Content uploaded by Lynn Fainsilber Katz
Author content
All content in this area was uploaded by Lynn Fainsilber Katz on Dec 25, 2018
Content may be subject to copyright.
Child Development, March/April 2002, Volume 73, Number 2, Pages 636–652
Hostility, Hostile Detachment, and Conflict Engagement in Marriages:
Effects on Child and Family Functioning
Lynn Fainsilber Katz and Erica M. Woodin
This study examined the relations between patterns of marital communication, child adjustment, and family func-
tioning. Couples with a 4- or 5-year-old child were divided into three groups (
N
5
126) based on observed patterns
of emotional communication: Hostile couples showed a cumulative increase in negative speaker behaviors
over the course of a high-conflict marital discussion; hostile-withdrawn couples showed a cumulative increase
in both negative speaker and negative listener behaviors over the course of the interaction; and engaged cou-
ples showed a cumulative increase in both positive speaker and listener behaviors over the course of the inter-
action. The families of these three types of couples were then compared on child outcomes (i.e., peer relations,
behavior problems), parenting quality, co-parenting quality, and family-level functioning. Differences in mari-
tal violence and marital satisfaction between marital couples were also examined in relation to family risk.
Families in which couples were hostile-detached showed the most negative outcomes. Hostile-detached cou-
ples were more likely than hostile or conflict-engaging couples to use more power-assertive methods of disci-
pline; to be ineffective in co-parenting their child; to have family units that were less cohesive, less playful, and
more conflictual; and to have children that exhibited behavior problems. Results also indicated that marital ty-
pology still accounted for significant variance in child outcome after controlling for marital violence and mari-
tal satisfaction. Differences in the absolute degree of negative behaviors also did not account for results. Find-
ings are discussed in terms of the detrimental impact of marital conflict on child and family functioning.
INTRODUCTION
There is considerable evidence that conflict within the
marriage is a powerful predictor of child behavior
problems. Marital conflict has been associated with
an array of adjustment difficulties in children, includ-
ing conduct problems, poor peer interaction, poor
health, depression, anxiety, attachment insecurity,
and low self-esteem (Cummings & Davies, 1994;
Emery, 1982, 1999; Emery & O’Leary, 1982, 1984; Gott-
man & Katz, 1989; Grych & Fincham, 1990; Howes &
Markman, 1989; Jouriles et al., 1991; Katz & Gottman,
1993). Little is known, however, about the specific
characteristics of bad marriages that are most caustic
to children’s well-being.
Recent efforts have highlighted key structural
characteristics of marital conflict that are related to
children’s functioning. Properties such as the inten-
sity, frequency, and content of marital conflict moder-
ate the level of distress that children demonstrate
within the laboratory setting (Fincham, 1994; Fin-
cham & Osborne, 1993; Grych & Fincham, 1990). Mar-
ital conflict that is intense, frequent, and child related
causes children to have feelings of fear and distress
and a desire to intervene in the conflict (Grych & Fin-
cham, 1993). Lack of resolution to the conflict has also
been related to negative affective responses such as
anger, sadness, and distress in children (Cummings,
Ballard, El-Sheikh, & Lake, 1991; Cummings, Vogel,
Cummings, & El-Sheikh, 1990).
Although the link between these dimensions of
marital conflict and children’s subjective feelings and
behavioral indices of distress has been established,
two limitations remain that need to be addressed.
First, there is little evidence that links specific charac-
teristics in the marriage to children’s behavior prob-
lems or interpersonal interactions with others outside
the laboratory setting. Second, a focus on specific
structural characteristics of marital conflict does not
sufficiently capture the multidimensional and dy-
namic nature of marital interaction. In marriages, it is
the relative balance of different relationship charac-
teristics rather than the absolute level of any one char-
acteristic that is important to relationship satisfaction.
Gottman and Levenson (1992) found that a ratio of
five positive behaviors to one negative behavior was
sufficient in order to sustain marital stability; couples
who fell below this ratio were at greater risk for
divorce. The relative balance between negative and
positive marital behaviors may be similarly impor-
tant in predicting child outcomes. If this is true, a
multidimensional means of capturing the couple’s
style of marital communication that focuses on the
balance between negativity and positivity may pro-
vide richer information about individual differences
in children’s reactions to marital conflict.
There is a venerable tradition of identifying com-
© 2002 by the Society for Research in Child Development, Inc.
All rights reserved. 0009-3920/2002/7302-0019
Katz and Woodin 637
munication patterns that differentiate between dis-
tressed and nondistressed couples (Christensen &
Heavey, 1990; Gottman, 1994; Gottman, Coan, Car-
rère, & Swanson, 1998; Heavey, Christensen, & Mala-
muth, 1995; Kurdek, 1995). Both negative reciprocity
and the balance between negative and positive be-
haviors are related to marital satisfaction and divorce
(Gottman & Levenson, 1992; Levenson & Gottman,
1985). There is also growing evidence that couples
have distinct ways of managing conflict in their rela-
tionship (Fitzpatrick & Indvik, 1982; Gottman, 1993a).
For example, some relationships are characterized by
hostility, in which couples attack each other through
contemptuousness, belittling, and blaming (Brown &
Smith, 1992; Gottman, 1993a; Newton & Burgoon,
1990). Other couples withdraw during conflict by
turning away from their partner, avoiding eye con-
tact, refraining from back channeling (e.g., nodding,
saying “uh-huh”), and changing the topic of conver-
sation (Christensen, 1987; Christensen & Heavey,
1990; Gottman, 1993a). Several typologies of marriage
have been proposed, but little is known about how
these different marital styles relate to children’s ad-
justment and family relationships. These stylistic in-
teraction patterns form the fabric of daily family life,
and serve as models of close interpersonal relation-
ships for children. What may influence how children
handle conflict with others are the specific behaviors
that are modeled by parents as they resolve their own
marital disputes.
The present study built on this research tradition
by examining the differential relations between four
types of marriages and child adjustment. Consistent
with a family systems approach (Haley, 1965; Mi-
nuchin, 1974), a second goal of this study was to un-
derstand how these four types of marriages relate to
other subsystems within the family. Given the impact
of marital distress on other family relationships such
as the parent– child relationship (Erel & Berman,
1995) and the co-parenting relationship (McHale,
1997), as well as family-level dynamics (McHale &
Cowan, 1996), the links among these four types of
marriages and parenting, co-parenting, and family-
level dynamics were explored.
Identifying a Typology of Marriage
Several different typologies of marriage have been
proposed. Some typologies have been based on cou-
ples’ own reports of their marriage. For example, Fitz-
patrick and Indvik (1982) characterized marriages
based on each spouse’s perception of their degree of
interdependence within the relationship. Similarly,
Margolin and colleagues (Burman, John, & Margolin,
1992; Margolin, John, & Gleberman, 1988) classified
couples based on the degree of physical or verbal ag-
gression present in their relationship.
Classification systems have also been based on
observations of marital interaction. Observed pat-
terns of marital conflict rather than overall frequen-
cies of certain behaviors have been identified, be-
cause often it is the relative frequencies of different
behaviors and their pattern over time, rather than the
absolute total of any one behavior, that is the most im-
portant level of analysis (Bradbury & Karney, 1993;
Gottman, 1994).
Consistent with this approach, Gottman and Lev-
enson (1992) examined the pattern of positive and
negative speaker behaviors exhibited over the course
of a high-conflict discussion. Couples were divided
into regulated and nonregulated groups based on the
ratio of positive to negative behaviors over time.
Nonregulated couples showed a negative speaker
slope, indicating that the relative frequency of nega-
tive behaviors outweighed the relative frequency of
positive behaviors over time. Regulated couples
showed a positive speaker slope, indicating that the
relative frequency of positive behaviors outweighed
the relative frequency of negative behaviors over
time. Gottman and Levenson found that nonregu-
lated couples had lower levels of marital satisfaction,
more negative and less positive emotional expres-
sion, more stubbornness and withdrawal, more de-
fensiveness, and a greater likelihood of divorce com-
pared with regulated couples.
This classification scheme was then expanded to
include withdrawn listening behaviors, given evi-
dence that withdrawal is destructive to marital well-
being (Christenson & Heavey, 1990). The inclusion of
hostile speaking behavior and withdrawn listening
behavior produced four possible groups: Hostile
(HO) couples were nonregulated speakers, but were
regulated in their listening behaviors; Avoiders (AV)
were nonregulated listeners, but regulated speakers;
Hostile-Detached (HD) couples were both nonregu-
lated listeners and nonregulated speakers; and Conflict-
Engagers (EN) were both regulated listeners and
speakers. The advantage of this typology was that it
allowed for the distinction between couples who ex-
hibited one form of negative behavior (hostility or
withdrawal) versus couples who exhibited both
forms. Problem-solving and affective differences were
found between these four groups. Conflict-Engagers
displayed more negative as well as positive behavior
and affect than did AV couples, whereas AV couples
displayed more listener withdrawal than did EN cou-
ples. In addition, HD couples exhibited more nega-
tive and less positive behavior than HO couples. Only
638 Child Development
those couples whose speaking behavior was nonreg-
ulated (i.e., HO, HD), however, were unstable and
prone to divorce (Gottman, 1993a, 1994).
