Psychiatry Research 86 Ž1999.

41]57

Etiology of the impulsivityraggression relationship:

Genes or environment?

Alesha D. Seroczynski a,U , C.S. Bergemana , Emil F. Coccaro b

a
Department of Psychology, Uni¨ ersity of Notre Dame, 118 Haggar Hall, Notre Dame, IN 46556, USA
b
Clinical Neuroscience Research Unit, Department of Psychiatry, Medical College of Pennsyl¨ ania at Eastern Pennsyl¨ ania
Psychiatric Institute, 3200 Henry A¨ enue, Philadelphia, PA 19129, USA

Received 6 October 1998; received in revised form 9 December 1998; accepted 12 January 1999

Abstract

Genetic and environmental influences on the phenotypic relationship between the Barratt Impulsiveness Scale and
the aggression scales from the Buss]Durkee Hostility Inventory in adult males were examined. This study used 182
pairs of male MZ twins and 118 pairs of male DZ twins from the Vietnam Era Twin Registry. Phenotypic
relationships between the measure of impulsivity and subscales of the measure of aggression Ždirect assault, verbal
assault, indirect assault, and irritability . ranged from 0.22 to 0.51. Genetic and environmental mediation of the
phenotypic relationship between impulsivity and aggression were approximately the same for all four models.
Multivariate model-fitting analysis indicated that irritability and impulsivity had a larger phenotypic relationship, as
well as a greater portion of shared genes and environment than the other three subscales of aggression. This
suggests, for example, that there are more overlapping genetic and environmental influences accounting for the
relationship between irritability and impulsivity than between direct assault and impulsivity. The effects of such
findings on our understanding of impulsive aggression are discussed. Q 1999 Elsevier Science Ireland Ltd. All rights
reserved.

Keywords: Impulsivity; Aggression; Behavior]genetic; Twins

U
Corresponding author. Tel.: q1 219 631 5373; fax: q1 219 631 8883; e-mail: alesha.d.seroczynski.1@nd.edu

