–environment correlations occur in both generations. Coefficients include additive genetic [a, b],
shared environmental [c] and unique environmental [e] sources of variance; subscripts m and f for males and females [e.g., am and af]; phenotypic cultural transmission from
the parent’s phenotype to the offspring’s shared environment, which may differ by sex of parent and child [m, n, o ,p]; genotype–environment covariance between gendercommon and male-specific genetic factors and shared environmental factors [s, v]; phenotypic assortment [i]. Correlation between gender-common and male-specific
genetic factors [r]. Variances of latent factors [q, u, x] are constrained to be equal across generations. Estimates of path coefficients are shown in Figure 2b.
Twin Research and Human Genetics February 2007
143
Hermine H. Maes, Judy L. Silberg, Michael C. Neale, and Lindon J. Eaves
Table 4
Prevalence Rates — N cases (per cent) — for Conduct Disorder and Antisocial Personality Disorder in Young Adult Twins and Parents of Twins,
and Their Tetrachoric Correlation (r)
N
Bio male
Bio female
All male
All female
Total
ASP
r CD-ASP
301 (30)
153 (13)
454 (20)
139 (6)
.44
6 (13)
266 (23)
113 (8)
379 (15)
89 (4)
.45
41 (30)
4 (10)
254 (23)
111 (8)
365 (15)
84 (3)
.46
4 (17)
0 (0)
81 (20)
38 (7)
119 (13)
20 (2)
.47
7 (28)
2 (33)
14 (31)
2 (15)
16 (28)
4 (7)
.31
BM male BM female
Young adults
2244
All parents
2533
220 (22)
107 (8)
46 (28)
Parents S (w1/w2)
2475
213 (22)
107 (8)
Parents S2 (w3/w4)
949
77 (21)
38 (7)
58
7 (35)
0 (0)
Parents ASP
Note: Bio: twin families with both biological parents; BM: twin families with biological mother and nonbiological father;
S: S-section of interview on adult and childhood antisocial behavior used in Wave 1 and Wave 2;
S2: S-section of interview on adult and childhood antisocial behavior used in Wave 3 and Wave 4;
ASP: ASP-section of interview on adult and childhood antisocial behavior used in Wave 3 if no S-section available from Wave 1 or 2
CD: Conduct disorder
ASP: Antisocial personality disorder
1999 for description of the Mx script) and fitted to
the raw ordinal data to obtain maximum likelihood
estimates of the model parameters. We now describe
these parameters in detail with reference to the path
diagram. The variance of the phenotype in twins and
parents (Pm, male twin; Pf, female twin; bPF, biological
father; bP M , biological mother; nP F , nonbiological
father) is partitioned in additive genetic [a,b], shared
environmental [c] and unique environmental [e]
sources of variance. These can vary in magnitude
between males and females [e.g., a m and a f ].
Furthermore, to allow a different set of genes in males
and females, we modeled them as gender-common [a]
and male-specific [b] additive genetic factors, where
the latter would contribute to the phenotypic variances of males only. Note that the paths from the
genetic factors in the parental generation to those in
the offspring generation are fixed to .5, including the
paths from the mother’s male-specific genes to her
male and female offspring. In addition to the genetic
transmission from parents to offspring, parents may
also influence their children through nongenetic pathways. In the current specification, these influences are
modeled as phenotypic cultural transmission from the
parent’s phenotype to the offspring’s shared environment, and they are allowed to differ by sex of parent
and child [m, n, o, p]. Note that we can test whether
these environmental transmission paths are equal for
biological and nonbiological fathers. Besides parental
environmental influences, other aspects of the environment may generate similarity between twins or
siblings, which are modeled as residual covariances,
and referred to as nonparental shared environment
[rc]. The combined presence of genetic and cultural
transmission generates genotype–environment covariance, which may exist between shared environmental
factors and both gender-common and male-specific
genetic factors [s, v]. Finally, a design including
spousal information can also assess the degree of
assortment, here modeled as phenotypic assortment
[i]. We can also test the equality of assortment
144
between the biological parents and between a biological and a nonbiological parent. As a consequence of
assortment, sources of variance may become correlated, for example, gender-common and male-specific
genetic factors [r].
Results
Response Frequencies
Twenty per cent of young adult twins endorsed more
than three conduct items retrospectively, resulting in a
diagnosis of conduct disorder (Table 4). Rates are significantly higher in males (30%) than females (13%). The
prevalence of conduct disorder — when asked retrospectively — is somewhat lower in parents (15%), with
fathers about three times more likely to have a diagnosis
of conduct disorder than mothers. As the parental data
resulted from the combination of slightly different
instruments administered at different times, we also
report the prevalence rates separately by instrument.