Marital Types, Family Functioning,
and Child Adjustment
Building on this work, we sought to apply a typol-
ogy of marriage to better understand the relation
among marital conflict, child behavior problems, and
family functioning. We examined whether conflict en-
gagement, conflict avoidance, hostility, and hostile
detachment in the marriage was associated with dif-
ferential aspects of child and family adjustment. Cou-
ples were divided into EN, AV, HO, and HD groups
based on speaker and listener behavior during obser-
vations of marital interaction. Child adjustment,
parenting, and family functioning were then assessed
in a variety of ways. First, children’s peer interaction
and behavior problems were examined, given evi-
dence that both are impacted by marital conflict
(Cummings & Davies, 1994; Emery, 1982; Gottman &
Katz, 1989; Katz & Gottman, 1995). These child out-
comes were also selected because they index social
processes operating outside the laboratory setting.
Second, we were also interested in whether these four
distinct marital styles were associated with unique
parenting qualities. Because it is well established that
marital conflict is associated with harsh and/or in-
consistent parenting processes (Erel & Burman, 1995),
we examined whether couples with these four dis-
tinct marital communication styles differed in their
use of discipline.
Third, we examined whether these different mari-
tal conflict styles were associated with other relational
dynamics within the family, such as co-parenting
processes and family-level dynamics (McHale, 1997;
McHale & Rasmussen, 1998). Co-parenting processes
refer to the ways in which partners support one an-
other in their joint role as leaders of the family. From
the child’s perspective, observing the quality of the
co-parenting relationship provides rich information
about relationship skills, such as the exchange and ex-
pression of affect and conflict resolution strategies.
Emphasis on whole-family processes is rooted in fam-
ily systems thinking, that is, that family-level pro-
cesses affect children’s development over and above
any individual subsystem, and that interactions at the
family level are a unique social force that goes beyond
the sum of its parts (e.g., Minuchin, 1974).
There is a growing body of evidence that links both
co-parenting behavior and family-level processes
with marital quality. Cowan and Cowan (1987) found
that happily married couples showed more warmth,
less conflict, and more cooperation with each other
when interacting in front of their child. Happily
married couples also showed greater investment,
sensitivity, support, and warmth during family inter-
actions than unhappily married couples (Cowan,
Cowan, Schultz, & Heming, 1994; McHale, 1997). The
importance of investigating interaction at the family
level is heightened by evidence that parent and child
behavior changes between the dyadic and triadic
(mother–father–child) contexts. Gjerde (1986) found
that the quality of the father–son relationship dimin-
ished in the context of mothers, but the mother–son
relationship improved while interacting in the triad.
Thus, one contribution of this study was the exam-
ination of how different marital interaction styles re-
late not only to child functioning but also to other
relational dynamics within the family, including parent-
ing, co-parenting, and family-level processes. We hy-
pothesized that living in a family in which couples
were either HO or HD would be associated with poor
child adjustment, harsh parenting, poor co-parenting,
and lack of cohesion in the family as a whole. This
prediction stemmed from the observation that both
HO and HD couples show a dysregulated pattern in
which they express increasing levels of negative affect
as they attempt to solve marital differences. Because
withdrawal from marital conflict represents an end
stage in the progression towards marital dissolution
(Gottman, 1994), we also hypothesized that families
in which couples are both hostile and detached may
be at greatest risk for poor child adjustment and fam-
ily functioning. Two alternative hypotheses were also
explored: (1) that the absolute degree of negative be-
haviors, rather than the balance of negative and posi-
tive conflict, would account for differences in child
adjustment; and (2) that marital violence and marital
satisfaction might account for any observed relations
between marital typology and increased child risk.
METHOD
Participants
One hundred and thirty families with a preschool-
age child were recruited for participation in a study of
children at risk for the development of conduct prob-
lems. Four- to six-year-old children and their families
were recruited through advertisements in preschools,
newspapers, and doctors’ offices. Children were, on
average, 60.5 months old (
range
5
49–71 months). In-
clusion into the study was determined by scores on a
telephone version of the Eyberg Child Behavior In-
ventory (ECBI; Robinson, Eyberg, & Ross, 1980). Fol-
lowing established cutoffs, children who received a
Katz and Woodin 639
score of 11 or higher on the ECBI were assigned to the
conduct-problem group and children who received a
score of 7 or lower were assigned to the control group.
Sixty-five families had children who met criteria for
the conduct-problem group (41 male, 24 female), and
65 families had children who met criteria for the con-
trol group (40 males, 25 females). Groups were case
matched for age, socioeconomic status, and neighbor-
hood. With the exception of 1 child, groups were also
case matched for gender. The conduct-problem group
was further subdivided into two groups based on ma-
ternal report on the Diagnostic Interview Schedule
for Children (DISC; Fisher, Wicks, Shaffer, Piacentini,
& Lapin, 1993): (1) The Oppositional Defiant Disor-
der (ODD) group consisted of children who both ex-
ceeded established cutoffs on the ECBI and also met
criteria for ODD or Conduct Disorder (CD) on the
DISC (
N
5
32; 20 males, 12 females); and (2) the high-
risk group, which consisted of children who exceeded
established cutoffs on the ECBI but did not meet cri-
teria for ODD or CD on the DISC (
N
5
33; 21 males, 12
females). The control group consisted of children who
scored below standard cutoffs on the ECBI and did
not meet diagnostic criteria for any psychiatric diag-
noses. Mother’s report on the Child Behavior Check-
list (CBCL; Achenbach & Edelbrock, 1983) was also
obtained. Mean
T
scores on the aggression subscale
were 68 for the conduct-problem group and 53 for the
control group. The predominant ethnic identification
was Anglo-European (88.5%), and the remainder of
the sample was Hispanic American (6.2%), African
American (3.8%), and Native American (1.5%). Sixty-
two percent of the sample was male (
N
5
81), and
38% was female (
N
5
49).
Procedure
The procedure consisted of observations of mari-
tal, family, and parent –child interactions, and par-
ent reports of child adjustment, marital quality, and
parental psychopathology.
Marital interaction.
To assess marital interaction
patterns, couples engaged in a 15-min conversation
about two areas of continuing disagreement in their
relationship (Katz & Gottman, 1993). Interactions
were videotaped for subsequent observational coding.
Family interaction.
A triadic family interaction was
conducted to assess parenting, co-parenting, and
family-level processes. After parents completed the
marital interaction, the family was seated in a room
that contained two shelves of toys—one shelf of
attractive, age-appropriate toys, and one shelf of unat-
tractive, age-inappropriate toys. The parents were in-
structed to allow their children to play only with the
unattractive toys because the attractive toys belonged
to another research group that requested their toys be
left intact. This procedure is similar to forbidden toy
paradigms used in previous research (e.g., Sears,
Maccoby, & Levin, 1957; Toner, Parke, & Yussen,
1979). The interaction was designed to simulate a
common dilemma that parents sometimes find them-
selves in, wherein they may need to discipline or
otherwise engage with their child after they them-
selves have been arguing. The 10-min interaction
was videotaped.
Peer interaction.
To assess peer functioning, chil-
dren were audiotaped for 45 min playing with a
same-sex best friend in their home. Best friends were
chosen because best friend interaction is an index of a
child’s maximal social competence (Gottman & Parker,
1986). Same-sex peers were used because social pro-
cesses during cross-sex peer interaction differ mark-
edly from those that occur during same-sex peer in-
teraction (Gottman & Parker, 1986).
Measures
Observational coding of marital interaction, parent-
ing quality, co-parenting, family-level processes, and
peer interaction was conducted by independent
groups of coders.
Observational coding of marital interaction.
Marital
interactions were coded using the Specific Affect
Coding System (SPAFF; Gottman, 1994). The SPAFF is
a gestalt coding system that identifies specific posi-
tive and negative emotions (e.g., joy, anger, and sad-
ness) as well as positive and negative emotional be-
haviors (e.g., validation, stonewalling). The SPAFF
contains 12 negative and 5 positive codes. In this
study, only codes that theoretically indexed engage-
ment, hostility, and detachment were included in the
analyses. Other SPAFF codes were not included
based on the consideration that they did not index
these relevant processes (e.g., defensiveness). In addi-
tion, codes were divided into two categories: those
performed by the speaker and those performed by
the listener. The speaker codes were contempt, bellig-
erence, criticism, interest, affection, high-intensity
validation (i.e., understanding and acceptance), tense
humor, and humor. The listener codes were stone-
walling (i.e., avoiding eye contact, the absence of nor-
mal conversational back channeling, monitoring gaze)
and low-intensity validation (i.e., acceptance and
openness to partner’s viewpoint; listening to partner
in affirmative fashion by using physical or vocal as-
senting behaviors; see Table 1 for a brief definition of
each code). This is necessarily a limited assessment
of listener behaviors because only two SPAFF codes
640 Child Development
adequately tap what the listener is doing. Behaviors
were coded on a second-by-second basis using a
computer-automated system that synchronizes ob-
servational coding with the original marital inter-
action. Each parent was coded by independent ob-
servers and Cohen’s
k
s were computed for each spouse.
