Annu. Rev. Psychol. 1998.

49:169–97
Copyright © 1998 by Annual Reviews Inc. All rights reserved

PSYCHOLOGY AND THE STUDY

OF MARITAL PROCESSES
John Mordechai Gottman
Department of Psychology, University of Washington, Seattle, Washington 98195;
e-mail: johng@u.washington.edu

KEY WORDS: marriage, relationships, divorce, children, sex

ABSTRACT

The divorce rate in the United States is extremely high. It is estimated that
between 50% and 67% of first marriages end in divorce. For second mar-
riages, failure rates are even higher. There are strong negative consequences
to separation and divorce on the mental and physical health of both spouses,
including increased risk for psychopathology, increased rates of automobile
accidents, and increased incidence of physical illness, suicide, violence,
homicide, significant immunosuppression, and mortality from diseases. In
children, marital distress, conflict, and disruption are associated with de-
pression, withdrawal, poor social competence, health problems, poor aca-
demic performance, and a variety of conduct-related difficulties. Though
intervention techniques might be expected to reduce these grim statistics, our
best scholars have concluded that marital therapy is at a practical and
theoretical impasse. This article discusses the progress of research on the
study of marriage.

CONTENTS
WHY STUDY MARRIAGE? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
PSYCHOLOGY’S EARLY YEARS IN STUDYING MARRIAGE . . . . . . . . . . . . . . . . . . . 171
THE GLOP PROBLEM AND REFUSING TO LIVE WITH GLOP . . . . . . . . . . . . . . . . . . 172
THE RELUCTANCE TO OBSERVE COUPLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
LABORATORY OBSERVATION OF COUPLES START GETTING CONSISTENT
RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
New Methods, New Concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
New Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
NEGATIVE AFFECT RECIPROCITY AS THE FAILURE OF REPAIR ATTEMPTS . . . 179

169
0066-4308/98/0201-0169$08.00
170 GOTTMAN

WHAT IS DYSFUNCTIONAL IN AILING MARRIAGES? . . . . . . . . . . . . . . . . . . . . . . . . 181

The Criterion of Correlates with Marital Satisfaction . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
The Longitudinal Criterion: Divorce Prediction Research . . . . . . . . . . . . . . . . . . . . . . . 183
WHAT WE HAVE LEARNED: THE PSYCHOLOGY OF MARRIAGE . . . . . . . . . . . . . . 184
The View from Observing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
The View from Cognitive Psychology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
The View from Psychophysiology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
The View from Intervention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
SUMMARY: AN INTEGRATIVE MODEL OF HOW MARRIAGES MAY
DYSFUNCTION AND FUNCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
The Stable Phenomena of Marital Process and Outcomes . . . . . . . . . . . . . . . . . . . . . . . . 190
What Is Dysfunctional in Unhappy, Unstable Marriages? . . . . . . . . . . . . . . . . . . . . . . . . 190
An Intergrative Account: the Bank Account Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191

WHY STUDY MARRIAGE?

The divorce rate remains extremely high in the United States. Current esti-
mates of the chances of first marriages ending in divorce range between 50%
and 67% (Martin & Bumpass 1989). Failure rates for second marriages are
about 10% higher than for first marriages.
We now know that separation and divorce have strong negative conse-
quences for the mental and physical health of both spouses. These negative ef-
fects include increased risk for psychopathology; increased rates of automo-
bile accidents including fatalities; and increased incidence of physical illness,
suicide, violence, homicide, significant immunosuppression, and mortality
from diseases (for a review, see Bloom et al 1978, Burman & Margolin 1992).
Marital distress, conflict, and disruption are also associated with a wide range
of deleterious effects on children, including depression, withdrawal, poor so-
cial competence, health problems, poor academic performance, and a variety
of conduct-related difficulties (Cowan & Cowan 1987, 1990; Cowan et al
1991; Cummings & Davies 1994; Easterbrooks 1987; Easterbrooks & Emde
1988; Emery 1982, 1988; Emery & O’Leary 1982; Forehand et al 1986; Gott-
man & Katz 1989; Hetherington 1988; Hetherington & Clingempeel 1992;
Hetherington et al 1978, 1982; Howes & Markman 1989; Katz & Gottman
1991a,b; Peterson & Zill 1986; Porter & O’Leary 1980; Rutter 1971; Shaw &
Emery 1987; Whitehead 1979). There is evidence from two US national prob-
ability samples that adults who experienced a divorce as a child are under con-
siderably more stress than those who did not (Glenn & Kramer 1985, Kulka &
Weingarten 1979). These adults report less satisfaction with family and
friends, greater anxiety, that bad things more frequently happen to them, and
that they find it more difficult to cope with life’s stresses in general. In a recent
report based on the Terman longitudinal study of gifted children (Friedman et
al 1995), survival curves show that the combination of one’s parents having di-
 MARITAL PROCESSES 171