0165-1781r99r$ - see front matter Q 1999 Elsevier Science Ireland Ltd. All rights reserved.
PII: S 0 1 6 5 - 1 7 8 1 Ž 9 9 . 0 0 0 1 3 - X
42 A.D. Seroczynski et al. r Psychiatry Research 86 (1999) 41]57
1. Introduction
Kendall and Clarkin Ž1992. have described the
study of comorbidity as ‘the premier challenge
facing mental health professionals in the 1990s’
Žp. 833.. One issue pertaining to this challenge
that needs attention is the etiological factors that
underlie comorbid disorders in children and
adults. Of the many symptoms of psychopathology
that can co-exist within an individual, impulsivity
and aggression are among the most common.
Whether conceptualizing the constructs with a
categorical Že.g. diagnositic. or dimensional ap-
proach Že.g. traits; Plomin et al., 1991; Nigg and
Goldsmith, 1994., the co-occurrence of impulsive
and aggressive behaviors cannot be denied. In
fact, attention-deficit hyperactivity disorder
ŽADHD. and disorders of conduct wi.e. oppositio-
nal defiant disorder ŽODD., conduct disorder
ŽCD.x have been found to co-occur in 30]50% of
cases in both epidemiological and clinical samples
ŽBiederman et al., 1991; Nottelmann and Jensen,
1995., although it should be noted that the bulk
of this work has been with child and adolescent
populations. The purpose of this study was to
address the etiology of the phenotypic relation-
ship between impulsivity and aggression with a
sample of adult males using a non-clinical sample.
Despite the fact that aggression and impulsivity
often co-exist in many individuals, little effort has
been made to distinguish between impulsive be-
havior with an aggressive flair and such behavior
without aggressive tendencies ŽGray et al., 1983..
Among those studying aggression, however, an
increasingly popular trend has been to identify
different types of aggressive behavior. One such
distinction has been made between two types of
aggression: reactive and proactive ŽDodge, 1991..
Reacti¨ e aggression has been defined as a hostile,
angry reaction to perceived frustration. The reac-
tive-aggressive individual, for example, overreacts
to minor provocations and is viewed as short-tem-
pered and volatile. Proacti¨ e aggression, on the
other hand, appears to be an instrumental behav-
ior that is prompted by its anticipated benefits.
The proactive aggressor is often a bully to peers
and a criminal threat to society ŽDodge, 1991..
Thus, reactive aggression is more strongly tied to
impulsive behavior, whereas proactive aggression
is, by definition, more premeditated.
Although the distinction between reactive and
proactive aggression was first suggested over 30
years ago ŽBerkowitz, 1962; Bandura, 1973., stud-
ies on aggressive behavior have rarely attempted
to separate the two. If such a distinction exists,
then it would be possible to be aggressive without
being impulsive, and not surprisingly, to be impul-
sive without being aggressive ŽBarkley, 1990.. For
this to be true, the etiologies of impulsivity and
aggression would necessarily be distinct. That is,
the underlying factors Žgenetic, environmental, or
some combination. influencing impulsivity and
aggression would necessarily be unique to each
characteristic, and an impulsive-aggressive indi-
vidual would simply inherit or learn both traits.
Research with impulsive and meditative criminals
suggests that this may be the case. For example,
Barratt et al. Ž1997. found that impulsive]aggres-
sive inmates differed from non-impulsive inmates
on neuropsychological and cognitive psychophysi-
ological measures of information processing, sug-
gesting that the two types of antisocial individuals
may have distinct etiologies.
Results of previous research on the etiology of
individual differences in impulsive and aggressive
behavior have indicated that both genetic and
environmental influences are important for the
development of each of these characteristics
ŽRushton et al., 1986; Pedersen et al., 1988;
Plomin et al., 1988, 1990a; Rose, 1988; McCart-
ney et al., 1990; Loehlin, 1992; Eaves et al., 1993;
Nigg and Goldsmith, 1994; Lyons et al., 1995;
DiLalla et al., 1996.. The present study, employ-
ing a behavior-genetic perspective, was designed
to assess the etiology of the phenotypic relation-
ship between self-reported impulsivity and self-
reported dimensions of aggression. Although ex-
isting measures of aggression make it difficult to
tease apart impulsive from aggressive behavior,
this study attempted to overcome this limitation
by using measures of four types of aggressive
behavior exhibiting varying degrees of reactivity:
direct assault; indirect assault; verbal assault; and
irritability, as well as a distinct measure of impul-
 A.D. Seroczynski et al. r Psychiatry Research 86 (1999) 41]57
sivity. The following is a brief review of the eti-
ology of both impulsivity and aggression, as well
as the impulsive]aggressive relationship.
1.1. Impulsi¨ ity
Shapiro Ž1965. has defined impulsi¨ ity as a ten-
dency to act on the spur of the moment, without
planning or a clear sense of decision or desire, as
if ‘the regular executive apparatuses or generally
operative modes of functioning are bypassed or
broken through’ Žp. 135.. Several behavior-genetic
studies, focusing on a variety of age groups, have
assessed genetic and environmental influences on
measures of impulsivity. In one such study, using
twins reared together and apart, researchers found
a broad heritability estimate for impulsivity of
0.45, and the influence of being reared in the
same family Žshared family environment. was
minimal ŽPedersen et al., 1988.. Results of this
study suggest that most of the phenotypic vari-
ance in impulsivity in these adult twins results
from a combination of genetic and non-shared
environmental influences Žinfluences that make
family members different from each other.. Other
researchers, using family or adoption studies, re-
port substantially lower estimates of heritability
for impulsivity than do twin studies. One such
study found that only 16% of the phenotypic
variance in impulsivity, as measured by the corre-
lation between the genetically related mid-parent
score Žor average of the two biological parents.
and the child, was attributable to genetic influ-
ence. Similarly, analyses based on biologically
related siblings and adopted siblings revealed a
heritability estimate of 0.30 ŽScarr et al., 1981..
Why such a large discrepancy in heritability
estimates? One reason could be that the genetic
influence on impulsivity appears to be of the
non-additive type. That is, the genes that influ-
ence behavior in a non-linear, dominant manner
would be partially shared by siblings and DZ
twins, fully shared by MZ twins, but not shared by
parents and their offspring. The result of using a
parent]offspring design to assess the etiology of a
behavior that is due Žat least partially. to non-ad-
ditive genetic influences would be an under-
estimation of the genetic influences on behavior
43
ŽScarr et al., 1981.. In fact, researchers in the field
argue that early behavior]genetic studies in per-
sonality failed to consider the importance of
non-additive genetic variance ŽRose, 1988.. This
omission could feasibly account for the lower
parent]child correlations, and would justify using
a methodology that can account for non-additive
genetic influence.
1.2. Aggression
Like impulsivity, much has been hypothesized
about the role that genes and the environment
play in the development and maintenance of ag-
gressive behavior. In his review of reactive aggres-
sion, Dodge Ž1991. maintains that reactive aggres-
sion results from environmental influences, such
as early chronic, life-threatening or traumatic ex-
periences that promote feelings of anger, fear,
and hyperactivity. He also stresses the importance
of the attachment relationship which plays a role
in the development of feelings of security and the
learning of empathy, the basis for controlling
reactive aggression. He acknowledges the impor-
tance of neurological mechanisms, but mentions
little about genetic influences.
Unlike most personality traits, including impul-
sivity, aggression shows no consistent pattern of
genetic influence ŽPlomin et al., 1990b.. In a
thorough review of the literature on the heritabil-
ity of aggressive behavior across the lifespan,
Plomin et al. Ž1990b. concluded that the results of
a variety of studies, using a myriad of measure-
ment techniques, are mixed. Some studies have
shown that genetic influences account for approx-
imately 50% of the variance in aggression ŽRush-
ton et al., 1986; Tellegen et al., 1988., whereas
others have found little difference in resemblance
between MZ and DZ twins on measures of ag-
gressive behavior ŽVandenberg, 1962; Plomin et
al., 1981.. Age also appears to be an important
factor. Genes appear to account for individual
differences in childhood and adult aggression, but
explain little variance in adolescent aggression
ŽLyons et al., 1995.. The majority of studies have
consistently found, however, that the remaining
variance can be attributed to specific factors id-
iosyncratic to the particular twin Žnon-shared en-
44 A.D. Seroczynski et al. r Psychiatry Research 86 (1999) 41]57
vironment.. What is especially interesting is that
being reared in the same environment does not
appear to make twins or other family members
similar to one another.
Many have stressed the importance of consis-
tency in the measurement of aggression Že.g.
Plomin et al., 1990b; Bergeman and Seroczynski,
1998.; no two studies have used the same mea-
sure of aggression to assess environmental and
genetic influences. Pencil and paper measures
have consistently shown moderate to strong ge-
netic influences and a lack of shared environmen-
tal influences, whereas studies employing obser-
vational techniques have produced mixed results
ŽPlomin et al., 1990b.. The differences in method-
ology may be reflected in the discrepant results.
Furthermore, observational measures have been
used more often with children Že.g. Ghodsian-
Carpey and Baker, 1987., whereas self-report
questionnaires are more likely to be used with an
adult population Že.g. Rushton et al., 1986..
Therefore, age differences seen in heritability es-
timates of aggressive behavior could be the result
of increasing genetic influence across the life-
span, methodological differences, or some combi-
nation of the two.
A recent study of twins from the Vietnam Era
Twin Registry ŽCoccaro et al., 1997. indicated
significant genetic influence for four subscales of
the Buss]Durkee Hostility Inventory ŽBuss and
Durkee, 1957., one of the most widely used mea-
sures of aggression. The direct assault scale as-
sesses physical violence directed toward others
Že.g. ‘I get into fights about as often as the next
person’.. The indirect assault scale taps into both
roundabout and undirected aggression, such as
malicious gossip and temper tantrums Že.g. ‘I
sometimes spread gossip about people I don’t
like’.. Quick temper, grouchiness, and exaspera-
tion are assessed with the irritability scale Že.g. ‘I
often feel like a powder keg ready to explode’..
Verbal assault measures negative affect expressed
in both the style and content of speech Že.g.
‘When I get mad, I say nasty things’.. Model-fit-
ting results indicated a heritability estimate of
0.47 for the direct assault subscale, 0.40 for indi-
rect assault, 0.28 for verbal assault, and 0.37 for
irritability. Coccaro et al. Ž1997. also reported
that non-shared environmental influences ac-
counted for 50]70% of the variance for the four
subscales.
The Coccaro et al. Ž1997. study is one of the
few to specifically delineate sub-components of
aggressive behavior and assess genetic and envi-
ronmental influences on each. The next step in
this line of research is to determine whether
behaviors such as impulsivity, which show pheno-
typic correlations with aggression exhibit etiologic
overlap as well. This study is an extension of this
direction of research.
1.3. Impulsi¨ ity r aggression relationship
Individuals who exhibit impulsive behaviors of-
ten have co-existing behavior problems. Although
this study is based on the assumption that behav-
ior operates along a continuum, much of the
research in this area has been based on a cate-
gorical Že.g. diagnostic. view. Two positions have
been proposed concerning the relationship
between impulsivity and aggressive behavior prob-
lems. One suggests that impulsivity and conduct
disorders are indistinguishable, and the other sug-
gests that they are either partially or completely
independent. Support can be found for both. For
example, proponents of the overlapping position
contend that individuals with ADHD and CD
often show similar correlates, outcomes, and
treatment responses ŽQuay et al., 1987; Barkley et
al., 1989.. Moreover, symptoms of ADHD and
CD Ždisruptive and non-compliant behaviors. are
often inter-correlated in factor-analytic studies of
children with behavior disorders ŽQuay, 1986;
Campbell and Werry, 1986..
In contrast, proponents of the independent po-
sition maintain that impulsivity and aggression
are separate dimensions. This view is based on a
series of studies by Loney et al. Ž1981., Bieder-
man et al., 1990, 1996a,b and Faraone et al.
Ž1995.. The results of the work by Loney et al.
indicated that symptoms of ADHD and aggressive
behaviors were not highly correlated and showed
different patterns of concurrent and predictive
validity. In a series of studies, Biederman and
Faraone et al. showed that ADHD and antisocial
disorders appeared to be co-segregated, that is,
 A.D. Seroczynski et al. r Psychiatry Research 86 (1999) 41]57
dependently transmitted, in families of children
with ADHD and CD. They suggest that children
with ADHD and CD may be etiologically distinct
from those without CD. Their work, however, is
confounded by the fact that the majority of their
etiological research has been conducted using
parents and siblings of probands, not twins; thus
environmental and genetic effects cannot be sta-
tistically isolated.
Taken together, these studies suggest that there
is a moderate phenotypic Žobservable. relation-
ship between impulsivity and aggression. The
question that needs to be asked, then, is whether
impulsive aggression is an entity in and of itself,
or whether it is more of a marriage of two traits
within an individual, and if it is a marriage, to
what extent are the same or different genetic and
environmental influences mediating the relation-
ship between these two behaviors? The purpose
of the present study was two-fold. First, to assess
the phenotypic, or observable, relationship
between impulsivity and aggression. The pheno-
typic relationship between impulsivity and the
subscales of the aggression measure was expected
to be moderate. The second purpose was to assess
the extent to which genes and environment medi-
ate the impulsivityraggression relationship, and
then determine whether the underlying influences
contributing to impulsivity were shared with ag-
gression. Due to the fact that some subscales of
