Family Discord and Conduct Disorder

Journal of Family Psychology
1994, Vol. 8. No. 2, 170-186
Copyright 1994 by the American Psychological Association, Inc.
0893-3200/94/$3.00
Family Discord and Conduct Disorder:
Cause, Consequence, or Correlate?
Michael Rutter
The inferences to be drawn from statistical associations between family discord and conduct
disorder in children are discussed with respect to the need to differentiate between risk
indicators and risk mechanisms, the conceptualization of risk mechanisms, measurement
issues, and the research strategies needed to test causal hypotheses. Such testing needs to
indicate genetic research strategies, as well as to focus on children's effects on parents,
person-environment interactions, nonshared environmental effects, causal chain effects,
and the need to use natural experiments.
The background to the current interest in the
effects on children of marital conflict derives
from the old observation that "broken homes"
are associated with delinquency. Following
Wootton's (1959) scathing critique of this field
of research, studies in the 1960s and 1970s
showed that the main risks accompanied divorce or separation, rather than parental death,
and it was suggested that family conflict, not
separation as such, might constitute the key risk
factor (Rutter, 1971). The finding that conflict
and discord in intact families also carried a
raised risk of psychopathology was consistent
with the hypothesis, as were findings in the
1980s that the psychological risks associated
with divorce preceded family breakup (Block,
Block, & Gjerde, 1986; Cherlin et al., 1991).
Subsequent evidence that continuing conflict
between parents postdivorce was also associated with psychopathology in children if they
felt caught in the conflict also suggests that the
risk mechanisms involve conflict in some way
(Buchanan, Maccoby, & Dornbusch, 1991;
Hetherington, 1989, 1991).
Clearly, there has been much progress since
those early contradictory findings on "broken
homes." Even so, there are many unresolved
issues, and numerous key questions remain unanswered, as is well brought out by the four
thoughtful articles in this special section. Fincham, Grych, and Osborne (1994) focus on two
Correspondence concerning this article should be
addressed to Michael Rutter, Department of Child
and Adolescent Psychiatry, Institute of Psychiatry,
DeCrespigny Park, Denmark Hill, London SE5 8AF
England.
main issues. First, they highlight the crucial
point that the concept of conflict is a very broad
one and, hence, that it is necessary to determine
which aspect of interparental conflict constitutes the risk factor. Does the risk lie in the
model of aggression and hostility provided by
the parents or, instead, is it that conflict indexes
a risk actually brought about by ineffective or
maladaptive disciplinary practices, by aggression directed against the child, or by a poor
parent-child relationship? Second, they note
the importance of researchers moving from the
finding of a statistical association between conflict and maladjustment to the rigorous testing
of the hypothesis that this association represents
a causal mechanism by which conflict predisposes to maladjustment. They emphasize that
any testing of the causal hypothesis demands a
clear conceptualization of postulated risk processes, and they draw attention to the important
point that the factors involved in the origins of
a disorder are not necessarily the same as those
that influence its course thereafter. They draw
attention to the need to consider causal chains
over time, to consider the possibility of threshold effects, and to examine the role of individual differences in response to conflict together
with the mechanisms involved in such differences. The situation that they rightly portray is
undoubtedly a complex one, but this does not
mean that it is not susceptible to rigorous empirical examination and they present a variety of
practical ways of moving forward.
Kitzmann and Emery (1994) report the findings of a study designed to determine whether
mediation to settle child custody disputes following divorce reduces problem behavior in
170
SPECIAL SECTION: FAMILY DISCORD AND CONFLICT DISORDER
children. The question is of interest in its own
right but, in addition, as the authors point out,
the intervention strategy also provides an important means of testing the hypothesized
causal relationship between parental conflict
and child disorder. Insofar as there was no difference in outcome 1 year later between the
mediation and litigation groups, the findings are
noncontributory to the causal hypothesis testing. However, the finding that a decrease in
conflict over that year was associated with a
lower rate of behavioral problems in the children provides some indirect support for the
causal hypothesis. Intervention strategies potentially provide an important experimental means
of putting causal hypotheses to the test, but their
success is crucially dependent on the therapeutic intervention actually having the effects that
are intended. As numerous studies have shown,
treatment research is a difficult enterprise for a
host of practical reasons. There may well be
other benefits of mediation over custody disputes, but the results of the study undertaken by
Kitzmann and Emery suggest that something
more than (or different from) brief psychological interventions are needed to reduce parental
conflict. If an effective intervention can be developed, this would provide a most useful experimental testing of the notion that interparental conflict plays a significant role in the
causation of childhood psychopathology, but
that day has not yet arrived.
The other two articles in the special section
focus on the question of the mechanisms that
might be involved in individual differences in
children's responses to marital conflict. Both
also focus on children's immediate reactions to
conflict, leaving open the question of the extent
to which such short-term responses reflect more
enduring features. Cummings, Davies, and
Simpson (1994) focus on children's perception
of marital conflict and on their styles of coping
with it. The findings are most striking and informative in the light that they throw on gender
differences in response to marital conflict.
Boys' perceptions of marital discord tended to
agree with that of their mothers', whereas girls'
perceptions did not; boys who coped well
tended to be better adjusted, whereas this association did not apply in girls; and boys' perception of threat in the marital discord was associated with their own externalizing problems,
whereas girls' perception of self-blame was associated with their own internalizing problems.
171
As the authors bring out, further research is
needed to determine whether these gender differences mean that boys are more sensitive to
marital conflict than are girls or that boys tend
to be drawn into the conflict to a greater extent
than are girls, or whether boys' tendency to
respond to conflict with oppositional behavior
means that coercive interchanges between parent and child are more likely to develop and
persist.
Easterbrooks, Cummings, and Emde (1994)
observed children's responses during marital
interaction in which a revealed differences approach was used to bring out conflict. Some two
dozen families were observed when their infants
were 12, 18, 24, and 36 months of age. The
main findings were that, although there was
stability over time in the style of the couples'
interaction, there was not stability in the patterns of child behavior, that the relationship
between the two varied across time periods, and
that the children showed a complex mixture
of positive and interfering behaviors during
the marital interactions. Within this sample of
generally well-functioning middle-class families with very young children, it was clear
that parental discussions of disagreements between them could be constructive interactions.
Whether it is appropriate to label these conflict
may be a matter that is open to question, but it
is clear that it would be an error to equate
constructive discussions designed to resolve
disputes with hostile discordant relationships.
Previous research has indicated that the latter
constitutes risk factors for psychopathology in
the children, whereas the former may not.
This set of four articles succeeds admirably in
opening up the topic of the possible psychopathological risks for children resulting from
their exposure to marital conflict. In seeking to
develop a better understanding of the risk processes that may be involved, research designs
will need to pay attention to the conceptualization of risk mechanisms, measurement issues,
and the devising of effective means of putting
causal hypotheses to the test. These three key
issues constitute the focus of this article. The
discussion of these issues concentrates on the
association between family discord and conduct disorder in children, but closely similar
considerations apply to other forms of psychopathology.
172
MICHAEL RUTTER
The Distinction Between Risk Indicators
and Risk Mechanisms
Until relatively recently, most research into
the psychological risks associated with psychosocial features has focused mainly on the identification of risk indicators. There is, as a result,
a large literature documenting the huge range of
psychosocial variables that are statistically associated with an increased risk of conduct disorder (Farrington, 1992, in press; Patterson,
1982; Robins, 1991; Rutter & Giller, 1983).
However, researchers know much less about
what is involved in the psychosocial risk mechanisms underlying this association. This lack of
information is very crucial, because in the absence of an adequate understanding of the mediating mechanisms, researchers are in severe
danger of drawing the wrong conclusions and,
thereby, advocating the wrong policies and
practices (as well as, of course, making the
wrong inferences about processes of psychological development).
For example, in the 1950s sweeping recommendations were made that mothers should not
go out to work and that group daycare should be
abolished (Baers, 1954; World Health Organization Expert Committee on Mental Health,
1951). Subsequent systematic research has
made it clear that neither maternal employment
nor group care per se constitutes a major risk
factor for children. There may be slight risks
associated with both in certain circumstances,
but, if so, they seem to derive from the particular features of the experiences involved rather
than from maternal employment or daycare
(Belsky, 1988; Gottfried & Gottfried, 1988;
Heyns & Catsambis, 1986; McCartney & Galanopoulos, 1988; Milne, Myers, Rosenthal, &
Ginsburg, 1986; Terr, 1989; Zigler & Gordon,
1982). In more recent times, the confusion of
risk indicators with risk mechanisms may be
seen in the expectation that combating poverty
will prevent mental disorder. There is good evidence that poverty makes good family functioning more difficult, and hence it plays a role
in the causal chain leading to conduct disorder.