Reliability was adequate, with mean
k
s of .634 for
husbands and .642 for wives.
Observational coding of co-parenting and family level
interaction.
The Family-Level and Co-Parenting In-
teraction Coding System was used to assess co-
parenting and family-level interaction (FICS; Katz,
Low, Young, & Kahm, 1997). The FICS is a global
coding system containing five scales that assess co-
parenting behaviors and five scales that assess family-
level processes. Co-parenting codes assessed the
degree of negativity, disengagement/withdrawal,
neutral conversation, cooperation, and positive affect
between parents. Family-level codes assessed the de-
gree of playfulness, cohesiveness, adaptability, and
conflict/negativity within the family as a whole. Each
scale was scored on a 5-point Likert scale ranging
from low to high. Interrater reliability was calculated
using intraclass correlations and ranged from .55 to
.85 with a mean of .73. The nature of alliance forma-
tions within the family unit was also assessed. The al-
liance formation scale assessed the degree of closeness,
influence, and affection of the dyadic relationships
within the triadic interaction. The scale was categori-
cal and consisted of balanced, marital, parent– child,
and disengaged formations. Reliability on this scale
was moderate, with a Cohen’s
k
of .56.
Observational coding of parenting behavior.
The For-
bidden Toy Coding System (FTCS; Mittmann & Katz,
1997) was used to assess parent– child interaction.
Behaviors were coded on a second-by-second basis
using a computer-automated system that synchro-
nizes observational coding with the original parent–
child interaction. For the purposes of this study, only
codes that theoretically indexed discipline attempts
during triadic interaction were included in the analy-
ses. Commands indicated overt attempts to gain com-
pliance through issuing directives. This measure was
computed by combining the command code for
mother and father. Less directive attempts to encour-
age children to alter their behavior were indexed by
combining mother and father codes of bargaining
and distraction.
Observational coding of peer interaction.
Peer inter-
action was coded using the Rapid-MACRO coding
system (R-MACRO; Gottman, 1983). The R-MACRO
identifies 43 positive and negative behaviors that are
indicative of children’s coordination of play. Behav-
iors were coded every 3 min by independent observ-
ers. A total for each code was computed by summing
the frequency of the code over each 3-min segment.
Only codes that theoretically indexed externalizing
behavior were included in the analyses. Externaliz-
Table 1 Brief Definition of Specific Affect Coding System Codes
Code Definition
Speaker codes
Contempt Attempt to insult or communicate a lack of respect about partner. Includes sarcasm, mockery, insults,
hostile humor.
Belligerence Provocative comments aimed at getting a rise out of spouse. Includes taunting questions, dares
(e.g., What are you going to do about it ?”).
Criticism Attacking spouse’s personality or character, rather than a specific behavior, often with blame. Includes
character attacks, “kitchen sinking” (i.e, complaining about multiple different concerns), betrayal
statements, negative mind reading.
Tense humor Nervous laughter by both spouses at the same time. Includes nervous giggling, tense joking, tense smile.
Interest Positive energy in relation to what spouse has said/done. Includes seeking elaboration, questioning
opinion.
Affection Verbal expressions of love; praising spouse.
High-intensity validation Acceptance and openness to partner’s viewpoint. Includes paraphrasing, finishing sentences.
Humor Moments of shared laughter that are not tense. Includes joking/good-natured teasing, wit, giggling, fun,
exaggeration.
Listener codes
Stonewalling Individual ceases to attend to what partner is saying. Includes away behaviors (i.e., focusing on
something trivial to avoid eye contact), automanipulation (e.g., touching face), absence of normal
conversational back channeling, monitoring gaze.
Low-intensity validation Acceptance and openness to partner’s viewpoint. Includes validation listening to partner in an
affirmative fashion by using paralinguistic cues (e.g., head nods, “umm-hmms”) or other physical
or vocal assenting behaviors. Must include eye contact.
Katz and Woodin 641
ing behaviors were examined because they have con-
sistently been linked to marital conflict, and because
they reflect the salient adjustment difficulty of this
sample. A negative affect/noncompliance score was
achieved by summing the following codes: conflict,
unable to manage conflict, anger, bossiness, crying,
fighting, negative statements about friend, negative
teasing, sadness, whining, noncompliance, and par-
ent intervention. Reliability was high, with a correla-
tion of .90.
Maternal ratings of child behavior problems.
Mothers
completed the CBCL to assess the presence of behav-
ior problems.
T
scores for the externalizing and inter-
nalizing subscales as well as the total score of the
CBCL were used to index child behavior problems.
The CBCL has a test–retest reliability of .84 and an in-
traclass correlation of .98 between parents (Achen-
bach & Edelbrock, 1983).
Self-report of marital quality and violence.
Parents in-
dividually completed the Marital Adjustment Test
(MAT; Locke & Wallace, 1959) to assess the quality of
the marital relationship. The MAT is a widely used
global assessment of marital satisfaction that has
well-established reliability and validity. In addition,
the Conflict Tactics Scale (CTS; Straus, 1979) was used
to measure the amount of reasoning, verbal aggres-
sion, and physical aggression between spouses. The
CTS is the most widely used measure of aggression
between spouses. The subscales have adequate reli-
ability, with
a
s ranging from .50 to .76 for the reason-
ing subscale, .79 to .88 for the verbal aggression sub-
scale, and .82 to .88 for the physical aggression subscale.
RESULTS
Analyses were conducted in four stages. The first
stage involved creating the marital typology. In the
second stage, marital groups were compared on
parenting, co-parenting, and family-level processes,
and on child outcome. The third stage examined
whether these marital groups could explain unique
variance in child outcomes, controlling for the effects
of parenting and co-parenting/family-level processes
on child outcomes. The fourth stage addressed the al-
ternative hypotheses that (1) absolute differences in
negative behavior between marital groups account
for child outcomes, and (2) marital violence or marital
satisfaction may account for observed relations be-
tween marital typology and child outcome.
Creation of a Marital Typology
A typology of marital communication was created
following Gottman’s (1993a) procedures, with modi-
fications made to accommodate use of a different ob-
servational coding system. The SPAFF speaker and
listener codes were used to classify couples into a 2
3
2 factorial design. Weights were assigned to each
SPAFF code based on several factors. First, weights
were assigned based on theoretical considerations
having to do with the relation between different affec-
tive displays and marital decline and dissolution.
Those codes that have previously been found to be
strong predictors of marital decline and dissolution
received negative weightings, whereas those codes
that were more associated with positive relationship
quality received more positive weightings. Using the
prediction of marital decline as a criterion for assign-
ing weights acknowledges the impact of those behav-
iors that are most destructive to family relationships.
Second, codes that more strongly indexed key theo-
retical constructs (i.e., hostility, withdrawal) were
given greater weightings. Weights were
2
4.0 for con-
tempt and stonewalling;
2
2.0 for belligerence and
criticism;
1
2.0 for tense humor, interest, and low-
intensity validation; and
1
4.0 for high-intensity vali-
dation, humor, and affection.
Because the SPAFF was coded on a second-by-
second basis, a SPAFF code was available for each sec-
ond of time. The weighted speaker SPAFF codes were
summed at each second in a cumulative fashion, yield-
ing a time series of cumulative speaker codes for each
spouse over the duration of the conversation. This time
series was plotted for each spouse, and the slopes of
the speaker plots were determined using linear regres-
sion analysis (see Figure 1). This yielded a positive or
negative speaker slope for each spouse. An identical
procedure was used to determine the listener slope for
each spouse. These slope values represent the overall
Figure 1 Examples of cumulative point graphs of positive
and negative speaker behavior over time.
642 Child Development
trajectory of speaking and listening behavior over the
duration of the conversation for each spouse. Positive
slopes indicated a greater ratio of positive to negative
behaviors over the course of the conversation, and
negative slopes indicated a greater ratio of negative to
positive behaviors over time. A negative slope is also
indicative of a dysregulated pattern of interaction
that reflects a consistently high degree of negative
marital behaviors relative to positive behaviors over
the course of the interaction. The steeper the negative
slope, the more negative affect that is being expressed
and the greater likelihood that the negativity is being
expressed in a hostile fashion. A positive slope in-
dexes a regulated interactive style that is balanced to-
ward more neutral or positive affect over time.
