Development and Psychopathology, 14 (2002), 731–759

Copyright  2002 Cambridge University Press

Printed in the United States of America
DOI: 10.1017.S0954579402004054

Interventions as tests of family systems

theories: Marital and family relationships
in children’s development
and psychopathology

PHILIP A. COWAN AND CAROLYN PAPE COWAN

University of California, Berkeley

Abstract
This paper addresses the role of family-based studies of preventive and therapeutic interventions in our
understanding of normal development and psychopathology. The emphasis is on interventions designed to improve
parent–child and/or marital relationships as a way of facilitating development and reducing psychopathology in
children and adolescents. Intervention designs provide the gold standard for testing causal hypotheses. We begin by
discussing the complexity of validating these hypotheses and the implications of the shift from a traditional
emphasis on theories of etiology to developmental psychopathology’s newer paradigm describing risks → pathways
→ outcomes. We summarize correlational studies that document the fact that difficult and ineffective parent–child
and marital relationships function as risk factors for children’s cognitive, social, and emotional problems in
childhood and adolescence. We then review prevention studies and therapy evaluation studies that establish some
specific parenting and marital variables as causal risk factors with respect to these outcomes. Our discussion focuses
on what intervention studies have revealed so far and suggests an agenda for further research.

A 10-year-old boy is referred for therapy chosocial problems in imbalances within the
because of his aggressive behavior in the child or adolescent’s intrapsychic system will
schoolyard. A girl of the same age is referred probably recommend play therapy. Others who
for therapy because she seems very sad at focus on reinforcements as the source of devi-
school and is having trouble making friends. ant behavior might suggest skills training. Cli-
What kind of treatment should be recom- nicians whose theoretical orientation focuses
mended for these children: play therapy or so- on the intergenerational transmission of be-
cial skills training for the child, parenting havior patterns through parent–child relation-
classes for the mother and father, or conjoint ships are more likely to get involved in parent
family treatment? Clinicians generally base support, education, or training. Those whose
their decisions concerning interventions for theory assumes that the identified patient’s
children and adolescents on their theories behavior represents a signal that the marital
about the causal mechanisms that affect nor- or family system is dysfunctional will proba-
mal development and psychopathology. Ther- bly opt for a family therapy approach.
apists whose theory locates the source of psy- This Special Issue of Development and
Psychopathology allows us to focus on this
issue in reverse, by showing how studies with
Address correspondence and reprint requests to: Philip A. intervention designs are essential for the vali-
Cowan, PhD, or Carolyn Cowan, PhD, Institute of Human
Development, 1203 Tolman Hall, University of California dation of theories of normal development and
@ Berkeley, Berkeley, CA 94720-1690; E-mail: pcowan psychopathology. Our focus here is specifi-
@socrates.berkeley.edu or ccowan@uclink4.berkeley.edu. cally on what can be learned from interven-

731
732 P. A. Cowan and C. P. Cowan

tion studies about the role of family relation- some of the empirical justification for creat-
ships—specifically, marital and parent–child ing interventions that target marital and par-
relationships—in children’s development of ent–child relationships with the aim of facili-
both adaptive and dysfunctional behavior pat- tating children’s development and reducing the
terns. The central line of our argument is that incidence and severity of psychopathology.
longitudinal studies of family risks and chil- Next, we explore what has been learned
dren’s outcomes are essential to the task of about efficacy or effectiveness of marital or
identifying factors to target in conceptually parenting intervention programs designed to
based interventions, but they cannot reveal reduce family risk factors and enhance chil-
whether the risk factors or the hypothesized dren’s academic or social skills. We then
mediating mechanisms have causal connec- summarize what family-based interventions
tions with the outcomes. By contrast, if an in- can teach us about theories of development
tervention reduces the toxicity of specific and developmental psychopathology and about
family relationship patterns (e.g., lowers the the family system contexts of children’s adap-
level of marital conflict or authoritarian par- tation. In a final section, we discuss some
enting) and if declines in the negativity of conceptual and methodological issues involved
these relationship processes are associated in the enterprise of risk research, prevention
with more positive adaptation for the child, trials, and theory testing and conclude with a
then we have learned three important things. list of promising directions for future research.
First, we have confirmed the hypothesis that
the intervention is effective, at least under the
Correlations and Causality
tested conditions. Second, we have obtained
evidence to suggest that marital and parent– Three well-accepted sets of “facts” about fam-
child relationships can play a causal role in ily relationships and child development are
nudging the child along the path toward adap- each subject to methodological ambiguities
tation or dysfunction. Third, such results pro- that make it difficult to interpret what they
vide empirical support for choosing a family mean. First, correlations have been found con-
systems approach to intervention in addition sistently between aspects of parents’ behavior
to, or instead of, individual therapy for the and their young children’s level of develop-
troubled boy or girl we described. ment (e.g., cognitive stage) or adaptation (e.g.,
Because the conclusions of this paper rest depression, aggression). When parents are
centrally on the argument that intervention warm, responsive, structured, and limit setting,
studies yield important knowledge about cau- while at the same time granting their children
sality, let us spell out some of the difficulties age-appropriate autonomy—a constellation that
involved in validating causal hypotheses in Baumrind (1980) calls an authoritative parent-
general and causal hypotheses about family ing style—children and adolescents show more
processes and children’s outcomes in particu- advanced academic achievement and social
lar. We begin by highlighting the fact that an competence and fewer indicators of problem
unrecognized paradigmatic shift in concep- behaviors than peers whose parents have au-
tions of causality occurred when the field of thoritarian or permissive parenting styles
developmental psychopathology adopted a (Conger, Ge, et al., 1994; Cowan, Powell, &
risk research paradigm. We then turn to the Cowan, 1998; Parke & Buriel, 1998; Stein-
literature on marital and parent–child rela- berg, 2001). Of course, if one is interested in
tionships to describe factors that have been psychopathology and deviations from normal
identified in correlational studies as risks for development, the story can be told the other
children’s adaptation from preschool through way around: parents who are authoritarian,
middle childhood, up to about age 10. This punitive, and harsh (structured and limit set-
brief review paints a broad-brush picture of ting without warmth and responsiveness) or
what has been established in systematic re- permissive (warm but laissez-faire) tend to
search about correlations between family pro- have offspring who are less academically and
cesses and children’s adaptation. It provides socially competent and more likely to have
Test of family systems theories 733

behavior problems as described by teachers, dence of parental influence. Although her

parents, peers, or the children and adolescents skepticism about parenting effects has been
themselves. challenged in a response by Collins, Mac-
Another consistent set of findings is that coby, Steinberg, Hetherington, and Bornstein
these same negative outcomes are associated (2000) and a number of other investigators
with high, unresolved marital conflict be- (see Borkowski, Ramey, & Bristol–Power,
tween the parents. This is the case, regardless 2002), Harris was correct in her admonition
of whether the parents are in intact marriages that concurrent measures of parenting style
(Cowan, Cowan, Schulz, & Heming, 1994; and child outcomes cannot be used to support
Cummings & Davies, 1994) or separated or the causal hypothesis that parents are to blame
divorced (e.g., Hetherington & Clingempeel, for their children’s undesirable behavior. It is
1992). A third set of facts, perhaps expectable equally plausible that the child’s behavior
from the first two, is that “virtually every elicits negative responses from the parents or
study examining associations between mar- that genetic inheritance is responsible for the
riage and parenting has found that the quality correlations. Harris did not write about mari-
of parent–child relationships and the quality tal conflict, but a similar argument could be
of marital relationships are linked within fam- made in this domain; the correlations between
ilies” (Lindahl, Clements, & Markman, 1997, parents’ marital quality and children’s out-
p. 385; see also Erel & Berman, 1995). That comes could be attributable to the disruptive
is, when parents are maritally distressed, one impact of a child’s negative behavior on the
or both are less likely to be authoritative with parents’ relationship as a couple.
their children. We are not arguing that correlational data
It is tempting to conclude, as many have are irrelevant to causality. If a correlation be-
done, that an ineffective or harsh parenting tween A and B is not found, then one can con-
style and conflict between the parents play clude that there is no direct causal connection
a causal role in children’s and adolescents’ between them. Correlational designs represent
academic and social difficulties, behavior a first step in the process of testing causal
problems, and psychopathology. Because the hypotheses. The next step involves the estab-
marital relationship usually precedes the par- lishment of a temporal relationship between
ent–child relationship, it is also tempting to the variables. Causes necessarily come before
conclude that marital disruption “spills over” effects, but if parent and child behaviors are
into parent–child relationships. However, as measured at the same time, it is not possible
students are taught in “Statistics 101,” no con- to determine which “comes first.” Despite fre-
clusions about causation or direction of ef- quent claims to the contrary, this difficulty
fects should be drawn from simple corre- cannot be completely solved with longitudinal
lations, regardless of how pervasive and research designs, because even if marital con-
consistent the findings are. flict or parenting behavior measured at Time
Nowhere has the issue of causality been 1 predicts a child outcome at Time 2, it is
more difficult to establish than in studies of possible that the parent or the couple have
family influences on children’s development. been responding to the child’s pattern of be-
Judith Harris (1998) provided a startling chal- havior since before the study began at Time 1
lenge to accepted wisdom by asserting that (Cowan & Cowan, 2002). What could help to
parents have little long-term impact on their determine the direction of effects in this situa-
children’s personality development and be- tion is a family-based intervention that fo-
havior, especially when that behavior is as- cuses on parent–child relationships and/or mar-
sessed in contexts outside the nuclear family ital quality, with randomized assignment to
(e.g., at school or in the peer group). Harris conditions and adequate controls. If participa-
was critical of family socialization researchers tion in the intervention produces a positive
for interpreting thousands of studies revealing change in parent–child or marital relationship
correlations between parents’ behavior and quality, and if change in these two central
children’s outcomes as if they provided evi- family relationships is associated with im-
734 P. A. Cowan and C. P. Cowan

provement in the child’s adaptation, that would digm in the study of psychopathology, with a
constitute strong support, though not absolute more differentiated and dynamic approach to
proof, of the causal relevance of family rela- the understanding of cause and effect.
tionship patterns to the child outcomes under
study.
Causation and etiology
We seem to be caught in a “Catch-22.”
Clinical and preventive interventions de- In traditional approaches to understanding
signed to improve the lot of children and ado- psychopathology, a great deal of effort has
lescents should be based on the best available been devoted to determining etiology, which
causal models of how targeted variables are is the cause or causes of various diseases or
related to the outcomes we wish to change. disorders. There are four basic problems with
But we cannot legitimately use correlational how this search has been conceptualized and
data to make claims about causal models be- conducted. First, as we have stated, scientists
fore an intervention is systematically evalu- have difficulty defining cause and agreeing on
ated. This is the kind of dilemma that gave the criteria that must be met before a causal
rise to the field of prevention science (Coie, relation between A and B can be established
Watt, West, & Hawkins, 1993), in which (Freedman, 1999). Second, despite the fact
there is a that traditional causal models often describe
linear main effects in which variable A is pre-
complementary interplay of science and practice. sumed to have a direct link with variable C,
Basic research on risk and protective factors should contemporary ideas of causality often involve
inform the design of preventive interventions. Field indirect effects in which A affects C through
trials of these interventions, in turn, should yield its effect on B (mediated or moderated mod-
insights about the causes of disorder and the devel- els; e.g., Simons, Lorenz, Wu, & Conger,
opmental processes that contribute to risk or recov- 1993).
ery. (Coie et al., 1993, p. 1013)
Third, investigators using retrospective,
case control, or archival data usually start
This systematic, circular approach—risk re- with the diagnostic category or indicator of
search, intervention trials, theory testing, and distress and then work backward to determine
back again to more risk research—helps potential antecedents or causes of the diagno-
avoid the Catch-22 by using one perspective sis in question. The problem with this ap-
repeatedly to inform the other. proach is that even if researchers find a signif-
icantly higher incidence of marital conflict or
Causal Models, Risk Models, ineffective parenting in the family back-
Developmental Psychopathology, ground of children who are depressed than in
and Intervention families of children who are not depressed, it
should be clear by now that they cannot con-
Since the first definitions of the field of clude from these findings that family relation-
developmental psychopathology by Rutter, ship difficulties caused the children’s depres-
Garmezy, and Sroufe (Rutter & Garmezy, sion (Cummings, DeArth–Pendley, Du Rocher
1983; Sroufe & Rutter, 1984), there has been Schudlich, & Smith, 2001). The laws of infer-
a decided shift in the language of prevention ence are often ignored in these situations. The
science from an emphasis on causality and fact that B (depression) implies A (family dif-
etiology to a focus on risk, buffering, and ficulty) does not mean that A implies B.
vulnerability.1 This shift involves a new para- Fourth, if researchers begin their studies,
not with the identification of children’s de-
pression (B) but with families who have inef-
1. The discussion in this section is based in part on a fective or dysfunctional marital or parent–
chapter by Cowan, Cowan, and Schulz (1996) and in
part on extensive discussions and collaboration with child relationships (A), they will typically find
Diana Baumrind on an unpublished paper on causal that some of the children are highly de-
inference in the social sciences. pressed, others show different kinds of psy-
Test of family systems theories 735