But economic hardship as such does not seem to
constitute the direct risk mechanism for psychopathology in children (Conger et al., 1992,
1993; Lempers, Clark-Lempers, & Simons,
1989). Thus it is striking that in the several
decades following the Second World War, eco-
nomic conditions in most western countries improved markedly, but, nevertheless, this improvement was accompanied by a rise in the
rate of juvenile delinquency and other forms of
psychosocial disorder in young people (Rutter
& Smith, in press).
It is important to emphasize that even quite
strong associations do not imply that a risk
factor measures a risk mechanism. For example,
one way to pick out from the general population
those who are most likely to be exhibiting antisocial behavior would be to focus on Black
male teenagers (Smith, in press). But what risk
mechanisms are involved? What is it about
maleness that is associated with antisocial behavior: the Y chromosome, prenatal hormonal
effects on the developing brain, testosterone
postpuberty, or societies' reactions to and expectations of men? Black skin per se obviously
does not constitute a risk mechanism in that
within the United Kingdom, crime rates in
Asians (who are equally subject to racial discrimination) are below those of Whites. Also, in
the rather different situation of the United
States, Blackness equally does not constitute an
explanation, let alone a mechanism (Wilson,
1987). Also, what is it about the teenage period
that constitutes the risk factor? Age as such
provides no explanation (Rutter, 1989). Risk is
not likely to be caused by puberty, because the
precursors of crime in the form of antisocial
behavior go back to the childhood years (Moffitt, 1993a; Patterson & Bank, 1989; Patterson,
Capaldi, & Bank, 1991; Robins, 1991; White,
Moffitt, Earls, Robins, & Silva, 1990). Also,
during the time period when the peak age of
crime in male adolescents has gone up (Farrington, 1986), the average age of puberty has
been going down (Tanner, 1989).
The dangers of neglecting the importance of
differentiating risk indicators and risk mechanisms are obvious and need no further discussion. However, part of the reason for that neglect stems from an overready acceptance of
the scientific defeatism that seems to be implicit in the teaching that correlations do not
mean causation. Of course, the warning that
causation cannot be inferred from even the
strongest statistical correlations is appropriate
and certainly needs heeding. Nevertheless,
there are many ways in which nonexperimental data can be used to test causal hypotheses
(see Rutter, 1981, in press), and it is crucial
that these methods be used effectively. Before
SPECIAL SECTION: FAMILY DISCORD AND CONFLICT DISORDER
these methods are considered in greater detail,
however, it is necessary to clarify the causal
hypotheses regarding family discord that require testing.
Conceptualization of Risk Mechanisms
Associated With Family Discord
As Fincham et al. (1994) emphasize, any systematic testing of the hypothesis that family
discord or marital conflict causes conduct disorders in children requires further specification
of the possible mechanisms postulated to be
involved in the conflict or discord. It may be
helpful, perhaps, to break down the issue into
three interlinked questions: Wherein lies the
risk (i.e., which aspect of conflict or discord
carries the risk)? How does the risk mechanism
operate? and How are the effects carried forward in time?
Several possibilities need to be considered in
relation to the first question. The risk could
derive from children witnessing strongly negative interchanges between their parents. If this
does constitute the risk variable, it would be
necessary to go on to consider whether this
factor applies across the range (e.g., whether
there is a greater risk associated with occasional
disputes between parents than no disputes at all)
or, rather, whether the risk stems only from
frequent disputes that involve the expression of
dislike, hostility, or aggression. Alternatively,
the risk might apply only when the child is
drawn into the marital discord or when the
discord is between parent and child. It could
also be that it is not the presence of quarreling
per se that matters, but rather a lack of a supportive, affectionate relationship between the
parents or between the parent and child. Yet
another alternative could be that it is not the
overall level of discord or lack of loving relationships in the home that matters, but rather
that either of these tends to lead to differential
treatment of the children. That is, when family
relationships are strained, parents tend to express their negative feelings more on one child
in the family than on others (Rutter, 1978). It
could be that it is the scapegoating of children
or the presence of a small, consistent, but persistent tendency to favor one child over another
that does the damage (Dunn & Plomin, 1990).
Alternatively, the risk might be mediated by
173
maladaptive parenting practices that are fostered by family discord rather than by marital
conflict (Fauber & Long, 1991).
These distinctions are far more than mere
academic niceties. There are hugely important
implications for policy in practice, as well as for
theories concerning the processes involved in
the development of psychopathology. Two examples may serve to illustrate the matter. For
many years there was vigorous dispute in the
literature about whether parental loss in childhood predisposed to depression in adult life
(Brown & Harris, 1978; Crook & Eliot, 1980;
Tennant, Bebbington, & Hurry, 1980). On the
one hand, numerous studies had shown statistical associations, but, on the other hand, the
associations tended to be quite weak and there
were many conflicting findings in the literature.
Two sets of findings have done much to clarify
the issue. First, Harris, Brown, and Bifulco
(1986) showed that parental loss was not associated with an increased risk of depression unless poor parenting followed the loss; equally,
poor parenting was associated with an increased
risk of depression even in the absence of loss.
The implication is that loss is important only
because it tends, in some instances, to predispose to poor parenting. Loss itself is not, however, the risk mechanism. Also, Kendler, Neale,
Kessler, Heath, and Eaves (1992) showed in a
genetically informative twin design that parental loss due to divorce or separation was associated with a substantial increase in the risk for
emotional disorders in adult life, whereas parental death was not to any substantial extent.
The implication here is that some aspect of
discordant relationships is likely to carry the
risk rather than loss as such. The second example is provided by Fergusson, Horwood, and
Lynskey's (1992) analyses of the Christchurch
longitudinal study. They showed that although
both multiple family changes and family discord were associated with psychopathological
risks for children, discord remained as a substantial risk factor in the absence of family
changes whereas family changes constituted a
risk only when they reflected discord. When
discord was held constant, family changes did
not add to the risk. The implication is that the
risk mechanism is likely to involve something
to do with discord rather than with changes in
family composition. These findings indicate
that progress can be made in differentiating one
postulated risk mechanism from another, pro-
174
MICHAEL RUTTER
vided the right measurements and comparisons
are made. The same need applies to the various
alternatives suggested with respect to family
discord and marital conflict.
At first, it might be supposed that the second
question about risk mechanisms (How do they
operate?) is much the same as the first question,
but it is not. To begin with, human beings are
not passive recipients of their experiences.
Rather, they interact in an active fashion with
their environment, and they actively process
such experiences. It is for that reason that
Kagan (1984a, 1984b) suggested that the sequelae of adverse experiences in infancy tended
to be rather less than those associated with the
same experiences in middle childhood, because
the cognitive processes possible in the infancy
period are so. much more limited. Numerous
studies have shown that there is only a very
modest level of agreement between experiences
as objectively measured, or as perceived by
parents, and the same experiences as perceived
by children. Thus, Cummings et al. (1994)
found that the perceptions of boys were much
more consonant with those of their mothers than
was the case with girls. Also, it was boys'
perception of the threat involved in marital discord that was associated with both internalizing
and externalizing problems, whereas in girls it
was self-blame that was associated with internalizing problems and no associations were
found with externalizing problems. Similarly,
Buchanan et al. (1991) found only a modest
level of agreement between parents' report of
discord and their children's expressed feelings
of being caught between parents. The latter was
related to depression/anxiety and deviant behavior in the offspring, but discord was so related only indirectly through the adolescents'
feelings of being caught.
A second issue with respect to the operation
of risk mechanisms derives from the findings on
the heterogeneity of conduct disorder. Many
studies have shown that persistent conduct disorders and recidivist delinquency are associated
with a particularly early onset that is associated
with hyperactivity, inattention, impulsivity, and
aggression. These features are much less apparent with respect to delinquency or conduct disorders that develop in adolescence and which
are more likely to be transient. Moffitt (1993a)
suggested that this distinction within the overall
group of conduct disorders may well prove to be
a fundamental one with rather different causal
pathways involved in the two types. The suggestion has yet to be put to the test in a systematic fashion, but it is consistent with the available evidence and implies that the mechanisms
by which family discord operates as a risk factor
may not be the same in these two circumstances. A further issue that derives from the
same set of findings concerns the question of
the extent to which family discord and conflict
predispose to the occurrence of this composite
early pattern of disorder involving a combination of multiple features or whether, instead,
family discord and conflict only operate as a
mechanism for translating the vulnerability associated with early hyperactivity, inattention,
and impulsivity into disorder. That is to say,
does family discord predispose to disorder in
the absence of these individual characteristics
(Hinshaw, Lahey, & Hart, 1993; Moffitt, 1990,
1993b)?