the Buss]Durkee Hostility Inventory indicated
more impulsive-like behavior Že.g. ‘I lose my tem-
per easily but get over it quickly’ in the irritability
scale., the bivariate relationships between each
aggression subscale and impulsivity were analyzed
separately. It was hypothesized that the measure
with the least amount of reactive aggression Že.g.
direct assault subscale . would show the least over-
lapping genetic and environmental influences with
impulsivity, whereas the subscale that is most
reflective of true reactive aggression Že.g. irritabil-
ity subscale . would show the most. There were no
specific hypotheses for the indirect and verbal
assault subscales of the Buss]Durkee. Environ-
mental influence of the non-shared type was ex-
pected to mediate the largest portion of the
phenotypic relationships observed between each
set of variables.
45
2. Method
2.1. Participants
Participants in the present study were adult
male twins from the Vietnam Era Twin ŽVET.
Registry. In light of gender differences in aggres-
sive behavior consistently noted in the literature
Že.g. Loeber and Keenan, 1994., the use of a male
population, then, becomes important in studying
aggressive behavior because some of these dif-
ferences may be due to etiologically distinct
phenomena. Many social problems involving ag-
gression are also related directly to men: most of
our prisons are overcrowded with aggressive men;
the majority of juvenile delinquents are young
men. To account for this, an adult, all male sam-
ple was used in this study.
The VET Registry was constructed from the
Defense Manpower Data Center of the United
States Department of Defense ŽDMDC; Eisen et
al., 1987.. The DMDC maintains a computer file
of servicemen discharged from active military duty
beginning in 1968. The VET Registry consists of
7375 male]male twin pairs born between 1939
and 1957 with both brothers having served in the
US military during the Vietnam War. A pilot
study conducted to assess ascertainment bias re-
vealed that these twins were representative of all
twins who served in the military during the Viet-
nam War on a wide variety of sociodemographic
and other variables ŽGoldberg et al., 1987.. Zy-
gosity for the VET Registry was determined using
a questionnaire technique supplemented with
blood group typing data from military records
ŽEisen et al., 1989.. Zygosity assignment was given
based on a logistic regression technique using a
different set of discriminating variables depend-
ing on race Ži.e. eye color is a good discriminating
variable in Caucasians, but not in African-
Americans.. This approach was similar to that
used by Magnus et al. Ž1983. with Norwegian twin
pairs.
Subjects were selected from the VET Registry
based on the residency of both members of the
pair in one of five states: Delaware, Maryland,
New Jersey, New York, or Pennsylvania. This
geographical region was chosen in order to facili-
46 A.D. Seroczynski et al. r Psychiatry Research 86 (1999) 41]57
tate future in-person follow-up. The subjects in
the present study were 182 pairs of MZ twins and
118 pairs of DZ twins obtained from the Vietnam
Era Twin ŽVET. Registry. Mean age of the sam-
ple was 43.97 years ŽS.D.s 2.9; range s 36]54
years.. The sample was 94.1% Caucasian and
5.9% African-American. The number of African-
Americans in the sample is slightly under-repre-
sentative of Blacks because there is a higher
percentage of Blacks in the general population.
In addition, African-Americans also experience a
higher incidence of twinning ŽFarber, 1981.. How-
ever, the sample is reflective of the distribution
for race found in the entire VET Registry Ž93.0%
Caucasian and 6.5% African-American; Hender-
son et al., 1990..
Subjects were mailed an introductory letter and
questionnaire packet which included a large bat-
tery of questionnaires. One week after this first
packet, a thank-yourreminder postcard was sent
to all subjects. VET Registry members not re-
sponding within 3 weeks of the postcard were sent
a second questionnaire. Six weeks after the third
and final questionnaire mailing, subjects were
contacted by telephone. Of the 1208 solicited, a
total of 796 responded Ž65.9%., 169 refused
Ž16.2%., 214 were lost to follow Ž18%., and two
were deceased Ž0.2%.. Of the 796 respondents,
628 Ž78.9%. were part of a twin pair. Therefore,
approximately 50% of both twins in a pair re-
sponded to the survey. Because the twins had
recently received a large survey from another
source, the response rate was probably reduced
due to subject fatigue.
2.2. Measures
2.2.1. Impulsi¨ ity
Items to assess impulsivity came from four sub-
scales of the Barratt Impulsiveness Scales Že.g.
BIS 7B, BIS 10, BIS 11; Barratt, 1959, 1965;
Patton et al., 1995.. Because estimates of internal
consistency reliability for our sample were low for
each of the subscales Ži.e. alphas ranged from 0.39
to 0.75., a principal components factor analysis
was conducted on all 26 items from the four
subscales. ŽThere were no overlapping items..
Items that had an aggressive or hostile compo-
nent based on content were removed. Results of
the factor analysis indicated that 17 of the items
formed a principal component that accounted for
49% of the variance. These 17 items were then
summed to create a total score of impulsivity.
Items such as ‘I plan trips well ahead of time’, ‘I
buy things on impulse’, and ‘I am restless at the
theater or at lectures’ were included. The internal
consistency reliability of this measure was 0.82.
The mean total score for the BIS was 31.76
ŽS.D.s 6.97.. There were no significant mean or
variance differences by zygosity which fulfills a
basic assumption of the twin method ŽLykken et
al., 1987.. Likewise, there was no significant cor-
relation with age Ž r s y0.07.; however, prior to
model-fitting analyses, scale scores were standard-
ized using a regression technique to correct for
any possible main effects of age that could inflate
twin correlations ŽMcGue and Bouchard, 1984..
2.2.2. Aggression
The Buss]Durkee Hostility Inventory was used
to assess aggression ŽBuss and Durkee, 1957.. A
complete overview of the use of the Buss]Durkee
Hostility Inventory with the present sample is
available elsewhere ŽCoccaro et al., 1997.. Four of
the seven subscales Ž44 items., which were dis-
cussed in detail in the introduction, were used:
direct assault, indirect assault, irritability, and
verbal assault. Cronbach’s alphas were 0.74, 0.65,
0.73, and 0.67 for assault, indirect hostility, irri-
tability, and verbal hostility, respectively, indicat-
ing reasonable internal consistency ŽCoccaro et
al., 1997.. As reported in Coccaro et al. Ž1997.,
there were no significant mean or variance dif-
ferences by zygosity. All four subscales were,
however, significantly correlated with age and
were standardized to correct for any main effects
of age.
2.3. Analyses
2.3.1. Correlations
This study used the classic twin design, which
compares identical and fraternal twins reared
together. It is assumed that monozygotic ŽMZ.
twins share 100% of their genes, whereas dizy-
gotic ŽDZ. twins share only 50% of genetic influ-
 A.D. Seroczynski et al. r Psychiatry Research 86 (1999) 41]57
ences that operate additively. Thus, if the intra-
class correlation between scores on a given trait
for MZ twins is higher than that for DZ twins,
then twice the difference is attributable to genetic
influence ŽRushton et al., 1986.. If the DZ corre-
lation is about half that of MZ twins, then an
additive influence is assumed. If, however, the DZ
correlation is less than half that of the MZ twin
correlation, then a non-additive genetic influence
must be considered ŽPlomin et al., 1990a.. In a
twin design, it is also possible to calculate the
effects of both shared rearing environment, which
makes the twins or other family members similar
to one another, and non-shared rearing environ-
mental influences, which contribute to the pheno-
typic differences observed in the twins or other
family members. The non-shared estimate also
includes measurement error ŽPlomin et al., 1990a..
2.3.2. Uni¨ ariate model-fitting
Estimates of genetic and environmental influ-
ences based on intraclass correlations have rela-
tively low power and large standard errors
ŽPedersen et al., 1988.. Model-fitting techniques
are useful because they make assumptions ex-
plicit, test the fit of a particular model with its set
of assumptions, and make it possible to provide
appropriate parameter estimates and standard er-
rors of the estimates given the assumptions of the
model and to compare the fit of alternative mod-
els ŽLoehlin, 1987..
A detailed overview of the use of model-fitting
with the classical twin method has been provided
elsewhere Žsee Neale and Cardon, 1992.. Esti-
mates of the additive and non-additive genetic
parameters, and the shared and non-shared envi-
ronmental parameters, were obtained using LIS-
REL VI ŽJoreskog and Sorbom, 1985.. This com-
puter program generates maximum-likelihood
estimates of the model parameters by using
observed mean squares within and between pairs
for the MZ and DZ twins. LISREL also provides
a x 2 value, indicating the goodness of fit of the
model. Because the four models tested were not
nested, Akaike’s Information Criterion ŽAIC. was
used to determine the best-fitting model ŽAkaike,
1987..
47
2.3.3. Multi¨ ariate model-fitting
A major objective of multivariate genetic analy-
sis is to assess the genetic and environmental
causes of phenotypic covariation between two
variables, in this case, measures of impulsivity and
aggression ŽFulker et al., 1983.. One way to test
the genetic and environmental influences on the
covariance between two measures is to use the
common factor model. Fig. 1 depicts the covaria-
tion between two observed variables Ž P . for one
member of the twin pair. By predicting one twin’s
aggression score Ž PX . from the co-twin’s impul-
sivity score Ž PY ., genetic and environmental in-
fluences mediating the phenotypic relationship
can be determined. As in univariate model-fitting,
multivariate model-fitting using mean-squares and
cross-products provides a more powerful test of
the genetic and environmental parameters. Fig. 1
represents the bivariate model for impulsivity and
aggression.
Path subscripts h11 and h21 are the genetic
effects on X Žaggression. and Y Žimpulsivity., re-
spectively, from the common G factor. Genetic
mediation between PX and PY is represented by
these paths through the common genetic factor
G. Path subscripts s12 and s22 are the shared
environmental effects on X and Y from the com-
mon Es factor; n13 and n23 are the non-shared
environmental effects on X and Y from the com-
mon En factor. Environmental mediation
between Px and Py is represented through both
the common shared rearing environmental factor
Ž s12 and s22 . and through the common non-
shared rearing environmental factor Ž n13 and
n23; Bergeman et al., 1991..
3. Results
3.1. Uni¨ ariate model-fitting
Intraclass correlations for the Barratt Impul-
siveness Scale total score indicate a pattern of
non-additive genetic influence. The MZ correla-
tion for the Barratt Total Score is 0.43, whereas
the DZ correlation is 0.17, or less than half the
MZ correlation, suggesting non-additive genetic
influence.
48 A.D. Seroczynski et al. r Psychiatry Research 86 (1999) 41]57
Fig. 1. Path model depicting common and unique tactors for
genetic and environmental sources or variance and covariance
for one twin in the pair. Abbre¨ iations: Gs the common
non-additive genetic factor; Ess the common shared environ-
mental factor; Ens the common non-shared environmental
factor; gl s the specific non-additive genetic factor for aggres-
sion; esl s the specific shared environmental factors for ag-
gression; enls the specific non-shared environmental factors
for aggression; g 2 s the specific non-additive genetic factor
for impulsivity; es2 s the specific shared environmental fac-
tors for impulsivity; and en2 s the specific non-shared envi-
ronmental factors for impulsivity.
Model-fitting analysis confirms the trend noted
with the intra-class correlations. For the BIS total
score, five models were fit to the data. The model
that allowed for non-additive genetic Ž Gd ., and
non-shared environmental Ž En. parameters, pro-
vided the best fit to the data Ž x 2 s 0.41, P Ž2. s
0.82, AICs 4.41.. The model that allowed for
additive genetic influence Ž Ga. and non-shared
environmental influence Ž En. provided the worst
fit Ž x 2 s 30.40, P Ž2. s 0.00, AICs 34.40.. The
model with just environmental parameters Ž Es
and En. also provided a poor fit to the data
Ž x 2 s 6.67, P Ž2. s 0.036, AICs 10.67.. Models
that included Es, En, and Ga fit the data well
Ž x 2 s 0.28, P Ž1. s 0.595, AICs 6.28., as did the
model that included Ga, En, and Gd Ž x 2 s 0.00,
P Ž1. s 0.946, AICs 6.0., although not as well as
the one with only non-additive genetic and non-
shared environmental parameters.
Table 1 indicates that the Barratt total score
showed significant non-additive genetic influence
Ž44%.; however, non-shared environmental in-
fluences accounted for the largest portion of the
variance Ž56%.. Results of the Buss]Durkee
model-fitting, reported in detail in Coccaro et al.
Ž1997., are also provided in Table 1. All four
subscales showed a significant genetic influence.
The best fitting model for direct assault showed
an additive genetic parameter, whereas the other
three aggression measures indicated the impor-
tance of non-additive genetic influences. None of
the four subscales showed an influence of shared
rearing environment, but the non-shared environ-
ment parameter was significant for all four vari-
ables.
3.2. Multi¨ ariate model-fitting
In order to investigate an etiological relation-
ship between two variables, a phenotypic relation-
ship between those variables must first be es-
tablished. Pearson product]moment correlations
were used to assess the phenotypic relationship
between aggression and impulsivity; analyses
showed that the BIS total score correlated 0.22
with direct assault, 0.32 with indirect assault, 0.29
with verbal assault, and 0.51 with irritability. All
of these correlations are significant at P- 0.001.
Cross-twin correlations between one twin’s
score on impulsivity and the other twin’s score on
direct assault, indirect assault, irritability, and
verbal assault were 0.14, 0.12, 0.23, and 0.14,
respectively, for MZ twins, and y0.05, 0.09, 0.02,
and y0.02, respectively, for DZ twins. Because
on average the DZ correlations are less than half
the MZ correlations, a non-additive genetic medi-
ation of the phenotypic relationship between im-
pulsivity and the subscales of the aggression mea-
sure is indicated. Table 2 shows the maximum-
likelihood estimates for each multivariate analy-
sis. Fig. 2 provides a graphic representation of the
multivariate model used to test the genetic and
environmental mediation of the covariance
between the Buss]Durkee irritability scale and
the Barratt measure of impulsivity Ždata provided
in Table 2..
 A.D. Seroczynski et al. r Psychiatry Research 86 (1999) 41]57 49