chiatric or behavior problems but not depres- ment and intervention is not the search for ul-
sion (C), some show no signs and symptoms timate first causes, but the discovery of which
of diagnosable disorders (D), and a few are conditions nudge people toward adaptation
functioning exceptionally well (E). To make and away from disorder and the design of in-
matters even more complex, we could also ex- terventions to create those conditions. In some
pect to find the same range of outcomes, dis- cases, this enterprise may address questions of
tributed in different proportions, in families etiology, but it is equally or more likely to
with warm and supportive relationships among address questions about factors that maintain
its members. Thus, as Cicchetti and Rogosh psychopathology, regardless of the original
(1996) note, follow-forward designs reveal causes.
complex patterns of equifinality (different paths Despite the almost universal tendency to
lead to the same end) and multifinality (simi- assume that theories of normal development,
lar risks are associated with different outcomes). psychopathology, and intervention come in
The method of beginning with risks and trac- interlinked sets, there are no necessary con-
ing developmental pathways to outcomes is nections between one’s theory of how a symp-
the hallmark of developmental psychopathol- tom, syndrome, or disorder is caused and
ogy (Cicchetti & Cohen, 1995a, 1995b). one’s theory of intervention (Cowan et al.,
Consistent with the public health models 1998). Biologically caused conditions are
from which developmental psychopathologists sometimes not susceptible to biochemical in-
adopted the risk–outcome paradigm, risks are terventions but may respond well to psycho-
considered markers that are associated with an therapy, whereas psychologically caused con-
increased chance of a disorder occurring in a ditions may be resistant to environmental
defined population. For some purposes, the treatment but responsive to drugs. If we
calculation of statistical association is suffi- mount an intervention to reduce specific risks
cient, without assuming causality. For exam- and if that intervention “works,” we have not
ple, if we want to know how many children necessarily isolated the natural etiology of the
in a school are in need of mental health ser- disorder. What we have done is identified spe-
vices, a statistical estimate of the risks associ- cific causal risk or protective factors that can
ated with the outcomes would give sufficient and should be targeted in future interventions.
information to make decisions about provid- It may still be important to establish the natu-
ing services. However, if we want to inter- ral etiology of the disorder. For example, even
vene to reduce the incidence of depression if we have good drug treatments for yellow
and antisocial behavior at school, it is neces- fever, it would be helpful to identify the
sary to do two things. We must study the path- swamps that serve as spawning grounds for
ways traveled by individuals under conditions disease-carrying mosquitoes.
of low and high risk, in order to identify (a)
the sequence of risk factors associated with
Methods of identifying causal risk
movement toward depression and antisocial
and protective factors
behavior and (b) the protective factors associ-
ated with movement toward adaptation, de- There are two ways to identify causal risk and
spite the presence of risk. Furthermore, it protective factors in families that are associ-
would be necessary to identify causal risk ated with children’s adaptation and maladap-
factors, which are the ones believed to func- tation. The first nonexperimental method is to
tion as generative causes of the outcomes. The subject the risk–outcome correlations to sys-
hypothesis would be that when the generative tematic statistical controls in order to reject
causes are eliminated or reduced, or when the the hypothesis that the connection is random
usual paths between cause and outcome are or artifactual. For example, in order to con-
interrupted, the incidence of depression and clude that parents’ divorce has a harmful
antisocial behavior will be reduced. (causal) impact on children, one must rule out
From a developmental psychopathology the possibility that the children were showing
perspective, what is important for both treat- behavior or emotional problems before the di-
736 P. A. Cowan and C. P. Cowan

vorce (Block, Block, & Gjerde, 1989) or that Parent–Child Relationships Associated
marital conflict and other associated risk fac- With Children’s Adaptation
tors, both predivorce (Cherlin, Chase–Lans- and Dysfunction
dale, & McRae, 1998) and postdivorce (John-
Consistent with a prevention science ap-
ston, Kline, & Tschann, 1989), might account
proach, we begin to map the territory of risk–
for significant variation in the child outcomes
outcome connections between family relation-
under investigation. That is, rejection of alter-
ships and children’s adaptation with a brief
native hypotheses using analyses with appro-
summary of correlational studies. In effect,
priate statistical controls helps us to evaluate
these “Step 1” studies provide the background
causal arguments, even though they fall short
data for hypotheses about causality, which
of providing definitive proof.
can then be tested further with intervention
A second method of identifying the causal
designs. We make no claims of providing an
nature of family risk factors is to perform a
exhaustive review, especially because a num-
controlled experiment, with randomized as-
ber of other authors cited in the sections that
signment of participants to intervention and
follow have covered this ground effectively.
control conditions. The intervention should be
targeted to the hypothesized causal risk factor
(in the present context, the quality of marital Parent–child relationships
or parent–child relationships), and appropriate and children’s development
statistical tests should be performed to evalu-
ate whether (a) the intervention had a positive Here we summarize three longstanding ap-
impact on the outcomes and (b) the positive proaches to describing variations in the qual-
outcome was associated with change in the ity of parent–child relationships—attachment,
quality of marital or parent–child relationships. parenting styles, and parenting practices—
Our discussion of etiology and causality and meta-emotion coaching, a fourth, rela-
may help to clarify what intervention research tively new, approach. Note that the prepon-
can and cannot accomplish. Intervention stud- derance of systematic studies have examined
ies establish an association between risk and mother–child relationships; father–child rela-
outcome, a temporal sequence in which the tionships have not received the attention they
change in risk precedes the change in out- deserve in light of fathers’ central contribu-
come, with some kind of dose–response or tion to their children’s development in mar-
functional relation between the two. For these ried, divorced, and remarried families (Bron-
reasons, intervention designs allow us to test stein & Cowan, 1988; Hetherington, 1999).
hypotheses about cause and effect and to dis- Note also that there is still little information
tinguish correlations from causal relations. about cultural variations in the links between
What intervention designs cannot do is test parenting behavior and children’s outcomes
hypotheses about etiology. To demonstrate (Parke & Buriel, 1998).
that increased use of an authoritative parent-
ing style is followed by improved academic Attachment. Taking their framework from
performance does not prove that the child’s Bowlby (1980) and the Strange Situation labo-
academic problems were caused by ineffec- ratory measurement paradigm by Ainsworth
tive parenting. Similarly, if the administration and Wittig (1969), a number of investigators
of phenothiazines improves schizophrenic focused on how infants and young children re-
symptoms, it does not follow that schizophre- act to their parent when reunited after a separa-
nia is caused by a phenothiazine deficit. We tion: a potentially anxiety-provoking condition
will show that this limitation does not consti- that elicits the operation of an attachment be-
tute a serious restriction on the contribution havioral system to maintain optimal closeness
of intervention research to our understanding and distance from a “secure base.” Results
of links between the quality of family rela- show that children categorized as “securely
tionships and children’s development. attached” may initially raise a fuss at both
Test of family systems theories 737

separation and reunion, but they quickly find children are insecurely attached as infants,
a way to reengage with the parent in mutually they are more likely to have low scores on
satisfying interaction. Some children charac- measures of social and academic competence
terized as “insecurely attached” avoid the par- in high school (Carlson et al., 1999) and more
ent anxiously, others react angrily and resist likely to have high scores on an overall index
the parent’s overtures, and a few show disor- of psychopathology in adolescence, after ac-
ganized behavior after the parent’s return. De- counting for measures of temperament and
spite the fact that these are all descriptions of specific parenting behaviors (Carlson, 1998).
the child’s behavior, both the theory and em- In sum, the quality of parent–child relation-
pirical evidence suggest that these reactions ships, as reflected in assessments of attach-
reflect the ongoing transactions in the rela- ment, is reliably associated with many mark-
tionship between parent and child. Belsky’s ers of child and adolescent development,
careful review (1999) concluded from many adaptation, and dysfunction.
studies that (a) securely attached infants have
caretakers who are observed to be warmer and Parenting style. Parenting style is generally
more responsive and sensitive than caretakers assessed in terms of a small number of broad-
of insecurely attached infants; (b) infants with brush dimensions encompassing central atti-
avoidant attachments have primary caretakers tudes and behaviors that define the emotional
who are observed to be intrusive, controlling, climate of the relationship between parent and
and excessively stimulating; and (c) infants child. Most investigators of parenting behav-
with resistant attachments have caretakers ior focus on at least some of the dimensions
who appear to be unresponsive and underin- identified by Baumrind (1971, 1989, 1991) as
volved with them. components of parenting style: warmth or re-
Results of longitudinal studies, most nota- sponsiveness, control (structure, limit setting),
bly those by Sroufe, Weinfield, Egeland, and and developmentally appropriate autonomy
colleagues (Sroufe, Duggal, Weinfield, & encouragement. Many also use Baumrind’s
Carlson, 2000; Weinfield, Sroufe, & Egeland, typologies, which are formed by the intersec-
2000), reveal that infants with an early history tion of the warmth and control dimensions:
of secure attachment, usually to the mother, authoritative parents are warm and set limits;
are likely to be more self-reliant and indepen- authoritarian parents are cold, hostile, and
dent, more empathic, less anxious, and more grant little autonomy; and permissive parents
socially competent than their less securely are warm and exert little control. In addition
attached peers throughout childhood and ado- to these three styles, Maccoby and Martin
lescence. A review of empirical studies (Green- (1983) suggested the fourth logical outcome
berg, 1999) indicates that insecure attach- of combining high and low warmth and con-
ments are risk factors for the development of trol: disengaged parents, who are low in af-
problematic behavior and diagnosable psy- fective expression (neither warm nor cold)
chopathology. Several longitudinal research and low in control.
programs found that infants with early inse- Steinberg (2001, p. 8) argued that the evi-
cure attachment were more likely to show ex- dence is in: “adolescents raised in authorita-
ternalizing behaviors up to 15 years later. tive homes achieve more in school, report less
Other longitudinal studies showed a predic- depression and anxiety, score higher on mea-
tive connection between anxious, avoidant sures of self-reliance and self-esteem, and are
attachment in infancy and later anxiety or de- less likely to engage in antisocial behavior,
pression (see Cassidy, 1995; Graham & East- including delinquency and drug use (Olds et
erbrooks, 2000). Disorganized attachment at al., 1998).” This generalization, Steinberg as-
12–18 months is a risk factor for dissociation serts, holds for families representing many
in adolescence and young adulthood (Ogawa, cultural or ethnic groups. There is comparable
Sroufe, Weinfield, Carlson, & Egeland, 1997). evidence of links between authoritative par-
Carlson and her colleagues found that when enting and outcomes for younger children
738 P. A. Cowan and C. P. Cowan

(Maccoby & Martin, 1983). Yet, despite the not the effectiveness of the discipline; and the
impressive number of studies cited in support child escalates until the parent gives in. The
of authoritative parenting, in our view there investigators argue that this pattern reinforces
are two reasons why the question is not yet the child for responding negatively and rein-
settled. There have been few observational forces the parent temporarily by ending the
studies of families who are not white and mid- struggle until it begins again. Not surpris-
dle class, and our search of the literature sug- ingly, such coercive cycles between parent
gests that studies of the connection between and child are related to the child’s overt ag-
parenting and psychopathology tend to focus gression and antisocial behavior both at home
on the separate dimensions of warmth and and at school.
control rather than on Baumrind’s typology. Internalizing behavior or depression. Cic-
chetti and his colleagues (Cicchetti & Carlson,
Parenting practices. Darling and Steinberg 1989; Cicchetti, Toth, & Maughan, 2000) re-
(1993) made a useful distinction between par- mind us that not all forms of legally defined
enting styles, which focus on broad dimen- abuse have the same impact. Physical abuse by
sions like warmth and control, and parenting parents is more often associated with aggres-
practices, which are the specific parental be- sive, acting out behavior in children, in contrast
haviors associated with positive and negative with sexual abuse and psychological neglect,
outcomes. which are implicated in children’s depression.
Externalizing behavior or aggression. Per- Investigators have made an important dis-
haps the single, most frequently investigated tinction between two types of parental control
parenting behavior associated with various that are less extreme than abuse: (a) behav-
categories of psychopathology has been harsh ioral control, which is seen in consistent en-
physical punishment. There is overwhelming forcement of rules and monitoring and super-
evidence that early physical punishment by vision of the child’s whereabouts; and (b)
parents that reaches the legal definition of psychological control, which is seen in con-
abuse (causing injury or leaving marks on the trolling the child by manipulating the attach-
skin for more than a day) is a serious risk fac- ment bond with threats of withdrawal of love,
tor for children, especially for the develop- guilt induction, shaming, or overprotective-
ment of aggression and oppositional defiant ness (Cummings et al., 2001). In general, de-
behavior (Bates, Pettit, & Dodge, 1995). Some velopmentally appropriate levels of behav-
investigators argue that any physical punish- ioral control are associated with lower levels
ment whatsoever, including light or moderate of externalizing (e.g., Herman, Dornbusch,
spanking used as a disciplinary technique, in- Herron, & Herting, 1997): Parents who set de-
creases the risks of a host of negative out- velopmentally appropriate limits have chil-
comes (Gershoff, in press; Straus, 2001). A dren who are less aggressive with their peers.
recent review and reanalysis of the literature By contrast, parents who use high levels of
on spanking (Baumrind, Larzelere, & Cowan, psychological control that arouse guilt and
in press) indicates that, whereas physical abuse shame are more likely to have children who
is clearly associated with negative outcomes for are anxious, depressed, and show physical
children, there is no evidence that moderate symptoms (Barber & Harmon, 2002). The dif-
spanking, in itself, has detrimental effects on ference in outcomes associated with the two
children’s development or adaptation. types of control is not absolute: some children
Less extreme than harsh physical punish- of parents with low levels of behavioral con-
ment is coercive parenting, a pattern described trol are internalizing, and some children of
by Patterson and his colleagues (Dishion, Pat- parents with high levels of psychological con-
terson, & Kavanagh, 1992; Patterson, Reid, & trol are angry and aggressive.
Dishion, 1998). The child begins with an ag-
gressive or disobedient act; the parent fails to Meta-emotion coaching. Gottman, Katz, and
respond or responds ineffectively; the child Hooven (1997) proposed a new parenting di-
escalates; the parent escalates the hostility but mension or typology; unlike the Baumrind
Test of family systems theories 739

formulation, it focuses very specifically on sequence of not doing well academically in