It is necessary to go on to ask as well how the
experience of family discord produces conduct
disorder in the child. Several rather different
possibilities may be suggested. For example, it
could be that there is a direct modeling of behavior. Thus, by observing their parents deal
with interpersonal difficulties by aggressive
hostile behavior, children may come to do the
same. A different possibility is that the experience of family discord leads to disorder because
of its effect on social information processing.
Numerous studies have shown that children
who display hostile attributional biases are
likely to behave aggressively (Dodge, 1980;
Quiggle, Garber, Panak, & Dodge, 1992). The
possibility arises, therefore, that this form of
maladaptive social information processing
might constitute the mediating variable. Dodge
and his associates (Dodge, Bates, & Pettit,
1990; Weiss, Dodge, Bates, & Pettit, 1992)
have examined this possibility with respect to
young children's experience of early harsh discipline. Their results suggested that there was
both a direct effect of the harsh discipline on
aggression in the home and an indirect effect on
aggression at school through a hostile attributional bias.
Another suggestion is that family discord predisposes to later conduct disorder because it
impairs the security of children's attachment
relationships (Greenberg, Speltz, & DeKlyen,
1993; Waters, Posada, Crowell, & Lay, 1993).
The suggestion here is that the risk mechanism
lies in the damaged potential for later relation-
SPECIAL SECTION: FAMILY DISCORD AND CONFLICT DISORDER
ships and implies that the connections between
damaged early attachment relationships and
poor peer relationships in adolescence (Hodges
& Tizard, 1989) may be important. Numerous
studies have shown the associations between
poor peer relations and various forms of later
psychopathology (Parker & Asher, 1987), and
peer rejection seems to make it more likely that
youngsters will become involved in antisocial
peer groups (Dishion, Patterson, Stoolmiller, &
Skinner, 1991; Patterson & Bank, 1989; Patterson et al., 1991). Yet another alternative is that
because parents have dealt with difficulties by
quarreling and shouting, the children fail to
learn effective coping or social problem-solving
strategies (see Cummings et al., 1994).
The third aspect of the conceptualization of
risk mechanisms concerns the question of how
the effects of family discord are carried forward
in time. This is an important issue because there
is abundant evidence that conduct disorders
tend to be strongly persistent, not only through
childhood, but with a substantial proportion
continuing into adult life in the form of personality disorders (Loeber, 1982; Olweus, 1979;
Robins, 1978, 1991). The question is why they
are so persistent. Is it because, once the disorder
becomes established, it becomes self-perpetuating in some fashion (and if so, how), or is it that
the disorder is maintained over time through
environmental circumstances of one sort or another? In a follow-up into adult life of children
reared by mentally ill parents, Quinton, Rutter,
and Gulliver (1990) showed that persistent parental discord was quite strongly associated
with the persistence of psychopathology from
childhood into adult life. Further analyses on
the different links in the causal chain over time
(based on two other long-term follow-ups from
childhood into adult life) showed that the persistence of conduct disorder was quite strongly
dependent upon involvement in an antisocial
peer group and, later on, by a discordant marriage to an unsupportive deviant spouse (Quinton, Pickles, Maughan, & Rutter, 1993). Family
relationships continued to play a part in these
later links in the chain, as well as in the earlier
ones. Thus, young people from disharmonious
families were more likely to move into deviant
peer groups, and a supportive harmonious family provided some protection against marriage
to a deviant spouse.
175
Measurement Issues
The majority of studies of family discord and
conduct disorder have relied on questionnaire
measures of both, and most have dealt with
associations in terms of single measures. Some
studies have sought to obtain questionnaire assessments of different aspects of the family.
Thus, Kitzmann and Emery (1994) used a measure of parent-parent relationships as well as a
measure of parent conflict, and Cummings et al.
(1994) used both a measure of marital discord
and of marital conflict. Provided the appropriate
multivariate analyses are undertaken, the use of
questionnaire measures to tap different aspects
of family functioning can be informative. Thus,
Jouriles, Murphy, et al. (1991) found that boys'
behavior problems were more strongly associated with child rearing disagreements between
the parents than with a measure of global marital adjustment (there was no measure of family
discord or marital conflict). Nevertheless, the
reliance on single questionnaire measures does
not seem optimal. Not only is there substantial
error variance in any one questionnaire score
but also there will be systematic variance
unique to the perspective of the rater. For some
purposes, of course, it may be important to tap
individual perspectives, but unless measures are
combined in some way, it will not be possible to
sort out the aspects of the score that reflect
individual perspectives as distinct from the
broader construct that the measure also taps.
Schwarz, Barton-Henry, and Pruzinsky (1985)
have shown the advantages of aggregating
scores across raters. Aggregated scores are not
anything special to family measures but rather
are a general reflection of the psychometric
properties of instruments (Epstein, 1983). Also,
although simple aggregated scores improve reliability, they do not deal with error variance.
For this purpose, a latent variable approach has
many advantages (Cook & Goldstein, 1993).
This statistical approach may be problematic
with small samples, but with larger samples, it
is likely to be the method of choice.
In addition, it is very dubious whether it is
satisfactory to rely on questionnaire measures
when it is necessary to make distinctions between different aspects of family interaction, as
is certainly required in tackling the questions on
risk mechanisms discussed above (Fincham &
Bradbury, 1987; Kurdek, 1992; Sabatelli,
1988). Questionnaires are reasonably good at
176
MICHAEL RUTTER
differentiating good and poor marriages but are
much less satisfactory for separating, say, marital satisfaction and marital conflict. In order to
assess different aspects of family interaction,
there is much to be said for the use of more
discriminating standardized interview measures
(Brown & Rutter, 1966; Jacob & Tennenbaum,
1988; Rutter & Brown, 1966), observational
measures (Reid, Baldwin, Patterson, & Dishion,
1988), or both.
The same issues arise with respect to the
assessment of child psychopathology. Questionnaires are reasonably satisfactory as a means of
detecting emotional-behavioral disturbance,
but they are not very good at making distinctions between different types of psychopathology. Thus, neither questionnaire measures of
depression nor questionnaire measures of anxiety have acceptable discriminative diagnostic
validity in relation to clinical concepts and measures of disorder (Fristad, Weller, Weller,
Teare, & Preskorn, 1991; Perrin & Last, 1992;
Roberts, Lewinsohn, & Seeley, 1991). There is
much to be said, as with measures of family
interaction and relationships, for the use of standardized interview and observation measures,
and not just questionnaires. Also, there is the
same need to use multiple measures from multiple informants (Epstein, 1983). In addition, for
many purposes, it will also be advantageous to
combine data across different points in time
(Rutter, 1977). Although this combining of data
means a loss of the distinction between influences on onset and influences on persistence of
disorder, it avoids the problem of the high proportion of children in the general population
who show transient disorders or disturbances
from time to time. Thus, Rutter and Quinton
(1984) found that the associations between parental mental disorder and psychopathology in
the children (mainly mediated through family
discord) were much more strongly evident in
relation to persistent psychopathology than with
measures of disorder in the child at any one
point in time. Again, too, as with the assessment
of family features, latent construct methods
based on structural equation modeling constitute the best means of taking into account both
systematic situational or rater variance and error
variance (Fergusson & Horwood, 1988; Rutter
& Pickles, 1990). Not only do latent construct
approaches provide a better differentiation between different types of psychopathology, but
also they bring out more clearly the associations
with risk factors of various kinds (Fergusson &
Horwood, 1987, 1989; Fergusson, Horwood, &
Lloyd, 1991).
Testing Causal Hypotheses With Respect
to Family Discord
The need for rigorous testing of causal hypotheses, together with careful consideration of
complicating assumptions, is well discussed by
Fincham et al. (1994). However, although the
testing of causal hypotheses is more difficult
outside the experimental laboratory, the combination of epidemiological and longitudinal data
can carry the testing a long way, provided that
the most appropriate research designs are used.
Such designs seek to find naturally occurring
situations that approximate to the controlled
experimental situation in the laboratory in terms
of separating variables or mechanisms that ordinarily go together (Rutter, in press; Susser,
1991). Fincham et al. concentrate on the need to
test competing hypotheses on particular environmental risk mechanisms. However, a prior
concern is to determine whether there is any
environmental risk effect of family discord on
child psychopathology, regardless of the particular environmental mechanism that may be involved. Two chief alternatives need to be considered and tested for: genetic mediation and the
reverse causal direction (i.e., an effect of children on parents).