Rates were somewhat lower at the second assessment of
CD (S2) than at the first (S) but showed a similar pattern
of sex differences. Parents who had not participated in
the first two waves of the study showed markedly higher
rates of CD in both males and females. This is partly due
to the greater proportion of nonbiological parents in this
group, whose rates for CD are consistently higher than
those of biological parents. We assessed the reliability of
CD using polychoric correlations, both at the level of
symptom counts (.73) and diagnosis (.72). The reliability
was slightly higher in males (.71) than in females (.62).
Rates for adult antisocial behavior (ASP) are about a
third of CD rates, with tetrachoric correlations over time
ranging from .31 to .47, indicating that only a subset of
individuals with CD continue to show antisocial behavior as adults, and that not all individuals with ASP
presented with CD in adolescence. Note that for these
analyses we did not include the presence of CD before
age 15 as a prerequisite for a diagnosis of ASP according
to DSM-III-R. Raw CD symptom count shows marked
skewness. To improve the metric properties of the scale,
Twin Research and Human Genetics February 2007
Genetic and Cultural Transmission of Conduct Disorder
Table 5
Sample Sizes, Mean and Variance for Conduct Disorder (After Square Root Transformation and Regression of Sex, Age and Their Interaction)
Both biological parents
Biological mother + nonbiological father
Twin 1
Twin 2
Father
Mother
Twin 1
Twin 2
Father
N MZM
213
210
194
222
31
32
22
Mother
35
N DZM
118
116
110
122
23
24
12
26
N MZF
284
260
236
270
62
63
37
63
23
N DZF
124
129
115
130
20
21
7
N DZO
177
182
171
185
34
33
16
36
Male twin
Female twin
Dad
Mom
Male twin
Female twin
Dad
Mom
–0.011
–0.044
–0.010
–0.017
0.158
0.185
0.168
0.135
0.581
0.456
0.593
0.401
0.568
0.477
0.522
0.423
Mean
Variance
Note: MZM: monozygotic male twins; DZM: dizygotic male twins; MZF: monozygotic female twins; DZF: dizygotic female twins; DZO: dizygotic opposite-sex twins
the count data were square root transformed and the
effects of sex, age and their interaction regressed out
within each generation.
Maximum Likelihood Estimation of Correlations
Table 5 presents sample sizes by zygosity and family
type, as well as means and variances for twins and
parents by sex and family type. We fitted a baseline
model with free parameters for all means, variances and
covariances. Means could be equated for male twins
across both members of a twin pair and across zygosity
within nuclear families and within step families without
significant loss of fit. This was also true for all female
twins, all fathers and all mothers. However, the CD
means for twins and parents in nuclear families were
significantly lower than those in twin families with a
biological mother and nonbiological father. Similar tests
for the equality of variances across order of the twins
and zygosity were nonsignificant both within and
across family type. Tests for equality of parent–offspring and spousal correlations across twin order and
zygosity within family type, that is, for example, equality of all father–son pairs, including father–male Twin
1, father–male Twin 2 in both MZ and DZ pairs, were
also nonsignificant. Maximum likelihood estimates of
the correlations by type of relative (twin, parent–offspring, spousal) and family are presented in Table 6.
The twin correlations were consistent with the contribution of both additive genetic and shared
environmental influences to antisocial behavior in addition to specific environmental influences. They were
slightly but not significantly higher in step families than
in nuclear families. Parent–offspring correlations were
considerably lower than DZ twin correlations.
Nonbiological father–offspring correlations were negative but not significant. Mother–offspring correlations
did not differ significantly by family type . The spousal
correlation was significant in nuclear families but not in
step families.
Maximum Likelihood Estimation of Genetic and Environmental
Contributions
A twin–parent model estimating the contributions of
additive genetic, shared and specific environmental
factors was fitted to conduct disorder measured retrospectively in twins and either their biological parents
or their biological mother and nonbiological father.
This model includes assortative mating and cultural
transmission and allows tests for sex differences in all
the effects. Given the inclusion of nonbiological
fathers, we allowed the assortment and paternal cultural transmission paths to be different in the two
types of families. Path coefficients are presented in
Figure 2b; variance components, both unstandardized
and standardized, and their confidence intervals (CIs),
are presented in Table 7. Genetic factors and unshared
environmental factors were the major sources of variance in antisocial behavior. Although shared
Table 6
Maximum Likelihood Estimates of Twin, Parent–Offspring and Spouse Correlations for Retrospective Reports of Conduct Disorder
Twin correlations
Bio
Parent–offspring correlations
Spouse
MZM
DZM
MZF
DZF
DZO
FS
FD
MS
MD
MF
.590
.351
.557
.361
.399
.123
.151
.169
.108
.175
BM
.681
.345
.623
.416
.438
-.035
-.015
.051
.211
.123
All
.598
.356
.570
.366
.409
.107
.139
.149
.124
.170
Note: MZM: monozygotic male twins; DZM: dizygotic male twins; MZF: monozygotic female twins; DZF: dizygotic female twins; DZO: dizygotic opposite-sex twins; FS: father–son;
FD: father–daughter; MS: mother–son; MD: mother–daughter; MF: mother–father; Bio: twin families with both biological parents; BM: twin families with biological mother and
nonbiological father.