Table 1
Maximum-likelihood model-fitting: parameter estimates, S.E., percent variances, and x 2 goodness of fit indices for the Barratt
Impulsiveness Scale and the Buss]Durkee Hostility Inventory

Scale Parameter estimates " S.E. Ž% variances.

Ga Gd Es En x2 Ž2 d.f.. P

Barratt
a
Total score 0.66" 0.05 0.75" 0.04 0.41 0.82
Ž44%. Ž0%. Ž56%.
Buss]Durkee
a
Direct assault 0.60" 0.05 0.73" 0.04 2.8 0.25
Ž47%. Ž0%. Ž53%.
a
Indirect assault 0.63" 0.06 0.77" 0.04 1.93 0.38
Ž40%. Ž0%. Ž60%.
a
Verbal assault 0.53" 0.07 0.85" 0.04 0.13 0.94
Ž28%. Ž0%. Ž72%.
a
Irritability 0.61" 0.06 0.80" 0.04 3.20 0.22
Ž37%. Ž0%. Ž63%.

Note: Gas additive genetic; Gds non-additive genetic; Ess shared rearing environment; Ens non-shared rearing environment.
Buss]Durkee Hostility Inventory results are from Coccaro et al. Ž1997..
a
Parameter estimate is zero.

This model included common and unique ge-
netic and non-shared environmental influences
on each of the two traits. Common factor load-
ings for the aggression variable and impulsivity in
each model were constrained to be equal, which
allowed the unique parameter estimates to be
calculated for each variable ŽLoehlin, 1996.. A
shared environmental parameter was not in-
cluded because shared environment was not found
to be a significant influence on either trait in the
univariate analyses, and if the shared environ-
ment is not found to be an important source of

Table 2
Maximum likelihood parameter estimates and S.E. for the commom and specific factor loadings for four bivariate models with the
Barratt total score of impulsivity and subscales of the Buss]Durkee Hostility Inventory

Common factor loadings Specific factor loadings

Gd En Gd En
mle S.E. mle S.E. mle S.E. mle S.E.

Irritability 0.49 0.06 0.52 0.05 0.37 0.08 0.60 0.05

with impulsivity 0.49 0.06 0.52 0.05 0.45 0.06 0.54 0.04
Direct assault 0.35 0.07 0.31 0.06 0.60 0.06 0.65 0.04
with impulsivity 0.35 0.07 0.31 0.06 0.56 0.07 0.68 0.04
Indirect assault 0.38 0.07 0.41 0.05 0.51 0.07 0.66 0.05
with impulsivity 0.38 0.07 0.41 0.05 0.54 0.06 0.63 0.05
Verbal assault 0.36 0.07 0.40 0.06 0.39 0.09 0.74 0.05
with impulsivity 0.36 0.07 0.40 0.06 0.55 0.06 0.63 0.05

Note: Gs the common non-additive genetic factor; Ens the common non-shared environmental factor; g s the specific non-ad-
ditive genetic factor; ens the specific non-shared environmental factor; mle s maximum-likelihood parameter estimate; S.E.s
standard error. The common Gd and En factor loadings were constrained to be equal. x 2 Ž6. s 5.48, Ps 0.48 for impulsivity wifh
irritability; x 2 Ž6. s 8.78, Ps 0.l9 for impulsivity with direct assault; x 2 Ž6. s 4.06, P s 0.67 for impulsivity with indirect assault;
x 2 Ž6. s 2.23, Ps 0.90 for impulsivity with verbal assault. The degrees of freedom reflect the use of mean; squares and
cross-products data to estimate the model.
50 A.D. Seroczynski et al. r Psychiatry Research 86 (1999) 41]57
Fig. 2. Path diagram of the multivariate model used to test
genetic and environmental mediation of the covariation
between impulsivity and irritability. Abbre¨ iations: Gds the
common non-additive genetic factor; Ens the common non-
shared environmental factor; g1 and g2 s the specific genetic
factors for aggression and impulsivity, respectively; en1 and
en2s the specific non-shared environmental factors for ag-
gression and impulsivity, respectively. The specific factors as-
sess the unique variation in the variables with which they are
associated by the arrow paths.
variance for each trait individually, then it cannot
mediate the relationship between two variables.
Genetic and environmental mediation estimates,
as well as genetic and environmental correlations,
provide useful information for showing how and
why two variables are etiologically related. Esti-
mates of genetic and environmental mediation
indicate the proportion of genetic and environ-
mental influences on the phenotypic resemblance.
Genetic and environmental correlations are also
important because they provide information about
the extent to which the genetic and environmen-
tal factors that affect one variable Ži.e. aggression.
overlap, or are related to those affecting the
other variable Ži.e. impulsivity..
Using the results presented in Table 2 and
shown in Fig. 2, the phenotypic correlation
between irritability and impulsivity can be calcu-
lated. This is done by multiplying the common
factor loadings for genetic and non-shared envi-
ronmental influences for each variable and then
summing the products. For example, to determine
the model-fitting estimates of the phenotypic cor-
relation between impulsivity and irritability, the
common genetic and common non-shared envi-
ronmental factor loadings are multiplied together
and summed. That equals Ž0.49. 2 q Ž0.52. 2 wor
Žhxhyrg. q Žexeyre.x, which is 0.51. To find the
percent of phenotypic covariation due to common
genetic influences Žgenetic mediation., the esti-
mate of common genetic influence Ž0.49 2 or
hxhyrg. is divided by the total phenotypic correla-
tion between impulsivity and irritability Ž0.51..
Thus, 47% of the covariation between the two
variables is due to common genetic influences,
and 53% wŽ0.52. 2r0.51x is due to common non-
shared environmental influences Ženvironmental
mediation., or non-shared events affecting both
variables. These might be influences, such as peer
relationships or significant life events, that affect
both aggression and impulsivity for one member
of the twin pair, but do not contribute to twin
similarity for these traits.
Table 3 provides the phenotypic correlations,
as well as the genetic and environmental media-
tion estimates, for all four models. The pheno-
typic correlations ranged from 0.22 for direct
assaultrimpulsivity to 0.51 for irritabilityrimpul-
sivity. As expected, the observable relationship
between irritable aggression and impulsivity was
stronger than the other three comparisons. The
estimates of genetic and environmental mediation
are described in the example above. The pheno-
typic correlation between direct assault and im-
pulsivity was 0.22; of that, 56% was mediated
genetically and 44% was mediated by non-shared
environmental influences. Likewise, 45% of the
phenotypic relationship between verbal assault
and impulsivity was mediated genetically and 55%
was mediated by non-shared environmental in-
fluences Ž rp s 0.29., 46% of the phenotypic rela-
tionship between indirect assault and impulsivity
was mediated genetically, and 54% was mediated
by non-shared environmental influences Ž rp s
0.31..
Genetic and environmental mediation esti-
mates indicate the proportion of genes and envi-
ronment contributing to the observable relation-
ship between two traits. Likewise, the genetic Ž rg .
and environmental Ž re . correlations provide help-
 A.D. Seroczynski et al. r Psychiatry Research 86 (1999) 41]57 51