one content area: how parents regulate their early elementary school (Hinshaw, 1992).
own and their child’s negative emotion. The
construct is coded from detailed, open-ended Moderation and mediation. The connection
interviews with parents about emotional mo- between parenting style and child or adoles-
ments in their interactions with their child. cent outcomes is not a simple one. A number
Parents are described as “emotion coaching” of important variables seem to moderate the
when they are aware of their own and their size of the correlations. For example, Darling
child’s sadness and anger; empathize with or and Steinberg (1993) made the useful point
validate these emotions; view the child’s ex- that the “same” parenting behavior (e.g., high
pression of these emotions as an opportunity levels of monitoring an adolescent’s behavior)
for intimacy; and teach, label emotions, or can have different impacts on the child, de-
solve the problems that give rise to the emo- pending on the general context of the relation-
tions. By contrast, “emotion-dismissing” par- ship. If the child’s parents are warm and re-
ents are those who fail to hear and discuss sponsive, their frequent checks on where the
their child’s sadness or anger, dismiss the im- child is and when he or she will be coming
portance of these feelings (“That’s not some- home may be interpreted as caring; if the par-
thing to get upset about”), or react in deroga- ents are generally cold and angry, monitoring
tory fashion when such feelings are expressed may be experienced as intrusive, distrustful,
(“Don’t be a baby”). Research on these con- and hostile. The gender of parent and child
structs is in the beginning phase. Gottman et and culture have been shown to affect whether
al. (1997) reported that the interview measures parenting behavior and children’s adaptation
of parents’ meta-emotion coaching are signifi- are connected, but we believe that it is prema-
cantly correlated with (a) observational mea- ture to make generalizations about specific
sures of behavior in parent–child interactions, connections characteristic of fathers and sons,
(b) physiological measures of children’s abil- fathers and daughters, mothers and sons, and
ity to down-regulate their own upset, (c) mea- mothers and daughters. Children’s temperament
sures of 8-year-olds’ interactions with friends, is also a frequently cited moderator (Bates,
and (d) academic achievement in school. 2001), although it is not always clear whether,
for example, a child’s difficult or easy tempera-
A note on academic achievement and parent- ment changes the relation between parenting
ing. Children’s problems in meeting the aca- style and outcomes or whether parenting style
demic challenges of school have been associ- is the moderator, changing the relation be-
ated with all the parenting styles and practices tween temperament and outcome.
we have described: insecure attachments, au- Children’s processing of social and emo-
thoritarian and permissive parenting styles, tional cues can function as a mechanism that
abuse, coercive cycles, and dismissing emo- mediates the link between parenting quality and
tions (Baumrind, 1989; Cicchetti & Carlson, child outcomes. For example, Dodge and his
1989; Dornbusch, Ritter, Leiderman, & Rob- colleagues (Dodge, Pettit, Bates, & Valente,
erts, 1987; Gottman et al., 1997; Melby & 1995) suggested that harsh physical punish-
Conger, 1996; Parke & Buriel, 1998). If causal ment by parents is associated with the child’s
links are established between parent–child rela- tendency to interpret peer behavior as hostile,
tionship qualities and child outcomes, these which, in turn, is associated with higher levels
findings inform us about the direct effects of of aggression in the child’s relationships with
parenting on children’s development. There are peers. Parke and Buriel (1998) summarized
also indirect effects of parents’ styles of inter- several studies showing that children’s skills in
acting with their children. When parent–child processing emotional cues and regulating emo-
relationships go awry, children and adolescents tion mediate the links between qualities of par-
are more likely to have difficulties with their ent–child interaction and the child’s social
peers and to be aggressive and/or depressed, skills with peers. Measelle (in press) showed
difficulties that may be both a cause and a con- that ineffective parenting during the preschool
740 P. A. Cowan and C. P. Cowan

period was more likely to be followed by low more recent, and consequently less differenti-
achievement and internalizing problems in first ated, than the literature on parent–child rela-
grade children who had a negative view of tionship quality. Despite the fact that clini-
themselves before they entered school. cally based family theorists since the 1950s
have been drawing attention to the marital re-
lationship as pivotal in creating the atmo-
Summary of what we know about
sphere in which children develop (Jackson,
parent–child relationships
1963; Lidz et al., 1985; Satir, 1972), studies
and children’s adaptation
of links between marital quality and chil-
The studies cited above represent only a frac- dren’s adaptation did not begin until three de-
tion of the thousands of publications revealing cades later, when they were stimulated by
significant correlations between parent–child studies of the impact of parents’ divorce on
relationship quality and adaptation in child- their children (Emery, 1988; Hetherington,
hood and adolescence. We have looked at four Cox, & Cox, 1982; Wallerstein & Kelly, 1980).
central approaches to the description of par- Emery and O’Leary (1984) were the first
ent–child relationships: attachment, parenting of many investigators (for reviews see Cum-
styles, parenting practices, and meta-emotion mings & Davies, 1994; Emery, 1999; Fin-
coaching. Although these approaches are cham, 1998) to show that a high level of mari-
somewhat different, it is clear that parenting tal conflict or overt marital violence, especially
research focuses primarily on variations in when the disruption occurs in front of the
parenting warmth and control. It is not clear child and is not resolved, are risk factors for
whether warmth by itself, without responsive- both depression and aggression in children
ness, accounts for variations in positive out- (Cowan, Cowan, Heming, & Miller, 1991)
comes for children. It is clear that there are and adolescents (Conger, Conger, Elder, &
important differences in outcomes associated Lorenz, 1992). The conflict does not have to
with parents’ behavioral and psychological be overt; cold, distanced withdrawal in either
control that range from supportive limit set- or both parents can place children at higher
ting to guilt induction and abuse. risk for problematic behavior (Cox, Paley, &
Consistent with the literature on parenting, Harter, 2001). Although it seems obvious that
we have emphasized what parents do when difficulties in the relationship between the par-
they are interacting with their children. We ents could be associated with less desirable out-
should note that parents also have important comes for children, the mechanisms of such
indirect influences on their children as they linkage need to be explained. It is quite proba-
provide opportunities for interaction outside ble that there is a direct effect: watching their
the family (Parke, Burks, Carson, Neville, & parents hit or scream at each other tends to
Boyum, 1994) and serve as models of peer produce powerful emotional responses and
relationships in their own interactions with impulses to act in children (Cummings & Da-
spouse and friends (Parke & Buriel, 1998). vies, 1994), though few constructive options
We have referred to children’s “outcomes” for action are available to them. Children’s or
in spite of the fact that the correlational de- adolescents’ perceptions of the frequency and
signs of all the studies we have cited so far intensity of parental conflict and their ten-
do not allow us to make statements about dency to assume that they are to blame for it
cause and effect. We return to this point after appear to mediate the links between observed
considering correlational studies of links be- marital conflict and both aggression and de-
tween parents’ marital relationship quality and pression in the children (Dadds, Atkinson,
their children’s development. Turner, Blums, & Lendich, 1999; Grych, Fin-
cham, Jouriles, & McDonald, 2000; Rogers &
Holmbeck, 1997).
Marital Relationships, Children’s
There is strong correlational evidence for
Adaptation, and Psychopathology
indirect effects of marital dysfunction on chil-
Research on the links between marital rela- dren, suggesting that the couple’s inability to
tionship quality and children’s adaptation is regulate their negative affect in productive
Test of family systems theories 741

ways spills over into the parent–child rela- Thus, the marital conflict and parenting prob-
tionships. A meta-analysis by Erel and Bur- lems that predict children’s outcomes in a nu-
man (1995) and a recent review by Cox and clear family may be markers of an unfolding
her colleagues (2001) found consistent corre- process of intergenerational transmission.
lations between marital conflict and measures Second, the parents’ own adjustment, includ-
of ineffective or disrupted parent–child rela- ing depression and serious mental illness, is
tionships. Researchers have reported connec- associated with marital problems, disruptions in
tions between marital conflict and each of the parent–child relationships, and children’s
aspects of parent–child relationships discussed maladjustment (Seifer & Dickstein, 2000).
above: parent–child attachment (Davies & Cum- Third, marital conflict and ineffective parent-
mings, 1994); parenting styles (Cowan et al., ing may combine to disrupt sibling relation-
1994); parenting practices (Holden & Miller, ships (Deal, Hagan, Bass, Hetherington, &
1999); and meta-emotion coaching (Gottman Clingempeel, 1999), and sibling relationships
et al., 1997). may play an important training role in chil-
A number of investigators have begun to dren’s later aggression and antisocial behavior
recognize that how two parents work with, or with peers (Patterson, Dishion, & Bank, 1984).
against, each other as they parent their child, a Fourth, socioeconomic factors and other
construct called coparenting (Gable, Crnic, & stressors outside the family have an impact on
Belsky, 1994; McHale & Rasmussen, 1998), (or, at least, an association with) the quality of
deserves recognition as a distinct aspect of marital and parent–child relationships (Conger,
family relationships. In part, this relationship Elder, Lorenz, Simons, & Whitbeck, 1994;
at the intersection of marital and parent–child McLoyd, 1990).
relationships had received little attention be- As family systems theories suggest, studies
cause of the tendency to examine dyadic rela- reveal that multiple aspects of family life
tionships rather than the whole family system combine to explain variance in development
(Fivaz–Depeursinge & Corboz–Warnery, 1999). and psychopathology. In a study of nonclini-
Evidence so far suggests that spousal conflict cal families with children entering kindergar-
and undermining in the coparenting relation- ten (Cowan & Cowan, in press), we used
ship is associated with the same problematic structural equation models to examine the
behaviors in children as marital conflict dur- power of five domains of family life to predict
ing a dyadic discussion (McHale & Cowan, variation in children’s academic achievement,
1996). We believe that the distinctions among externalizing, and internalizing as they made
marital, parenting, and coparenting relation- the transition to elementary school in kinder-
ships will prove useful in designing family- garten and first grade. Each domain (individ-
based interventions to facilitate children’s de- ual, marital, parent–child, three-generational,
velopment and adaptation. and outside the family stressors) accounted
for a modest amount of unique variance in the
children’s academic and social competence.
Family Factors in Children’s
In combination, these multidomain risk as-
Adjustment Beyond Marital
sessments in the preschool period predicted
and Parenting Relationships
from 32 to 65% of the variance in children’s
Given space limitations, we do not focus on adaptation 1 and 2 years later.
all the domains of family life that can possi- Here we make two points based on the
bly affect children’s development and adapta- findings cited in this section. First, not all the
tion. Here we simply acknowledge the poten- variance in children’s adaptation can be attrib-
tial contributions to children’s adaptation from uted to marital and parent–child relationships.
domains outside marital and parent–child re- Other aspects of family life and life outside
lationships. First, there is evidence from cor- the nuclear family unit function as risk and
relational studies that both negative and posi- protective factors that are associated with
tive marital and parent–child relationships children’s development and psychopathology.
tend to be repeated across generations (Amato, Second, the correlations suggest that marital
1996; Cowan, Bradburn, & Cowan, in press). and parent–child relationships are two “ther-
742 P. A. Cowan and C. P. Cowan

mostats” in the family system that amplify or women becoming parents and mothers and
reduce the impact of external forces (e.g., children after divorce.
poverty), intergenerational patterns, individ- Our own review of the parenting interven-
ual psychopathology, and sibling relationships tion literature (Cowan et al., 1988) was some-
on children’s ability to meet the academic and what more optimistic. Like Weissberg and
social challenges of life outside the home. Bell, we noted methodological difficulties
This last statement, which presupposes a with the design of preventive intervention
causal role for marital and parent–child rela- studies, some of which had no systematic
tionships, is an assumption for which correla- evaluation component. Nevertheless, we found
tional studies cannot provide adequate sup- rays of hope, primarily in interventions ori-
port. We turn now to the question of whether ented toward early cognitive development and
dysfunctional marital and parent–child rela- in parenting interventions for children already
tionships can be considered as causal risk fac- identified as having behavioral or academic
tors that affect children’s development or problems (see below). In addition, some newer
whether they function as markers of dysfunc- parent-focused interventions published since
tion that are useful as predictors of later prob- Durlak and Wells’ meta-analysis and our own
lems but not necessarily as designated targets recent intervention studies provide some sup-
of intervention. We will see that intervention port for the idea that interventions with par-
designs offer partial answers but still leave ents can affect children’s academic and social
unanswered questions to be addressed in fu- adaptation.
ture research.
Parenting classes. Anyone with the time to do
an informal survey of bookstores, community
Parenting Interventions
college courses, community throw-away ad-
We first describe prevention and treatment in- vertisers, and on-line bulletin boards will be-
terventions that focus on the relationship be- come aware of a vast number of self-help
tween parents and children and then describe books and parenting classes in every major
prevention and treatment interventions that urban setting. Some focus on familiar “brand
focus on the relationship between the parents. names,” such as Parent Effectiveness Training
(Gordon, 1975) or STEP (Adams, 2001), while
others are more loosely described as parent
Prevention or promotion studies
education, or parent training. Unfortunately,
with a parenting focus
despite their apparent popularity, parenting
Durlak and Wells (1997) used meta-analysis interventions have rarely been evaluated sys-
to summarize the results of 177 primary pre- tematically. Most published studies do not
vention programs designed to reduce prob- have random assignment and rely entirely on
lems and promote competence in children and parents’ perspectives on their family function-
adolescents. In the context of this paper, it is ing before and after the intervention. Virtually
noteworthy that only 10 (5.6%) programs no studies of parenting classes measure child
were described as “parent training” and that outcomes independent of parents’ (usually
this was the only category of intervention that mothers’) reports. Gottman et al. (1997) de-
produced no statistically significant effects on scribed some pilot work on classes focusing
children. In their commentary on Durlak and specifically on helping parents become better
Wells’ paper, Weissberg and Bell (1997) meta-emotion coaches for their children, but
noted that the small number of such studies outcome results of this study are not yet avail-
and the lack of detail about their scope, de- able.
sign, and sample characteristics precluded
conclusions about the efficacy of parent train- Early interventions for high-risk mothers. In
ing interventions. The authors also included studies of high-risk populations, we find sys-
more successful interventions for parents in tematic information about successful interven-
another category called “life transitions”: tions with parents, again almost entirely from
Test of family systems theories 743

studies of mothers. Perhaps the best-known programs for parents were designed to reduce
early intervention programs using a random- this risk. Wolchik and her colleagues (Wol-
ized research design (Olds et al., 1998) pro- chik, West, Westover, & Sandler, 1993) eval-
vided home visiting nurses to low income uated a 12-session program by comparing
mothers from pregnancy and up to 2 years divorced mother and child intervention partic-
after their babies were born. Early follow-up ipants with waiting list controls. Noting that
results found advantages, not only for pro- there had been only two previous quasiexperi-
gram mothers compared with controls but mental evaluations of parent-based postdi-
also for the intellectual development of their vorce programs (Stolberg & Garrison, 1985;
children. Moreover, 15-year follow-ups com- Warren & Amara, 1984), neither of which re-
paring participants with controls revealed a sulted in positive effects on the children, Wol-
lower incidence of criminal behavior, school chik et al. (1993) designed a theory-based
suspensions, behavior problems, and use of program to target five potentially modifiable
substances in children of mothers who had risk factors that have been correlated with chil-
participated in the program. dren’s psychological adjustment. Ten group
In terms of children’s intellectual function- and two individual sessions were held for a
ing, Olds and Kitzman’s (1993) review of 15 total of 48 divorced mothers of 8- to 15-year-
randomized studies of home visiting programs olds, with the goal of improving (a) the qual-
for low income mothers found that only 6 pro- ity of the custodial parent–child relationship,
duced significant benefits for children. The (b) contact with the noncustodial parent, (c)
successful programs were more likely to be negative divorce-related events such as inter-
staffed by professionals and graduate students parental conflict, (d) support from noncusto-
rather than paraprofessionals. We should note dial, nonparent adults, and (e) parental disci-
that although Olds’ own program provides in- pline strategies. A control group composed of
tervention for parents (mothers), it focuses 46 mother–child pairs was randomly assigned
primarily on health-related behaviors rather to a waiting list condition.
than on the quality of the parent–child rela- There were significant effects of the inter-
tionship. vention on two of the five targeted variables.
Other home visiting programs have been At posttest, program participants were more
designed specifically to instruct low income accepting and warm in their interactions with
mothers in reading and language games to their children and reported fewer negative di-
play with their preschoolers (Levenstein, vorce-related events than controls, according
1988; Whitehurst, Arnold, Epstein, & Angell, to reports of both mothers and children. Moth-
1994). These programs also have measurable ers’ reports, though not children’s, showed sig-
effects on the children’s cognitive and lan- nificant effects of the intervention on disci-
guage skills, but it is not clear how long the pline. In general, the program benefited those
impact lasts. Both the Olds programs and the who were at highest risk at pretest. A Barron
reading programs show that an intervention and Kenny–type mediational analysis of inter-
with parents can have an impact on young vention to outcome effects was conducted in an
children’s intellectual and language skills. innovative way, using structural equation mod-
What is missing from this research, and from els. In these models, the direct path linking in-
most of the intervention research, is a design tervention with mothers’ report of total behav-
that seeks to establish the mechanisms by ior problems (outcome) was not statistically
which interventions have their positive effects. significant, but the indirect paths—intervention
→ improved quality of the mother–child rela-
Parenting interventions in divorced families. tionship and improved mother–child relation-
Parents’ divorce elevates the risk of academic, ship → child outcome—accounted for 43% of
externalizing, and internalizing problems in the reduction in behavior problems between
their children, at least over a period of ap- pretest and posttest.
proximately 2 years (Hetherington & Kelly, Another postdivorce program with positive
2002). Two successful preventive intervention intervention results was reported by Forgatch
744 P. A. Cowan and C. P. Cowan