Genetic Research Strategies
The need to take seriously the possibility that
the effects of family discord on conduct disorder in children might be mediated genetically is
emphasized by the findings from longitudinal
studies that conduct disturbance in childhood is
strongly associated with a much increased rate
of stress experiences in adult life (Champion,
Goodall, & Rutter, 1994) and especially of marital discord (Quinton et al., 1993). The consistent finding that conduct disorder in childhood
is also associated with parental psychopathology and criminality (Farrington, in press; Robins, 1991) underlines the possibility of genetic
mediation. The evidence in the Frick et al.
(1992) study that antisocial personality disorder
accounts for much of the association between
poor parenting and conduct disorder points in
the same direction. Nevertheless, there are rea-
SPECIAL SECTION: FAMILY DISCORD AND CONFLICT DISORDER
sons for thinking that genetic factors are unlikely to explain the whole of the effect. Thus,
the genetic contribution to conduct disorder in
childhood seems modest, although it may be
somewhat stronger for that which persists into
adult life (DiLalla & Gottesman, 1989). Also,
there is evidence, as already noted, that the
persistence of childhood conduct disorder into
adult personality disorder is dependent upon
various intervening experiences (Quinton et al.,
1993). Nevertheless, the matter needs to be settled, not by assumptions, but by direct testing.
There are many ways in which this may be
done, but four may be mentioned. All require
the inclusion of specific measures of the postulated environmental risk factor. First, twin designs may be used to examine the effects of a
specific environmental risk factor, after taking
account of genetic influences. This was the approach followed, for example, by Kendler et al.
(1992) in their examination of the role of parental loss as a risk factor for depressive disorders in adult life. Second, adoptee designs may
similarly be used to separate the effects of nature and nurture. Thus, Cadoret, Troughton,
Merchant, and Whitters (1990) showed that although there was indeed genetic mediation of
the association between antisocial problems in
parent and offspring, there was also an environmental effect by which parental antisocial disorder increased the risk of depressive symptomatology in children. It should be noted that a
complete design is not essential for the use of
adoptee samples to study environmental effects.
Much can be learned from the study of associations between family features and child outcomes within adoptee samples (e.g., see Tienari
et al., 1990) without parallel data on the biological families, provided that selective placement
biases can be excluded. Third, environmental
effects may be examined by a comparison of
their operation in the presence and in the absence of particular psychiatric disorders in the
parent. This design does not provide the same
control of genetic factors, but it may be helpful
in indicating whether the genetic possibility is
strong or weak. Thus, Rutter and Quinton
(1984) showed that the effects of family discord
on conduct disorder in sons still operated after
taking account of personality disorder in the
parents. Fourth, nonfamily situations may be
used as analogues. Thus, Rutter, Maughan,
Mortimore, and Ouston (1979) examined school
characteristics in relation to pupil behavior. One
177
of the findings was that the frequent use of
unofficial physical punishments, such as slapping or cuffing the children, was associated
with greater behavioral disturbance in the pupils. Similarly, Sinclair (1971) found that boys'
success in probation hostels was greatest in
those with warm but strict wardens whose wives
agreed with them on how the hostel should be
run. Of course, neither of these studies provides
an exact parallel with marital discord, but the
finding that discordant relationships within nonfamily environments is associated with disturbed behavior makes the environmental mediation hypothesis with respect to marital
discord more plausible.
Children's Effects on Parents
The possibility that the association between
family discord and conduct disorder represents
an effect of children's behavior on parents also
needs to be taken seriously. There is quite a lot
of evidence that children with conduct disorder
differ from others in their neuropsychological
characteristics (Moffitt, 1993b), in their temperamental features (Maziade, 1989), and in their
autonomic responsivity (Lahey, McBurnett,
Loeber, & Hart, in press). Also, conduct disorders, particularly those that persist, are frequently first manifest in the preschool years
(White et al., 1990). All of these findings may
apply more with respect to hyperactivity than to
conduct disorder in the absence of overactive
behavior, but, nevertheless, they do provide the
need to examine possible child effects.
Bell and Chapman (1986) summarized some
of the key ways in which child effects may be
examined, and Lytton (1990) summarized some
of the key findings with respect to boys' conduct disorder. Five main research strategies
have been used. First, there is the confederate
participant design. For example, Brunk and
Henggeler (1984) trained 10-year-old child confederates to exhibit oppositional or socially
withdrawn behavior while playing a board
game with a range of different mothers. The
findings showed that mothers behaved quite
differently in these conditions and, even more
surprisingly, that they did so even in response to
neutral behaviors. Over the course of just a few
minutes, the adults tended to react in terms of
their anticipations of the meaning of child behaviors. Second, there is the altered behavior
178
MICHAEL RUTTER
design. For example, stimulant medication has
been used to alter the behavior of children with
hyperkinetic problems. Both Barkley and Cunningham (1979) and Schachar, Taylor, Wieselberg, Thorley, and Rutter (1987) found that
the mothers of drug responders altered their
style of interaction with their children, although
there was not an effect on marital conflict.
The third design of altered perception is characterized by the "baby X" experiments in which
toddlers are dressed in unisex snowsuits and
given names that indicate gender, with the
names only coinciding with their own gender
half the time. This strategy is a useful one but is
not particularly applicable in the case of conduct disorder. The fourth strategy of sample
selection is exemplified by a design that was
probably first used by Gardner (1977) with autistic children and later used by Anderson,
Lytton, and Romney (1986) with conductdisordered boys. Mothers of normal and of conduct-disordered children were observed interacting with their own child, with someone else's
normal child, and with someone else's conductdisordered child. By these means it was possible
to separate the effects of the mothers' style of
the child being their own or someone else's and
of the child being normal or conduct disordered.
A substantial effect of the child's behavior on
maternal behavior was found. Finally, longitudinal investigations provide a less rigorous testing of the direction of causal relations but have
the advantage of studying effects over time in
real-life settings. Thus, Martin (1981) in his
longitudinal study of children from 10 to 42
months showed that demanding behavior in infancy was followed by a tendency for mothers
to back away, which in turn was associated with
an increased tendency for boys' noncompliant
behavior to persist. Two-way, coercive cycle
effects were also found in Patterson's observational studies of conduct-disordered boys and
their parents (Patterson, 1982; Patterson &
Bank, 1989; Patterson et al., 1991).
There seems little doubt that there are true
child effects on parents, but it is much more
questionable whether this accounts for the association between marital conflict or family discord and conduct disorder in offspring. To begin with, most of the demonstrations of child
effects concern the immediate situation in relation to parental responses and not the longer
term effects on parental behavior or family relationships. Moreover, as already noted, Scha-
char et al. (1987) found no effects of improved
child behavior on family discord, in spite of
beneficial effects on parent-child interaction.
Also, in at least some instances, it is clear from
longitudinal data that the family discord must
have preceded the children's conduct disorder
(Rutter & Quinton, 1984). Bidirectional influences seem likely (Patterson & Bank, 1989), but
further research is required to determine the
relative strength of effects in each direction.
Treatment strategies could be informative, but
only if the design was such as to examine the
effects on child behavior of therapeutically induced reductions in family discord. Unfortunately, although there is substantial evidence of
the efficacy of behaviorally oriented treatments
designed to improve parental management techniques, the data are not such as to indicate what
change in the family or in parenting has brought
about changes in the children's behavior
(Kazdin, 1993; but see also Patterson & Bank,
1989). Furthermore, the main focus of treatments has tended to be on parental management
techniques rather than on marital relationships.
Person-Environment Interactions
Although person-environment interactions
are widespread in biology and medicine (Rutter
& Pickles, 1991), they have received relatively
little attention up to now in studies examining
the effects of family discord. Nevertheless,
there are various pointers suggesting that they
may be important. Several studies have suggested that the effects of family disruption, and
associated stressors, may be greater in genetically vulnerable children (Cadoret, 1985). Also,
it is clear from studies of children with a hyperkinetic disorder (generally thought to have a
strong constitutional component) that they are
very vulnerable to the adverse effects of family
discord and disruption (Moffitt, 1990; Taylor,
Sandberg, Thorley, & Giles, 1991). Vorria
(1991) found that children reared in institutions
in Greece because they were orphaned or because of parental poverty did not show the
marked behavioral disturbance that was characteristic of a high proportion of children in the
same institutions because of a breakdown in
parenting stemming from family discord and
disruption. It is possible, too, that the findings
on multiple stressors reflect the same phenomenon. Thus, several studies (e.g., Kolvin, Miller,
SPECIAL SECTION: FAMILY DISCORD AND CONFLICT DISORDER
Scott, Gatzanis, & Fleeting, 1990; Rutter, 1979)
have shown that a single family stressor (including discord) does little to raise the risk of disorder in children but that two or more stressors
substantially increase the risk. This relationship
has usually been interpreted as an interactive
effect between environmental risk factors, but it
could equally well mean that environmental effects tend to be greater when the children are
genetically vulnerable. The reason for laboring
this point is that the possibility of person-environment interactions has been so little investigated up to now. If the causal hypothesis is to be
tested most effectively, it will be necessary to
use strategies that examine such effects. Comparisons of children who are high and low in
genetic vulnerability and comparisons of children who differ in terms of personal risk factors—such as hyperactivity, neuropsychological
deficits, and autonomic underresponsivity—
would be particularly informative.