Table 3
Mediating influences on the relationship between the Barratt total score and subscales of the Buss]Durkee Hostility Inventory

hxhyrg exeyre Phenotypic Genetic Environmental

correlation mediation mediation

Impulsivity with:
Irritability 0.24 0.27 0.51 47% 53%
Direct assault 0.12 0.10 0.22 56% 44%
Indirect assault 0.14 0.17 0.31 46% 54%
Verbal assault 0.13 0.16 0.29 45% 55%

Note: hxhyrg is the genetic path representing the square root of the heritability estimates for each aggression subscale and
impulsivity Ž h x and h y , respectively. and the genetic correlation Ž rg .; exeyre is the non-shared environmental path which represents
the square root of the non-shared environmental influences on each aggression subscale and impulsivity Ž e x and e y , respectively.
and the environmental correlation Ž re .; the phenotypic correlation is the estimate of the observed relationship between the two
variables. Genetic and environmental mediation scores estimate the proportion of the phenotypic correlation accounted for by
genetic and non-shared environmental influences.

ful etiological information. The genetic correla-
tion is an indication of the degree of overlap
between genes affecting both traits. The environ-
mental correlation, showing the degree of related
non-shared environmental influences affecting
both traits, can be calculated in the same manner.
The genetic correlation between impulsivity and
irritability can be calculated by multiplying the
common genetic influence on both impulsivity
and aggression, and dividing that product by the
square root of the product of the common and
unique genetic influences on each latent variable
separately. Using the path subscripts in Fig. 1,
this formula can be written as follows:
rgxy s Ž h11.Ž h21. r6 wŽ h11. q Ž h14. xwŽ h21 .
2 2
q Ž h27. x
2
for the model between direct assault and impul-
sivity to 0.59 for the model depicting the relation-
ship between impulsivity and irritability. The
non-shared environmental correlations ranged
from 0.18 to 0.45, and also indicate that the
relationship between irritability and impulsivity
shows the greatest overlap of non-shared environ-
mental influences. Compared to other genetic
and environmental correlations reported in the
literature Že.g. Bergeman et al., 1991., these are
fairly low.
4. Discussion
This study was designed for two purposes. The
2
first purpose was to assess the phenotypic rela-