vorced and one’s own divorce reduced longevity by an average of approxi-

mately eight years.
Intervention techniques such as marital therapy might be expected to re-
duce these grim statistics. Our best scholars have concluded that, unfortu-
nately, marital therapy is at a practical and a theoretical impasse (Jacobson &
Addis 1993). Outcome results suggest the following. 1. Most couples (75%)
report improvement in marital satisfaction immediately following marital
therapy. 2. All therapies are about equally effective in this regard, regardless of
the “school” of therapy, once replication studies have been done. 3. There is a
strange effect that separate “components” of an intervention are often equally
effective, and about as effective as the combined treatment; thus, it has proven
difficult to build a theory of change based on dismantling a complex interven-
tion. 4. There is a large relapse effect. In general, after long-term follow-up,
only between 30% and 50% of couples stay improved. In fact, the relapse data
are probably much more grim than these conclusions suggest. As far as we
know, long-term follow-up is likely to yield evidence of even greater relapse.

PSYCHOLOGY’S EARLY YEARS IN STUDYING

MARRIAGE
Psychology was a latecomer to the study of marriage. Sociologists had been
studying marriages for 35 years before psychologists became interested in the
topic, although the first published study on marriage was by a psycholo-
gist—Louis Terman—in 1938 (Terman et al 1938). What psychologists ini-
tially brought to the study of marriage was the use of observational methods,
the design and evaluation of intervention programs, and an unbridled opti-
mism that changing marriages was going to be easy and quick work. These
contributions have had an enormous impact on the study of marriage. Psy-
chologists were initially skeptical about studying marriage, in part because
personality theory was facing a severe challenge in the 1970s from the work of
Walter Mischel. In Personality and Assessment (Mischel 1968), Mischel re-
viewed research on personality and suggested that personality theory had
come far short of being able to predict and understand behavior. He concluded
that correlations were quite low on the whole, that the field was plagued with
common method variance (mostly self-reports predicting self-reports), and
that the best predictors of future behavior were past behavior in similar situa-
tions. This book was a great stimulus to many researchers. It encouraged a new
look at personality measurement, validity, and reliability (Wiggins 1973) and
stimulated new kinds of research in personality. It contributed, however, to a
pessimistic view that research in interpersonal psychology would have very
little payoff. In hindsight, however, this view was wrong. We are in fact learn-
172 GOTTMAN

ing that much of the order in individual personality exists at the interpersonal
level. For example, Patterson (1982), in his conclusion that there is a great deal
of consistency across time and situations in aggression, suggested that the ag-
gressive trait should be rethought in interpersonal terms as the aggressive
boy’s recasting people in his social world to play out dramatic coercive scenes
shaped in his family. The same is true of gender differences: They appear to
emerge primarily in the context of relationships (Maccoby 1990).
The use of self-report measures, including personality measures, had initially
dominated the field of marriage research. Unfortunately, even with the prob-
lem of common method variance, the self-report paper-and-pencil personality
measures yielded relatively weak correlations of marital satisfaction (Burgess
et al 1971). As an example, psychoanalysis initially embraced Winch’s comple-
mentarity of need theory (Winch 1958), which proposed that happy marriage
would be associated with spouses who have complementary needs (such as she
needs to dominate and he needs to be dominated). No other theory in this field
has ever been as soundly rejected as Winch’s. Not until studies asked spouses
to fill out questionnaires about their spouse’s personality were substantial cor-
relations discovered with marital satisfaction. Unhappily married couples
were found to endorse nearly every negative trait as characteristic of their
spouses (the negative halo effect), whereas happily married spouses were
found to endorse nearly every positive trait as characteristic of their spouses
(the positive halo effect) (Nye 1988).