Couples were then divided into four groups based
on these regression slopes (Figure 2): (1) EN couples
were those in which both spouses displayed overall
positive speaker and listener slopes; (2) AV couples
were those in which both spouses displayed positive
speaker slopes, but in which at least one spouse dis-
played a negative listener slope (i.e., indicative of
withdrawal); (3) HO couples were those in which
both spouses displayed positive listener slopes, but in
which at least one spouse displayed a negative
speaker slope; and (4) HD couples were those in
which both spouses displayed overall negative speaker
and listener slopes. Thus, this typology reflected both
the degree to which couples escalated or de-escalated
negative affect (i.e., how dysregulated they were), as
well as the presence of both hostility and withdrawal
in the marriage.
This classification process yielded 50 EN (38%), 5
AV couples (4%), 44 HO couples (34%), and 31 HD
couples (24%). Due to the disproportionately small
Listener Slope
Both Spouses
Positive
One Spouse
Negative
Both Conflict-
Engagers AvoidersSpouses
Positive
One
Hostile Hostile-
Detached
Spouse
Negative
Speaker Slope
Figure 2 Marital typology as a function of listener and
speaker cumulative point-graph slopes.
Table 2 Zero-Order Correlations Between Family Variables and Child Outcome
2 3 4 5 6 7 8 9 10 11 12 13 14 15
a
Parenting
1. Bargaining/
distraction .31*** .19*
2
.04
2
.18* .12 .05 .09 .22**
2
.22** .14 .13 .14 .13 .15
2. Commands
2
.05 .06 .09
2
.05
2
.01
2
.20* .36***
2
.16 .32*** .20* .36*** .27** .35***
Co-parenting
3. Cooperation
2
.30**
2
.64*** .30** .52*** .71***
2
.17 .11
2
.03
2
.07 .01
2
.09
2
.24**
4. Negativity .24**
2
.20*
2
.25**
2
.38*** .33***
2
.06 .24** .27** .22** .10 .33***
5. Disengagement/
withdrawal
2
.25**
2
.40***
2
.60*** .18*
2
.10
2
.01 .02
2
.05 .05 .23**
6. Positive affect .27** .43***
2
.20* .07
2
.02 .07
2
.03
2
.10
2
.15
Family level
7. Playfulness .65***
2
.28*** .17 .09 .08 .11
2
.01
2
.22**
8. Cohesiveness
2
.44*** .25**
2
.07
2
.04
2
.07
2
.17
2
.33***
9. Conflict/negativity
2
.34*** .22** .14 .28** .11 .23**
10. Adaptability
2
.01 .09 .07
2
.14
2
.22**
Child outcomes
11. CBCL— total .82*** .94*** .25*** .21*
12. CBCL— internalizing
subscale .69*** .10 .12
13. CBCL— externalizing
subscale .25** .23*
14. Negative affect/
noncompliance
with peers .25**
15. Marital typology
a
Note:
CBLL
5
Child Behavior Checklist.
a
Spearman
r
used to compute correlations because of categorical data.
*
p
,
.05; **
p
,
.01; ***
p
,
.001.
Katz and Woodin 643
number of AV couples, this group was dropped from
further analyses. Correlations between family vari-
ables and child outcomes are presented in Table 2.
Relation between Marital Typology Group
and Parenting Processes
Table 3 summarizes parenting differences between
the marital groups. In the context of interacting with
their children, HD couples used more commands
than did EN couples, and HO parents used more
bargaining/distraction than did EN parents.
Relations between Marital Typology Group
and Co-Parenting and Family-Level Processes
Table 3 summarizes the one-way ANOVAs used to
explore differences between marital groups in co-
parenting and family-level processes. In co-parenting
behaviors, HD couples displayed more negativity and
more disengagement and withdrawal than did EN and
HO couples. HD couples also displayed less coopera-
tion than did EN couples. At the family level, HD fam-
ilies were less playful and displayed more conflict and
negativity than did EN families. In addition, the HD
families were less cohesive than both EN and HD fam-
ilies. Table 4 reports the
x
2
analyses used to analyze dif-
ferences between HD, HO, and EN families in alliance
formation. Results were significant,
x
2
(1,
N
5
6)
5
Table 3 ANOVAs Comparing Marital Typology Groups on Parenting, Co-Parenting, Family-Level Behaviors, and Child Outcomes
Engaged (EN)
a
Hostile (HO)
b
Hostile-
Detached (HD)
c
M SD M SD M SD
F Post Hoc Scheffé
d
Parenting
Commands 4.02 7.43 8.48 9.38 10.13 10.37 5.29** HD
.
EN
Bargaining/distraction 1.66 3.01 3.50 4.15 2.10 2.82 3.57* HO
.
EN
Co-parenting (FICS)
Negativity 1.18 .75 1.16 .37 1.74 1.18 6.19** HD
.
EN, HO
Disengagement/withdrawal 1.76 .85 1.75 .87 2.65 1.40 8.86*** HD
.
EN, HO
Neutral conversation 1.78 .74 2.02 1.15 1.74 .73 1.17
Cooperation 3.66 1.06 3.48 1.00 2.94 1.24 4.36* HD
,
EN
Positive affect 2.06 .91 2.02 .88 1.71 .78 1.73
Family level (FICS)
Playfulness 2.96 1.15 2.68 .98 2.29 1.04 3.73* HD
,
EN
Cohesiveness 3.72 1.03 3.34 1.06 2.71 1.19 8.37*** HD
,
EN, HO
Adaptability 4.16 1.08 3.89 1.06 3.65 1.11 2.25
Conflict/negativity 1.34 .69 1.55 .82 1.81 .91 3.32* HD
.
EN
Child outcomes
Total externalizing 48.52 11.48 49.36 10.60 56.73 13.16 5.22** HD
.
EN, HO
Total internalizing 45.68 9.20 46.50 9.53 49.47 11.71 1.40
Total problems 47.24 11.91 48.20 10.70 55.13 12.31 4.76* HD
.
EN, HO
Negative affect/
noncompliance with peers 8.06 7.92 10.40 9.97 14.00 12.26 3.26* HD
.
EN
a
n
5
50.
b
n
5
44.
c
n
5
31.
d
Post hoc Scheffé significant differences,
p
,
.05, represented by less than sign (
,
). FICS
5 Family-Level and Co-Parenting Interaction
Coding System.
*p , .05; ** p , .01; ***p , .001.
Table 4 Chi-Square Significance Test on Marital Type 3 Al-
liance Formation Family-Level Process
Engageda
n (%)
Hostileb
n (%)
Hostile-
Detachedc
n (%)
Balanced 42 (84) 29 (66) 19 (61)
Marital 0 (0) 1 (2) 0
Parent– child 3 (6) 10 (22) 3 (10)
Disengaged 5 (1) 4 (9) 9 (29)
Note: x2 (1, N 5 6) 5 15.13, p , .05.
an 5 50.
bn 5 44.
cn 5 31.
644 Child Development
15.13, p , .05, and an inspection of cells indicated that
EN couples were more likely to have more balance al-
liances compared with the other groups, and HD cou-
ples were more likely to have a disengaged alliance
pattern compared with the other groups.
Associations between Marital Type
and Child Functioning
Marital typology groups were compared on a
number of different indexes of child maladjustment.
One-way ANOVAs were conducted on CBCL T scores
to determine the relative frequency of behavior prob-
lems across marital typology groups (see Table 3).
Children of HD couples were higher on the external-
izing factor score and total problems score of the CBCL
relative to children of EN and HO couples. In terms of
their interactions with peers, children of HD couples
also exhibited more negative affect/noncompliance
during interactions with their best friends than did
children of EN couples.
A comparison of child diagnoses across marital
groups was also conducted using x2 analysis to de-
termine the relative proportion of ODD, high-risk,
and control children in each of the marital group-
ings. The x2 was significant, x2 (4, N 5 125) 5 11.24,
p , .05, and an inspection of cell means indicated
that HD couples were less likely to have children in
the control group than in the high-risk or ODD
group, and EN families were less likely to have chil-
dren in the ODD group than in the high-risk or con-
trol group.
Predicting Unique Variance in Child Outcome
Associated with Marital Type
The next set of analyses examined whether marital
type explained unique variance in predicting child
outcome above and beyond that explained by parent-
ing, co-parenting, and family-level processes. Only
those parenting, co-parenting, and family-level vari-
ables that were significantly associated with the mar-
ital typology were entered into regression equations.
This is a conservative approach, because collinearity
between predictors mitigates against the possibility
of a unique contribution of the marital typology
groups. In all regression analyses, parenting variables
were entered in the first step, co-parenting and fam-
ily-level variables were entered in the second step,
and the marital typology was entered in the third step.
Three separate regression analyses were conducted
using the CBCL total problem score, the CBCL exter-
nalizing factor score, and the peer interaction nega-
tive affect/noncompliance score as outcome variables.