and DeGarmo (1999). In an intervention de- substantial drop-off in positive effects over
sign with random assignment to experimental time. Behavioral treatments for externalizing
and control conditions, groups of recently di- problems appear to be more effective when
vorced mothers with sons in grades 1–3 met the treatment emphasizes parent–child rela-
weekly for 14 weeks. The behaviorally ori- tionships as well as behavioral contingency
ented intervention produced significant reduc- management principles (Miller & Prinz, 1990).
tions in coercive parenting and improvements The treatments are more effective for children
in positive parenting. Using structural equation younger than 7 than for those between 7 and
models, the investigators found that improve- 11, but there is some suggestion of more posi-
ment in parenting practices was associated with tive effects again in adolescence (Durlak &
improvements in children’s adjustment, as re- Wells, 1998). In most cases, parent-focused
ported by teachers, mothers, and the children treatment is contrasted with no treatment at
themselves. all or with waiting list controls. We are aware
Both of these prevention studies examined of only one study (Bank et al., 1991) that
a theoretically determined set of intervening compares parent training with more individu-
mechanisms and found statistically significant ally focused treatment. These investigators
effects, with intervention-induced changes in found that behavioral interventions with the
parenting explaining some of the variance in adolescents alone did not affect their aggres-
outcomes in the children. In both studies, sive behavior, but the addition of parent train-
some targeted variables were not affected by ing yielded significant positive effects. Within
the intervention and not all the affected vari- the context of individual therapy and other in-
ables functioned as mediators of positive out- terventions, we definitely need more studies
comes. What could we learn about the other of the valued-added impact of working with
variables that were hypothesized as risk fac- parents on their children’s development and
tors by these investigators? Are they not caus- adaptation.
ally related to child outcomes? Further re-
search is clearly necessary before we accept Parents and depressed children. Children and
the null hypothesis. Some potentially impor- adolescents of depressed parents are at risk
tant mechanisms implicated in children’s be- for depression, and the prevailing explanation
havior problems were not investigated, partic- is that parenting quality in these families is
ularly the quality of relationship between the often compromised (Downey & Coyne, 1990;
fathers and their children. Moreover, answers Gotlib & Goodman, 1999). Regardless of par-
to questions obtained in one study raise new ents’ depression, psychologically controlling
questions to be investigated in further re- and guilt-inducing parenting practices are as-
search. sociated with depression in both children and
adults. Given these facts, we would have ex-
pected to find that there were interventions for
Treatment studies with a parenting focus
parents of depressed children just as there are
Parenting style and parenting practices with interventions for parents of aggressive chil-
aggressive children. Over the past 30 years, a dren.
number of well-evaluated intervention studies What we find instead is that the major
have examined what happens when therapists treatment approaches to child and adolescent
work with parents, primarily mothers, to help depression involve individual therapy with the
them become more consistent, responsive, en- identified patient using a cognitive–behavior-
gaged, structured, and limit setting in order al, interpersonal, or psychodynamic approach,
to reduce their children’s aggressive behavior sometimes in conjunction with psychotropic
(e.g., Bank, Marlowe, Reid, & Patterson, 1991; medication (Garber, 2000; Weisz, Valeri,
Webster–Stratton, 1984; Wierson & Fore- McCarty, & Moore, 1999). An emerging ap-
hand, 1994). In general, these programs result proach to depression prevention involves
in posttreatment improvement in from half to school-based discussion groups with children
two-thirds of the children, although there is a and teenagers (Stark & Boswell, 2000). De-
Test of family systems theories 745

spite theoretical arguments that treatment of with 13% of the children with nondepressed
childhood depression is a family affair (Gott- mothers. Sixteen months later, the proportion
man, Ryan, Carrere, & Erley, 2002; Teich- of insecurely attached children of the non-
man, 1989), our impression is that even when treated depressed mothers remained high,
parents are seen “adjunctively” in connection whereas the proportion of insecurely attached
with their child’s treatment, there has been lit- children in the intervention condition had
tle separate evaluation of the effects of help- dropped to the level of the nondepressed con-
ing parents alter their parenting styles, prac- trols (Cicchetti et al., 1999). Furthermore,
tices, or approaches to coaching their children children with mothers in the intervention
when they are emotionally upset. We found group maintained their IQ scores over time,
two exceptions. A study by Lewinsohn and as did the control children with nondepressed
his colleagues (Lewinsohn, Clarke, Rohde, mothers, whereas children with untreated
Hops, & Seeley, 1996) found that parent mothers showed declining IQ scores over the
training added significantly to the effective- same period.
ness of a cognitive–behavioral intervention
for depressed adolescents. In addition, based Parenting as part of a multimodal prevention
on preliminary evidence, Stark and his col- or treatment package. In the past decade, it
leagues advocate both parent and family has become evident that serious childhood
involvement in the therapy for adolescent de- problems are not likely to be prevented or al-
pression (Stark et al., 1996). leviated by interventions that target a single
risk factor or family domain. Taking an ex-
Attachment-based interventions. Three pro- plicit or implicit lead from family systems
grams used an attachment framework to pro- theories (Wagner & Reiss, 1995) or Bronfen-
vide interventions to mothers at high risk brenner’s (1979) ecological model of devel-
because of poverty and socioeconomic cir- opment, investigators have designed large-
cumstances (Erickson, Korfmacher, & Ege- scale collaborative intervention studies with
land, 1992; Heinicke et al., 1996) or severe interventions in different contexts: the child
psychopathology in their infant (Lieberman & alone, the child with peers at school or sum-
Pawl, 1993). Although none of these pro- mer camp, the child in the classroom, and the
grams demonstrated a significant change in parents individually or in groups (Group,
the attachment status of parent or child, all 2002; Hinshaw et al., 2000; Reid, Eddy, Fe-
three revealed that the mother’s attachment to trow, & Stoolmiller, 1999). The examples all
the intervenor is an important factor in involve behavioral approaches to teaching
whether the intervention has a positive impact both the child and the parents.
on the mother, the mother–child relationship, Aggressive children. Reid and his col-
or the child. A fourth program, Toddler– leagues (Eddy, Reid, & Fetrow, 2000; Reid et
Parent Psychotherapy (TPP), was based on al., 1999) developed a “universal” preventive
the Infant Parent Program developed by intervention by targeting schools in high de-
Fraiberg (1975) and extended by Lieberman linquency areas, with the novel approach of
and Pawl (1993). This intervention for moth- including all the families with children attend-
ers who are depressed included weekly meet- ing those schools. By a random pairing of
ings for almost a year that focused on schools, they developed intervention and con-
mother–child interactions, the mother’s repre- trol samples of 671 first and fifth grade ele-
sentations, and helping her connect her own mentary school boys. Interventions included a
early experiences with her parents to her be- classroom-based social skills program, a play-
havior with her child (Cicchetti, Rogosch, & ground behavioral program, a school–parent
Toth, 2000; Cicchetti, Toth, & Rogosch, communication component, and 6-week groups
1999). At pretest, about 40% of 20-month-old for parents focused on behaviorally oriented
children with depressed mothers were inse- parenting skills. To date, the program has af-
curely attached in both the randomly assigned fected the targeted antecedents positively.
intervention and control groups, compared During the 3 years following the program,
746 P. A. Cowan and C. P. Cowan

participants showed a delay in becoming in- ruptive behavior at school, to the point that
volved with antisocial peers during middle intervention children were no longer signifi-
school, in using alcohol or marijuana, and in cantly different from their classmates.
first police arrest. There are several noteworthy aspects of
In an approach with similar components, these findings. First, although both the com-
the FAST Track Conduct Problems Preven- bined behavioral and medication treatments
tion Research Group (Conduct Problems Pre- and the behavioral treatment alone produced a
vention Group, 2002; McMahon & Slough, in significant reduction in negative or ineffective
press) targeted the most aggressive 10% of parenting, only in the combined treatment was
kindergarten children identified by both par- this reduction associated with children’s im-
ents’ and teachers’ descriptions. Here, too, provement in school. The authors argue that
there were components directed to the child
alone, with peers, and in the classroom and to the effects of pharmacologic intervention added to
the parents. This study, still ongoing, included behavioral therapy were at least partially explained
interventions not only at the beginning of the by psychological processes regarding the parent–
study but also in subsequent years. In both child relationship. Indeed the field is now begin-
these ambitious studies there were significant, ning to appreciate that interventions of a “biological”
although modest, effects of the intervention nature may well exert their effects via “psychologi-
cal” means and vice versa. (Hinshaw et al., 2000,
package on the children. The FAST Track
p. 566)
program has also documented some positive
effects on observed parent–child interaction.
However, at this point, neither study has ana- The authors also acknowledge that although
lyzed the potential connections between inter- the results provide strong support for a causal
vention effects and child outcomes and whether interpretation of parenting on ADHD symp-
changes in parent–child relationships could toms, a third variable such as changed paren-
be partially responsible for the improvements tal expectations could have produced change
observed in the children. in both the parents’ and children’s behavior
Children with Attention Deficit Hyperac- or that rapid changes in the child as a result
tivity disorder (ADHD). Hinshaw and col- of the medication could be responsible for the
leagues (2000) recently reported on a multi- change in parenting quality. A third point that
site, multimodal study of 579 children between the authors did not make but that is relevant
7 and 10 years of age who were diagnosed to our analysis is that when results show im-
with ADHD. They compared typical commu- provements in the parent–child relationship
nity treatment (the control condition) with a accompanied by improvements in children’s
medication management condition, an inten- behavior, they do not implicate ineffective
sive behavior therapy condition, and an inter- parenting as a root cause in the etiology of
vention that combined behavior therapy and inattention and hyperactivity. Rather, they in-
medication management. The behavior ther- dicate that parents can play a causal role in
apy was itself a psychosocial “package” that changing the symptoms of ADHD, regardless
included interventions with the child alone of how these symptoms originate.
and in a summer camp, with his or her teach-
ers in school, and with the parents, both indi-
Marital Interventions
vidually and in groups. Using growth curve
analyses, Hinshaw et al. showed that interven- It seems obvious that if we want to enhance a
tion-induced improvements in negative or in- child’s development or prevent or treat child
effective discipline mediated improvements in psychopathology, we should focus some as-
children’s social skills at school when the be- pects of an intervention for parents on parent-
havioral interventions were combined with ing. What is apparently less obvious to chil-
medication treatment. In that condition alone, dren’s well-being is the idea of an intervention
a reduction in negative discipline by parents focused on helping mothers and fathers make
was followed by a reduction in children’s dis- their relationship as a couple less troubled and
Test of family systems theories 747

more satisfying. Yet, the correlational results style following the discussion of conflict.
summarized above are consistent with family Family warmth and cohesion were improved
system theories in suggesting that increases in after the couple’s discussion of an enjoyable
parents’ ability to cooperate as partners and topic, but only in families in which mothers
feel more satisfied with their relationship as a reported high marital satisfaction. No specific
couple or to decrease their patterns of unregu- child outcomes were examined in this study.
lated conflict or icy withdrawal could have
important benefits for their children’s well- Adding marital therapy
being. to parenting interventions
There are many studies of couples or mari-
Three different sets of investigators who be-
tal therapy that reveal that, compared to cou-
gan with parent-training programs for mothers
ples with no treatment, couples who take ad-
of young aggressive children noted that the
vantage of various preventive interventions
treatment was not as effective when the moth-
(marital preparation or enhancement) and var-
ers were in high-conflict or unsatisfying rela-
ious types of marital therapy receive measur-
tionships with their husbands. For example, in
able benefits in relationship satisfaction and
two small studies (Dadds, Schwartz, & Sand-
positive change in their behavior patterns
ers, 1987; Griest, 1982), adding a marital
(Gottman et al., 2002; Shadish, Ragsdale,
communication segment to the regular behav-
Glaser, & Montgomery, 1995; Silliman &
ior management program for parents of ag-
Schumm, 2000; Stanley, Blumberg, & Mark-
gressive boys resulted in significantly reduced
man, 1999). Because most of these studies fo-
aggression in the children, especially in the
cus on couples 20–40 years of age, it is likely
families with maritally distressed parents. A
that a substantial number of them are parents,
larger study by Webster–Stratton (1994) com-
although this is not always mentioned. Al-
pared her standard parenting program with a
though correlational studies would lead us to
new program that included an advanced unit
expect that the children benefited from their
focusing on the parents’ relationship as a cou-
parents’ improved relationships, none of the
ple. Although both interventions produced
treatment studies of which we are aware as-
significant reductions in children’s aggression,
sessed the impact of the parents’ shifts on
the advanced unit produced additional signifi-
their children. We focus next on the few stud-
cant improvements in parents’ communication,
ies we found, including one that we con-
problem-solving skills, and satisfaction, as well
ducted, that examined the impact of a couple-
as increases in children’s knowledge about pro-
focused intervention on marital and parenting
social solutions to social problems.
quality and on the child’s adaptation.
These programs all examined the value-
added impact of a marital intervention, over
A short-term laboratory marital intervention and above a parenting intervention. We know
of no studies that added a parenting interven-
An innovative study by Kitzmann (2000) il-
tion to a marital intervention. However, in a
lustrates what can be done within a labora-
small study of 24 participants, Brody and
tory–experimental framework to test causal
Forehand (1985) assessed changes in marital
theories about family relationships with an in-
satisfaction during a parent-training interven-
tervention. Forty married couples with 6- to
tion. They found that the impact of parent
8-year-old sons engaged in two discussions of
training on marriage made a troubled mar-
marital issues (in counterbalanced order), fol-
riage better but it did not make a satisfying
lowed immediately by an interaction as a cou-
marriage better.
ple with their child. In one discussion the cou-
ple was asked to discuss an enjoyable topic
and in the other to discuss a disagreement. Fa- A couples group intervention for parents
thers showed significantly less engagement or of young children
support to their sons. Couples tended to shift In a recently completed monograph, we and
toward a less mutually supportive coparenting our colleagues (Cowan, Cowan, Ablow, John-
748 P. A. Cowan and C. P. Cowan