THE GLOP PROBLEM AND REFUSING TO LIVE WITH

GLOP
Bank et al (1990) wrote an important paper subtitled “Living with Glop.” Glop
refers to high correlations among variables obtained using a common method
of measurement, usually with just one reporter. In the marital field, this usually
refers to self-report data obtained from a single reporter. The point to be made
about glop is an old one. Bank et al noted that reviews by Rutter (1979) and
Emery (1982) showed consistent correlations between marital discord and
child adjustment problems, but that Emery & O’Leary (1984) pointed out that,
in most of these studies, both variables come from one person (usually the
mother), who uses self-report measures. It was the experience of Emery &
O’Leary and other investigators that the correlations between parents’ and
teachers’ reports of child behaviors were essentially zero. Structural Equations
Modeling (SEM) was developed in part to deal with the problem of increasing
reliability and validity in measurement of constructs. The SEM methods hark
back to a classic paper in measurement by Campbell & Fiske (1959), in which
they argued for the use of a multitrait-multimethod matrix. It is quite common
 MARITAL PROCESSES 173

to be able to create beautiful SEM models when all the data are collected by the
same agent and self-report data are employed. However, once the issue of va-
lidity is introduced to the model (just adding the requirement that the data cor-
relate with measurement of the same construct using a different reporter or dif-
ferent method), the models typically fall apart.
In studying marriage, the glop problem is also of great theoretical concern.
For example, it is commonly observed that measures of neuroticism are corre-
lated with marital satisfaction (Kurdek 1993). It is, however, difficult to dem-
onstrate that one is really measuring neuroticism rather than distress associ-
ated with being unhappily married. It is well known that unhappily married
people are more depressed, more anxious, more distressed, less optimistic, and
so on. Is the personality measure assessing an enduring trait or a state of well-
ness, distress, or the current quality of life, which is essentially the same thing
that the marital satisfaction measure taps?
The glop problem also occurs when one uses different reporters, as Kelly &
Conley (1987) did when they had friends and acquaintances of the couple fill
out the personality measures. Clearly, friends are likely to be familiar with the
general dysphoria of their unhappily married friends, and they are likely to use
the items of the personality tests they are given to report this fact. The conclu-
sion we must reach from these studies is clear: We should refuse to accept con-
structs as explanatory if they have not dealt adequately with the glop problem.
It is absolutely critical that any theory of marriage be very careful about how a
construct was measured in drawing conclusions. This is an admonition that all
studies of marriage employ multiple methods to operationalize constructs.

THE RELUCTANCE TO OBSERVE COUPLES

Researchers have been reluctant to use systematic observation to study cou-

ples, primarily because it is very costly and frustrating. In addition, it takes lots
of time and experience to develop a good coding system for marital interac-
tion, and then more time to obtain actual numbers from tapes of the interaction.
The researcher must persistently deal with issues of reliability of measurement
and problems with defining categories and interobserver reliability drift and
decay (Reid 1970). It is so much easier to hand out a packet of questionnaires.
In addition, observational measures are often somewhat atheoretical. They
usually are designed to exhaustively describe all the behavior within a particu-
lar framework that can be observed in a particular situation. In contrast, a ques-
tionnaire measures a specific set of constructs, such as egalitarianism in the
marriage, paranoid ideation in each partner, and so on. After an observation, it
is often unclear what has been measured. Consequently, a purely descriptive,
174 GOTTMAN