Twin Research and Human Genetics February 2007
145
Hermine H. Maes, Judy L. Silberg, Michael C. Neale, and Lindon J. Eaves
environmental factors significantly contribute to individual differences in antisocial behavior, when split into
parental and nonparental sources, only the nonparental
factors remain significant. Cultural transmission
accounted for only 2% of the variance in both male
and female offspring. Genotype–environment covariance resulting from the combined presence of genetic
and cultural transmission was negative, given the negative causal paths from parents to offspring, but also
nonsignificant. Although the genetic consequences of
assortment only explained 3% of the variance, their
effects were significant. Estimates of assortment and
paternal cultural transmission were lower in step families compared to nuclear families, resulting in less
genotype-environment covariance and larger estimated
total variance. However, these parameters were not
significantly different from those of nuclear families.
When aggregated across sources, additive genetic
factors accounted for 38% (95% CI 16%–55%) of
the variance in antisocial behavior in males and 40%
(CI 17%–56%) in females. The contribution of shared
environmental factors was 23% (CI 11%–41%) in
males and 18% (CI 8–35%) in females, and genotypeenvironment covariance was estimated at negative
6%–7% (CI males –15%–2%, females –18%–1%).
Specific environmental factors explained the remainder of the variance, 39% (CI 30%–48%) and 42%
(CI 33%–51%) in males and females respectively.
Note that for the twin–parent analyses, the standardized estimates were obtained by dividing each of the
variance components by the total variance minus the
genotype-environment covariance.
We evaluated sex differences in (unstandardized)
variance components by fitting a series of submodels.
The contributions of cultural transmission did not
differ according to the sex of the parent or the sex of
the child. Although the total shared environmental
variance or the additive genetic variance could be
equated across sexes, both could not be equated
simultaneously without significant loss of fit. The
unique environmental variance components also differed significantly by sex.
Discussion
Results of our analyses extend previous evidence on
the contribution of genetic and environmental factors
on antisocial behavior. Both additive genetic and
shared environmental factors appear to contribute
significantly to the variance of CD in males and
females. In addition to data collected from the classical twin design, the twin–parent design allows us to
identify the effects of assortative mating and cultural
transmission. Both appear to explain a very modest
proportion of the total variance in CD, and only the
consequences of assortment were statistically significant. There is greater power to detect the effects of
assortment, which directly derive from the spousal
correlation, than there is to detect cultural trans146
Table 7
Standardized Variance Components from the Twin–Parent Model for
Antisocial Behavior
Variance components
Standardized variance
components
Male
Female
Male
Female
.018
.012
.031
.002 to .067
.028
006 to .061
.016
.010
.026
.005 to .059
.024
.005 to .056
Common additive genetic
.258
.171
.445
.061 to .681
.398
.152 to .620
Male-specific genetic
.000
—
.000
.000 to .313
—
Unique environment
.232
.192
.400
.335 to .479
.448
.387 to .519
Shared environment
.113
.081
.194
.043 to .381
.189
.045 to .367
Cultural transmission
.012
.008
.003
.004
.020
.000 to .082
.005
.000 to .046
.018
.000 to .076
.008
.000 to .050
–.052
–.035
–.019
–.017
Assortative mating
Genotype–environment
correlation
Total Variance
.581
.429
.603
.441
–.090
–.082
–.247 to .011 –.228 to .011
–.031
–.039
–.103 to .174 –.144 to .054
Note: Estimates in italics for families with nonbiological father
mission, which is inferred from the relative magnitude of parent–offspring and twin correlations.
At face value, it appears that adding the parental
data to the twin design does not provide much additional information. However, this may depend of the
phenotype of interest. If the assumptions of the twin
design are (mostly) met, that is, no assortative mating,
no genotype–environment covariance (which is theoretically possible but would imply that biological
parents who provide the genetic make-up of their children do not have any environmental influence on their
offspring for a phenotype that is at least in part
genetic), and if it is not the case that both shared environment and dominance contribute to variability of
the phenotype (which cannot be simultaneously estimated with data from the classical twin study), then
one would not expect the inclusion of phenotypic data
of the parents to alter the partitioning of the variance.