Nonshared Environmental Effects
Plomin and Daniels (1987), on the basis of
behavior-genetic evidence, argued that, for
most outcomes, family influences that impinge
differentially on children within the same family tend to make a greater impact than family
influences that impinge on the family as a whole
to a roughly equal degree. Conduct disorders
appear to constitute somewhat of an exception
to this general rule. It is frequent for several
children in the same family to show conduct
disorder, and the limited genetic data available
so far also point to the operation of shared, as
well as nonshared, environmental effects (DiLalla & Gottesman, 1989). Even so, there is a
need for research into family discord to obtain
measures that explicitly assess features that may
be relevant for nonshared effects. Thus, it would
be helpful to quantify the extent to which each
child is embroiled in the family conflict and
perceives himself or herself as caught up in the
discord and to assess the extent to which the
family conflict leads to differential treatment of
children or even to scapegoating. Jenkins and
Smith (1990) found that a good relationship
with an adult inside or outside the family was
associated with less psychopathology when
children were exposed to marital discord. This
finding has been interpreted as a protective effect—and probably does represent this with re-
179
spect to relationships outside the family—but it
may well imply that the main risk to children
derives from poor parent-child relationships
rather than from discord per se. Thus, there was
very little effect of a poor parental marriage
when the child's relationship with the mother
was a good one. The implication could be that it
is a discordant relationship between parent and
child that matters (i.e., a nonshared effect)
rather than a general effect of family discord.
Causal Chain Effects
As discussed above, there is a growing body
of evidence that the persistence of conduct disorder from early childhood into adult life is
much influenced by indirect causal chain effects. Thus, Patterson and his colleagues
(Patterson & Bank, 1989; Patterson et al, 1991;
Patterson, DeBaryshe, & Ramsey, 1989) have
argued that there is a developmental progression for antisocial behavior. They found that
this behavior started with poor parental discipline and monitoring, leading to child conduct
problems that, in turn, led to both rejection by
normal peers and academic failure; both of
these features then predisposed to involvement
in a deviant peer group and this predisposed to
delinquency. Quinton et al. (1993) also demonstrated the key mediating role of a deviant peer
group in the persistence of conduct disorder
over time. Both studies suggested that family
influences continued to be important in adolescence, but it may well be that they operate in a
somewhat different way from influences in
early childhood. The importance of the finding
of indirect causal chain effects is that it points to
the need both for longitudinal data and for
methods of analysis that provide for tests of the
postulated causal mechanisms at each point in
the chain.
The Need to Use Natural Experiments
Rutter (in press) has argued that an experimental approach is just as necessary for testing
causal hypotheses in epidemiological-longitudinal studies as it is in the controlled conditions
of the experimental laboratory. With the exception of therapeutic interventions (see above), the
experimental variations have to be brought
about through naturally occurring circumstances, rather than through the control of con-
180
MICHAEL RUTTER
trived experimental conditions. Nevertheless,
the aim is exactly the same in terms of finding
circumstances that separate risk variables that
ordinarily go together. Genetic designs do this
with respect to the separation of nature and
nurture, and other designs do so in relation to
determining the direction of causal influences
from parent to child or the other way round.
Broadly similar issues apply with respect to the
testing of competing hypotheses about postulated environmental mechanisms associated
with family discord.
Mention has already been made of the pitting
of family discord and family separations as alternative mechanisms. Other studies have
sought to do the same with respect to discord
and poor parenting. Some of the results suggest
that parenting practices exert a greater effect
than does family conflict (e.g., Fauber & Long,
1991; Patterson & Bank, 1989), but other data
suggest a direct effect of family conflict that is
not mediated by parenting practices (Emery,
Fincham, & Cummings, 1992). These findings
all rely on multivariate analyses of one kind or
another that seek to partial out the effects of one
variable when examining the effects of another.
Inevitably, this provides a weak test, and experiments involving either contrasting circumstances or changes in circumstances are always
to be preferred.
For example, Hodges and Tizard (1989) contrasted two groups of children who spent their
first few years in a residential nursery, one of
which returned to the biological parents and one
of which was adopted. Both showed a rather
similar pattern of peer relationships (suggesting
that these had been influenced by the early
experiences in the institution rather than by the
very different family circumstances in adolescence), but conduct disorders were more prevalent in the children restored to their biological
parents than in those who had been adopted.
Because biological families provided a much
more disrupted, disorganized, and disharmonious environment (at least as far as could be
judged from the limited data available), the implication is that the family circumstances in
adolescence and middle childhood had indeed
had an effect on the children's conduct disorder.
It is unlikely to have been a selection bias
because, if there had been such a bias, it is likely
to have worked in such a way that the children
from less disturbed homes would have gone
back to their biological parents. The outcome
findings that showed a higher level of conduct
disorder, therefore, run against any such selection bias.
One way of tackling the alternative that the
risks stem from a lack of consistent, secure
attachment relationships rather than discord is
provided by the natural experiment of an institutional rearing (Quinton & Rutter, 1988). Discord is not a particular feature of well-run group
foster homes. On the other hand, they are characterized by a huge turnover in parenting figures, so that there is little opportunity for children reared in them to develop close attachment
relationships. The key group for this comparison, therefore, is that of children admitted to
institutions in very early childhood who remained there until leaving school. Quinton and
Rutter's (1988) findings indicated that this
group did as poorly in terms of social outcome
in early adult life as did those with a more
prolonged exposure to family discord. The result does not indicate a lack of effect from
family discord, but it does show that there is a
risk of broadly similar degree associated with
parenting provided on a roster basis in the absence of marked discord.
It is sometimes supposed that much of the
effect of family conflict, especially that following divorce, is a function of associated economic hardship. The natural experiment of parental remarriage provides a convenient way of
researchers examining this possibility, because
it is usually associated with a very marked improvement in economic circumstances. The
finding that remarriage is not accompanied by
any consistent improvement in children's behavior (Hetherington & Clingempeel, 1992)
suggests either that economic hardship is not the
key mechanism or that remarriage brings new
stressors that counterbalance any benefits that
stem from improved living conditions. The fact
that sometimes, following divorce, different
children in the same family are placed in different homes provides another convenient contrast (Monahan, Feldman, Hammer, & Rosario,
1993). The finding that siblings who lived apart
after divorce were behaviorally more different
than those who lived together suggests a family
effect, and potentially this design provides a
useful way of examining the impact of family
discord.
Rutter (in press) has argued that reversal effects provide a useful means of testing causal
hypotheses, quoting the example of unemploy-
SPECIAL SECTION: FAMILY DISCORD AND CONFLICT DISORDER
ment. The finding that there tends to be a relief
of distress in unemployed individuals who obtain work, as well as an increase in distress
when people lose their jobs, provides substantial support for the causal hypothesis. Fincham
et al. (1994) express some skepticism on the
value of reversals of key experiences in illuminating causal processes. Quite rightly, they note
that the factors that lead to the development of
the disorder are not necessarily the same as
those that maintain it. Accordingly, whether one
would expect a reduction in family conflict to
have beneficial effects on conduct disorder necessarily depends on how the risk mechanisms
are thought to work. If, for example, the maintenance of conduct disorder is dependent on the
role of hostile attributional biases leading to
aggressive behavior or to involvement in a deviant peer group, it may well be that a reduction
in discord may not help a great deal. Also, if
there are threshold effects, it may well be that
modest reductions in conflict will make little
difference (Richman, Stevenson, & Graham,
1982), although a major family change could be
beneficial (Rutter, 1971). If the peer group constitutes a key variable in the prolongation of
conduct disorder in middle childhood and adolescence (which seems to be the case), it may
well be that parental supervision and monitoring may be more important at this stage than
discord (Dishion, 1990; Dishion et al., 1991;
Patterson & Bank, 1989). If this model is correct, it might be supposed that a reduction in
family discord might make its greatest impact if
it occurs in earlier childhood rather than later.
Undoubtedly, there are difficulties in the assessment of reversal effects, but the finding from
naturalistic studies (Fincham et al., 1994) and
intervention studies (Kitzmann & Emery, 1994)
showing that reductions in family conflict are
associated with reductions in deviant behavior
implies that reversal effects should operate.