hypothesis-generating phase of research is required to validate the observa-

tional measures—an added time-consuming phase that is often skipped. While
observational data are often a richer source of hypotheses and are more satisfy-
ing to the truly voyeuristic researcher, they also require much more psycho-
metric work to know what one has actually measured.
Moreover, underlying the selection of categories for any observational sys-
tem are a set of assumptions and some rudimentary theory about what behav-
iors are important to observe. For example, the Marital Interaction Coding
System (MICS) emerged from behavioral marital therapy in which it was con-
sidered important to “pinpoint” the marital issue; as such, for example, vague
statements of the problem were considered “negative,” whereas specific state-
ments were considered “positive.” Only recently have Heyman et al (1995)
been able to conduct an important study of the structure of the MICS and pro-
vide an empirically based guide about what it may measure. They used an ar-
chival data set of 995 couples’ interactions coded with the Marital Interaction
Coding System–IV (MICS). Their factor analysis yielded four factors: hostil-
ity, constructive problem discussion, humor, and responsibility discussion.
Furthermore, observational approaches to the study of families in the 1960s
by General Systems theorists failed miserably. (The General Systems theorists
were a group of researchers and clinicians who began to view the family as an
interactive system and to emphasize communication patterns as a potential eti-
ology of psychopathology.) The story of this failure is very dramatic. First, a
few maverick psychologists, psychiatrists, and anthropologists in the 1950s
observed a strange pattern of interaction between adult schizophrenics and
their mothers when they visited them in the hospital. The mothers would greet
their children with warmth and then stiff coldness, all packaged in one em-
brace. The researchers called this the “double bind” theory of schizophrenia
(Bateson et al 1956), which maintained that the mother sent a mixed message
to her child and that this message put the child in a “double bind,” meaning that
he was damned if he responded to one part of the message and damned if he re-
sponded to the other (contradictory) part of the message. The General Systems
theorists thought that the resulting double bind created the emotional with-
drawal we have come to associate with schizophrenia. The way out of the dou-
ble bind was to “meta-communicate,” that is, to raise the communication to a
new level through commenting on the communication itself.
In 1964, a new journal, Family Process, was formed and was dedicated to
research on possible family origins of mental disorders. A lot of this research
was observational, because the theories focused on communication. Unfortu-
nately, this early quantitative observational research was very weak. Not a sin-
gle hypothesis of these General Systems theorists received clear support from
research, except for the finding that the communication of families with a
 MARITAL PROCESSES 175

schizophrenic member was more confusing than that of normal families (see
Jacob 1987). This hardly encouraged others to do observational research on
families.
In the early 1970s, however, several psychology laboratories did begin to
use direct observation to study marital interaction.1 These investigators were
motivated by the idea that marriages could be helped with a behavioral ap-
proach that essentially taught couples new social skills for resolving conflict.
Because these researchers all came from a behavioral perspective, observa-
tional methods were part of the assumed assessment battery. The historical
reason for this is worth noting. Early in the intervention research on families
with children who had problems, researchers (Robert Weiss, Hyman Hops, JR
Patterson) at the Oregon Social Learning Center had noted that even if the chil-
dren did not improve, the parents’ claimed that their children were greatly im-
proved in self-reports of their satisfaction with treatment. These outcomes
bred a distrust among researchers of parental reports of child status and a reli-
ance on direct observation. They also decided to design an intervention for
marriages in distress and naturally employed observational methods to help
distressed couples resolve their conflicts. Thus, from a classic book chapter
published by Weiss et al (1973), observational research on marriage was
born.The chapter was titled “A Framework for Conceptualizing Marital Con-
flict: A Technology for Altering It, Some Data for Evaluating It.” It was an op-
timistic attempt to define an entire field in one paper. It did just that.
The first question that marital researchers tackled was the same question
raised by Terman: “What makes some marriages happy and others miserable?”
Although it continues to be criticized today, marital satisfaction had shown it-
self to be a venerable criterion variable, and so the search began for the obser-
vational correlates of marital satisfaction. There was a faith that an adequate
theoretical background would be provided by social learning theory—broadly
conceptualized and coupled with principles of communication that had been
described by General Systems theorists and by behavior exchange theorists
(Thibaut & Kelley 1959). The hope was that by bringing precision to the study
of marriage, a superior theory would emerge and lead the way to effective
marital therapies. Two paths were simultaneously pursued: the attempt to an-
swer Terman’s question with psychological constructs, and the design and
evaluation of new therapies.