son, & Measelle, in press) report on a longitu- Berkeley Puppet Interview (Measelle, Ablow,
dinal preventive intervention study of families Cowan, & Cowan, 1998) and their academic
with children making the transition to elemen- achievement was tested with the Peabody In-
tary school. From preschools and child care dividual Achievement Test–Revised (Mark-
centers in 27 communities in northern Califor- wardt, 1989).
nia, we recruited 100 families whose first When group couples were compared with
child was 4 years old. A randomized clinical consultation controls, there were significant
trial design was used to assign couples during positive effects of both the maritally focused
the year before their child entered kindergar- and parenting-focused group interventions.
ten to (a) maritally focused couples groups Parents in the maritally focused groups were
meeting weekly for 16 weeks, (b) parenting- observed to fight less in front of their children
focused couples groups meeting weekly for in the year after the groups ended. Parents in
16 weeks, or ) a consultation control in which the parenting-focused groups were observed
parents were offered one consultation meeting to be warmer and more structuring (more au-
each year for 3 years with the mental health thoritative) than they were earlier and more
professional couples who led the couples authoritative than the controls. Furthermore,
groups. Discussions in the two types of groups the children whose parents were in the groups
were similar, but in the open-ended, unstruc- fared better than children of controls in adapt-
tured portion of each evening, staff leaders fo- ing to kindergarten and first grade (higher ac-
cused more on the marital issues in some ademic achievement, less aggression and de-
groups and more on parenting issues in other pression). We were able to show that the
groups. Even when the content was the same intervention affected the theorized media-
(a problem in discipline), the leaders in the tors—marital and parenting quality—and that
marital groups helped the couple explore con- changes in these mediators were associated
flicts that the issue raised between them with how the children were faring 1 and 2
whereas leaders in the parenting groups ex- years later.
plored alternative ways to modify the child’s How large are the effects? Comparisons of
behavior. structural equation models that excluded and
The families were assessed with question- then included the intervention as a variable
naires, videotaped couple interaction sessions, (in effect, an estimate of the increase in R2)
and family interaction sessions before the showed that the intervention groups ac-
children entered kindergarten and then fol- counted for an added 12% of the variance in
lowed again when the children were in kin- children’s academic achievement scores and
dergarten and first grade. Outcome data con- an additional 21% of the variance in their self-
cerning externalizing and internalizing behavior reported adjustment in kindergarten. In first
in the classroom were collected by having grade, the intervention accounted for an addi-
teachers describe each child in his or her class tional 31% of the variance in teachers’ reports
on the Child Adaptive Behavior Inventory of externalizing and 12% of the variance in
(Cowan, Cowan, & Heming, 1995), a 106- internalizing. Our more recent, as yet unpub-
item checklist of adaptive and maladaptive lished, analyses of the effects of the preschool
behaviors adapted from a number of sources, intervention reveal that these positive benefits
including Achenbach and Edelbrock (1983) extend to the children’s academic achieve-
and Schaefer and Hunter (1983). Teachers did ment, externalizing, and internalizing behav-
not know which child in their class was in our ior in fourth grade when they are 9 to 10 years
study. Their ratings of more than 1400 chil- of age.
dren in 80 classrooms revealed that the chil- This study shows that, beyond its status as
dren in our study were no different from their a predictor of academic and behavior prob-
classmates in mean scores or standard devia- lems in kindergarten and first grade, marital
tions in aggression, hyperactivity, depression, conflict as we observed it in the preschool pe-
or somaticizing. Children’s self-reported ad- riod when the child was present was a causal
justment was assessed with items from the risk factor for a number of indices of chil-
Test of family systems theories 749

dren’s subsequent adaptation to school. Simi- alliance between the parents, we are aware of
larly, low levels of authoritative parenting in no specific evaluations of whether this alli-
mother–child and father–child interaction ance actually changes or whether changes in
functioned as a causal risk factor for chil- the parents’ relationship are connected with
dren’s later adaptation. Specific links between the positive outcomes of the therapy for the
marital and parent–child relationships and children.
children’s outcomes depended on which out-
come measure we focused on and when the
What We’ve Learned
outcome was measured. These results suggest
and Where We Need to Go
that, even with the precision of a randomized
clinical trial in which one domain of family
Signs of progress
life was highlighted by the group leaders,
principles of equifinality and multifinality op- In our view, important progress has been
erate, meaning that there is no lockstep, one made over the last three decades in the corre-
to one correspondence among the targets of lational study of linkages between family rela-
the intervention, the family processes af- tionships and children’s development. Espe-
fected, and the specific child outcomes that cially in the last decade, intervention studies
follow. have helped the field draw some conclusions
An interesting pattern of findings revealed about the causal forces operating from family
in the differential impact of the marital and processes to child and adolescent outcomes.
parenting interventions has important implica- The first and most simple, but clearest, mes-
tions for our understanding of the direction of sage from intervention studies calls Harris’
effects at transition times, when family rela- (1998) skepticism about parent–child rela-
tionships are shifting. When the men and tionships into question: parents do indeed
women in the intervention emphasizing par- have effects on their children and adolescents,
enting issues showed improved parenting not only in the family but also at school and
skills, we observed no comparable improve- in their peer group. Parenting intervention
ment in their marital interaction. By contrast, studies not only buttress claims about parents’
when couples in the groups emphasizing mar- influence on their children’s development and
ital issues reduced their conflict in front of adaptation, but they also provide specific in-
the children, their parenting style also became formation about particular parent behaviors
more effective. If this finding were to be rep- that could be targeted to reduce coercive cy-
licated in different intervention studies, we cles and negative or ineffective parenting and
could conclude that the tone of the relation- increase parents’ responsiveness and setting
ship between the parents has a more powerful of developmentally appropriate limits with
effect on the parent–child relationships than their children.
parenting quality does on the marital relation- A second conclusion, more tentative be-
ship. cause of the relatively few systematic studies,
is that a negative relationship between the
parents should be considered a causal risk fac-
Family therapy and children’s outcomes
tor in children’s adjustment, because of its di-
For the sake of completeness, we should men- rect impact on children and its indirect impact
tion the fact that family therapy interventions on other family relationships. This conclusion,
have been shown to be effective for a variety which is supported by the correlational stud-
of diagnostic conditions in children, including ies of marital conflict and child or adolescent
antisocial behavior (Miller & Prinz, 1990), outcomes, awaits results of further interven-
drug abuse (Liddle et al., 2001), and schizo- tion studies.
phrenia (Hahlweg & Goldstein, 1987). Al- A third conclusion is more subtle. An often
though the clinical family theories on which unspoken subtext of claims suggests that if con-
many of these interventions are based empha- nections between family characteristics and
size the importance of shoring up the working children’s development are attributable to ge-
750 P. A. Cowan and C. P. Cowan

netic or biological factors, then children’s mal- in the structure and process of interventions
adaptive behaviors are not subject to psycho- are based on theories about how individuals
social influence. Results of intervention studies and relationships change. However, the links
clearly refute this assertion. between developmental theory and interven-
Fourth, growth curves and hierarchical lin- tion designs are often intuitive and implicit.
ear modeling (Burchinal, Bailey, & Snyder, We believe that a more explicit consideration
1994; Stoolmiller, Duncan, Bank, & Patter- of developmental theories can enrich the de-
son, 1993) represent new and powerful meth- sign of interventions and make the results more
ods of assessing change and have been used, useful and generalizable; deriving principles
often in error, to infer causal connections concerning the mechanisms that link risks and
from longitudinal data (Raudenbush, 2000). outcomes makes it possible to focus the in-
Used in conjunction with an intervention de- tervention on modifiable risks with the best
sign, the analysis of individual growth curves chance of inducing positive effects.
increases the power of what are often studies Suggestion 2 is to increase the variety of
with small numbers of subjects to detect inter- theories to be tested. Most of the well-controlled
vention effects. systematic studies published so far use a be-
The outcome of intervention studies is par- havioral or cognitive–behavioral theoretical
ticularly important in the context of contem- frame to guide the intervention. This is true
porary policy debates. Social science data are of evaluations of both preventive interven-
immediately relevant to policies only as far tions (Durlak & Wells, 1997, 1998) and child
as they convey information about causal risks and adolescent psychotherapy (Weisz, Weiss,
related to the outcomes of concern. For exam- Han, & Granger, 1995). More well-designed
ple, as we write, there is a national debate on studies are needed of interventions based on
government encouragement of marriage in low frameworks emphasizing attachment, parent-
income populations (Toner, 2002). The pro- ing style, and meta-emotion and those that tar-
posal to encourage men and women to marry, get specific behaviors that the correlational
especially if they are parents, is explicitly studies suggest as causal risk factors for de-
based on correlational data showing that chil- velopmental psychopathology. Encouraging
dren in two-parent families fare better than signs of a move in this direction come from
those growing up with a single parent (Waite the attachment-based intervention studies we
& Gallagher, 2000). The correlational and in- described (Cicchetti et al., 1999; Erickson et
tervention data summarized here suggest that al., 1992; Heinicke et al., 1996; Lieberman &
the situation is more complex: While satisfy- Pawl, 1993). More effort could be devoted to
ing marriages—in which partners are able to combining biological and psychosocial inter-
convey their feelings and solve their problems ventions, as the Hinshaw et al., project (2000)
productively—can contribute to children’s did, especially when the disorder is conceptu-
well-being, marriages with high levels of un- alized as having both biological and psycho-
resolved conflict between parents place chil- social components.
dren at risk for a variety of undesirable out- Suggestion 3 is to fill in the gaps between
comes. data from correlational studies and evidence
from intervention studies. Some of the gaps
between family theory and intervention find-
Directions for future research
ings are attributable to the fact that the theo-
On the basis of this review, we offer six sug- ries are now more complex than our statistical
gestions for intervention research that aspires techniques can handle. Real families assessed
to be relevant to theories of development and at one point in time are complex; changing
psychopathology. families, in which individual and relationship
Suggestion 1 is to make the links between adaptation can shift in different directions, are
theories of development and theories of inter- even more so. The regressions or structural
vention more explicit. We have simply as- equation models that we use to assess the con-
serted in this paper that the choices involved nections among different domains of family
Test of family systems theories 751

life simply cannot handle the number of inter- are more likely to have single mothers as head
acting variables that may be playing important of the household in which children live. Re-
roles in determining the pathways from child- cent studies make it clear that a majority of
hood to adulthood and old age. Family sys- single mothers of young children are either
tems theories are dynamic and circular, with living with or involved with a romantic part-
each “effect” becoming a cause in a new cas- ner, often the child’s father (Harknett, Hard-
cade of processes and events, but analytic man, Garfinkel, & McLanahan, 2001), and
models tend to be static and unidirectional. many others are in coparenting relationships
This discrepancy between real life and our with other adults such as their own mothers,
models of real life is endemic in any area in fathers, grandparents, and friends (Burton,
which basic and applied research address sim- 1995; Harknett et al., 2001). Continuing to fo-
ilar questions. As scientists, we face a “Goldi- cus only on mothers in both research and in-
locks” task of reducing the complexity of our tervention studies risks a return to the 1940s
theoretical models in order to test them—and and 1950s when mothers were blamed for
we must do so, not too much, not too little, their child’s psychopathology, mothers but
but just right. not fathers were seen when their child was in
It seems like a daunting task to mount a therapy, and, with the exception of studies of
large-scale intervention study in order to test father absence, the role of fathers was not in-
one or two hypotheses about the connections vestigated in any systematic way.
between family relationships and children’s Suggestion 5 is to pay more attention to
development. A relatively overlooked strategy mediating and moderating mechanisms. Cen-
is that suggested by the Kitzmann (2000) tral to formulations at the heart of develop-
study described briefly above. After identify- mental psychopathology is the notion that
ing a hypothesized mechanism in correlational three possible types of mechanisms link risks
studies (in this case, the emotional tone of a with outcomes: (a) those that unfold so that
husband–wife discussion), one can conduct a initial risks lead to dysfunction and distress;
short-term laboratory experiment to determine (b) those that buffer the child in ways that
whether systematic manipulation of the puta- prevent risks from having a negative impact;
tive causal variable makes a difference to the and (c) those that amplify the impact of risks
subsequent emotional tone of family interac- or create vulnerabilities, so that even low lev-
tion. Small scale parenting interventions can els of risk result in negative outcomes. Thus
be tested in similar fashion. We are not ar- far, correlational studies of family–child con-
guing that a very brief intervention in a labo- nections have tended to focus on type (a)
ratory is sufficient to alter long-standing fam- mechanisms, for example, negative cycles re-
ily processes (although it might be in some peating across generations or consistencies
cases) but rather, that small, short-term stud- and spillover across family domains. However,
ies might provide increased support for causal as we know, the correlation between risk and
hypotheses that could then be tested further in outcome is far from perfect. A few models that
large-scale interventions. we described above specify the moderator ef-
Suggestion 4 is to routinely include fathers fects of buffers or vulnerabilities and reveal
in family studies. The single, most frequent that the effects are different for boys versus
omission in family-based intervention studies girls, for authoritative versus authoritarian par-
is a focus on mothers and children that ex- ents, and for maritally satisfied versus unsatis-
cludes fathers. Studies of parenting and mar- fied spouses. In the few intervention studies
riage, using both correlational and interven- that focus on mediators, the investigators have
tion designs, indicate that the quality of the not yet begun to identifying moderators that
relationship between the parents affects par- would illuminate who is most likely to benefit
ent–child relationships and children’s devel- and who might be vulnerable to harm from
opment in both traditional and nontraditional this intervention. Some studies, including ours,
family arrangements. Researchers and service seem to suggest that the marital relationship
providers often argue that high-risk families has more influence on the parent–child rela-
752 P. A. Cowan and C. P. Cowan