In all other instances, however, the addition of the
parents provides at least a check of some of the
assumptions or allows for the simultaneous estimation
of shared environment and genetical dominance
assuming no cultural transmission.
Table 8 compares the unstandardized and standardized variance components from the twin–parent
analysis with those obtained from fitting traditional
models using twin data only. The estimates for the
Twin Research and Human Genetics February 2007
Genetic and Cultural Transmission of Conduct Disorder
Table 8
Comparison of Variance Components Estimates for Conduct Disorder Obtained from the Twin–Parent Analyses with Those from the Twin Data Alone
Unstandardized variance components
Twins and parents
Male
Total variance
Twins
Female
Male
Standardized variance components
Twins and parents
Female
Twins
Male
Female
Male
Female
58
43
58
47
Additive genetic
23
9 to 37
19
7 to 29
20
7 to 31
14
4 to 23
38
16 to 55
40
17 to 56
34
13 to 52
29
8 to 47
Shared environment
14
7 to 24
8
4 to 15
15
6 to 27
14
6 to 23
23
11 to 41
18
8 to 35
26
10 to 44
29
12 to 47
Unique environment
24
20 to 28
19
17 to 22
23
20 to 28
20
17 to 23
39
30 to 48
42
33 to 51
40
33 to 48
42
36 to 50
Genotype–environment covariance
–3
–10 to 1
–3
–9 to 1
–6
–15 to 2
–7
–18 to 1
unique environmental component were almost identical
in the twin–parent and twin-only analyses, as would be
expected. The slightly wider confidence intervals in the
twin–parent analyses may reflect a larger number of
parameters estimated in that model. The additive
genetic variance component is greater in the twin–
parent analysis than in the twin analysis, but includes
the consequences of assortative mating. Conversely, the
shared environmental component (including the effects
of cultural transmission) is smaller in the twin–parent
analysis. This comparison supports the notion that the
estimates obtained from a classical twin study are
biased by assortment, such that the additive genetic
component is underestimated and the shared environmental component overestimated. The negative
estimate for the genotype–environment covariance,
which further biases the estimate of the shared environmental component, could reflect the presence of
dominance or age by genotype interaction.
The standardized estimates for additive genetic,
shared environmental and unique environmental contributions to variation in antisocial behavior from this
analysis are consistent with those from a meta-analysis
of 51 twin and adoption studies (Rhee & Waldman,
2002), when considering the sum of the additive and
dominance effects. However, the shared environmental
component appears to be slightly greater in our sample
than in the meta-analysis. We are aware of only one
other study that includes twins and their parents, but it
focused on general versus disorder-specific externalizing
disorders rather than CD specifically (Hicks et al.,
2004). That study did not explicitly model and estimate
genetic and environmental variance components;
instead it estimated correlations between different types
of relatives and calculated two kinds of heritability estimates based on the twin correlations and the
parent–offspring correlations.
The current analyses augment the literature on the
genetics of antisocial behavior by refining our estimates
of the contributions of additive genetic, shared environmental and unique environmental factors to the
variability in CD. However, these analyses need to be
combined with the prospective longitudinal data in
twins and the data on ASP in the young adults and
parents, to provide a comprehensive understanding
about the role of genes and environment in the developmental trajectory of antisocial behavior. Our results are
consistent with those of many twin studies of CD in
adolescence, whether assessed concurrently or retrospectively, in finding a significant effect of the shared
family environment. The parent–offspring data
augment this finding, and affect assessment of its clinical significance in several ways. First, allowing for any
genetic consequences of assortative mating accounts for
very little of the apparent shared environmental effect.
Indeed, the parent–offspring correlations, though
genetic, are relatively small and contribute little to the
prediction of juvenile behavior. No less important,
perhaps, is the implication that how parents behaved as
adolescents themselves has little long-term impact on
how they function as parents later in life. This finding is
consistent with the view that a substantial part of adolescent misconduct is short-term, adolescence-limited
and without pervasive influence on subsequent life
course. Indeed, the apparent negative passive genotypeenvironment correlation, though theoretically possible,
is psychologically unlikely and suggests that many
genetic effects on adolescent CD are not only transient
developmentally but also transient culturally. That is,
genetic effects on behavior in one generation do not
translate readily to genetic effects on the same behavior assessed in the next.