As indicated, the key point of natural experiments is for researchers to search for circumstances that split variables to test competing
hypotheses and that involve changes that either
are associated with a systematic increase in the
risk experience or, alternatively, with systematic decreases in it. Of course, it is not enough
to use designs involving natural experiments
that dissociate possible mechanisms and therefore allow tests of competing hypotheses on
mechanisms. Testing for dose response relationships, examining effects specificity, considering
181
biological plausibility, and assessing the
strength of effects are also all very important in
checking the likelihood that associations reflect
the causal mechanisms that have been hypothesized (Rutter, in press).
Conclusion
There is now abundant evidence that serious
family discord provides a good risk indicator for
conduct disorder in children. For the most part,
the associations found in the literature are of
only moderate strength (Jouriles, Bourg, & Farris, 1991), but the weakness of effect is quite
likely to reflect a reliance on single questionnaire measures of family functioning and single
questionnaire measures of child psychopathology, so that the error variance in the studies will
have been very large. On the other hand, it is
equally possible that the weak associations
mean that, although family discord is indeed a
risk indicator, it does not represent a risk mechanism. Rather, it could be that the risk processes
involve features that are associated with discord, instead of the discord itself. The research
undertaken to date clearly indicates the importance of investigating the mechanisms involved,
and there are ample leads to follow. It has been
argued that progress is likely to come through a
clear conceptualization of the hypothesized risk
factors, a careful attention to measurement issues, and the use of designs that provide a
rigorous testing of causal hypotheses.
References
Anderson, K. E., Lytton, H., & Romney, D. M.
(1986). Mothers' interactions with normal and conduct-disordered boys: Who affects whom? Developmental Psychology, 22, 604-609.
Baers, M. (1954). Women workers and home responsibilities. International Labor Review, 69, 338355.
Barkley, R., & Cunningham, E. (1979). The effects
of methylphenidate on mother-child interaction of
hyperactive children. Archives of General Psychiatry, 36, 201-208.
Bell, R. Q., & Chapman, M. (1986). Child effects in
studies using experimental or brief longitudinal
approaches to socialization. Developmental Psychology, 22, 595-603.
Belsky, J. (1988). Infant day care and socioemotional
development: The United States. Journal of Child
Psychology and Psychiatry, 29, 397-406.
Block, J. H., Block, J., & Gjerde, P. F. (1986). The
182
MICHAEL RUTTER
personality of children prior to divorce: A prospective study. Child Development, 57, 827-840.
Brown, G. W., & Harris, T. O. (1978). Social origins
of depression: A study of psychiatric disorder in
women. London: Tavistock.
Brown, G. W., & Rutter, M. L. (1966). The measurement of family activities and relationships: A
methodological study. Human Relations, 19, 241263.
Brunk, M. A., & Henggeler, S. W. (1984). Child influences on adult controls: An experimental investigation.
Developmental
Psychology,
20,
1074-1081.
Buchanan, C. M., Maccoby, E. E., & Dornbusch,
S. M. (1991). Caught between parents: Adolescents' experience in divorced homes. Child Development, 62, 1008-1029.
Cadoret, R. J. (1985). Genes, environment and their
interaction in the development of psychopathology. In T. Sakai and T. Tsuboi (Eds.), Genetic
aspects of human behavior (pp. 165-175). Tokyo:
Igaku-Shoin.
Cadoret, R. J., Troughton, E., Merchant, L. M., &
Whitters, A. (1990). Early life psychosocial events
and adult affective symptoms. In L. Robins & M.
Rutter (Eds.), Straight and devious pathways from
childhood to adulthood (pp. 300-313). Cambridge,
England: Cambridge University Press.
Champion, L., Goodall, G., & Rutter, M. (1994).
Behaviour problems in childhood and stressors in
early adult life: A twenty year follow-up of London
school children. Manuscript submitted for publication.
Cherlin, A. J., Furstenberg, F. F., Jr., Chase-Lansdale, P. L., Kiernan, K. E., Robins, P. K., Morrison, D. R., & Teitler, J. O. (1991). Longitudinal
studies of effects of divorce on children in Great
Britain and the United States. Science, 252, 13861389.
Cook, W. L., & Goldstein, M. J. (1993). Multiple
perspectives on family relationships: A latent variables model. Child Development, 64, 1377-1388.
Conger, R. D., Conger, K. J., Elder, G. H., Lorenz,
F. O., Simons, R. L., & Whitbeck, L. B. (1992). A
family process model of economic hardship and
adjustment of early adolescent boys. Child Devel-
opment, 63, 526-541.
Conger, R. D., Conger, K. J., Elder, G. H., Lorenz,
F. O., Simons, R. L., & Whitbeck, L. B. (1993).
Family economic stress and adjustment of early
adolescent girls. Developmental Psychology, 29,
206-219.
Crook, T., & Eliot, J. (1980). Parental death during
childhood and adult depression: A critical review
of the literature. Psychological Bulletin, 87, 252259.
Cummings, E. M., Davies, P. T., & Simpson, K. S.
(1994). Marital conflict, gender, and children's appraisals and coping efficacy as mediators of child
adjustment. Journal of Family Psychology, 8,
141-149.
DiLalla, L. F., & Gottesman, 1.1. (1989). Heterogeneity of causes for delinquency and criminality:
Lifespan perspective. Development and Psychopathology, 1, 339-349.
Dishion, T. J. (1990). The family ecology of boys'
peer relations in middle childhood. Child Development, 61, 874-892.
Dishion, T. J., Patterson, G. R., Stoolmiller, M., &
Skinner, M. L. (1991). Family, school, and behavioral antecedents to early adolescent involvement
with antisocial peers. Developmental Psychology,
27, 172-180.
Dodge, K. A. (1980). Social cognition and children's
aggressive behavior. Child Development, 51, 162170.
Dodge, K. A., Bates, J. E., & Pettit, G. S. (1990).
Mechanisms in the cycle of violence. Science, 250,
1678-1683.
Dunn, J., & Plomin, R. (1990). Separate lives: Why
siblings are so different. New York: Basic Books.
Easterbrooks, M. A., Cummings, E. M., & Emde,
R. N. (1994). Young children's responses to constructive marital disputes. Journal of Family Psy-
chology, 8, 160-169.
Emery, R. E., Fincham, F. D., & Cummings, E. M.
(1992). Parenting in context: A reply to Fauber and
Long. Journal of Consulting and Clinical Psychology, 60, 909-912.
Epstein, S. (1983). Aggregation and beyond: Some
basic issues on the prediction of behavior. Journal
of Personality, 51, 360-392.
Farrington, D. P. (1986). Age and crime. In M. Tonry
& N. Morris (Eds.), Crime and justice (Vol. 7, pp.
189-250). Chicago: University of Chicago Press.
Farrington, D. P. (1992). Explaining the beginning,
progress and ending of antisocial behavior from
birth to adulthood. In J. McCord (Ed.), Facts,
frameworks and forecasts: Advances in criminological theory (Vol. 3, pp. 253-286). New Brunswick, NJ: Transaction.
Farrington, D. P. (in press). The challenge of teenage
antisocial behavior. In M. Rutter (Ed.), Psychosocial disturbances in young people: Challenges
for prevention. New York: Cambridge University
Press.
Fauber, R., & Long, N. (1991). Children in context:
The role of the family in child psychotherapy.
Journal of Consulting and Clinical Psychology, 59,
813-820.
Fergusson, D. M., & Horwood, L. J. (1987). The trait
and method components of ratings of conduct disorder: Part I. Maternal and teacher evaluations of
conduct disorder in young children. Journal of
Child Psychology and Psychiatry, 28, 249-260.
Fergusson, D. M., & Horwood, L. J. (1988). Structural equation modelling of measurement processes
in longitudinal data. In M. Rutter (Ed.), Studies of
SPECIAL SECTION: FAMILY DISCORD AND CONFLICT DISORDER
psychosocial risk: The power of longitudinal data
(pp. 325-353). Cambridge, England: Cambridge
University Press.
Fergusson, D. M., & Horwood, L. J. (1989). Estimation of method and trait variance in ratings of
conduct disorder. Journal of Child Psychology and
Psychiatry, 30, 365-378.
Fergusson, D. M , Horwood, L. J., & Lloyd, M.
(1991). Confirmatory factor models of attention
deficit and conduct disorder. Journal of Child Psychology and Psychiatry, 32, 251-21 A.
Fergusson, D. M., Horwood, L. J., & Lynskey, M. T.
(1992). Family change, parental discord and early
offending. Journal of Child Psychology and Psychiatry, 33, 1059-1075.