1
1Only a handful of laboratories have consistently included observational methods in their study
of couples. These researchers include J Alexander, D Baucom, S Beach, G Birchler, T Bradbury,
A Christensen, F Fincham, FJ Floyd, M Fitzpatrick, WC Follette, M Forgatch, H Hops, G Howe,
N Jacobson, G Margolin, H Markman, C Notarius, KD O’Leary, J Vincent, R Weiss, C Schaap
(in Holland), the Max Planck group (in Germany, including L Schindler, K Hahlweg, and D
Revenstorf), HB Vogel, K Halford (in Australia), and P Noller.
176 GOTTMAN

LABORATORY OBSERVATION OF COUPLES START

GETTING CONSISTENT RESULTS
New Methods, New Concepts
In the 1970s, developmental psychology converged methodologically with the
psychological study of marriage, and a common approach to the study of social
interaction began to emerge. Important works included Lewis & Rosenblum’s
(1974) The Effect of the Infant on Its Caregiver, and RQ Bell’s (1968) paper on
the bidirectionality of effects between parent and child. These works led re-
searchers to view interaction as sequences of behavior unfolding over time. In
1974, the landmark Communication, Conflict, and Marriage appeared; the
authors, Raush et al, had followed a cohort of newlyweds completing the tran-
sition to parenthood. Raush et al introduced the use of Markov models of se-
quential interaction and an idea they called “adaptive probabilism.” They
claimed that marriages and families should be studied as systems, and sug-
gested how this could be done mathematically using “information theory”
(Shannon & Weaver 1949) for the study of sequential patterns of interaction,
showing that the mathematics were a new approach to interaction. Instead of
this systems concept remaining a vague metaphor, or a mathematical proce-
dure for analyzing data, Raush et al realized that it was a whole new way of
thinking. It was their intention to introduce the idea of stochastic models,
which are uniquely designed for thinking in terms of systems rather than indi-
vidual behavior. Stochastic models refer to the conceptualization of behavior
sequences in terms of probabilities and the reduction of uncertainty in predict-
ing patterns of interaction. (For a systematic development of these concepts
and their mathematics, see Bakeman & Gottman 1986, Bakeman & Quera
1995, and Gottman & Roy 1990.)

New Findings
Because of space limitations, only a sampler, some general conclusions, and a
flavor of the findings are presented. Focusing on studies that included some se-
quential analyses of the data, this review is restricted to the research of Weiss
(Oregon), Raush (Massachusetts), Gottman (Washington), Schapp (Holland),
Ting-Toomey (New Jersey), the Max Planck group (Revenstorf, Hahlweg,
Schindler, and Vogel in Munich), and Fitzpatrick (Wisconsin). What were the
results of various laboratories that investigated the Terman question, and in
particular, what were the results of sequential analyses?
Of the many studies that have observed marital interaction, few have em-
ployed sequential analyses. There are only two such studies that use the MICS.
Margolin & Wampold (1981) reported the results of interaction with 39 cou-
 MARITAL PROCESSES 177

ples, combined from two studies conducted in Eugene, Oregon, and Santa Bar-
bara, California. Codes were collapsed into three global categories: positive
(problem-solving, verbal and nonverbal positive), negative (verbal and non-
verbal negative), and neutral. Distressed couples showed negative reciprocity
through Lag 2, whereas nondistressed couples did not demonstrate this effect
to any significant extent. For positive reciprocity, Margolin and Wampold
found that, “whereas both groups evidenced positive reciprocity through Lag
2, this pattern appears to continue even into Lag 3 for distressed couples” (p.
559). Thus, reciprocating positive acts were more likely between distressed
than for nondistressed couples. Gottman (1979) had reported similar results,
suggesting that distressed couples showed greater rigidity and interactional
structure than nondistressed couples.
Margolin & Wampold also defined a sequence called “negative reactivity,”
which involves a positive response to a negative antecedent by one’s spouse.
They proposed that there is a suppression of positivity following a negative an-
tecedent in distressed couples. They found this for all four lags for distressed
couples, but they found no evidence for this suppression of positivity by nega-
tivity for any lag for nondistressed couples.
Revenstorf et al (1980), studying 20 German couples, collapsed the MICS
categories into six summary codes. These codes were positive reaction, nega-
tive reaction, problem solution, problem description, neutral reaction, and
filler; interrupts, disagrees, negative solution, and commands were considered
negative. They employed both lag sequential analyses that allowed them to ex-
amine sequences out for four lags and the multivariate information theory that
Raush et al (1974) had employed. From the multivariate information analysis,
Revenstorf et al (p. 103) concluded that
In problem discussions distressed couples respond differently from nondis-
tressed couples….In particular [distressed couples] are more negative and
less positive following positive (+) and negative (-) reactions. At the same
time they are more negative and more positive, that is more emotional, fol-
lowing problem descriptions (P) of the spouse. Above all distressed couples
are more negative and less positive in general than nondistressed couples.