Limitations
The results of these analyses of the VTSABD data
should be considered in the light of five potential limitations. First, the results may not be generalizable to
the general population, as participation was limited to
Caucasian twin families. Second, we used retrospective assessment of CD, which for the young adult
twins corresponds to a 3- to 15-year recall and for
their parents a 15- to 45-year recall. The accuracy
Twin Research and Human Genetics February 2007
147
Hermine H. Maes, Judy L. Silberg, Michael C. Neale, and Lindon J. Eaves
with which one recalls adolescent behavior may vary
with the time interval. However, the advantage of the
current analysis is that the exact same measure was
used for the parents and the offspring that is the
optimal situation for parent–offspring designs. These
models are based on the assumption that sources of
variance as well as their magnitude are equal across
generations. This assumption seems reasonable when
the phenotype is comparable, as in the current example
where we were asking parents and young adults retrospectively about a behavior during the same period of
the life span. However, the instrument used to assess a
particular behavior may differ across children, adolescents or adults. Potential failure of the assumption of
equal phenotypic variances and variance components in
parents and offspring might be addressed by adding
genetically informative data from the two generations
involved, or by adding longitudinal data that span the
period from childhood/adolescence to adulthood. Such
data are available in the VTSABD/YAFU, and future
analyses will make use of these data. Third, while we
included families in which a biological mother and
nonbiological father was interviewed, the uneven split
of families with two versus one biological parent(s)
balances the power and precision of the estimates in
favor of the nuclear families. Fourth, because of the
high level of skewness in the symptom count for CD,
we used a square root transformation of the data
which were treated as continuous and normally distributed. Further analyses will include an ordinal
treatment of the data. Finally, we fitted only twin–
parent models which modeled phenotypic assortment
and phenotypic cultural transmission, as they seem
most plausible for CD. Other mechanisms of assortment and transmission, or a model including
nonadditive genetic effects could be specified.
Acknowledgments
This research has been supported by grants
MH45268, MH55557, MH57761 and MH068521
from NIH, and grants from the JM Templeton
Foundation. The authors would also like to thank the
twins and their families for their participation in this
project. The first author is supported by grants
MH068521, DA016977, DA018673, CA93423 and
the Virginia Tobacco Youth Project.
References
Achenbach, T. M. (1991). Manual for the Child Behavior
Checklist/4-18 and 1991 Profile. Burlington:
University of Vermont Department of Psychiatry.
American Psychiatric Association. (1987). Diagnostic and
statistical manual of mental disorders (3rd ed., rev.).
Washington, DC: Author.
Angold, A., Prendergast, M., Cox, A., Harrington, R.,
Simonoff, E., & Rutter, M. (1995). The Child and
Adolescent Psychiatric Assessment (CAPA).
Psychological Medicine, 25, 739–753.
148
Buss, A. H., & Plomin, R. (1975). A temperament theory
of personality development. New York: Wiley.
Carbonneau, R., Eaves, L. J., Silberg, J. L., Simonoff, E., &
Rutter, M. (2002). Assessment of the within-family environment in twins: Absolute versus differential ratings,
and relationship with conduct problems. Journal of
Child Psychology and Psychiatry, 43, 1064–1074.
Caspi, A., McClay, J., Moffitt, T. E., Mill, J., Martin, J.,
Craig, I. W., Taylor, A., & Poulton, R. (2002). Role of
genotype in the cycle of violence in maltreated children. Science, 297, 851–854.
Costello, E. J., & Angold, A. (1988). Scales to assess child
and adolescent depression: Checklists, screens and
nets. Journal of the American Academy of Child and
Adolescent Psychiatry, 27, 726–737.
Daniels, D., & Plomin, R. (1985). Differential experience
of siblings reared in the same family. Developmental
Psychology, 21, 747–760.
Eaves, L. J., Heath, A. C., Martin, N. G., Neale, M. C.,
Meyer, J. M., Silberg, J. L., Corey, L. A., Truett, K., &
Walters, E. (1999). Comparing the biological and cultural inheritance of stature and conservatism in the
kinships of monozygotic and dizygotic twins. In C. R.
Cloninger (Ed.), Proceedings of 1994 APPA Conference,
269–308.
Eaves, L., Rutter, M., Silberg, J., Shillady, L., Maes, H., &
Pickles, A. (2000). Genetic and environmental causes
of covariation in interview assessments of disruptive
behavior in child and adolescent twins. Behavior
Genetics, 30, 321–334.
Eaves, L. J., Silberg, J. L., Meyer, J. M., Maes, H. H.,
Simonoff, E., Pickles, A., Rutter, M., Neale, M. C.,
Reynolds, C., Erickson, M., Heath, A., Loeber, R.,
Truett, K. R., & Hewitt, J. K. (1997). Genetics and
Developmental Psychopathology: 2. The main effects of
genes and environment on behavioral problems in the
Virginia Twin Study of Adolescent Behavioral
Development. Journal of Child Psychology and
Psychiatry, 38, 965–980.
Eaves, L. J., Silberg, J. L., & Erkanli, A. (2003).
Interaction and correlation of genes and life events in
the development of depression in adolescent girls.