tionship than vice versa, but this conclusion will probably be in Piaget’s Preoperational
remains to be tested in a variety of contexts stage and the older one in his Concrete opera-
(e.g., in clinical and nonclinical high-risk and tional stage.
low-risk samples). We urge researchers to Although some argue that depression may
choose mediators and moderators to test on look different in a 5-year-old, a 15-year-old,
the basis of the theory associated with the in- and a 25 year-old (Kaufman, Martin, King, &
tervention. Unless this happens, all we have Charney, 2001), current practice has been to
is a random collection of variables that seem use the same diagnostic criteria for child, ado-
to make a difference but we do not know how lescent, and adult depression. In an evolving
or why. set of conceptual papers, Cicchetti and his col-
Suggestion 6 is to place more emphasis on leagues (Cicchetti & Schneider–Rosen, 1984;
assessing developmental outcomes. In our view, Cicchetti & Toth, 1995, 1998) presented a de-
there has been a curious underemphasis on velopmental perspective on depression. They
development in the field of developmental describe a transactional model in which reor-
psychopathology. Certainly, there has been an ganizations of biological, psychological, and
emphasis on change and on following individ- social processes intersect as individuals at-
uals over time to chart risk–outcome pathways. tempt to accomplish specific developmental
However, as the focus shifts to outcomes, de- tasks (e.g., the formation of attachment rela-
velopmental constructs are often left behind. tionships, the emergence of self-identity). That
Studies of prevention and treatment of chil- is, stage-salient developmental issues present
dren almost all adopt one of two kinds of out- challenges. Depending on past and current de-
come measures. A few use the diagnostic cat- velopmental strengths and vulnerabilities in
egories of DSM (II, III, or IV) and ask whether biopsychosocial domains, individuals move
the intervention has a clinically significant ef- toward or away from depressive states. In
fect. The question is whether a preventive in- sum, depression as a phenomenon is not de-
tervention results in a lower incidence of case- scribed in developmental terms; but, in Cic-
ness in an at-risk sample or whether treatment chetti and colleagues’ view, it is essential to
results in a change from a clinical to a non- understand the etiology, onset, and course of
clinical diagnosis of depression or Oppositional depression in terms of the trajectory by which
Defiant disorder. With a very few exceptions the individual arrives at this particular phase
(e.g., Oppositional Defiant Disorder and Anti- of the life cycle.
social Personality), the categories for children There are some implicit developmental
and adults are similar or identical. We are not distinctions made in the DSM, in that Opposi-
aware of studies of developmental psycho- tional Defiant Disorder is a category applied
pathology that chronicle systematic changes to younger children and Antisocial Personal-
within each diagnostic category that might be ity is not used until the teen years. Generally,
associated with age or developmental stage. though, the widespread use of continuous
Most studies evaluate the impact of the in- quantitative measures typified by the Child
tervention on continuous measures of achieve- Behavior Check List (Achenbach & Edel-
ment, externalizing, or internalizing, with out- brock, 1983) means that we know very little
come measures normed with reference to a about whether and how the form of externaliz-
given population. That is, intervention studies ing disorders changes over the life span. Simi-
are concerned with measures of adaptation, lar to current conceptualizations of depression,
with the yardstick being explicit or implicit there are some indications that externalizing
comparisons of the child with his or her peers. symptoms or disorders are products of the un-
The idea that measures of adaptation are usu- folding of biological, psychological, and social
ally “development free” can most easily be development over time.
seen in studies of cognitive functioning. The In sum, we do not yet have a useful devel-
average 5-year-old and the average 10-year- opmental taxonomy of achievement, internal-
old will get the same score on an achievement izing, or externalizing outcomes for children
test (50th percentile), but the younger child or adolescents, but there are some intriguing
Test of family systems theories 753

conceptions of how a developmental analysis Although our emphasis here has been on
could help us to understand the emergence of family-based interventions, we do not mean
competence or psychopathology. Develop- to argue that family intervention is the only
mental psychopathologists (see Cicchetti & possible alternative when we are concerned
Cohen, 1995a, 1995b) write about the need to about children. Correlational data show clearly
pay attention to the developmental stage of that there are many risks outside the family
the client and to match the nature of the inter- that could affect children and reasonably be
vention to the child or adolescent’s level of the targets of intervention, especially in the
understanding (regardless of chronological age). peer group (Hinshaw et al., 2000; McMahon
One example of this approach can be seen in & Slough, in press) and at school (Perry &
a chapter by Noam (1998), who combined an Weinstein, 1998). Moreover, we are aware
ego–psychological developmental theory (Loev- that family-based interventions do not always
inger, 1997) with a Piaget–Kohlberg stage– work. Despite the rather optimistic notion
developmental theory (Kohlberg & Kramer, from family systems theory that positive
1969) to illustrate how different schools of change somewhere in the system will have re-
therapy could address specific behavior prob- verberating positive effects in other parts, in
lems differently, depending on the child’s fact, family relationships are difficult to
level of cognitive, social, and emotional de- change, and even when they do, children do
velopment. Other examples can be seen in de- not always show the improvements we hope
scriptions of a developmental stage approach for (Durlak & Wells, 1997).
to child therapy (Shirk & Russell, 1996; Toth Intervention studies are not always possi-
& Cicchetti, 1999; Weisz et al., 1999). A more ble to conduct. Some aspects of family life,
concerted effort to conceptualize psychopathol- especially maltreatment, abuse, or harsh phys-
ogy in developmental terms would require the ical punishment, cannot be examined in ran-
creation of new instruments to assess outcomes domized interventions. In the marital realm,
in terms of developmental levels. The advan- too, there are ethical limitations on the extent
tage of this or other developmental approaches to which theories about marital function and
is that it could help to bring together develop- children’s adaptation can be tested using ran-
mental psychopathology’s long-expressed in- domly controlled intervention designs. For
terest in the connections between normal and these aspects of parenting and marital rela-
pathological development within the context tionships, careful correlational studies will
of intervention studies. have to suffice.

Limitations of family-based A Final Word

intervention studies
The time has come to design family-based in-
Although intervention designs are the gold tervention studies that go beyond the simple
standard for making inferences about causal- question of “does the intervention work?” or
ity, those inferences are always open to chal- the more complex question of “what kind of
lenge. As Hinshaw and colleagues (2000) intervention works for particular people or
pointed out clearly, there are always unmeas- families, under what specified conditions?”
ured confounded factors that could account We join a growing chorus that advocates the
for the results and it is always possible that use of theory and existing research to focus
the order of effects implied by the interven- on specific intervention targets within the
tion is in error. As we have noted, interven- family (e.g., emotion regulation in marital and
tion studies have shown that changes in mari- parent–child relationships) and the inclusion
tal and parenting relationships can affect child of measures to test whether change in the hy-
outcomes. This fact provides support for theo- pothesized mediator accounts for variation in
ries of intervention and change, but not for the intervention outcome. This strategy will
theories about parents’ role in the etiology of have two important payoffs: it will contribute
normal development and psychopathology. to a more differentiated understanding of how
754 P. A. Cowan and C. P. Cowan

family factors affect children’s and adoles- clinicians in the design of more effective pre-
cents’ development and it will help to guide ventive and therapeutic interventions.

References
Achenbach, T. M., & Edelbrock, C. S. (1983). Manual Borkowski, J. G., Ramey, S., & Bristol–Power, M.
for the Child Behavior Checklist. Burlington, VT: (Eds.). (2002). Parenting and the child’s world: Influ-
Queen City. ences on intellectual, academic, and social–emo-
Adams, J. F. (2001). Impact of parent training on family tional development. Mahwah, NJ: Erlbaum.
functioning. Child & Family Behavior Therapy, 23, Bowlby, J. (1980). Attachment and loss (Vol. 3). New
29–42. York: Basic Books.
Ainsworth, M. D. S., & Wittig, B. A. (1969). Attachment Brody, G. H., & Forehand, R. (1985). The efficacy of
and exploratory behavior of one-year-olds in a strange parent training with maritally distressed and nondis-
situation. In B. M. Foss (Ed.), Determinants of infant tressed mothers: A multimethod assessment. Behav-
behavior (Vol. IV, pp. 113–136). London: Methuen. iour Research & Therapy, 23, 291–296.
Amato, P. R. B. (1996). A prospective study of divorce Bronfenbrenner, U. (1979). The ecology of human devel-
and parent–child relationships. Journal of Marriage opment: Experiments by nature and design. Cam-
and the Family, 58, 356–365. bridge, MA: Harvard University Press.
Bank, L., Marlowe, J. H., Reid, J. B., & Patterson, G. R. Bronstein, P., & Cowan, C. P. (Eds.). (1988). Fatherhood
(1991). A comparative evaluation of parent-training today: Men’s changing role in the family. New York:
interventions for families of chronic delinquents. Wiley.
Journal of Abnormal Child Psychology, 19, 15–33. Burchinal, M. R., Bailey, D. B., & Snyder, P. (1994).
Barber, B. K., & Harmon, E. L. (2002). Violating the Using growth curve analysis to evaluate child change
self: Parental psychological control of children and in longitudinal investigations. Journal of Early Inter-
adolescents. In B. K. Barber (Ed.), Intrusive parent- vention, 18, 403–423.
ing: How psychological control affects children and Burton, L. M. (1995). Intergenerational patterns of pro-
adolescents (pp. 15–52). Washington, DC: American viding care in African–American families with teen-
Psychological Association. age childbearers: Emergent patterns in an ethno-
Bates, J. E. (2001). Adjustment style in childhood as a graphic study. In V. L. Bengtson, K. W. Schaie, &
product of parenting and temperament. In T. D. L. M. Burton (Eds.), Adult intergenerational rela-
Wachs & G. A. Kohnstamm (Eds.), Temperament in tions: Effects of societal change (pp. 79–125). New
context (pp. 173–200). Mahwah, NJ: Erlbaum. York: Springer.
Bates, J. E., Pettit, G. S., & Dodge, K. A. (1995). Family Carlson, E. A. (1998). A prospective longitudinal study
and child factors in stability and change in children’s of attachment disorganization/disorientation. Child
aggressiveness in elementary school. In J. McCord Development, 69, 1107–1128.
(Ed.), Coercion and punishment in long-term per- Carlson, E. A., Sroufe, L. A., Collins, W. A., Jimerson,
spectives (pp. 124–138). New York: Cambridge Uni- S., Weinfield, N., Henninghausen, K., Egeland, B.,
versity Press. Hyson, D., Anderson, F., & Meyer, S. (1999). Early
Baumrind, D. (1971). Current patterns of parental author- environmental support and elementary school adjust-
ity. Developmental Psychology Monographs, 4, 1– ment as predictors of school adjustment in middle ad-
103. olescence. Journal of Adolescent Research, 14,
Baumrind, D. (1980). New directions in socialization re- 72–94.
search. American Psychologist, 35, 639–652. Cassidy, J. (1995). Attachment and generalized anxiety
Baumrind, D. (1989). Rearing competent children. In W. disorder. In D. Cicchetti & S. L. Toth (Eds.), Emo-
Damon (Ed.), Child development today and tomorrow tion, cognition, and representation (pp. 343–370).
(pp. 349–378). San Francisco, CA: Jossey–Bass. Rochester, NY: University of Rochester Press.
Baumrind, D. (1991). Effective parenting during the early Cherlin, A. J., Chase–Lansdale, P. L., & McRae, C.
adolescent transition. In P. A. Cowan & E. M. Heth- (1998). Effects of parental divorce on mental health
erington (Eds.), Advances in family research (Vol. 2, throughout the life course. American Sociological Re-
pp. 111–163). Hillsdale, NJ: Erlbaum. view, 63, 239–249.
Baumrind, D., Larzelere, R. E., & Cowan, P. A. (2002). Cicchetti, D., & Carlson, V. (1989). Child maltreatment:
Ordinary physical punishment—Is it harmful? Com- Theory and research on the causes and consequences
ment on Gershoff. Psychological Bulletin, 128, 580– of child abuse and neglect. New York: Cambridge
589. University Press.
Belsky, J. (1999). Interactional and contextual determi- Cicchetti, D., & Cohen, D. J. (Eds.). (1995a). Develop-
nants of attachment security. In J. Cassidy & P. R. mental psychopathology: Vol. 1. Theory and methods.
Shaver (Eds.), Handbook of attachment: Theory, re- New York: Wiley.
search, and clinical applications (pp. 249–264). New Cicchetti, D., & Cohen, D. J. (Eds.). (1995b). Develop-
York: Guilford Press. mental psychopathology: Vol. 2. Risk, disorder, and
Block, J., Block, J. H., & Gjerde, P. F. (1989). Parental adaptation. New York: Wiley.
functioning and the home environment in families of Cicchetti, D., & Rogosch, F. A. (1996). Equifinality and
divorce: Prospective and concurrent analyses. In S. multifinality in developmental psychopathology. De-
Chess & M. E. Hertzig (Eds.), Annual progress in velopment and Psychopathology, 8, 597–600.
child psychiatry and child development (pp. 192– Cicchetti, D., Rogosch, F. A., & Toth, S. L. (2000). The
207). New York: Brunner/Mazel. efficacy of toddler–parent psychotherapy for foster-
Test of family systems theories 755