They also found 17 sequences that differentiated the two groups. There is
some inconsistency in the group differences for sequences with similar names
(e.g. “reconciliation”), so I summarize only their clearest results. For what
might be called constructive interaction sequences, they found that nondis-
tressed couples engaged in more validation sequences (problem description
followed by positivity) and positive reciprocity sequences (positive followed
by positive). On the destructive side, they found that distressed couples en-
gaged in more devaluation sequences (negative follows positive), negative
178 GOTTMAN

continuance sequences (which they called “fighting on” or “fighting back” in

three-chain sequences), and negative startup sequences (which they called
“yes-butting,” meaning that somewhere in the four-chain sequence, negative
follows positive) than nondistressed couples. After an analysis of the se-
quences following a problem description, Revenstorf et al (p. 107) concluded
that
It appears as if the distressed couples would interact like nondistressed—had
they only higher positive response rates following a problem description of
the spouse. And vice versa. The nondistressed would react equally detrimen-
tally as the distressed—were they to respond more negatively to problem de-
scription of their spouse. The way they handle problems [problem descrip-
tion statements] seems to be the critical issue—not the sheer number of prob-
lems stated.

Revenstorf et al also continued their sequential analyses for five lags and
found that these reciprocity differences held across lags. They wrote (p. 109):
In summary, different patterns of response tendencies emerge for distressed
and nondistressed couples. After a positive statement the partner continues
to reciprocate it positively in nondistressed, whereas no immediate response
is likely in distressed couples. After a negative statement no immediate re-
sponse is most likely in nondistressed, whereas in distressed couples both
partners continue to reciprocate negatively. A problem description finally is
repeatedly followed by a positive response in nondistressed. In distressed
couples, negative statements follow repeatedly.
Revenstorf et al then described four types of sequences. The first type of se-
quence is continued negativity (they called it “distancing”). This sequence
measures the extent to which negativity becomes an absorbing state. The sec-
ond sequence type was positive reciprocity (which they called “attraction”).
This sequence measures the extent to which positivity becomes an absorbing
state. The third sequence consisted of alternating problem descriptions and
negativity (they called it “problem escalation”). The fourth type of sequence
consists of validation sequences—sequences of alternating problem descrip-
tions and positive responses to them (they called it “problem acceptance”). In
most instances (e.g. for positive reciprocity), the differences between the
groups were not very great. The evidence, however, was very clear that nega-
tivity represented an absorbing state for distressed couples but not for nondis-
tressed couples. By Lag 2, nondistressed couples begin to escape from the
negativity, but distressed couples could not escape. Graphs of the data provide
dramatic information of group differences reflected in sequential patterning of
MICS codes. The consistent findings in these two studies and other studies that
have employed sequential analysis (Fitzpatrick 1988, Gottman 1979, Raush et
al 1974, Schaap 1982, Ting-Toomey 1982; for a review, see Gottman 1994)
 MARITAL PROCESSES 179

are that (a) unhappily married couples appear to engage in long chains of re-
ciprocated negativity, and (b) there is a climate of agreement created in the in-
teraction of happily married couples.
Most of these studies collapsed their codes into a global positive or nega-
tive. However, using a new method developed by Sackett (see Bakeman &
Gottman 1986) called “lag sequential analysis,” Gottman, Notarius, and Mark-
man (see Gottman 1979) examined sequences using specific codes of their
Couples Interaction Scoring System (CISS). These analyses revealed the anat-
omy of distressed and nondistressed marital interaction, both in the laboratory
and at home. In a series of studies that combined observation with behavior ex-
change principles (using a “talk table” device in which spouses rated each ex-
change), and individual assessments of social competence, Gottman and his
students developed an empirically based intervention, which they evaluated.
In three separate studies significant results were obtained on marital satisfac-
tion and interaction (Gottman 1979).