Table 4
For the model depicted in Fig. 2, Genetic and environmental correlations for the relationship
between the subscales of the Buss]Durkee measure of
rgxy s Ž 0.49. 2r6 wŽ 0.49. Ž 0.37. xwŽ 0.49.
2 2 2
aggression and the Barratt total score of impulsivity
q Ž 0.45. x s 0.59
2
Genetic Environmental
correlation correlation
Thus, the genetic contribution to the pheno-
typic correlation weights the genetic correlation Impulsivity with:
by the product of the square-roots of heritability Irritability 0.59 0.45
estimates for each variable. Non-shared environ- Direct assault 0.26 0.18
Indirect assault 0.34 0.29
mental correlations can similarly be calculated for Verbal assault 0.37 0.25
each model.
Table 4 provides the genetic and non-shared Note: Genetic and environmental correlations represent the
degree of overlapping genes andror non-shared environmen-
environmental correlations for each of the four
tal influences affecting both a subscale of aggression and
models. The genetic correlations ranged from 0.26 impulsivity.
52 A.D. Seroczynski et al. r Psychiatry Research 86 (1999) 41]57
tionship between impulsivity and aggression. Cor-
relation coefficients suggest a moderate to strong
relationship between different kinds of aggressive
behavior and impulsivity. For this sample, direct
assault showed the weakest relationship with im-
pulsivity, verbal assault and indirect assault
showed moderate relationships with impulsivity,
and irritability showed the strongest relationship
with impulsivity. The second purpose of this study
was to determine the extent to which the genetic
and environmental influences that contribute to
impulsivity are shared with those contributing to
reactive aggression, and to assess environmental
and genetic mediation of the impulsivityraggres-
sion relationship.
Bivariate model-fitting results lend support to
our hypothesis: impulsivity and reactive aggres-
sion shared a large portion of genetic and envi-
ronmental variance, whereas impulsivity and
physically assaultive aggression shared very little
genetic and environmental factors. Although all
four models indicated that the relationship
between the measure of impulsivity and measures
of aggression were mediated genetically to a simi-
lar extent, the model of impulsivity and irritability
showed the strongest genetic correlation Ž0.59.,
indicating that these two variables have a larger
proportion of overlapping genetic influences than
the other three pairs tested. As expected, the
model of direct assault and impulsivity showed
the smallest genetic correlation Ž0.26..
The amount of non-shared environmental me-
diation was also approximately equal in all four
bivariate models. That is, non-shared environ-
mental influences made similar contributions to
all four phenotypic relationships. However, it is
interesting to note that although non-shared envi-
ronmental mediation was approximately equal for
all four relationships, the non-shared environ-
ment correlation for irritability and impulsivity
was much larger Ž0.45. than for the other three
models Ž0.18, 0.29, and 0.25 for direct assault,
indirect assault, and verbal assault with impulsiv-
ity, respectively., indicating that there are more
overlapping non-shared environmental influences
accounting for the phenotypic relationship
between irritability and impulsivity than for the
other three comparisons. As also expected, direct
assault and impulsivity shared the smallest num-
ber of non-shared environmental influences.
Results also indicate that the strongest pheno-
typic correlation Ž 0.51 . is between the
Buss]Durkee subscale of irritability and the Bar-
ratt total score of impulsivity. This, coupled with
the strong genetic and environmental correlations
between irritability and impulsivity, lends support
to a distinction between physical and irritable
reactive aggression. Indeed, one major compo-
nent of the concept of reactive aggression is im-
pulsivity, and it appears that impulsivity and irri-
table aggression are more etiologically related
than impulsivity and physical aggression. This
finding is not surprising considering that this
measure of irritability has also been found to
correlate strongly with other measures of motoric
impulsivity ŽCoccaro, 1989., and some have sug-
gested that irritability may represent the trait
condition existing in the absence of aversive sti-
muli in individuals prone to aggressive behavior.
In addition, Coccaro et al. Ž1997. found that the
measures of reactive aggression used in this study
showed only moderate genetic and environmental
correlations between themselves, further support-
ing the etiological distinction between physical
aggression and other types of reactive or impul-
sive aggression.
Another reason that this phenotypic relation-
ship between impulsivity and irritability is stronger
could be due to the overlap of impulsive-type
questions in the aggression measure, which rules
out the possibility of measuring aggression in a
‘pure’ sense. Items such as ‘I am always patient
with others’ make it difficult to determine what
exactly is aggression within a subscale, and raise
larger methodological issues addressed by others
regarding the concept of comorbidity or co-occur-
rence of behaviors or traits ŽCaron and Rutter,
1991; Lilienfeld et al., 1994.. The fact that the
irritability subscale contained a greater portion of
impulsive-like items than any of the other three
subscales of aggression may explain the larger
phenotypic, genetic, and environmental correla-
tions with impulsivity. This indicates that accu-
rately determining the etiology of comorbid be-
haviors may be as much of a measurement issue
as it is a scientific question, and highlights the
 A.D. Seroczynski et al. r Psychiatry Research 86 (1999) 41]57
need to create measures of behaviors that are
both internally and externally valid ŽWaldman et
al., 1995; Bergeman and Seroczynski, 1998..
There are several implications that need to be
considered in light of these results. The question
that researchers in the field of aggression need to
be asking is, ‘‘How should we be viewing this
construct labeled ‘impulsive aggression’?’’ Al-
though some would argue that impulsive behav-
iors, such as those found in ADHD, and disorders
of conduct, such as CD and ODD or Antisocial
Personality Disorder, are often overlapping and
show similar correlates, outcomes, and treatments
ŽBarkley et al., 1989; Quay et al., 1987., others
believe that impulsivity and aggression are sepa-
rate dimensions ŽLoney et al., 1981; Loney, 1983;
Biederman et al., 1990, 1996a,b; Faraone et al.,
1995.. Our results suggest that both may be cor-
rect. That is, the prototypical form of aggression
Že.g. physical. appears to share very little etiologi-
cally with impulsivity, whereas reactive or irrita-
ble aggression shares a great deal etiologically.
This lends support to findings that indicate the
necessity of a subtype or separate diagnosis for
impulsive]aggressive individuals who happen to
inherit andror develop both characteristics.
ŽFaraone et al., 1995; Barratt et al., 1997..
Evidence for genetic relatedness between ag-
gression and impulsivity comes from research on
the possible neural networks involved in the con-
trol of impulsivity. Some researchers have con-
cluded that the limbic system structures are re-
lated to an impulsive system of risk-taking, speed
of response, time judgments, and behavioral
timing ŽBarratt and Patton, 1983.. Others have
noted that several monoamine systems, such as
serotonin, dopamine, and norepinephrine, have
been implicated in the control of behavior, possi-
bly in the limbic system structures ŽPribram and
McGuinness, 1975; Crow, 1977.. Much of the
literature attempting to identify biological struc-
tures underlying aggression has also focused on
the monoamine systems. For instance, the most
consistent biological finding associated with im-
pulsive aggression is that reduction in the activity
of the central monoamine neurotransmitter sero-
tonin Ž5-HT. increases impulsive aggression ŽCoc-
caro, 1989; Coccaro and Astill, 1990.. The biologi-
53
cal system that is identified in impulsive individu-
als could be the same one mediating the impul-
siveraggressive relationship.
Non-shared environmental influences, those
factors that make the twins different from each
other Žincluding error variance., that might be
contributing to the relationship between impulsiv-
ity and aggression have rarely been systematically
investigated. Although not explicitly tested, these
results indicate that non-shared variables, such as
a close peer relationship, a medical trauma, or a
significant life event like moving to a new neigh-
borhood or participating in a war, influence the
acquisition of both aggressive and impulsive ten-
dencies for an individual, but do not contribute to
making the twins more similar for these charac-
teristics, especially when the measure of aggres-
sion taps into physical injury to others. For
example, in adolescence, the peer group has con-
sistently been shown to be related to both aggres-
sion ŽCairns et al., 1988; Dishion et al., 1991. and
impulsivity ŽWhite et al., 1994.. Unfortunately,
the directions of those relationships are unknown.
Involvement in a delinquent peer group might
influence one’s inclination to make hasty, impul-
sive decisions. The delinquent group might also
encourage the development of an aggressive re-
sponse style; however, it is just as likely that an
adolescent comes into a delinquent peer group
already possessing a measure of each trait. Future
studies should attempt to examine whether fac-
tors such as a delinquent peer group systemati-