In the first multiple regression analysis predicting
mother’s report of total behavior problems on the
CBCL (see Table 5), all three predictors together ex-
plained significant amounts of variance, R 5 .50,
F(10, 111) 5 3.64, p , .001. Parenting and co-parenting/
family-level processes each added unique variance in
predicting total behavior problems. Marital typology
also added unique variance, even after controlling for
parenting, co-parenting, and family-level processes,
DR2 5 .03, DF 5 4.18, p , .05.
Table 6 summarizes results of the second multiple
regression in which the mother’s report on the CBCL
externalizing factor was the outcome measure of in-
terest. As in the first regression, the parenting, co-
parenting, and marital typology variables together
explained a significant amount of variance in child
outcome, R 5 .53, F(10, 111) 5 4.39, p , .001. Parent-
ing and co-parenting/family-level processes each
added unique variance in predicting total behavior
problems. Marital typology also added unique vari-
ance even after controlling for parenting, co-parenting,
and family-level processes, DR2 5. 03, DRF 5 4.91,
p , .05.
In the last multiple regression, in which negative
affect/noncompliance with peers was the outcome
measure (see Table 7), parenting, co-parenting/family-
level processes, and marital typology together ex-
plained a significant amount of variance in child out-
come, R 5 .44, F(10, 107) 5 2.04, p , .05. Additional
Table 5 Regression Analysis Predicting Mother’s Report on
the Child Behavior Checklist Total Problems Factor
Overall Partial
b
WeightStep and Predictors RFR
2Ft
1. Parenting .33 7.50**
Commands .32 3.48**
Bargaining/
distraction .05 .59
2. Co-parenting-family-
level processes .47 3.48** .11 2.18*
Negativity .25 2.641
Disengagement/
withdrawal 2.08 2.75
Cooperation 2.09 2.70
Playfulness .20 1.751
Cohesiveness .07 .42
Alliance formation .16 1.40
Conflict/negativity .07 .61
3. Marital typology .50 3.64*** .03 4.18*
Couple type .19 2.04*
Note: df 5 (5, 112) N 5 118.
*p , .05; ** p , .01; ***p , .001; 1p , .10.
Katz and Woodin 645
variance accounting for by co-parenting/family-level
processes approached significance, as did the effects
of marital typology after controlling for parenting,
co-parenting, and family-level processes, DR2 5 .02,
DF 5 2.88, p , .10.
Do Absolute Differences in Negative Behavior
between Marital Typology Groups Account
for Differences in Child Outcome ?
One possible explanation for the relation between
marital typology and increased child maladjustment
is that the groups differ not only in the balance of neg-
ative and positive conflict behaviors, but also in the
absolute degree of negative behaviors, and that re-
sults may reflect higher levels of negativity in chil-
dren from HD homes. A one-way ANOVA was con-
ducted to examine differences between EN, HO, and
HD couples in negative interactive behaviors. A sum-
mary code consisting of husband and wife contempt,
belligerence, criticism, and stonewalling was com-
puted. Because HD couples were theoretically ex-
pected to show higher levels of detachment (e.g.,
stonewalling) than were other groups, it is perhaps
not surprising to find that group differences were ob-
served, with HD couples showing higher levels of
marital negativity than either EN or HO couples,
F(3, 126) 5 7.40, p , .001; M 5 51.4, 57.0, and 120.3 for
EN, HO, and HD couples, respectively. Regression
analyses were then computed to see whether marital
negativity alone could predict child adjustment. Hus-
band and wife contempt, belligerence, criticism, and
stonewalling were used as predictors. For mother
CBCL total behavior problem score and mother CBCL
externalizing problems, these negative marital behav-
iors did not significantly predict child adjustment, for
mother CBCL total problem score, R 5 .29, F 5 1.30,
p 5 .27; for mother externalizing score, R 5 .32, F 5
1.60, p 5 .13. For negative affect/noncompliance with
peers, negative marital behaviors significantly pre-
dicted child behavior, R 5 .46, F 5 3.72, p , .001.
Because negative marital behaviors alone pre-
dicted negative affect and noncompliance with peers,
an additional regression analysis was conducted to
see if negative marital behaviors explained unique
variance in predicting negative peer interaction above
and beyond that explained by parenting, co-parenting
and family-level processes. As in previous regression
analyses, parenting variables were entered in the first
step, and co-parenting, and family-level variables
were entered in the second step. The negative mari-
tal behaviors of contempt, belligerence, criticism,
and stonewalling were entered as a group in the
third step. Parenting significantly predicted nega-
tive affect/noncompliance with peers, R 5 .28, F 5
4.78, p , .01, and co-parenting/family-level processes
approached significance in predicting negative peer
interaction, R 5 .42, DR2 5 .10, DF 5 1.78, p , .10.
Negative marital behaviors also added unique vari-
ance even after controlling for parenting, co-parent-
Table 7 Regression Analyses Predicting Negative Affect/
Noncompliance with Peers
Overall Partial
b
WeightStep and Predictors RFR
2Ft
1. Parenting .28 4.78*
Commands .26 2.74**
Bargaining/
distraction .05 .56
2. Co-parenting-family-
level processes .42 2.50* .10 1.781
Negativity .08 .78
Disengagement/
withdrawal 2.09 2.74
Cooperation .05 .33
Playfulness .11 .91
Cohesiveness 2.12 2.71
Alliance formation .32 2.81**
Conflict/negativity 2.14 21.23
3. Marital typology .44 2.57** .02 2.881
Couple type .17 1.70
Note: df 5 (5, 122), N 5 128.
*p , .05; ** p , .01; ***p , .001; 1p , .10.
Table 6 Regression Analysis Predicting Mother’s Report on
the Child Behavior Checklist Externalizing Factor
Overall Partial
b
WeightStep and Predictors RFR
2Ft
1. Parenting .37 9.20***
Commands .35 3.93***
Bargaining/
distraction .04 .46
2. Co-parenting-family
level processes .50 4.19*** .12 2.52*
Negativity .21 2.30*
Disengagement/
withdrawal 2.14 21.25
Cooperation 2.07 2.57
Playfulness .22 2.01*
Cohesiveness .03 .18
Alliance formation .14 1.28
Conflict/negativity .15 1.42
3. Marital typology .53 4.39*** .03 4.91
Couple type .20 2.22*
Note: df 5 (5, 112), N 5 118.
*p , .05; ** p , .01; ***p , .001.
646 Child Development
ing, and family-level processes, R 5 .59, DR2 5 .17,
DF 5 3.24, p , .01.
It was possible to predict that HD couples differed
from other couples in overall negativity, because they
were, by definition, higher in stonewalling. Therefore,
two additional analyses were conducted to decompose
the negative marital behaviors into those that indexed
hostility and those that indexed withdrawal. First, the
last regression analysis was rerun by dropping stone-
walling from the set of negative marital behaviors to
see whether marital hostility alone was a key contribu-
tor to child outcomes. Once withdrawal was removed
from the analysis, negative marital behaviors no longer
added unique variance in predicting negative affect/
noncompliance with peers, R 5 .49, DR2 5 .06, DF 5
1.40, p 5 .22. To examine the unique role of withdrawal
in predicting negative affect/noncompliance with
peers, the last regression analysis was rerun and only
included husband and wife stonewalling to see
whether marital withdrawal alone contributed to peer
outcomes after controlling for parenting and co-
parenting/family-level processes. Marital withdrawal
no longer added unique variance in predicting nega-
tive affect/noncompliance with peers, R 5 .48, DR2 5
.009, DF 5 .62, p 5 .542. In combination with the data
indicating that hostility alone did not predict unique
variance in peer outcomes, this suggests that it was the
pattern of marital interaction rather than the presence
of any one set of negative marital behaviors that was
associated with negative peer interaction.
Do Differences in Marital Satisfaction or Marital
Violence between Marital Typology Groups
Account for Differences in Child Outcome?
Another possible explanation for the relation be-
tween marital typology and increased child malad-
justment is that the groups differ not only in the bal-
ance of negative and positive conflict behaviors, but
also in their level of marital satisfaction or marital vi-
olence, and that results may reflect lower levels of
marital satisfaction and higher levels of marital vio-
lence in HD homes.
To address this possibility, marital typology groups
were compared on marital satisfaction and violence
(see Table 8). One-way ANOVAs yielded significant
differences between marital groups on many of these
measures and Scheffe post hoc comparisons were
conducted to determine the source of group differ-
ences. The results showed that HD husbands had sig-
nificantly lower levels of marital satisfaction than did
EN and HO husbands, and HD wives had lower levels
of marital satisfaction than did EN wives. In terms of
marital violence, HD husbands reported more verbal
aggression than did EN husbands, and HD wives re-
ported more verbal aggression than did EN and HO
wives. In addition, both HD husbands and wives re-
ported more physical aggression than did EN and HO
husbands and wives. There were no group differences
in the reported use of reasoning to resolve conflict.