NEGATIVE AFFECT RECIPROCITY AS THE FAILURE OF

REPAIR ATTEMPTS
The basic sequential result that held across laboratories was that greater recip-
rocated negative affective interaction is an absorbing state for dissatisfied cou-
ples. This result has profound implications for interaction process. The result
means that negativity becomes an absorbing state for dissatisfied couples, that
is, it is a state that is difficult to exit once entered.
We need to know two additional facts about marital interaction to under-
stand the implications of this finding. Vincent et al (1979) studied the interac-
tion of distressed and nondistressed couples in the Inventory of Marital Con-
flicts, a problem-solving task. The two groups could be discriminated from
each other on five out of six MICS summary codes, positive problem solving,
and verbal and nonverbal positive and negative codes. Vincent et al then asked
the couples to try either to fake good or to fake bad during the next 10 minutes.
Both groups of couples were unable to fake their nonverbal behaviors. Hence,
nonverbal behavior may be a better discriminator of distressed and nondis-
tressed groups than verbal behavior alone. Second, Gottman (1979) found that
most couples express the most negative affect during the middle arguing phase
of the conflict resolution, and their major attempts at repair of the interaction
are usually delivered in this phase as well. Attempts at interaction repair are of-
ten delivered with negative affect. For example, statements like “Stop inter-
rupting me!” or “We’re getting off the subject” may be accompanied by irrita-
tion, tension, sadness, or some other form of distress. Thus, repair attempts
180 GOTTMAN

usually have two components, a negative affective nonverbal component and a

metacommunicative content component attempting to repair the interaction.
The implication of greater negativity being an absorbing state for dissatis-
fied couples is that they may attend primarily to the negative affect component
of repair attempts, whereas satisfied couples attend primarily to the repair
component. Thus, it can be concluded that repair processes do work very well
in dissatisfied marriages. Instead, what predominates in dissatisfied couples’
attempts to use these social processes is the negative affect. Hence, in various
sequential analyses of the stream of behavior, if one spouse attempts a repair
mechanism with negative affect, the other spouse is more likely to respond to
the negative affect component with reciprocated negative affect in a dissatis-
fied marriage than in a satisfied one. The usual social processes present during
conflict that repair the interaction (such as metacommunication) do not work
in unhappy marriages. These processes are the mechanisms used by satisfied
couples for exiting a negative state (Gottman 1979). They include metacom-
munication, feeling probes that explore feelings, information exchange, social
comparison, humor, distraction, gossip, finding areas of common ground, and
appeals to basic philosophy and expectations in the marriage. What goes hand
in glove with this phenomena is a constriction of social processes in distressed
couples. The constriction of available social processes is the fascinating struc-
tural dynamic that maintains the absorbing state.
How does this constriction work? For example, assume that a message has
two parts, one positive and one negative, such as “Stop interrupting me,”
which is an attempt to repair the interaction, but may have also been said with
some irritation. In a happy marriage, there is a greater probability that the lis-
tener will focus on the repair component of the message and respond by say-
ing, “Sorry, what were you saying?” In an unhappy marriage, there is a greater
probability that the listener will respond only to the irritation in the message
and say something like, “I wouldn’t have to interrupt if I could get a word in
edgewise.” In this case the attempted repair mechanism does not work. The re-
sponse to the negativity now continues for long chains of reciprocated nega-
tive affect in dissatisfied marriages. Negativity as an absorbing state implies
that all these social processes have less of a chance of working because what
people attend to, and respond to, is the negativity. An interesting side effect of
this analysis is that the interactions of dissatisfied couples show a higher de-
gree of interaction structure, more predictability of one spouse’s behaviors
from those of the other, and less statistical independence than is found in the
interactions of satisfied couples. The interaction of happy married couples is
more random than that of unhappily married couples. This is precisely what
Raush et al (1974) predicted. One finding that may be related to this phenome-
non is that greater structure may come to pervade positive as well as negative
 MARITAL PROCESSES 181

interaction. This latter result is not as consistently found across laboratories,

but this may not be so much a failure to replicate as the inconsistency across
laboratories in conceptualizing, generating, and measuring “positivity” and in
the lack of studies that do sequential analyses of data.