ing cognitive development in offspring. Journal of ences on academic, intellectual, and social–emotional
Abnormal Child Psychology, 28, 135–148. development (pp. 75–97). Mahwah, NJ: Erlbaum.
Cicchetti, D., & Schneider–Rosen, K. (1984). Toward a Cowan, P. A., & Cowan, C. P. (in press). “Mega-models”
transactional model of childhood depression. New Di- of parenting in context. In P. A. Cowan, C. P. Cowan,
rections for Child Development, 26, 5–27. J. Ablow, V. K. Johnson, & J. Measelle (Eds.), The
Cicchetti, D., & Toth, S. L. (1995). Developmental psy- family context of parenting in children’s adaptation
chopathology and disorders of affect. In D. Cic- to elementary school. Mahwah, NJ: Erlbaum.
chetti & D. J. Cohen (Eds.), Developmental psycho- Cowan, P. A., Cowan, C. P., Ablow, J., Johnson, V. K.,
pathology: Vol. 2. Risk, disorder, and adaptation (pp. & Measelle, J. (Eds.). (in press). The family context
369–420). New York: Wiley. of parenting in children’s adaptation to school. Mah-
Cicchetti, D., & Toth, S. L. (1998). The development of wah, NJ: Erlbaum.
depression in children and adolescents. American Cowan, P. A., Cowan, C. P., & Heming, G. (1995). Man-
Psychologist, 53, 221–241. ual for the Child Adaptive Behavior Inventory (CABI).
Cicchetti, D., Toth, S. L., & Maughan, A. (2000). An Unpublished manuscript, University of California,
ecological–transactional model of child maltreatment. Berkeley.
In A. J. Sameroff, M. Lewis, & S. M. Miller (Eds.), Cowan, P. A., Cowan, C. P., & Schulz, M. S. (1996).
Handbook of developmental psychopathology (2nd Thinking about risk and resilience in families. In E. M.
ed., pp. 689–722). New York: Kluwer Academic/Ple- Hetherington & E. A. Blechman (Eds.), Stress, cop-
num Press. ing, and resiliency in children and families: Family
Cicchetti, D., Toth, S. L., & Rogosch, F. A. (1999). The research consortium: Advances in family research
efficacy of toddler–parent psychotherapy to increase (Vol. 5, pp. 1–38). Hillsdale, NJ: Erlbaum.
attachment security in offspring of depressed mothers. Cowan, P. A., Cowan, C. P., Schulz, M. S., & Heming,
Attachment & Human Development, 1, 34–66. G. (1994). Prebirth to preschool family factors in chil-
Coie, J. D., Watt, N. F., West, S. G., & Hawkins, J. D. dren’s adaptation to kindergarten. In R. D. Parke &
(1993). The science of prevention: A conceptual S. G. Kellam (Eds.), Exploring family relationships
framework and some directions for a national re- with other social contexts: Family research consor-
search program. American Psychologist, 48, 1013– tium: Advances in family research (Vol. 4, pp. 75–
1022. 114). Hillsdale, NJ: Erlbaum.
Collins, W. A., Maccoby, E. E., Steinberg, L., Hethering- Cowan, P. A., Powell, D., & Cowan, C. P. (1998). Par-
ton, E. M., & Bornstein, M. H. (2000). Contemporary enting interventions: A family systems perspective. In
research on parenting: The case for nature and nur- W. Damon (Ed.), Handbook of child psychology (5th
ture. American Psychologist, 55, 218–232. ed., Vol. 4, pp. 3–72). New York: Wiley.
Conduct Problems Prevention Research Group. (2002). Cox, M. J., Paley, B., & Harter, K. (2001). Interparental
Evaluation of the first 3 years of the Fast Track pre- conflict and parent–child relationships. In J. H. Grych
vention trial with children at high risk for adolescent & F. D. Fincham (Eds.), Interparental conflict and
conduct problems. Journal of Abnormal Child Psy- child development: Theory, research, and applica-
chology, 30, 19–35. tions (pp. 249–272). New York: Cambridge Univer-
Conger, R. D., Conger, K. J., Elder, G. H., & Lorenz, sity Press.
F. O. (1992). A family process model of economic Cummings, E. M., & Davies, P. (1994). Children and
hardship and adjustment of early adolescent boys. marital conflict: The impact of family dispute and res-
Child Development, 63, 526–541. olution. New York: Guilford Press.
Cummings, E. M., DeArth–Pendley, G., Du Rocher
Conger, R. D., Elder, G. H., Jr., Lorenz, F. O., Simons,
Schudlich, T., & Smith, D. A. (2001). Parental de-
R. L., & Whitbeck, L. B. (Eds.). (1994). Families in
pression and family functioning: Toward a process-
troubled times: Adapting to change in rural America.
oriented model of children’s adjustment. In S. R. H.
New York: Aldine de Gruyter.
Beach (Ed.), Marital and family processes in depres-
Conger, R. D., Ge, X., Elder, G. H., & Lorenz, F. O.
sion: A scientific foundation for clinical practice (pp.
(1994). Economic stress, coercive family process, and
89–110). Washington, DC: American Psychological
developmental problems of adolescents, [Special is- Association.
sue: Children and poverty]. Child Development, 65, Dadds, M. R., Atkinson, E., Turner, C., Blums, G. J., &
541–561. Lendich, B. (1999). Family conflict and child adjust-
Cowan, C. P., Cowan, P. A., Heming, G., & Miller, ment: Evidence for a cognitive–contextual model of
N. B. (1991). Becoming a family: Marriage, parent- intergenerational transmission. Journal of Family Psy-
ing, and child development. In P. A. Cowan & E. M. chology, 13, 194–208.
Hetherington (Eds.), Family transitions (pp. 79–109). Dadds, M. R., Schwartz, S., & Sanders, M. R. (1987).
Hillsdale, NJ: Erlbaum. Marital discord and treatment outcome in behavioral
Cowan, P. A., Bradburn, I. S., & Cowan, C. P. (in press). treatment of child conduct disorders. Journal of Con-
Parents’ working models of attachment: The intergen- sulting & Clinical Psychology, 55, 396–403.
erational context of problem behavior in kindergarten. Darling, N., & Steinberg, L. (1993). Parenting style as
In P. A. Cowan, C. P. Cowan, J. Ablow, V. K. John- context: An integrative model. Psychological Bulle-
son, & J. Measelle (Eds.), The family context of par- tin, 113, 487–496.
enting in children’s adaptation to elementary school. Davies, P. T., & Cummings, E. M. (1994). Marital con-
Mahwah, NJ: Erlbaum. flict and child adjustment: An emotional security hy-
Cowan, P. A., & Cowan, C. P. (2002). What an interven- pothesis. Psychological Bulletin, 116, 387–411.
tion design reveals about how parents affect their chil- Deal, J. E., Hagan, M. S., Bass, B., Hetherington, E. M.,
dren’s academic achievement and behavior problems. & Clingempeel, G. (1999). Marital interaction in dy-
In J. G. Borkowski, S. L. Ramey, & M. Bristol– adic and triadic contexts: Continuities and discontinu-
Power (Eds.), Parenting and the child’s world: Influ- ities. Family Process, 38, 105–115.
756 P. A. Cowan and C. P. Cowan

Dishion, T. J., Patterson, G. R., & Kavanagh, K. A. Sameroff, M. Lewis, & S. M. Miller (Eds.), Hand-
(1992). An experimental test of the coercion model: book of developmental psychopathology (2nd ed., pp.
Linking theory, measurement, and intervention. In J. 467–490). New York: Kluwer Academic/Plenum
McCord & R. E. Tremblay (Eds.), Preventing antiso- Press.
cial behavior: Interventions from birth through ado- Gershoff, E. T. (2002). Parental corporal punishment and
lescence (pp. 253–282). New York: Guilford Press. associated child behaviors and experiences: A meta-
Dodge, K. A., Pettit, G. S., Bates, J. E., & Valente, E. analytic and theoretical review. Psychological Bulle-
(1995). Social information-processing patterns par- tin, 128, 539–579.
tially mediate the effect of early physical abuse on Gordon, T. (1975). P.E.T. Parent effectiveness training:
later conduct problems. Journal of Abnormal Psy- The tested new way to raise responsible children.
chology, 104, 632–643. New York: New American Library.
Dornbusch, S. M., Ritter, P. L., Leiderman, P. H., & Rob- Gotlib, I. H., & Goodman, S. H. (1999). Children of par-
erts, D. F. (1987). The relation of parenting style to ents with depression. In W. K. Silverman & T. H.
adolescent school performance. Child Development, Ollendick (Eds.), Developmental issues in the clinical
58, 1244–1257. treatment of children (pp. 415–432). Needham
Downey, G., & Coyne, J. C. (1990). Children of de- Heights, MA: Allyn & Bacon.
pressed parents: An integrative review. Psychological Gottman, J. M., Katz, L. F., & Hooven, C. (1997). Meta-
Bulletin, 108, 50–76. emotion: How families communicate emotionally. Hills-
Durlak, J. A., & Wells, A. M. (1997). Primary prevention dale, NJ: Erlbaum.
mental health programs for children and adolescents: Gottman, J. M., Ryan, K. D., Carrere, S., & Erley, A. M.
A meta-analytic review. American Journal of Com- (2002). Toward a scientifically based marital therapy.
munity Psychology, 25, 115–152. In H. A. Liddle & D. A. Santisteban (Eds.), Family
Durlak, J. A., & Wells, A. M. (1998). Evaluation of indi- psychology: Science-based interventions (pp. 147–
cated preventive intervention (secondary prevention) 174). Washington, DC: American Psychological As-
mental health programs for children and adolescents. sociation.
American Journal of Community Psychology, 26, 775– Graham, C. A., & Easterbrooks, M. A. (2000). School-
802. aged children’s vulnerability to depressive symptom-
Eddy, J. M., Reid, J. B., & Fetrow, R. A. (2000). An atology: The role of attachment security, maternal
elementary school–based prevention program target- depressive symptomatology, and economic risk. De-
ing modifiable antecedents of youth delinquency and velopment and Psychopathology, 12, 201–213.
violence: Linking the Interests of Families and Teach- Greenberg, M. T. (1999). Attachment and psychopathol-
ers (LIFT). Journal of Emotional & Behavioral Dis- ogy in childhood. In J. Cassidy & P. R. Shaver (Eds.),
orders, 8, 165–176. Handbook of attachment: Theory, research, and clini-
Emery, R. E. (1988). Marriage, divorce, and children’s cal applications (pp. 469–496). New York: Guilford
adjustment. Thousand Oaks, CA: Sage. Press.
Emery, R. E. (1999). Marriage, divorce, and children’s Griest, D. L. (1982). Effects of parent enhancement ther-
adjustment (2nd ed.). Thousand Oaks, CA: Sage. apy on the treatment outcome and generalization of a
Emery, R. E., & O’Leary, K. D. (1984). Marital discord parent training program. Behaviour Research & Ther-
and child behavior problems in a nonclinic sample. apy, 20, 429–436.
Journal of Abnormal Child Psychology, 12, 411–420. Grych, J. H., Fincham, F. D., Jouriles, E. N., & McDon-
Erel, O., & Burman, B. (1995). Interrelatedness of marital ald, R. (2000). Interparental conflict and child adjust-
relations and parent–child relations: A meta-analytic ment: Testing the mediational role of appraisals in the
review. Psychological Bulletin, 118, 108–132. cognitive–contextual framework. Child Development,
Erickson, M. F., Korfmacher, J., & Egeland, B. R. (1992). 71, 1648–1661.
Attachments past and present: Implications for thera- Hahlweg, K., & Goldstein, M. J. (Eds.). (1987). Under-
peutic intervention with mother–infant dyads. Devel- standing major mental disorder: The contribution of
opment and Psychopathology, 4, 495–507. family interaction research. New York: Family Pro-
Fincham, F. D. (1998). Child development and marital cess Press.
relations. Child Development, 69, 543–574. Harknett, K., Hardman, L., Garfinkel, I., & McLanahan,
Fivaz–Depeursinge, E., & Corboz–Warnery, A. (1999). S. S. (2001). The Fragile Families Study: Social poli-
The primary triangle: A developmental systems view cies and labor markets in seven cities. Children &
of mothers, fathers, and infants. New York: Basic Youth Services Review, 23, 537–555.
Books. Harris, J. (1998). The nurture assumption: Why children
Forgatch, M. S., & DeGarmo, D. S. (1999). Parenting turn out the way they do. New York: Free Press.
through change: An effective prevention program for Heinicke, C. M., Goorsky, M., Moscov, S., Dudley, K.,
single mothers. Journal of Consulting and Clinical Gordon, J., Schnieder, C., & Guthrie, D. (1996). Rela-
Psychology, 67, 711–724. tionship based intervention with at-risk mothers: Fac-
Fraiberg, S. (1975). Ghosts in the nursery: A psychoana- tors affecting variations in outcome. Unpublished
lytic approach to impaired infant–mother relation- manuscript, University of California at Los Angeles.
ships. Journal of the American Academy of Child and Herman, M. R., Dornbusch, S. M., Herron, M. C., &
Adolescent Psychiatry, 14, 387–421. Herting, J. R. (1997). The influence of family regula-
Freedman, D. (1999). From association to causation: tion, connection, and psychological autonomy on six
Some remarks on the history of statistics. Statistical measures of adolescent functioning. Journal of Ado-
Science, 14, 243–258. lescent Research, 12, 34–67.
Gable, S., Crnic, K., & Belsky, J. (1994). Coparenting Hetherington, E. M. (1999). Coping with divorce, single
within the family system: Influences on children’s de- parenting, and remarriage: A risk and resiliency per-
velopment. Family Relations, 43, 380–386. spective. Matwah, NJ: Erlbaum.
Garber, J. (2000). Development and depression. In A. J. Hetherington, E. M., & Clingempeel, G. (1992). Coping
Test of family systems theories 757