Because HD husbands and wives reported more
Table 8 Analysis of Variance Comparing Marital Typology Groups on Marital Quality
Engaged (EN)aHostile (HO)b
Hostile-
Detached (HD)c
Post HocSchefféd
Marital Quality MSDMSDMSDF
Locke-Wallace marital satisfaction
Husband 112.86 20.20 110.93 17.91 87.48 32.42 12.68*** HD , EN, HO
Wife 117.41 23.18 109.73 18.48 98.58 30.98 5.88*** HD , EN
Conflict tactics scale
Reasoning
Husband 20.10 13.39 22.45 14.19 22.87 14.52 .48
Wife 23.76 13.01 21.36 12.30 28.42 17.19 2.35
Verbal aggression
Husband 8.28 9.26 11.41 11.74 18.53 22.69 4.80* HD . EN
Wife 12.54 14.00 15.91 18.17 30.67 26.52 8.72*** HD . EN, HO
Physical aggression
Husband .44 1.51 .23 .60 2.59 5.56 6.72** HD . EN, HO
Wife .56 1.46 .84 2.33 2.90 5.50 5.67** HD . EN, HO
an 5 50.
bn 5 44.
cn 5 31.
dPost hoc Scheffé significant differences, p , .05, represented by less than sign (,).
*p , .05; ** p , .01; ***p , .001.
Katz and Woodin 647
physical aggression and lower marital satisfaction
than did EN and HO husbands and wives, the hy-
pothesis that these differences may account for the re-
lationship between marital typology and poor child
outcome was then tested. To address each alternative
hypothesis, two sets of analyses were conducted to
(1) examine whether group status predicted child ad-
justment after accounting for the level of marital sat-
isfaction; and (2) test whether group status predicted
child adjustment after controlling for the level of inter-
parental physical aggression. For each set of analyses,
three separate regression analyses were conducted
using the CBCL total problem score, the CBCL exter-
nalizing factor score, and the peer interaction nega-
tive affect/noncompliance score as outcome variables.
Marital satisfaction. For analyses of marital satis-
faction, mother and father report of marital satisfac-
tion was entered in the first step, and marital typol-
ogy was entered in the second step. For all three
indices of child outcomes, marital typology signifi-
cantly contributed to variance in child adjustment
above and beyond any variance accounted for by
marital satisfaction, for CBCL total problem score,
DR2 5 .03, DF 5 3.78, p , .05; for CBCL externalizing
factor score, DR2 5 .027, DF 5 3.46, p , .05; and for
negative affect/noncompliance with peers, DR2 5
.033, DF 5 3.93, p , .05.
Marital violence. For analyses of marital violence,
mother and father report of physical marital aggres-
sion was entered in the first step, and the marital ty-
pology was entered in the second step. For all three
indices of child outcomes, marital typology signifi-
cantly contributed to variance in child adjustment
above and beyond any variance accounted for by
physical marital aggression, for CBCL total problem
score, DR2 5 .03, DF 5 3.98, p , .05; for CBCL exter-
nalizing factor score, DR2 5 .035, DF 5 4.74, p , .05;
and for negative affect/noncompliance with peers,
DR2 5 .033, DF 5 3.88, p , .05.
Adding family processes. Two sets of regression
analyses were also conducted to see whether parent-
ing, co-parenting, and family-level processes continued
to add significant risk of child maladjustment above
and beyond marital typology, marital satisfaction, or
physical marital aggression. In these analyses, a third
step was added to the hierarchical regressions de-
scribed above, consisting of the parenting, co-parenting,
and family-level variables. To examine each alterna-
tive hypothesis, separate regressions were conducted
for marital satisfaction and marital violence.
Analyses related to marital satisfaction indicated
that for all three indices of child adjustment, parent-
ing, co-parenting, and family-level processes continued
to add unique variance to explaining child outcomes
after controlling for marital satisfaction and marital
typology, for CBCL total problem score, R 5 .51,
DR2 5 .18, DF 5 2.87, p , .05; for CBCL externalizing
factor score, R 5 .54, DR2 5 .21, DF 5 3.46, p , .05; and
for negative affect/noncompliance with peers, R 5 .44,
DR2 5 .14, DF 5 2.00, p , .05. Similarly, for analyses
related to marital violence, parenting, co-parenting,
and family-level processes also added significant risk
of child maladjustment above and beyond marital
violence and marital typology for all measures of
child outcome, for CBCL total problem score, R 5 .51,
DR2 5 .18, DF 5 2.87, p , .05; for CBCL externalizing
factor score, R 5 .54, DR2 5 .21, DF 5 3.46, p , .05; and
for negative affect/noncompliance with peers, R 5 .45,
DR2 5 .15, DF 5 2.12, p , .05.
DISCUSSION
The results of the present study suggest that it is
the combination of hostility and detachment within the
marital relationship that is the most destructive form
of marital conflict and is associated with maladjust-
ment throughout many levels of the family system.
When husbands and wives engage in both attacking
and withdrawing behavior in their communication
with one another, they tend to have children who dis-
play negative affect and noncompliance with peers
and who have higher levels of externalizing problems
than families in which couples are either hostile alone
or are conflict engaging. Furthermore, findings could
not be explained by the level of marital satisfaction or
marital violence.
Hostility and detachment in the marital relation-
ship also related to the couple’s ability to parent and
co-parent after an argument. Parents in HD marriages
tended to issue more commands when seeking child
compliance. The level of hostility and withdrawal
witnessed in the couples’ interactions with each other
also continued when they attempted to co-parent
their child. There was often an undercurrent of anger
and frustration, as well as a distance between the cou-
ple as they interacted with their child. In addition,
these couples were less likely to collaborate together
effectively in engaging their child in activities. This
continuance of disengagement and hostility points to
a spillover effect similar to that found by Katz and
Gottman (1996) in which difficulties in the marital re-
lationship relate to the couple’s difficulties in co-
parenting their child. Similarly, at the family level,
HD couples and their children appeared to interact in
a disjointed manner, with less playfulness and cohe-
sion and more conflict in their interactions. The fam-
ily itself appeared to have difficulty synchronizing
their interactions and engaging successfully as a unit.
648 Child Development
The coupling of hostility and withdrawal appears
to reflect a fundamentally different style of resolving
marital disputes than occurs when couples either en-
gage effectively in conflict resolution or use hostility
alone. Although the tendency to withdraw from con-
flict appears to be an important addition, it is not simply
withdrawal but rather withdrawal in the face of hos-
tility that is operating in HD couples. Conflict-avoiding
couples who are relatively more withdrawn even in
the absence of hostility have marriages that are just as
stable as conflict-engaging couples (Gottman, 1993a),
suggesting that withdrawal alone is not necessarily
destructive to relationship quality. To the extent that
this is true, children in these families may be experi-
encing and displaying the same type of emotional
dysregulation with peers and others that they ob-
serve their parents to display at home.
The use of withdrawal in the face of hostility may
be a more primitive coping strategy that is indicative
of one partner feeling so overwhelmed by the level of
negativity that he or she needs to shut down the con-
versation. Alternatively, the withdrawal may reflect a
more fundamental problem in the marriage. Al-
though both HO and HD couples showed a dysregu-
lated pattern in which they expressed increasing lev-
els of negative affect as they attempted to solve
marital differences, HD couples were not only dys-
regulated as speakers but also as listeners. Perhaps
once marital differences result in changes in the non-
verbal channel of emotional expression, couples are
at a stage at which more direct verbal communication
has been unsuccessful. Less satisfaction with their re-
lationship was reported by HD couples than HO cou-
ples, a finding consistent with a profusion of previous
studies that examined hostility and withdrawal in
marriages (Gottman & Krokoff, 1989; Heavey et al.,
1995; Kurdek, 1995; Noller, Feeney, Bonnell, & Callan,
1994; Smith, Vivian, & O’Leary, 1990). These HD cou-
ples were also more likely to report verbal and phys-
ical aggression by both husbands and wives than
were EN or HO couples.
There are several reasons why the mixture of hos-
tility and withdrawal appears to be most destructive
to child and family functioning. Given their difficulty
in resolving marital disputes, lower marital satisfac-
tion, and more verbal and physical aggression, HO
couples may be further along the trajectory toward
divorce. Gottman (1993b) proposed a cascade theory
in which couples at risk of dissolution start out mutu-
ally hostile toward one another but, as their relation-
ship continues to deteriorate, their behavior becomes
more and more withdrawn and hopeless. Associated
with this hopelessness are thoughts of divorce and in-
creased loneliness and isolation within the relation-
ship. One possible reason for why children in families
with HD marital relationships show the most nega-
tive child outcomes may be because the marital rela-
tionship is closer to dissolution.