Fincham, F. D., & Bradbury, T. N. (1987). The assessment of marital quality: A reevaluation. Journal of Marriage and the Family, 49, 797-809.
Fincham, F. D., Grych, J. H., & Osborne, L. N.
(1994). Does marital conflict cause child maladjustment? Directions and challenges for longitudinal research. Journal of Family Psychology, 8,
128-140.
Frick, P. J., Lahey, B. B., Loeber, R., StouthamerLoeber, M , Christ, M. A. G., & Hanson, K.
(1992). Family risk factors to oppositional defiant
disorder and conduct disorder: Parental psychopathology and maternal parenting. Journal of Consulting and Clinical Psychology, 60, 49-55.
Fristad, M. A., Weller, R. A., Weller, E. B., Teare,
M , & Preskorn, S. H. (1991). Comparison of the
parent and child versions of the children's depression inventory (CDI). Annals of Clinical Psychiatry, 3, 341-346.
Gardner, J. (1977). Three aspects of childhood autism: Mother-child interactions, autonomic responsivity, and cognitive functioning. Unpublished
doctoral dissertation, University of Leicester,
England.
Gottfried, A. E., & Gottfried, A. W. (Eds.). (1988).
Maternal employment and children's development,
longitudinal research. New York: Plenum.
Greenberg, M. T., Speltz, M. L., & DeKlyen, M.
(1993). The role of attachment in the early development of disruptive behavior problems. Development and Psychopathology, 5, 191-213.
Harris, T., Brown, G. W., & Bifulco, A. (1986). Loss
of a parent in childhood and adult psychiatric disorder: The role of lack of adequate parental care.
Psychological Medicine, 16, 641-659.
Hetherington, E. M. (1989). Coping with family transitions: Winners, losers, and survivors. Child Development, 60, 1-14.
Hetherington, E. M. (1991). Families, lies and videotapes. Journal of Research on Adolescence, 1,
323-348.
Hetherington, E. M., & Clingempeel, W. G. (with
Anderson, E. R., Deal, J. E., Hagan, M. S., Hollier,
E. A., & Lindner, M. S.). (1992). Coping with mar-
183
ital transitions. Monographs of the Society for Research in Child Development, 57(2-3, Serial No.
227).
Heyns, B., & Catsambis, S. (1986). Mother's employment and children's achievement: A critique.
Sociology of Education, 59, 140-151.
Hinshaw, S. P., Lahey, B. B., & Hart, E. L. (1993).
Issues of taxonomy and comorbidity in the development of conduct disorder. Development and Psychopathology, 5, 31-49.
Hodges, J., & Tizard, B. (1989). Social and family
relationships of ex-institutional adolescents. Journal of Child Psychology and Psychiatry, 30, 1197.
Jacob, T., & Tennenbaum, D. L. (1988). Family assessment methods. In M. Rutter, A. H. Tuma, &
I. S. Lann (Eds.), Assessment and diagnosis in
child psychopathology (pp. 196-231). New York:
Guilford Press.
Jenkins, J. M., & Smith, M. A. (1990). Factors protecting children living in disharmonious homes:
Maternal reports. Journal of the American Academy of Child and Adolescent Psychiatry, 29, 6 0 69.
Jouriles, E. N., Bourg, W. J., & Farris, A. M. (1991).
Marital adjustment and child conduct problems: A
comparison of the correlation across subsamples.
Journal of Consulting and Clinical Psychology, 59,
354-357.
Jouriles, E. N., Murphy, C. M., Farris, A. M., Smith,
D. A., Richters, J. E., & Waters, E. (1991). Marital
adjustment, parental disagreements about child
rearing, and behavior problems in boys: Increasing
the specificity of the marital assessment. Child
Development, 62, 1424-1433.
Kagan, J. (1984a). Continuity and change in the
opening years of life. In R. N. Emde & R. J. Harmon (Eds.), Continuities and discontinuities in development (pp. 15-39). New York: Plenum.
Kagan, J. (1984b). The nature of the child. New
York: Basic Books.
Kazdin, A. E. (1993). Treatment of conduct disorder:
Progress and directions in psychotherapy research.
Development and Psychopathology, 5, 277-310.
Kendler, K. S., Neale, M. C , Kessler, R. C , Heath,
A. C , & Eaves, L. J. (1992). Childhood parental
loss and adult psychopathology in women: A twin
study perspective. Archives of General Psychiatry,
49, 109-116.
Kitzmann, K. M., & Emery, R. E. (1994). Child and
family coping one year after mediated and litigated
child custody disputes. Journal of Family Psychology, 8, 150-159.
Kolvin, I., Miller, F. J. W., Scott, D. Mel., Gatzanis,
S. R. M., & Fleeting, M. (1990). Continuities of
Deprivation—Newcastle thousand families study.
Aldershot, England: Avebury.
Kurdek, L. A. (1992). Dimensionality of the dyadic
adjustment scale: Evidence from heterosexual and
184
MICHAEL RUTTER
homosexual couples. Journal of Family Psychology, 6, 22-35.
Lahey, B. B., McBurnett, K., Loeber, R., & Hart,
E. L. (in press). Psychobiology of conduct disorder. In G. P. Sholevar (Ed.), Conduct disorders in
children and adolescents: Assessments and interventions. Washington, DC: American Psychiatric
Press.
Lempers, J. D., Clark-Lempers, D., & Simons, R. L.
(1989). Economic hardship, parenting, and distress
in adolescence. Child Development, 60, 25-39.
Loeber, R. (1982). The stability of antisocial and
delinquent child behavior: A review. Child Development, 53, 1431-1446.
Lytton, H. (1990). Child and parent effects in boys'
conduct disorder: A reinterpretation. Developmental Psychology, 26, 683-697.
Martin, J. A. (1981). A longitudinal study of the
consequences of early mother-infant interaction: A
microanalytic approach. Monographs of the Society for Research in Child Development, 46(3, Serial No. 190).
Maziade, M. (1989). Should adverse temperament
matter to the clinician? In G. A. Kohnstamm, J. E.,
Bates & M. K. Rothbart (Eds.), Temperament in
childhood (pp. 421-435). Chichester, England:
Wiley.
McCartney, K., & Galanopoulos, A. (1988). Child
care and attachment: A new frontier the second
time around. American Journal of Orthopsychiatry, 58, 16-24.
Milne, A. M., Myers, D. E., Rosenthal, A. S., &
Ginsburg, A. (1986). Single parents, working
mothers, and the educational achievement of
school children. Sociology of Education, 59, 125139.
Moffitt, T. E. (1990). The neuropsychology of delinquency: A critical review of theory and research. In
N. Morris & M. Tonry (Eds.), Crime and justice
(Vol. 12, pp. 99-169). Chicago: Chicago University Press.
Moffitt, T. E. (1993a). Adolescence-limited and lifecourse-persistent antisocial behavior: A developmental taxonomy. Psychological Review, 100,
674-701.
Moffitt, T. E. (1993b). The neuropsychology of conduct disorder. Development and Psychopathology,
5, 135-151.
Monahan, S., Feldman, R. S., Hammer, M., & Rosario, M. (1993). Sibling differences in divorced families. Child Development, 64, 152-168.
Olweus, D. (1979). Stability of aggressive reaction
patterns in males: A review. Psychological Bulletin, 86, 852-875.
Parker, J. G., & Asher, S. R. (1987). Peer relations
and later personal adjustment: Are low-accepted
children at risk? Psychological Bulletin, 102, 357389.
Patterson, G. R. (1982). Coercive family process. Eugene, OR: Castalia.
Patterson, G. R., & Bank, L. (1989). Some amplifying mechanisms for pathologic processes in family.
In M. R. Gunner & E. Thelen (Eds.), The Minnesota Symposium on Child Psychology: Vol. 22.
Systems and development (pp. 167-209). Hillsdale,
NJ: Erlbaum.
Patterson, G. R., Capaldi, D., & Bank, L. (1991). An
early starter model for predicting delinquency. In
D. J. Pepler & K. H. Rubin (Eds.), The development and treatment of childhood aggression (pp.
139-168). Hillsdale, NJ: Erlbaum.
Patterson, G. R., DeBaryshe, B. D., & Ramsey, E.
(1989). A developmental perspective on antisocial
behavior. American Psychologist, 44, 329-335.
Perrin, S., & Last, C. G. (1992). Do childhood anxiety measures measure anxiety? Journal of Abnormal Child Psychology, 20, 567-578.
Plomin, R., & Daniels, D. (1987). Why are children
in the same family so different from one another?
Behavioral and Brain Sciences, 10, 1-15.
Quiggle, N. L., Garber, J., Panak, W. F., & Dodge,
K. A. (1992). Social information processing in aggressive and depressed children. Child Development, 63, 1305-1320.
Quinton, D., Pickles, A., Maughan, B., & Rutter, M.