cally shape an individual’s impulsive and aggres-
sive behavior, or whether such behavior is already
in place and the person is simply seeking environ-
ments that match his or her genetic predisposi-
tion ŽScarr and McCartney, 1983..
4.1. Ca¨ eats to the study
As in any study, certain caveats must be con-
sidered when attempting to interpret the results.
Although precautions were taken to ensure a
conceptually and methodologically valid research
design, some aspects inherent in the design could
not be avoided. For instance, the VET Registry is,
by its nature, composed of an all male, adult
sample. This, of course, limits generalization to
54 A.D. Seroczynski et al. r Psychiatry Research 86 (1999) 41]57
adult female populations, as well as male and
female child populations. Likewise, the VET
Registry is made up of twins who were involved in
military service during the Vietnam War. Because
it could be argued that combat experience might
spuriously inflate aggression scores, Coccaro et al.
Ž1997. assessed the extent to which twin similarity
was influenced by experiences, such as engaging
enemy in firelight, being wounded or captured,
serving in a river patrol or gunboat, retrieving
dead bodies from the field, or being a tunnel rat
in enemy camps. Results of Hierarchical Multiple
Regression ŽHMR, Cohen and Cohen, 1983. indi-
cated no significant differences in twin similarity
as a result of combat experience. This, of course,
does not nullify the fact that these twins may be
more aggressive than men of this generation who
did not serve in the Vietnam War. However,
considering that only 219 of the 628 individuals
involved in this study saw direct combat experi-
ence, this explanation seems unlikely.
As with any study using self-report measures,
questions of construct validity come to mind. With
this sample, a measure of impulsivity that both
exhibited distinct characteristics of impulsivity and
was shown to be consistent with existing measures
of impulsivity was used Že.g. Eysenck and Eysenck,
1977; Plomin et al., 1988.. The measure used to
assess aggression has long-standing, well-estab-
lished construct validity ŽBuss and Durkee, 1957..
Any attempt to improve the construct validity of
this study would necessarily include multiple mea-
sures of aggression and impulsivity assessed in a
variety of methodologies Ži.e. observational mea-
sures, measures from peers, parents, teachers,
spouses.. Although more objective measures
might provide different results, genetic influences
on measures of behavior do not merely reflect
how subjective they are. Reviews of the literature
indicate that both subjective and objective mea-
sures indicate genetic influence ŽPlomin and
Bergeman, 1991..
The present study is just one piece of the
puzzle aimed at understanding the etiological re-
lationship between impulsivity and aggression.
Future work in this area should focus on the
development of measures of aggression that do
not contain impulsive types of questions, thereby
allowing researchers in this area to more uniquely
consider impulsivity separately when investigating
reactive aggression. Moreover, if non-shared envi-
ronmental influences are to be assessed specifi-
cally, then measures of the non-shared environ-
ment need to be developed to accomplish this
goal. We cannot continue to assume that environ-
mental factors affect all people equally, and must
begin to explore which environmental variables
are systematically contributing to individual dif-
ferences in impulsivity and aggression. Research
of this type is currently underway.
Acknowledgements
The first author was supported by an NIH
Training Grant ŽHD-07184.. This work was also
partially supported by NIMH Grant 1 KO2-
MH500951 ŽRSDA Level II Award. to Dr. Coc-
caro. In addition, the authors wish to ac-
knowledge the efforts of the Vietnam Era Twin
Registry of the Hines VA Cooperative Studies in
Health Services, Health Services Research and
Development Service as identified below:
Department of Veterans Affairs, Health Ser-
vices Research and Development Service: Direc-
tor, Daniel Deykin, M.D.; Cooperative Studies in
Health Services: Program Manager, Charles
Welch, III, Ph.D.; Health Services Research and
Development: Deputy Director, Shirley Meehan,
M.A., Ph.D.
Hines VA Center for Cooperative Studies in
Health Services, Vietnam Era Twin Registry: Di-
rector, Wiliam G. Henderson, Ph.D., Epidemi-
ologist, Jack Goldberg, Ph.D.; Registry Coordina-
tor, Mary Ellen Vitek; Programmer, Kenneth
Bukowski; Survey Interviewers, Mary Biondic,
Kathleen Egan, Geraldine McStay; Data Entry,
Willie Armstrong and Diane Zullo.
VET Registry Advisory Committee: W.E.
Nance, M.D., Ph.D. ŽChairman., A.G. Beam, M.D.
Žpast., G. Chase, Sc.D. Žpast., T. Colton, Sc.D.,
R.S. Paffenbarger, Jr., M.D., Dr P.H., M.M. Weiss-
man, Ph.D. and R.R. Williams, M.D.
Human Rights Committee: Ann Hendrich,
R.N., M.S., Ph.D. ŽChairperson., Ada Cole, Ph.D.,
Ronald Hahn, William Juneau, Barbra L. Dancy,
Ph.D., Terry Bering, Sr. Susan M. Sanders, Ph.D.,
 A.D. Seroczynski et al. r Psychiatry Research 86 (1999) 41]57
Rev. Thomas S. Burrs, and Thomas M. Schmidt,
Ph.D.
Most importantly, the authors wish to ac-
knowledge and thank the members the Vietnam
Era Twin Registry for their participation and
cooperation. The Registry members were true
partners in this endeavor; they willingly provided
sensitive information and considerable time in
responding to the survey questionnaire. Without
their contribution this research project would not
have been possible.
References
Akaike, H., 1987. Factor analysis and the AIC. Psychometrika
52, 327]332.
Bandura, A., 1973. Aggression: A Social Learning Analysis.
Prentice-Hall, Englewood Cliffs, N.J.
Barkley, R.A., 1990. Attention-Deficit Hyperactivity Disorder:
A Handbook for Diagnosis and Treatment. The Guildford
Press, New York.
Barkley, R.A., McMurray, M.B., Edelbrock, C.S., Robbins, K.,
1989. The response of aggressive and non-aggressive ADHD
children to two doses of methylphenidate. Journal of the
American Academy of Child and Adolescent Psychiatry 28,
873]881.
Barratt, E., 1959. Anxiety and impulsiveness related to psy-
chomotor efficiency. Perceptual and Motor Skills 9,
191]198.
Barratt, E., 1965. Factor analysis of some psychometric mea-
sures of impulsiveness and anxiety. Psychological Reports
16, 547]554.
Barratt, E.S., Stanford, M.S., Kent, T.A., Felthous, A., 1997.
Neuropsychological and cognitive psychophysiological sub-
strates of impulsive aggression. Biological Psychiatry 41,
1045]1061.
Barratt, E.S., Patton, J.H., 1983. Impulsivity: cognitive, behav-
ioral, and psychophysiological correlates. In: Zuckerman,
M. ŽEd.., Biological Bases of Sensation Seeking, Impulsiv-
ity, and Anxiety. Erlbaum, Hillsdale, N.J., pp. 77]116.
Bergeman, C.S., Seroczynski, A.D., 1998. Genetic and environ-
mental influences on aggression and impulsivity. In: Maes,
M., Coccaro, E.F. ŽEds.., Neurobiology and Clinical Views
on Aggression and Impulsivity. John Wiley and Sons, New
York, pp. 63]80.
Bergeman, C.S., Plomin, R., Pedersen, N.L., McClearn, G.E.,
1991. Genetic mediation of the relationship between social
support and psychological well-being. Psychology and Ag-
ing 6, 640]646.
Berkowitz, L., 1962. Aggression: A Social Psychological Analy-
sis. McGraw-Hill, New York.
Biederman, J., Faraone, S., Keenan, K., Knee, D., Tsuang,
M.T., 1990. Family-genetic and psychosocial risk factors in
DSM-III attention deficit disorder. Journal of the Ameri-
55
can Academy of Child and Adolescent Psychiatry 29,
526]533.
Biederman, J., Newcorn, J., Sprich, S., 1991. Comorbidity of
attention deficit hyperactivity disorder with conduct, de-
pressive, anxiety, and other disorders. American Journal of
Psychiatry 148, 564]577.
Biederman, J., Faraone, S., Milberger, S., Guite, J., Mick, E.,
Chen, L., Mennin, D., Marrs, A., Ouellette, C., Moor, P.,
Spencer, T., Norman, D., Wilens, T., Kraus, I., Perrin, J.,
1996a. A prospective 4-year follow-up of attention-deficit
hyperactivity and related disorders. Archives of General
Psychiatry 53, 437]446.
Biederman, J., Faraone, S.V., Milberger, S., Jetton, J.G., Chen,
L., Mick, E., Greene, R.W., Russell, R.L., 1996b. Is child-
hood oppositional defiant disorder a precursor to adoles-
cent conduct disorder? Findings from a four-year follow-up
study of children with ADHD. Journal of the American
Academy of Child and Adolescent Psychiatry 35, 1193]1204.
Buss, A.H., Durkee, A., 1957. An inventory for assessing
different kinds of hostility. Journal of Consulting Psy-
chology 21, 343]349.
Cairns, R.B., Cairns, B.D., Neckerman, H.J., Gest, S.D.,
Gariepy, J.L., 1988. Social networks and aggressive behav-
ior: peer support or peer rejection? Developmental Psy-
chology 24, 815]823.
Campbell, S.B., Werry, J.S., 1986. Attention deficit disorder
Žhyperactivity.. In: Quay, H.C., Werry, J.S. ŽEds.., Psy-
chopathologic Disorders of Childhood. Wiley, New York.
Caron, C., Rutter, M., 1991. Comorbidity in child psy-
chopathology: concepts, issues and research strategies.
Journal of Child Psychology and Psychiatry 32, 1063]1080.
Coccaro, E.F., 1989. Central serotonin and impulsive aggres-
sion. Symposium of the XVIth Collegium Internationale
Neuro-psychopharmacologium: serotonin in behavioural
disorders. British Journal of Psychiatry 155, 52]62.
Coccaro, E.F., Astill, J.L., 1990. Central serotonergic function
in parasuicide. Joint Annual Meeting of the British Associ-
ation for Psychopharmacology and Canadian College of
Neuro-Psychopharmacology: perspective in British and
Canadian neuro-psychopharmacology. Progress in Neuro-
Psychopharmacology and Biological Psychiatry 14, 663]674.