with marital transitions. Society for Research in Child Predicting marital and parent functioning in dyads
Development Monograph, 57(2–3). and triads: A longitudinal investigation of marital pro-
Hetherington, E. M., Cox, M. J., & Cox, R. (1982). Ef- cesses. Journal of Family Psychology, 11, 139–151.
fects of divorce on parents and children. In M. E. Loevinger, J. (1997). Stages of personality development.
Lamb (Ed.), Nontraditional families. Hillsdale, NJ: In R. Hogan, J. A. Johnson, & S. R. Briggs (Eds.),
Erlbaum. Handbook of personality psychology (pp. 199–208).
Hetherington, E. M., & Kelly, J. (2002). For better or San Diego, CA: Academic Press.
for worse: Divorce reconsidered. New York: W. W. Maccoby, E. E., & Martin, J. A. (1983). Socialization in
Norton. the context of the family: Parent–child interaction. In
Hinshaw, S. P. (1992). Externalizing behavior problems E. M. Hetherington (Ed.), Handbook of child psychol-
and academic underachievement in childhood and ad- ogy: Socialization, personality and social develop-
olescence: Causal relationships and underlying mech- ment (4th ed., Vol. 4, pp. 1–101). New York: Wiley.
anisms. Psychological Bulletin, 111, 127–155. Markwardt, F. C. (1989). Peabody Individual Achieve-
Hinshaw, S. P., Owens, E. B., Wells, K. C., Kraemer, ment Test—Revised. Circle Pines, MN: American
H. C., Abikoff, H. B., Arnold, L. E., Conners, C. K., Guidance Service.
Elliott, G., Greenhill, L. L., Hechtman, L., Hoza, B., McHale, J. P., & Cowan, P. A. (Eds.). (1996). Under-
Jensen, P. S., March, J. S., Newcorn, J. H., Pelham, standing how family-level dynamics affect children’s
W. E., Swanson, J. M., Vitiello, B., & Wigal, T. development: Studies of two-parent families. San
(2000). Family processes and treatment outcome in Francisco, CA: Jossey–Bass.
the MTA: Negative/ineffective parenting practices in McHale, J. P., & Rasmussen, J. L. (1998). Coparental and
relation to multimodal treatment. Journal of Abnor- family group-level dynamics during infancy: Early
mal Child Psychology, 28, 555–568. family precursors of child and family functioning dur-
Holden, G. W., & Miller, P. C. (1999). Enduring and dif- ing preschool. Development and Psychopathology,
ferent: A meta-analysis of the similarity in parents’ 10, 39–59.
child rearing. Psychological Bulletin, 125, 223–254. McLoyd, V. C. (1990). The impact of economic hardship
Jackson, D. D. (1963). A study of family structure. Psy- on Black families and children: Psychological dis-
chological Reports, 13, 422–429. tress, parenting, and socioemotional development.
Johnston, J. R., Kline, M., & Tschann, J. M. (1989). On- Child Development, 61, 311–346.
going postdivorce conflict: Effects on children of McMahon, R. J., & Slough, N. M. (1996). Family-based
joint custody and frequent access. American Journal intervention in the Fast Track program. In R. D. Pe-
of Orthopsychiatry, 59, 576–592. ters & R. J. McMahon (Eds.), Prevention and early
Kaufman, J., Martin, A., King, R. A., & Charney, D. intervention: Childhood disorders, substance abuse,
(2001). Are child-, adolescent-, and adult-onset de- and delinquency. Thousand Oaks, CA: Sage.
pression one and the same disorder? Biological Psy- Measelle, J. R. (in press). The role of young children’s
chiatry, 49, 980–1001. self-perceptions in links between early family rela-
Kitzmann, K. M. (2000). Effects of marital conflict on tionships and social adjustment to kindergarten. In
subsequent triadic family interactions and parenting. P. A. Cowan, C. P. Cowan, J. Ablow, V. K. John-
Developmental Psychology, 36, 3–13. son, & J. Measelle (Eds.), The family context of par-
Kohlberg, L., & Kramer, R. (1969). Continuities and dis- enting in children’s adaptation to elementary school.
continuities in childhood and adult moral develop- Mahwah, NJ: Erlbaum.
ment. Human Development, 12, 3–120. Measelle, J. R., Ablow, J. C., Cowan, P. A., & Cowan,
Levenstein, P. (1988). Messages from home: The Mother– C. P. (1998). Assessing young children’s views of
Child Home program and the prevention of school their academic, social, and emotional lives: An evalu-
disadvantage. Columbus, OH: Ohio State University ation of the self-perception scales of the Berkeley
Press. Puppet Interview. Child Development, 69, 1556–
Lewinsohn, P. M., Clarke, G. N., Rohde, P., Hops, H., & 1576.
Seeley, J. R. (1996). A course in coping: A cogni- Melby, J. N., & Conger, R. D. (1996). Parental behaviors
tive–behavioral approach to the treatment of adoles- and adolescent academic performance: A longitudinal
cent depression. In E. D. Hibbs & P. S. Jensen (Eds.), analysis. Journal of Research on Adolescence, 6,
Psychosocial treatments for child and adolescent dis- 113–137.
orders: Empirically based strategies for clinical prac- Miller, G. E., & Prinz, R. J. (1990). Enhancement of so-
tice (pp. 109–135). Washington, DC: American Psy- cial learning family interventions for childhood con-
chological Association. duct disorder. Psychological Bulletin, 108, 291–307.
Liddle, H. A., Dakof, G. A., Parker, K., Diamond, G. S., Noam, G. (1998). Clinical–developmental psychology:
Barrett, K., & Tejeda, M. (2001). Multidimensional Toward developmentally differentiated interventions.
family therapy for adolescent drug abuse: Results of In I. E. Sigal & K. A. Renninger (Eds.), Handbook of
a randomized clinical trial. American Journal of Drug child psychology (Vol. 4, pp. 585–634). New York:
and Alcohol Abuse, 27, 651–688. John Wiley & Sons.
Lidz, T., Fleck, S., Cornelison, A. R., Alanen, Y. A., Carl- Ogawa, J. R., Sroufe, L. A., Weinfield, N. S., Carlson,
son, D. T., Cowgill, C. W., Feidelson, S. S., Freed- E. A., & Egeland, B. (1997). Development and the
man, D. X., & Lidz, R. W. (1985). Schizophrenia and fragmented self: Longitudinal study of dissociative
the family (2nd ed.). New York: International Univer- symptomatology in a nonclinical sample. Develop-
sities Press. ment and Psychopathology, 9, 855–879.
Lieberman, A. F., & Pawl, J. H. (1993). Infant–parent Olds, D. L., Henderson, C. R., Jr., Cole, R., Eckenrode,
psychotherapy. In C. H. Zeanah (Ed.), Handbook of J., Kitzman, H., Luckey, D., Pettit, L., Sidora, K.,
infant mental health (pp. 427–442). New York: Guil- Morris, P., & Powers, J. (1998). Long-term effects
ford Press. of nurse home visitation on children’s criminal and
Lindahl, K. M., Clements, M., & Markman, H. (1997). antisocial behavior: 15-year follow-up of a random-
758 P. A. Cowan and C. P. Cowan

ized controlled trial. JAMA: Journal of the American (2000). Relationships, development, and psychopath-
Medical Association, 280, 1238–1244. ology. In A. J. Sameroff, M. Lewis, & S. M. Miller
Olds, D. L., & Kitzman, H. (1993). Review of research (Eds.), Handbook of developmental psychopathology
on home visiting for pregnant women and parents of (2nd ed., pp. 75–91). New York: Kluwer Academic/
young children. Future of Children, 3, 53–92. Plenum Press.
Parke, R. D., & Buriel, R. (1998). Socialization in the Sroufe, L. A., & Rutter, M. (1984). The domain of devel-
family: Ethnic and ecological perspectives. In N. Eisen- opmental psychopathology. Child Development, 55,
berg (Ed.), Social, emotional, and personality devel- 17–29.
opment (5th ed., Vol. 3, pp. 463–552). New York: Stanley, S. M., Blumberg, S. L., & Markman, H. J.
Wiley. (1999). Helping couples fight for their marriages: The
Parke, R. D., Burks, V. M., Carson, J. L., Neville, B., & PREP approach. In R. Berger & M. T. Hannah (Eds.),
Boyum, L. A. (1994). Family–peer relationships: A Preventive approaches in couples therapy (pp. 279–
tripartite model. In R. D. Parke & S. G. Kellam 303). Philadelphia, PA: Brunner/Mazel.
(Eds.), Exploring family relationships with other so- Stark, K. D., & Boswell, J. (2000). Discussion of the
cial contexts (pp. 115–146). Hillsdale, NJ: Erlbaum. Penn Optimism Program: Recognizing its strengths
Patterson, G. R., Dishion, T. J., & Bank, L. (1984). Fam- and considerations for enhancing the program. In
ily interaction: A process model of deviancy training. J. E. Gillham (Ed.), The science of optimism and
Aggressive Behavior, 10, 253–267. hope: Research essays in honor of Martin E. P. Selig-
Patterson, G. R., Reid, J. B., & Dishion, T. J. (1998). man (pp. 235–256). Philadelphia, PA: Templeton
Antisocial boys. In J. M. Jenkins, K. Oatley, & N. L. Foundation Press.
Stein (Eds.), Human emotions: A reader (pp. 330– Stark, K. D., Swearer, S., Kurowski, C., Sommer, D., &
336). Malden, MA: Blackwell Publishers. Bowen, B. (1996). Targeting the child and the family:
Perry, K. E., & Weinstein, R. S. (1998). The social con- A holistic approach to treating child and adolescent
text of early schooling and children’s school adjust- depressive disorders. In E. D. Hibbs & P. S. Jensen
ment. Educational Psychologist, 33, 177–194. (Eds.), Psychosocial treatments for child and adoles-
Raudenbush, S. W. (2000). Comparing personal trajecto- cent disorders: Empirically based strategies for clini-
ries and drawing causal inferences from longitudinal cal practice (pp. 207–238). Washington, DC: Ameri-
data. Annual Review of Psychology, 52, 501–525. can Psychological Association.
Reid, J. B., Eddy, J. M., Fetrow, R. A., & Stoolmiller, Steinberg, L. (2001). We know some things: Parent–adoles-
M. (1999). Description and immediate impacts of a cent relationships in retrospect and prospect. Journal
preventive intervention for conduct problems. Ameri- of Research on Adolescence, 11, 1–19.
can Journal of Community Psychology, 27, 483–517. Stolberg, A. L., & Garrison, K. M. (1985). Evaluating a
Rogers, M. J., & Holmbeck, G. N. (1997). Effects of in- primary prevention program for children of divorce.
terparental aggression on children’s adjustment: The American Journal of Community Psychology, 13,
moderating role of cognitive appraisal and coping. 111–124.
Journal of Family Psychology, 11, 125–130. Stoolmiller, M., Duncan, T., Bank, L., & Patterson, G. R.
Rutter, M., & Garmezy, N. (1983). Developmental psy- (1993). Some problems and solutions in the study of
chopathology. In P. Mussen (Ed.), Handbook of child change: Significant patterns in client resistance. Journal
psychology (Vol. 4, pp. 775–911). New York: Wiley. of Consulting and Clinical Psychology, 61, 920–928.
Satir, V. (1972). Peoplemaking. Palo Alto, CA: Science Straus, M. A. (2001). Beating the devil out of them: Cor-
and Behavior Books. poral punishment in American families and its effects
Schaefer, E. S., & Hunter, W. M. (1983, April). Mother– on children. New Brunswick, NJ: Transaction.
infant interaction and maternal psychosocial predic- Teichman, Y. (1989). Childhood depression: A family
tors of kindergarten adaptation. Paper presented at perspective. Israel Journal of Psychiatry and Related
the Society for Research in Child Development, De- Sciences, 26, 46–57.
troit, MI. Toner, R. (2002, February 19). Welfare chief is hoping to
Seifer, R., & Dickstein, S. (2000). Parental mental illness promote marriage. New York Times, pp. A1, A16–17.
and infant development. In C. H. Zeanah, Jr (Ed.), Toth, S. L., & Cicchetti, D. (1999). Developmental psy-
Handbook of infant mental health (2nd ed., pp. 145– chopathology and child psychotherapy. In S. W.
160). New York: Guilford Press. Russ & T. H. Ollendick (Eds.), Handbook of psycho-
Shadish, W. R., Ragsdale, K., Glaser, R. R., & Montgom- therapies with children and families (pp. 15–44).
ery, L. M. (1995). The efficacy and effectiveness of New York: Kluwer Academic/Plenum Press.
marital and family therapy: A perspective from meta- Wagner, B. M., & Reiss, D. (1995). Family systems and
analysis. Journal of Marital and Family Therapy, 21, developmental psychopathology: Courtship, marriage,
345–360. or divorce? In D. Cicchetti & D. J. Cohen (Eds.), De-
Shirk, S. R., & Russell, R. L. (1996). Change processes velopmental psychopathology: Vol. 1. Theory and
in child psychotherapy: Revitalizing treatment and re- methods (pp. 696–730). New York: Wiley.
search. New York: Guilford Press. Waite, L. J., & Gallagher, M. (2000). The case for mar-
Silliman, B., & Schumm, W. R. (2000). Marriage prepa- riage: Why married people are happier, healthier,
ration programs: A literature review. Family Jour- and better off financially (1st ed.). New York: Dou-
nal—Counseling and Therapy for Couples and Fami- bleday.
lies, 8, 133–142. Wallerstein, J., & Kelly, J. (1980). Surviving the breakup.
Simons, R. L., Lorenz, F. O., Wu, C.-I., & Conger, R. D. New York: Basic Books.
(1993). Social network and marital support as media- Warren, N. J., & Amara, I. A. (1984). Educational groups
tors and moderators of the impact of stress and de- for single parents: The Parenting After Divorce pro-
pression on parental behavior. Developmental Psy- grams. Journal of Divorce, 8, 79–96.
chology, 29, 368–381. Webster–Stratton, C. (1984). Randomized trial of two
Sroufe, L. A., Duggal, S., Weinfield, N., & Carlson, E. parent-training programs for families with conduct-
Test of family systems theories 759

disordered children. Journal of Consulting and Clini- Weisz, J. R., Weiss, B., Han, S. S., & Granger, D. A.
cal Psychology, 52, 666–678. (1995). Effects of psychotherapy with children and
Webster–Stratton, C. (1994). Advancing videotape parent adolescents revisited: A meta-analysis of treatment
training: A comparison study. Journal of Consulting outcome studies. Psychological Bulletin, 117, 450–
and Clinical Psychology, 62, 583–593. 468.
Weinfield, N. S., Sroufe, L. A., & Egeland, B. (2000). Whitehurst, G. J., Arnold, D. S., Epstein, J. N., & Angell,
Attachment from infancy to early adulthood in a high- A. L. (1994). A picture book reading intervention in
risk sample: Continuity, discontinuity, and their cor- day care and home for children from low-income
relates. Child Development, 71, 695–702. families. Developmental Psychology, 30, 679–689.
Weissberg, R. P., & Bell, D. N. (1997). A meta-analytic Wierson, M., & Forehand, R. (1994). Parent behavioral
review of primary prevention programs for children training for child noncompliance: Rationale, concepts,
and adolescents: Contributions and caveats. American and effectiveness. Current Directions in Psychologi-
Journal of Community Psychology, 25, 207–214. cal Science, 3, 146–150.
Weisz, J. R., Valeri, S. M., McCarty, C. A., & Moore, Wolchik, S. A., West, S. G., Westover, S., & Sandler,
P. S. (1999). Interventions for depression: Features, ef- I. N. (1993). The Children of Divorce Parenting Inter-
fects, and future directions. In C. A. Essau & F. Peter- vention: Outcome evaluation of an empirically based
mann (Eds.), Depressive disorders in children and ado- program. American Journal of Community Psychol-
lescents: Epidemiology, risk factors, and treatment (pp. ogy, 21, 293–331.
383–435). Northvale, NJ: Jason Aronson.