Journal of Child Psychology and Psychiatry, 44,
1006–1014.
Eaves, L. J., Silberg, J. L., Foley, D., Bulik, C., Maes, H.,
Erkanli, A., Angold, A., Costello, E. J., & Worthman,
C. (2004). Genetic and environmental influences on
the relative timing of pubertal change. Twin Research,
7, 471–481.
Eysenck, H. J., & Eysenck, S. B. G. (1975). Manual of the
Eysenck Personality Questionnaire. San Diego, CA:
Digits.
Foley, D. L., Eaves, L. J., Wormley, B., Silberg, J. L.,
Maes, H. H., Kuhn, J., Riley, B. (2004). Childhood
adversity, monoamine oxidase a genotype, and risk for
conduct disorder. Archives of General Psychiatry, 61,
738–744.
Twin Research and Human Genetics February 2007
Genetic and Cultural Transmission of Conduct Disorder
Foley, D. L., Pickles, A., Rutter, M., Gardner, C.O., Maes,
H. H., Silberg, J. L., & Eaves, L. J. (2004). Risks for
conduct disorder symptoms associated with parental
alcoholism in stepfather families versus intact families
from a community sample. Journal of Child
Psychology and Psychiatry, 45, 687–696.
Foley, D. L., Rutter, M., Angold, A., Pickles, A., Maes, H.
M., Silberg, J. L., & Eaves, L. J., (2005). Making
sense of informant disagreement for overanxious disorder. Journal of Anxiety Disorders, 19, 193–210.
sixteen year old twins: The Virginia Twin Study of
Adolescent Behavioral Development. Journal of
Studies of Alcohol, 60, 293–305.
Malone, S. M., Taylor, J., Marmomrstein, N. R., McGue,
M., & Iacono, W. G. (2004). Genetic and environmental influences on antisocial behavior and alcohol
dependence from adolescence to early adulthood.
Development and Psychopathology, 16, 943–966.
Gau, J. S., Silberg, J. L., Erickson, M. T., & Hewitt, J. K.
(1992). Childhood behavior problems: A comparison
of twin and non-twin samples. Acta Geneticae
Medicae et Gemellologiae, 41, 53–63.
Meyer, J. M., Rutter, M. L., Silberg, J. L., Maes, H.,
Simonoff, E., Shillady, L. L., Pickles, A., Hewitt, J. K.,
& Eaves, L. J. (2000). Familial aggregation for
conduct disorder symptomatology: The role of genes,
marital discord, and family adaptability. Psychological
Medicine, 30, 759–774.
Fulker, D. W. (1988). Genetic and cultural transmission in
human behavior. In S. B. Weir, E. J. Eisen, M. M.
Goodman, & G. Namkoong (Eds.), Proceedings of the
Second International Conference on Quantitative
Genetics, 318–340.
Meyer, J. M., Silberg, J. L., Simonoff, E., Kendler, K. S.,
& Hewitt, J. K. (1996). The Virginia Twin Study of
Adolescent Behavioral Development: Assessing sample
biases in demographic correlates of psychopathology.
Psychological Medicine, 26, 1119–1133.
Heath, A. C., & Eaves, L. J. (1985). Resolving the effects
of phenotype and social background on mate selection. Behavior Genetics, 15, 15–30.
Meyer, J. M., Silberg, J. L., Eaves, L. J., Maes, H. H.,
Simonoff, E., Pickles, A., Rutter, M. L., & Hewitt, J.
K. (1997). Variable age of gene expression: implications for developmental genetic models. In M. C.
LaBuda & E. L. Grigorenko (Eds.), On the way to
individuality: Current methodological issues in behavioral genetics. Moscow: Progress Publishers.
Hewitt, J. K., Silberg, J. L., Rutter, M. L., Simonoff E,
Meyer, J. M., Maes, H. H., Pickles, A. R., Neale, M.
C., Loeber, R., Erickson, M. T., Kendler, K. S., Heath,
A. C., Truett, K. R., Reynolds, C. A., & Eaves, L. J.
(1997). Genetics and developmental psychopathology:
1. Phenotypic assessment in The Virginia Twin Study
of Adolescent Behavioral Development. Journal of
Child Psychology and Psychiatry, 38, 943–963.
Hicks, B. M., Krueger, R. F., Iacono, W. G., McGue, M.,
& Patrick, C. J. (2004). Family transmission and heritability of externalizing disorders: A twin-family
study. Archives of General Psychiatry, 61, 922–928.
Jacobson, K. C., Prescott, C. A., & Kendler, K. S. (2002).
Sex differences in the genetic and environmental influences on the development of antisocial behavior.
Development and Psychopathology, 14, 395–416.