(1993). Partners, peers and pathways: Assortative
pairing and continuities in conduct disorder. Development and Psychopathology, 5, 763-783.
Quinton, D., & Rutter, M. (1988). Parenting breakdown: The making and breaking of inter-generational links. Aldershot, England: Avebury.
Quinton, D., Rutter, M., & Gulliver, L. (1990). Continuities in psychiatric disorders from childhood to
adulthood in the children of psychiatric patients. In
L. Robins & M. Rutter (Eds.), Straight and devious
pathways from childhood to adulthood (pp. 259278). Cambridge, England: Cambridge University
Press.
Reid, J. B., Baldwin, D. V., Patterson, G. R., & Dishion, T. J. (1988). Observations in the assessment of
childhood disorders. In M. Rutter, A. H. Tuma, &
I. S. Lann (Eds.), Assessment and diagnosis in
child psychopathology (pp. 156-195). New York:
Guilford Press.
Richman, N., Stevenson, J., & Graham, P. (1982).
Preschool to school: A behavioural study. London: Academic Press.
Roberts, R. E., Lewinsohn, P. M., & Seeley, J. R.
(1991). Screening for adolescent depression: A
comparison of depression scales. Journal of the
American Academy of Child and Adolescent Psychiatry, 30, 58-66.
Robins, L. (1978). Sturdy childhood predictors of
adult antisocial behaviour: Replications from longitudinal studies. Psychological Medicine, 8, 611622.
Robins, L. N. (1991). Conduct disorder Annual Re-
SPECIAL SECTION: FAMILY DISCORD AND CONFLICT DISORDER
search Review, Journal of Child Psychology and
Psychiatry, 32, 193-213.
Rutter, M. (1971). Parent-child separation. Psychological effects on the children. Journal of Child
Psychology and Psychiatry, 12, 233-260.
Rutter, M. (1977). Prospective studies to investigate
behavioural change. In J. S. Strauss, H. M. Babigian, & M. Roff (Eds.), The origins and course of
psychopathology (pp. 223-247). New York: Plenum.
Rutter, M. (1978). Family, area and school influences
in the genesis of conduct disorders. In L. Hersov,
M. Berger, & D. Shaffer (Eds.), Journal of Child
Psychology and Psychiatry Book Series No. 1:
Aggression and antisocial behaviour in childhood
and adolescence (pp. 95-113). Oxford, England:
Pergamon.
Rutter, M. (1979). Protective factors in children's
responses to stress and disadvantage. In M. W.
Kent & J. E. Rolf (Eds.), Primary prevention in
psychopathology: Social competence in children
(Vol. 3, pp. 95-113). Hanover, NH: University
Press of New England.
Rutter, M. (1981). Epidemiological/longitudinal
strategies and causal research in child psychiatry.
Journal of the American Academy of Child Psychiatry, 20, 513-544.
Rutter, M. (1989). Age as an ambiguous variable in
developmental research: Some epidemiological
considerations from developmental psychopathology. International Journal of Behavioural Development, 12, 1-34.
Rutter, M. (in press). Beyond longitudinal data:
Causes, consequences, changes and continuity.
Journal of Consulting and Clinical Psychology.
Rutter, M. L., & Brown, G. W. (1966). The reliability
and validity of measures of family life and relationships in families containing a psychiatric patient. Social Psychiatry, 1, 38-53.
Rutter, M , & Giller, H. (1983). Juvenile delinquency: Trends and perspectives. Harmondsworth, England: Penguin Books.
Rutter, M , Maughan, B., Mortimore, P., & Ouston, J.
(with Smith, A.). (1979). Fifteen thousand hours:
Secondary schools and their effects on children.
Cambridge, MA: Harvard University Press.
Rutter, M., & Pickles, A. (1990). Improving the
quality of psychiatric data: Classification, cause
and course. In D. Magnusson & L. R. Bergman
(Eds.), Data quality in longitudinal research (pp.
32-57). Cambridge, England: Cambridge University Press.
Rutter, M., & Pickles, A. (1991). Person-environment interactions: Concepts, mechanisms, and implications for data analysis. In T. D. Wachs & R.
Plomin (Eds.), Conceptualization and measurement of organism-environment interaction (pp.
105-141). Washington, DC: American Psychological Association.
185
Rutter, M., & Quinton, D. (1984). Parental psychiatric disorder: Effects on children. Psychological
Medicine, 14, 853-880.
Rutter, M., & Smith, D. J. (Eds.), (in press). Psychosocial disorders in young people: Time trends and
their causes. Chichester, England: Wiley.
Sabatelli, R. M. (1988). Measurement issues in marital research: A review and critique of contemporary survey instruments. Journal of Marriage and
the Family, 50, 891-915.
Schachar, R., Taylor, E., Wieselberg, M., Thorley,
G., & Rutter, M. (1987). Change in family function
and relationships in children who respond to methylphenidate. Journal of the American Academy of
Child and Adolescent Psychiatry, 26, 728-732.
Schwarz, J. C , Barton-Henry, M. L., & Pruzinsky, T.
(1985). Assessing child-rearing behaviors with the
CRPBI: A comparison of ratings by mother, father,
student, and sibling. Child Development, 56, 462479.
Sinclair, I. A. C. (1971). Hostels for probationers.
London: Her Majesty's Stationery Office.
Smith, D. J. (in press). Youth crime and conduct
disorders: Trends, patterns, and causal explanations. In M. Rutter & D. Smith (Eds.), Psychosocial disorders in young people: Time trends and
their causes. Chichester, England: Wiley.
Susser, M. (1991). What is a cause and how do we
know one? A grammar for pragmatic epidemiology. American Journal of Epidemiology, 133, 635648.
Tanner, J. M. (1989). Foetus into man: Physical
growth from conception to maturity (2nd ed.).
Ware, England: Castlemead Publications.
Taylor, E., Sandberg, S., Thorley, G., & Giles, S.
(1991). The epidemiology of childhood hyperactivity (Maudsley Monograph No. 33). Oxford,
England: Oxford University Press.
Tennant, C , Bebbington, P., & Hurry, J. (1980).
Parental death in childhood and risk of adult depressive disorders: A review. Psychological Med-
icine, 10, 289-299.
Terr, L. C. (Series ed.). (1989). Resolved: Day care is
the best care for children under age 5 of working
Americans. Journal of the American Academy of
Child and Adolescent Psychiatry, 28, 130-135.
Tienari, P., Lahti, I., Sorri, A., Naarala, M., Moring,
J., Kaleva, M., Wahlberg, K-E., & Wynne, L. C.
(1990). Adopted-away offspring of schizophrenics
and controls: The Finnish adoptive family study of
schizophrenia. In L. Robins & M. Rutter (Eds.),
Straight and devious pathways from childhood to
adulthood (pp. 365-379). Cambridge, England:
Cambridge University Press.
Vorria, P. (1991). Children growing up in Greek
institutions: Their behaviour and relationships at
school and in the institution. Unpublished doctoral
186
MICHAEL RUTTER
dissertation, University of London, Institute of
Education.
Waters, E., Posada, G., Crowell, J., & Lay, K-L.
(1993). Is attachment theory ready to contribute to
our understanding of disruptive behavior problems? Development and Psychopathology, 5, 215—
224.
Weiss, B., Dodge, K. A., Bates, J. E., & Pettit, G. S.
(1992). Some consequences of early harsh discipline: Child aggression and a maladaptive social
information processing style. Child Development,
63, 1321-1335.
White, J., Moffitt, T. E., Earls, F., Robins, L. N., &
Silva, P. A. (1990). How early can we tell? Preschool predictors of boys' conduct disorder and
delinquency. Criminology, 28, 507-533.
Wilson, W. J. (1987). The truly disadvantaged: The
inner city, the underclass and public policy. Chicago: University of Chicago Press.
Wootton, B. (1959). Social science and social pathology. London: Allen & Unwin.
World Health Organization Expert Committee on
Mental Health. (1951). Report on the second session, 1951. Geneva, Switzerland: World Health
Organization.
Zigler, E. R, & Gordon, E. W. (Eds.). (1982). Day
care: Scientific and social policy issues. Boston:
Auburn House.
Received February 11, 1994
Revision received February 14, 1994
Accepted February 14, 1994 •
Reviewers for the Journal of Family Psychology Special Section
on Contexts of Interparental Conflict and Child Behavior
Frank D. Fincham, Guest Editor
Bonnie Burman
Phillip Cowan
JohnH . Grych
Ernest N. Jouriles
Gay Kitson
Lawrence A. Kurdek