Coccaro, E.F., Bergeman, C.S., Kavoussi, R.J., Seroczynski,
A.D., 1997. Heritability of aggression and irritability: a twin
study of the Buss]Durkee aggression scales in adult male
subjects. Biological Psychiatry 41, 273]284.
Cohen, J., Cohen, P., 1983. Applied Multiple RegressionrCor-
relation Analysis for the Behavioral Sciences, 2nd ed. Erl-
baum, Hillsdale, N.J.
Crow, T., 1977. Neurotransmitter-related pathways: the struc-
ture and function of central monoamine neurons. In:
Davidson, A. ŽEd.., Biochemical Correlates of Brain Struc-
ture and Function. Academic Press, New York.
DiLalla, D.L., Carey, G., Gottesman, I.I., Bouchard, T.J., 1996.
Heritability of MMPI personality indicators of psy-
chopathology in twins reared apart. Journal of Abnormal
Psychology 105, 491]499.
Dishion, T.J., Patterson, G.R., Stoolmiller, M., Skinner, M.L.,
56 A.D. Seroczynski et al. r Psychiatry Research 86 (1999) 41]57
1991. Family, school, and behavioral antecedents to early
adolescent involvement with antisocial peers. Developmen-
tal Psychology 27, 172]180.
Dodge, K.A., 1991. The structure and function of reactive and
proactive aggression. In: Pepler, D.J., Rubin, K.H. ŽEds..,
The Development and Treatment of Childhood Aggression.
Erlbaum, Hillsdale, N.J.
Eaves, L., Silberg, J., Hewitt, J.K., Meyer, J., Rutter, M.,
Simonoff, E., Neale, M., Pickles, A., 1993. Genes, personal-
ity, and personality: a latent class analysis of liability to
symptoms of attention-deficit hyperactivity disorder in twins.
In: Plomin, R., McClearn, G.E. ŽEds.., Nature, Nurture and
Psychology. American Psychological Association, Washing-
ton, D.C.
Eisen, S., Neuman, R., Goldberg, J., Rice, J., True, W., 1989.
Determining zygosity in the Vietnam era twin registry: an
approach using questionnaires. Clinical Genetics 35,
423]432.
Eisen, S., True, W., Goldberg, J., Henderson, W., Robinette,
C.D., 1987. The Vietnam era twin ŽVET. registry: method
of construction. Acta Geneticae, Medicae, et Gemellolo-
giae 36, 61]66.
Eysenck, S.B.G., Eysenck, H.J., 1977. The place of impulsive-
ness in a dimensional system of personality description.
British Journal of Social and Clinical Psychology 16, 57]68.
Faraone, S.V., Biederman, J., Chen, W.J., Milberger, S., War-
burton, R., Tsuang, M.T., 1995. Genetic heterogeneity in
attention-deficit hyperactivity disorder ŽADHD.: gender,
psychiatric comorbidity, and maternal ADHD. Journal of
Abnormal Psychology 104, 334]345.
Farber, S.L., 1981. Identical Twins Reared Apart: A Reanaly-
sis. Basic Books, New York, NY.
Fulker, D.W., Baker, L.A., Bock, R.D., 1983. Estimating com-
ponents of covariance using LISREL. Data Analyst: Com-
munications in Computer Data Analysis 1, 5]8.
Ghodsian-Carpey, J., Baker, L.A., 1987. Genetic and environ-
mental influences on aggression in 4- to 7-year-old twins.
Aggressive Behavior 13, 173]186.
Goldberg, J., True, W., Eisen, S., Henderson, W., Robinette,
C.D., 1987. The Vietnam era twin ŽVET. registry: ascer-
tainment bias. Acta Geneticae, Medicae, et Gemellologiae
36, 67]78.
Gray, J.A., Owen, S., Davis, N., Tsaltas, E., 1983. Psychological
and physiological relations between anxiety and impulsivity.
In: Zuckerman, M. ŽEd.. Biological Bases of Sensation
Seeking, Impulsivity, and Anxiety. Lawrence Erlbaum As-
sociates, Hillsdale, N.J., pp. 181]217.
Henderson, W.G., Eisen, S., Goldberg, J., True, W.R., Barnes,
J.E., Vitek, M.E., 1990. The Vietnam era twin registry: a
resource for medical research. Public Health Reports 105,
368]373.
Joreskog, K.G., Sorbom, P., 1985. LISREL VI User’s Guide.
Scientific Software, Morrisville, I.N.
Kendall, P.C., Clarkin, J.F., 1992. Introduction to special sec-
tion: comorbidity and treatment implications. Journal of
Clinical and Consulting Psychology 60, 833]834.
Lilienfeld, S., Waldman, I., Israel, A., 1994. A critical exami-
nation of the use of the term and concept of comorbidity in
psychopathology research. Clinical Psychology } Science
and Practice 1, 71]83.
Loeber, R., Keenan, K., 1994. Interaction between conduct
disorder and its comorbid conditions: effects of age and
gender. Clinical Psychology Review 14, 497]1423.
Loehlin, J.C., 1987. Latent Variable Models. Erlbaum, Hills-
dale, N.J.
Loehlin, J.C., 1992. Genes and Environment in Personality
Development. Sage Publications, London.
Loehlin, J.C., 1996. The Cholesky approach: a cautionary
note. Behavior Genetics 26, 65]69.
Loney, J., 1983. Research diagnostic criteria for childhood
hyperactivity. In: Guze, S.B., Earls, F.J., Barrett, J.E. ŽEds..,
Childhood Psychopathology and Development. Raven Press,
New York, pp. 109]137.
Loney, J., Kramer, J., Milich, R.S., 1981. The hyperactive child
grows up: predictors of symptoms, delinquency and
achievement at follow-up. In: Gadow, K.D., Loney, J. ŽEds..,
Psychosocial Aspects of Drug Treatment for Hyperactivity.
Westview Press, Boulder, C.O.
Lykken, D.T., McGue, M., Tellegen, A., 1987. Recruitment
bias in twin research: the rule of two-thirds reconsidered.
Behavior Genetics 11, 343]362.
Lyons, M.J., True, W.R., Eisen, S.A., Goldberg, J., Meyer,
J.M., Faraone, S.V., Eaves, L.J., Tsuang, M.T., 1995. Dif-
ferential heritability of adult and juvenile antisocial traits.
Archives of General Psychiatry 52, 906]915.
Magnus, P., Berg, K., Nance, W.E., 1983. Predicting zygosity in
Norwegian twin pairs born 1915]1960. Clinical Genetics
24, 103]112.
McCartney, K., Bernieri, F., Harris, M.J., 1990. Growing up
and growing apart: a developmental meta-analysis of twin
studies. Psychological Bulletin 107, 226]237.
McGue, M., Bouchard, T.J., 1984. Adjustment of twin data for
the effects of age and sex. Behavior Genetics 14, 325]343.
Neale, M.C., Cardon, L.R., 1992. Methodology for Genetic
Studies of Twins and Families. Kluwer Academic Publish-
ers, The Netherlands.
Nigg, J.T., Goldsmith, H.H., 1994. Genetics of personality
disorders: perspectives from personality and psy-
chopathology research. Psychological Bulletin 115, 346]380.
Nottelmann, E.D., Jensen, P.S., 1995. Comorbidity of dis-
orders in children and adolescents: developmental perspec-
tives. In: Ollendick, T.H., Prinz, R.J. ŽEds.., Advances in
Clinical Child Psychology, vol. 17. Plenum Press, New York.
Patton, J.H., Stanford, M.S., Barratt, E.S., 1995. Factor struc-
ture of the Barratt Impulsiveness Scale. Journal of Clinical
Psychology 51, 768]774.
Pedersen, N.L., Plomin, R., McClearn, G.E., Friberg, L., 1988.
Neuroticism, extraversion, and related traits in adult twins
reared apart and reared together. Journal of Personality
and Social Psychology 55, 950]957.
Plomin, R., Bergeman, C.S., 1991. The nature of nurture:
genetic influences on ‘environmental’ measures. Behavioral
and Brain Sciences 14, 373]427.
Plomin, R., Foch, T.T., Rowe, D.C., 1981. Bobo clown aggres-
 A.D. Seroczynski et al. r Psychiatry Research 86 (1999) 41]57
sion in childhood: environment, not genes. Journal of Re-
search in Personality 15, 331]342.
Plomin, R., Pedersen, N., McClearn, G.E., Nesselroade, J.,
Bergeman, C.S., 1988. EAS temperaments during the last
half of the life span: twins reared apart and twins reared
together. Psychology and Aging 3, 43]50.
Plomin, R., DeFries, J.C., McClearn, G.E., 1990a. Behavioral
Genetics, 2nd ed. Freeman, New York.
Plomin, R., Nitz, K., Rowe, D.C., 1990b. Behavioral genetics
and aggressive behavior in childhood. In: Lewis, M., Miller,
S.M. ŽEds.., Handbook of Developmental Psychopathology.
Plenum Press, New York, pp. 119]134.
Plomin, R., Rende, R., Rutter, M., 1991. Quantitative genetics
and developmental psychopathology. In: Cicchetti, D., Toth,
S.L. ŽEds.., Internalizing and Externalizing Expressions of
Dysfunction. Erlbaum, Hillsdale, N.J.
Pribram, H., McGuinness, D., 1975. Arousal, activation, and
effort in the control of attention. Psychological Review 82,
16]149.
Quay, H.C., 1986. Conduct disorder. In: Quay, H.C., Werry,
J.S. ŽEds.., Psychopathologic Disorders of Childhood. Wi-
ley, New York.
Quay, H.C., Routh, D.K., Shapiro, S.K., 1987. Psychopathology
of childhood: from description to validation. Annual Re-
view of Psychology 38, 491]532.
Rose, R.J., 1988. Genetic and environmental variance in con-
tent in the MMPI. Journal of Personality and Social Psy-
chology 55, 302]311.
57
Rushton, J.P., Fulker, D.W., Neale, M.C., Nias, D.K.B.,
Eysenck, H.J., 1986. Altruism and aggression: the heritabil-
ity of individual differences. Journal of Personality and
Social Psychology 50, 1192]1198.
Scarr, S., McCartney, K., 1983. How people make their own
environments: a theory of genotype}environment effects.
Child Development 54, 424]435.
Scarr, S., Webber, P.L., Weinberg, R.A., Wittig, M.A., 1981.
Personality resemblance among adolescents and their par-
ents in biologically-related and adoptive families. Journal
of Personality and Social Psychology 40, 885]898.
Shapiro, D., 1965. Neurotic Styles. Basic Books, New York.
Tellegen, A., Lykken, D.T., Bouchard, T.J., Wilcox, K.J., Segal,
N.L., Rich, S., 1988. Personality similarity in twins reared
apart and together. Journal of Personality and Social Psy-
chology 54, 1031]1039.
Vandenberg, S.G., 1962. The heredity abilities study: heredity
components in a psychological test battery. American Jour-
nal of Human Genetics 14, 220]237.
Waldman, I.D., Lilienfeld, S.O., Lahey, B.B., 1995. Toward
construct validity in the childhood disruptive behavior dis-
orders. In: Ollendick, T.H., Prinz, R.J. ŽEds.., Advances in
Clinical Child Psychology, vol. 17. Plenum Press, New York.
White, J.L., Moffitt, T.E., Caspi, A., Bartusch, D.J., Needles,
D.J., Stouthamer-Loeber, M., 1994. Measuring impulsivity
and examining its relationship to delinquency. Journal of
Abnormal Psychology 103, 192]205.