Johnson, J. H., & McCutcheon, S. (1980). Assessing life
stress in older children and adolescents: Development
of the Life Events checklist. In I. G. Garason & C. D.
Spielberger (Eds.), Stress and anxiety (Vol. 7).
Washington DC: Hemisphere.
Lyons, M. J., True, W. R., Eisen, S. A., Goldberg, J., Meyer,
J. M., Faraone, S. V., Eaves, L. J., & Tsuang, M. T.
(1995). Differential heritability of adult and juvenile
antisocial traits. Archives of General Psychiatry, 52,
906–915.
Moffitt, T. E. (1993). Adolescence-limited and life-coursepersistent antisocial behavior: A developmental
taxonomy. Psychological Review, 100, 674–701.
Nadder, T., Silberg, J., Rutter, M., Maes, H., & Eaves, L.
(2001). Comparison of multiple measures of ADHD
symptomatology: A multivariate genetic analysis.
Journal of Child Psychology and Psychiatry, 42,
475–486.
Nadder, T., Silberg, J., Rutter, M., Maes, H. H., & Eaves,
L. (2002). Genetic effects on the variation and covariation of ADHD and ODD/CD symptomatologies
across informant and occasion of measurement.
Psychological Medicine. 32, 39–53.
Neale, M. C., Boker, S. M., Xie, G., & Maes, H. H.
(2006). Mx: Statistical modeling (7th ed,). Richmond
Virginia: Department of Psychiatry, Virginia
Commonwealth University.
Neale, M. C., & Cardon, L. R. (1992). Methodology for
genetic studies of twins and families. Dortrecht:
Kluwer Academic Publishers.
Neale, M. C., & Fulker, D. W. (1984) A bivariate path
analysis of fear data on twins and their parents. Acta
Genetica Medica et Gemellologia, 33, 273–286.
Maes, H. H., Neale, M. C., Martin, N. G., Heath, A. C.,
& Eaves, L. J. (1999). Religious attendance and frequency of alcohol use: Same genes or same
environments: a bivariate extended twin kinship
model. Twin Research, 2, 169–179.
Ollandick, T., King, N., & Frary, R. B. (1989). Fears in
children and adolescents: Reliability and general liability across gender, age and nationality. Behavior
Research and Therapy, 27, 19–26.
Maes, H. H., Woodard, C. E., Murrelle, L., Meyer, J. M.,
Silberg, J. L., Hewitt, J. K., Rutter, M., Simonoff, E.,
Pickles, A., Carbonneau, R., Neale, M. C., & Eaves,
L. J. (1999). Tobacco, alcohol and drug use in eight to
Olsson, D. H., Sprenkle, D. H., & Russel, C. S. (1979).
Circumplex models of marital and family systems. I.
Cohesion and adaptability dimension, family types
and clinical adaptations. Family Processes, 18, 3–28.
Twin Research and Human Genetics February 2007
149
Hermine H. Maes, Judy L. Silberg, Michael C. Neale, and Lindon J. Eaves
Olweus, D. (1989). Prevalence and incidence in the study
of anti-social behavior: Definitions and measurement.
In M. W. Klein (Ed.), Cross-national research in selfreported crime and delinquency (pp. 187–201).
Dordrecht, the Netherlands: Kluwer.
Silberg, J. L., Pickles, A., Rutter, M. L., Hewitt, J. K.,
Simonoff, E., Maes, H., Carbonneau, R., & Eaves, L.
J. (1999). The influence of genetic factors and life
stress on adolescent depression. Archives of General
Psychiatry, 56, 225–232.
Pickles, A., Rowe, R., Simonoff, E., Foley D., Rutter, M.,
& Silberg, J. (2001). Child Psychiatric symptoms and
psychosocial impairment: Relationship and prognostic
significance. British Journal of Psychiatry, 179,
230–235.
Silberg, J. L., Rutter, M., Meyer, J. M., Simonoff, E.,
Hewitt, J., Loeber, R., Pickles, A., Maes, H., & Eaves,
L. J. (1996). Genetic and environmental influences on
the covariation between hyperactivity and conduct
disturbance among juvenile twins. Journal of Child
Psychology and Psychiatry, 37, 803–816.
Prom, E. C., Eaves, L. J., Foley, D. L., Gardner, C. O.,
Riley, B. P., Archer, K. J., Wormley, B. K., & Silberg, J.
L. (2006). Monoamine Oxidase-A and childhood
adversity as risk factors for conduct disorder in
females. Manuscript in preparation.
Raven, J. C. (1956). Progressive matrices. London: H. K.
Lewis
Reynolds, C. A., Hewitt, J. K., Erickson, M. T., Silberg, J.