Another possibility is that couples in HD mar-
riages are more likely to have difficulties co-parenting
their children, and family interactions are likely to be
more conflictual and asynchronous. There is now
growing evidence that low levels of warmth, har-
mony, and mutuality and high levels of conflict and
competitiveness in the co-parenting or whole family
system are associated with higher child externalizing
behaviors (Lindahl & Malik, 1999; McHale & Ras-
mussen, 1998) and poorer peer relations (Katz & Gott-
man, 1996; McHale, Johnson, & Sinclair, 1999). Unlike
marital interaction that can occur behind closed doors
or otherwise outside of the earshot of children, the in-
teractive processes that occur when parents work
jointly in parenting their children, by definition, can-
not be hidden from children. The lack of playfulness
and fun, greater degree of conflict, and lack of cohe-
sion seen at the co-parenting and family levels closely
parallel the hostility and withdrawal occurring in the
marriage. If children are exposed to these processes,
they not only may have poor models of successful
conflict resolution, but also may miss the experience
of intimacy and sense of security that comes with coor-
dinated and loving family engagement.
Another possibility is that couples in HD mar-
riages are more likely to use a more power-assertive
approach to parenting. The HD couples in the present
study tended to be more restrictive, but did not pro-
vide a rationale for their demands. In contrast, the
parents in HO marriages tried to coax their children
in an effort to gain compliance. These parents offered
some type of reward if the child engaged in a desired
behavior, and changed the focus of the child’s atten-
tion by substituting one undesirable activity with an-
other more appropriate one. This form of negotiation
appears to be a more sophisticated means of gaining
child compliance, and one in which the child’s desires
are solicited. One consequence of being in a family in
which couples are HD may be that the methods that
family members use to influence one another are
more power assertive, and there is considerable evi-
dence that coercive processes within the family are
associated with externalizing behaviors in children
(e.g., Patterson, 1982).
Several competing hypotheses were also exam-
ined. The present study showed that marital satisfac-
tion and marital violence did not explain differences
between marital typology groups in child outcomes.
Absolute differences in negative behavior between
marital groups did not account for differences in
Katz and Woodin 649
mother’s ratings of children’s behavior problems, but
did relate to children’s negative affect and noncom-
pliance with peers. It may be the case that children are
modeling individual negative marital behaviors in
their peer interaction as well as (like their parents)
failing to modulate the overall negativity of their
interaction over time. Children’s ability to regulate
emotion within the context of peer interaction could
be examined by creating similar cumulative point
graphs using peer interaction data, and relating their
regulatory abilities to their parents’ marital behavior.
These data extend previous research efforts to
identify constructive and destructive qualities of mar-
ital interaction (Cummings & Wilson, 1999; Katz &
Gottman, 1996). For example, Katz and Gottman (1996)
found that marital hostility predicted teacher ratings
of child externalizing difficulties 3 years later. Most
previous research, however, has taken a unidimen-
sional view of marriage and examined the degree to
which single dimensions of marital conflict are asso-
ciated with detrimental child effects. Although exam-
ining individual dimensions of marital conflict may
appear to provide a more parsimonious explanation
of the qualities of marital conflict associated with
poor child adjustment, balancing the goal of parsi-
mony with that of description is a key dialectic in sci-
entific research (Popper, 1968). The main drawback in
examining single dimensions of marital conflict is
that it not only fails to adequately describe the com-
plexity of marital interaction patterns, but also ig-
nores the fact that individual marital dimensions tend
to cluster together in meaningful ways.
In this study, a categorical approach was used to
create a typology of marriage. This typology reflected
not only the balance of negative and positive affect
that was expressed during conflict resolution, but also
the degree to which couples escalated or de-escalated
negative affect (i.e., how dysregulated they were),
and the presence of both hostility and withdrawal in
the marriage. Such profiling allowed us to explore in-
dividual differences between families and, in particu-
lar, compare families in which couples displayed only
attacking behaviors with those in which couples dis-
played both attacking and withdrawal behaviors. The
results point to a qualitative difference in child and
family functioning in families headed by couples in
HD and HO marital relationships.
The method used to create a typology of married
couples based on their observed affect during marital
conflict interactions has proven useful for examining
links between marital functioning and the associated
adjustment of the child and competency of the family
as a whole. Limitations to the current study must be
addressed, however. One limitation of this marital ty-
pology is that it was not possible to detect whether
one or both spouses were hostile or withdrawing, or the
gender of the spouse displaying the hostility or with-
drawal. We chose to collapse across gender in the ser-
vice of parsimony, because crossing gender by
listener and speaker behavior would have yielded a
16-cell design; however, that decision caused us to
lose interesting information about the effects that par-
ent gender might have.
Sample characteristics also present limitations to
the data. These findings were based on a sample of
families screened for conduct problems. By selecting
children with significant levels of conduct problems,
families with higher levels of marital discord may
have been oversampled. Oversampling for behavior-
ally disordered children may have also increased the
possibility that children’s disruptive parenting may
have placed a strain on parents’ child-rearing abili-
ties, co-parenting, and family processes. Directional-
ity of effects cannot be determined in these data, and
the possibility remains that such child characteristics
may also place a strain on marital relations and be
associated with hostility and withdrawal in the mar-
riage. Furthermore, because families with conduct-
problem children tend to engage in coercive inter-
actions (Patterson, 1982), this may also have contrib-
uted to the relatively low rate of avoiding couples in
this sample. Gottman (1993a) reported that 15% of
his sample consisted of conflict-avoiding couples
compared with 4% in the present study’s sample.
Replication using samples other than children at risk
for externalizing behavior is necessary to examine
whether there are other domains of child and family
functioning that are impacted by marital hostility
and detachment.
ACKNOWLEDGMENTS
This research was supported by a research grant from
the National Institute of Mental Health to the first au-
thor (1 R01 49141). The authors thank the families and
peers involved in this research for their participation.
Special thanks to Angela Mittmann, Sabina Low,
Anna Buskirk, Natali Kahm, Larry Young, and Julie
Udd for their hard work at observational coding.
ADDRESSES AND AFFILIATIONS
Corresponding author: Lynn Fainsilber Katz, De-
partment of Psychology, University of Washington,
Box 351525, Seattle, WA 98195; E-mail: katzlf@u.
washington.edu. Erica M. Woodin is also at the Uni-
versity of Washington.
650 Child Development
REFERENCES
Achenbach, T. M., & Edelbrock, C. S. (1983). Manual for
the Child Behavior Checklist and Revised Child Behavior
Profile. Burlington: University of Vermont Department
of Psychiatry.
Bradbury, T. N., & Karney, B. R. (1993). Longitudinal study
of marital interaction and dysfunction: Review and anal-
ysis. Clinical Psychology Review, 13, 15– 27.
Brown, P. C., & Smith, T. W. (1992). Social influence, mar-
riage, and the heart: Cardiovascular consequences of in-
terpersonal control in husbands and wives. Health Psy-
chology, 11, 88–96.
Burman, B., John, R. S., & Margolin, G. (1992). Observed
patterns of conflict in violent, nonviolent, and nondis-
tressed couples. Behavioral Assessment, 14, 15– 37.
Christensen, A. (1987). Detection of conflict patterns in cou-
ples. In K. Hahlweg & M. J. Goldstein (Eds.), Understand-
ing major mental disorder. The contribution of family interac-
tion research. New York: Family Process Press.
Christensen, A., & Heavey, C. L. (1990). Gender and social
structure in the demand/withdraw pattern of marital
conflict. Journal of Personality and Social Psychology, 59,
73– 81.
Cowan, P. A., & Cowan, C. P. (1987, April). Couples relation-
ships, parenting styles and the child’s development at three.
Paper presented at the biennial meeting of the Society
for Research in Child Development, Baltimore.
Cowan, P. A., Cowan, C. P., Schultz, M. S., & Heming, G.
(1994). Prebirth to preschool family factors in children’s
adaptation to kindergarten. In R. D. Parke & S. G. Kellam
(Eds.), Exploring family relationships with other social con-
texts. Family research consortium: Advances in family re-
search (pp. 75– 114). Hillsdale, NJ: Erlbaum.
Cummings, E. M., Ballard, M., El-Sheikh, M., & Lake, M.
(1991). Resolution and children’s responses to interadult
anger. Developmental Psychology, 27, 462–470.
Cummings, E. M., & Davies, D. (1994). Children and marital
conflict: The impact of family dispute and resolution. New
York: Guilford Press.
Cummings, E. M., Vogel, D., Cummings, J. S., & El-Sheikh, M.
(1990). Children’s responses to different forms of expression
of anger between adults. Child Development, 60, 1392–1404.
Cummings, E. M., & Wilson, A. (1999). Contexts of marital
conflict and children’s emotional security: Exploring the
distinction between constructive and destructive conflict
from the children’s perspective. In M. J. Cox & J. Brooks-
Gunn (Eds.), Conflict and cohesion in families: Causes and
consequences (pp. 105– 129). Mahwah, NJ: Erlbaum.