L., Rutter, M. L., Simonoff, E., Meyer, J. M., & Eaves,
L. J. (1996). The genetics of children’s oral reading
performance. Journal of Child Psychology and
Psychiatry, 37, 425–434.
Reynolds, C. R., & Richmond, B. O. (1978). What I
think and feel: A revised measure of children’s manifest anxiety. Journal of Abnormal Child Psychology, 6,
271–280.
Rhee, S. H., & Waldman, I. D. (2002). Genetic and environmental influences on antisocial behavior: A
meta-analysis of twin and adoption studies.
Psychological Bulletin, 128, 490–529.
Robins, L. N., Schoenberg, S. P., Homes, S. J., Ratcliff, K.
S., Benham, A., & Works, J. (1985). Early home environment and retrospective recall: A test for
concordance between siblings with and without psychiatric disorders. American Journal of
Orthopsychiatry, 55, 27–41.
Rowe, R., Pickles, A., Simonoff, E., Bulik, C., & Silberg,
J. (2002). Bulimic symptoms in the Virginia Twin
Study of Adolescent Behavioral Development:
Correlates, comorbidity, and genetics. Biological
Psychiatry, 51, 172–182.
Rutter, M., Tizard, J., & Whitmore, K. (1970).
Education, health, and behavior. London: Longman.
150
Silberg, J. L., Rutter, M. L., & Eaves, L. J. (2001a).
Genetic and environmental influences on the temporal
association between early anxiety and later depression
in girls. Biological Psychiatry, 49, 1040–1049.
Silberg, J. L., Rutter, M., Neale, M., & Eaves, L. (2001b)
Genetic moderation of environmental risk for depression and anxiety in adolescent girls. British Journal of
Psychiatry, 179, 116–121.
Simonoff, E., Pickles, A., Hervas, A., Silberg, J. L., Rutter,
M., & Eaves, L. J. (1998). Genetic influences on childhood hyperactivity: Contrast effects imply parental
rating bias, not sibling interaction. Psychological
Medicine, 28, 825–837.
Simonoff, E., Pickles, A., Hewitt, J. K., Rutter, M. L.,
Silberg, J. L., Eaves, L. J., & Meyer, J. M. (1995).
Multiple raters of disruptive child behavior: Using a
genetic strategy to examine shared views and bias.
Behavior Genetics, 25, 311–326.
Simonoff, E., Pickles, A., Meyer, J., Silberg J., & Maes, H.
(1998). Genetic and environmental influences on subtypes of conduct disorder behavior in boys. Journal of
Abnormal Child Psychology, 26, 495–509.
Simonoff, E., Pickles, A. R., Meyer, J. M., Silberg, J. L.,
Maes, H. H., Loeber, R., Rutter, M. L., Hewitt, J. K.,
& Eaves, L. J. (1997). The Virginia Twin Study of
Adolescent Behavioral Development: Influences of
age, gender and impairment on rates of disorder.
Archives of General Psychiatry, 54, 801–808.
Slosson, R. L. (1990). Slosson Oral Reading Test —
Revised. East Aurora, NY: Slosson Educational
Publications.
Silberg, J. L., Meyer, J. M., Pickles, A. R., Simonoff, E.,
Eaves, L. J., Hewitt, J. K., Maes, H. H., & Rutter, M. L.
(1996). Heterogeneity among juvenile antisocial behaviors: Findings from the Virginia Twin Study of
Adolescent Behavioral Development. In Genetics of
criminal and antisocial behavior (pp. 76–92).
Proceedings of the CIBA Foundation Symposium 194 on
the Genetics of Antisocial Behavior. Chichester: John
Wiley and Sons.
Spitzer, R. L., Williams, J. B. W., Gibbon, M., & First, M.
B. (1990). Structured Clinical Interview for DSM-IIIR,
Patient
Edition/Non-patient
Edition
(SCID-P/SCID-NP). Washington DC: American
Psychiatric Press.
Silberg, J. L., Parr, T., Neale, M. C., Rutter, M., Angold,
A., & Eaves, L. J. (2003). Maternal smoking during
pregnancy and risk to boys’ conduct disturbance: an
examination of the causal hypothesis. Biological
Psychiatry, 53, 130–135.
Topolski, T., Hewitt, J. K., Eaves, L., Meyer, J. M.,
Silberg, J. L., Simonoff, E., & Rutter, M. (1999).
Genetic and environmental influences on ratings of
manifest anxiety by parents and children. Journal of
Anxiety Disorders, 13, 371–397.
Spanier, G. B. (1976). Measuring dyadic adjustment: New
scales for assessing the quality of marriage and other
dyads. Journal of Marriage and the Family, 38, 15–28.
Twin Research and Human Genetics February 2007