Emery, R. E. (1982) Interparental conflict and the children of
discord and divorce. Psychological Bulletin, 92, 310–330.
Emery, R. E. (1999). Marriage, divorce and children’s adjust-
ment. Thousand Oaks, CA: Sage.
Emery, R. E., & O’Leary, D. (1982). Children’s perceptions of
marital discord and behavior problems of boys and girls.
Journal of Abnormal Child Psychology, 10, 11–24.
Emery, R. E., & O’Leary, D. (1984). Marital discord and child
behavior problems in a nonclinic sample. Journal of Ab-
normal Child Psychology, 12, 411–420.
Erel, O., & Burman, B. (1995). Interrelatedness of marital
relations and parent-child relations: A meta-analytic re-
view. Psychological Bulletin, 118, 108–132.
Fincham, F. D. (1994). Understanding the association be-
tween marital conflict and child adjustment: Overview.
Journal of Family Psychology, 8, 123–127.
Fincham, F. D., & Osborne, L. N. (1993). Marital conflict and
children: Retrospect and prospect. Clinical Psychology Re-
view, 13, 75–88.
Fisher, P., Wicks, J., Shaffer, D., Piacentini, J., & Lapin, J.
(1993). National Institute of Mental Health, Diagnostic In-
terview Schedule for Children, Version 2.3.
Fitzpatrick, M. A., & Indvik, J. (1982). The instrumental and
expressive domains of marital communication. Human
Communication Research, 8, 195– 213.
Gjerde, P. R. (1986). The interpersonal structure of family in-
teraction settings: Parent-adolescent relations in dyads
and triads. Developmental Psychology, 22, 297–304.
Gottman, J. M. (1983). How children become friends. Mono-
graphs of the Society for Research in Child Development, 48(2,
Serial No. 201).
Gottman, J. M. (1993a). The roles of conflict engagement, es-
calation, and avoidance in marital interaction: A longitu-
dinal view of five types of couples. Journal of Consulting
and Clinical Psychology, 61, 6–15.
Gottman, J. M. (1993b). A theory of marital dissolution and
stability. Journal of Family Psychology, 7, 57–75.
Gottman, J. M. (1994). What predicts divorce? Hillsdale, NJ:
Erlbaum.
Gottman, J. M., Coan, J., Carrère, S., & Swanson, C. (1998).
Predicting marital happiness and stability from newlywed
interactions. Journal of Marriage and the Family, 60, 5– 22.
Gottman, J. M., & Katz, L. F. (1989). Effects of marital dis-
cord on young children’s peer interaction and health.
Developmental Psychology, 25, 373–381.
Gottman, J. M., & Krokoff, L. J. (1989). Marital interaction
and satisfaction: A longitudinal view. Journal of Consult-
ing and Clinical Psychology, 57, 47–52.
Gottman, J. M., & Levenson, R. W. (1992). Marital processes pre-
dictive of later dissolution: Behavior, physiology, and health.
Journal of Personality and Social Psychology, 63, 221–233.
Gottman, J. M., & Parker, J. G. (Eds.). (1986). Conversations of
friends: Speculations on affective development. Cambridge,
U.K.: Cambridge University Press.
Grych, J. H., & Fincham, F. D. (1990). Marital conflict and
children’s adjustment: A cognitive contextual frame-
work. Psychological Bulletin, 108, 267–290.
Grych, J. H., & Fincham, F. D. (1993). Children’s appraisals
of marital conflict: Initial investigations of the cognitive-
contextual framework. Child Development, 64, 215– 230.
Haley, J. (1965). Control in brief psychotherapy. Archives of
General Psychiatry, 4, 139–153.
Heavey, C. L., Christensen, A., & Malamuth, N. M. (1995).
The longitudinal impact of demand and withdrawal
during marital conflict. Journal of Consulting and Clinical
Psychology, 63, 797–801.
Howes, P., & Markman, H. J. (1989). Marital quality and
child functioning: A longitudinal investigation. Child
Development, 60, 1044– 1051.
Katz and Woodin 651
Jouriles, E. N., Murphy, C. M., Farris, A. M., Smith, D. A.,
Richters, J. E., & Waters, E. (1991). Marital adjustment,
parental disagreements about child rearing, and behav-
ior problems in boys: Increasing the specificity of the
marital assessment. Child Development, 62, 1424– 1433.
Katz, L. F., & Gottman, J. M. (1993). Patterns of marital con-
flict predict children’s internalizing and externalizing
behaviors. Developmental Psychology, 29, 940–950.
Katz, L. F., & Gottman, J. M. (1995). Marital interaction
and child outcomes: A longitudinal study of mediating
and moderating processes. In D. Cicchetti & S. L. Sheree
(Eds.), Emotion, cognition, and representation. Rochester
symposium on developmental psychopathology, Vol. 6
(pp. 301– 342). Rochester, NY: University of Rochester
Press.
Katz, L. F., & Gottman, J. M. (1996). New directions for child
development: No., 74. Spillover effects of marital conflict: In
search of parenting and co-parenting mechanisms (pp. 57–
76). San Francisco: Jossey-Bass.
Katz, L. F., Low, S., Young, L., & Kahm, N. (1997). Family-
level interaction co-parenting system. Unpublished manu-
script, University of Washington, Seattle.
Kurdek, L. A. (1995). Predicting change in marital satisfac-
tion from husbands’ and wives’ conflict resolution
styles. Journal of Marriage and the Family, 57, 153– 164.
Levenson, R. E., & Gottman, J. M. (1985). Physiological and
affective predictors of change in relationship satisfac-
tion. Journal of Personality and Social Psychology, 49, 85–
94.
Lindahl, K. M., & Malik, N. M. (1999). Marital conflict, fam-
ily processes, and boys’ externalizing behavior in His-
panic American and European American families. Jour-
nal of Clinical Child Psychology, 28, 12–24.
Locke, H. J., & Wallace, K. M. (1959). Short marital adjust-
ment and prediction tests: Their reliability and validity.
Marriage and Family Living, 21, 251– 255.
Margolin, G., John, R. S., & Gleberman, L. (1988). Affective
responses to conflictual discussions in violent and non-
violent couples. Journal of Consulting and Clinical Psychol-
ogy, 56, 24–33.
McHale, J. P. (1997). Overt and covert co-parenting pro-
cesses in the family. Family Process, 36, 183– 201.
McHale, J. P., & Cowan, P. A. (Eds.). (1996). Understanding
how family-level dynamics affect children’s development:
Studies of two-parent families. San Francisco: Jossey-Bass.
McHale, J. P., Johnson, D., & Sinclair, R. (1999). Family dy-
namics, preschoolers’ family representations, and pre-
school peer relationships. Early Education and Develop-
ment, 10, 373– 401.
McHale, J. P., & Rasmussen, J. L. (1998). Coparental and
family group-level dynamics during infancy: Early fam-
ily precursors of child and family functioning during
preschool. Development and Psychopathology, 10, 39– 59.
Minuchin, S. (1974). Families and family therapy. Cambridge,
MA: Harvard University Press.
Mittmann, A. J., & Katz, L. F. (1997). The Forbidden Toy Para-
digm Coding System. Unpublished manuscript, Univer-
sity of Washington, Seattle.
Newton, D. A., & Burgoon, J. K. (1990). The use and conse-
quences of verbal influence strategies during interper-
sonal disagreements. Human Communication Research, 16,
477– 518.
Noller, P., Feeney, J. A., Bonnell, D., & Callan, V. J. (1994). A
longitudinal study of conflict in early marriage. Journal
of Social and Personal Relationships, 11, 233–252.
Patterson, G. (1982). Coercive family process. Castalia, OR.
Popper, K. R. (1968). The logic of scientific discovery. New
York: Harper & Row.
Robinson, E. A., Eyberg, S. M., & Ross, A. W. (1980). The stan-
dardization of an inventory of child conduct problem be-
haviors. Journal of Clinical Child Psychology, 9, 22–28.
Sears, R. R., Maccoby, E. E., & Levin, H. (1957). Patterns of
child rearing. Stanford, CA: Stanford University Press.
Smith, D. A., Vivian, D., & O’Leary, D. (1990). Longitudinal
prediction of marital discord from premarital expres-
sions of affect. Journal of Consulting and Clinical Psychol-
ogy, 58, 790–798.
Straus, M. A. (1979). Measuring intrafamily conflict and vi-
olence: The conflict tactics (CT) scales. Journal of Marriage
and the Family, 41, 75–88.
Toner, I. J., Parke, R. D., & Yussen, S. R. (1979). The effect of
observation of model behavior on the establishment and
stability of resistance to deviation in children. Journal of
Genetic Psychology, 132, 283–290.
This document is a scanned copy of a printed document. No warranty is given about the accuracy of the copy.
Users should refer to the original published version of the material.