WHAT IS DYSFUNCTIONAL IN AILING MARRIAGES?

Research on the correlates of marital satisfaction represents only one way to
address the Terman question. The second approach is to ask the longitudinal
question about which marital interaction patterns and other variables predict
marital stability and eventual happiness. There have been many suggestions
about what is dysfunctional about ailing marriages. For example, Lederer &
Jackson’s (1968) book Mirages of Marriage spelled out what is dysfunctional
about ailing marriages and how therapists should go about fixing them. They
said that the sine qua non of marriage was the quid pro quo, that in good mar-
riages there was a reciprocal exchange of positive behaviors, and that in bad
marriages there was for various reasons (like romanticism) the breakdown of
these agreements, these contracts. This point of view was consistent with an
economically based behavior exchange theory recommended 10 years earlier
by Thibaut & Kelly. (Romanticism, they argued, sets up false expectations that
lead to the quid pro quo being violated.) The Lederer & Jackson book had an
enormous impact. In marital behavior therapy, for example, it led to the
method of contingency contracting.
The claim about quid pro quo turned out to be totally wrong (Murstein et al
1977). Not only were happy marriages not characterized by the quid pro quo,
but it actually characterized unhappy marriages! This erroneous and untested
assumption not only spawned a new marital therapy, even when it was con-
firmed it continued on as a major ingredient of marital therapy.
Marital therapy must be guided by a theory of both what is dysfunctional in
ailing marriages and what is functional in marriages that are working. This
knowledge helps spell out the objectives of the treatment and the assessment of
the marriage, although not necessarily the methods for producing change. Two
important hypotheses about what is dysfunctional in ailing marriages were
suggested by Raush et al (1974). In their book, they raised two critical ques-
tions. The first was why in some marriages do minor conflicts “escalate far be-
yond their apparent triviality” (p. 2). Raush et al’s question was also more
broadly about what makes conflict constructive or destructive in marriages.
Raush et al gave the example of a couple given the task of deciding about
which television show to watch. They were impressed by the fact that many
couples got quite involved with the role-play improvisations. One wife be-
came extremely upset and said, “Damn it, you always watch what you want to
182 GOTTMAN

see. You’re always drinking beer and watching football. Nothing else seems
important to you, especially my wishes.” The seemingly small discussion of
which TV show to watch had led her to escalate and to express her complete
exasperation with her partner and with the marriage. Raush et al’s second
question was whether the avoidance of conflict in marriage was functional or
dysfunctional. They concluded that conflict avoidance is dysfunctional and
that conflict or bickering about trivial issues is also dysfunctional (indicative
of what they called “symbolic conflict”).
Many other hypotheses have been proposed, most of them stated as if they
were, without much empirical backing. In addition to the two proposed by
Raush, a baker’s dozen have been suggested: 1. a dominance structure is dys-
functional (Gottman 1979); 2. the lack of a dominance structure is dysfunc-
tional (Kolb & Straus 1974); 3. a “demand-withdraw” pattern or a “pursuer-
distancer” pattern is dysfunctional (e.g. Heavey et al 1995); 4. not being able to
change each other’s behavior is dysfunctional (Jacobson & Margolin 1979); 5.
a good marriage is characterized by acceptance, in which spouses accept each
other as they are and do not try to get behavior change (Jacobson & Chris-
tensen 1997); 6. poor problem solving is dysfunctional (Jacobson 1989); 7.
“mindreading,” or attributing motives or behaviors to one’s spouse is dysfunc-
tional (Watzlawick et al 1967); 8. not metacommunicating is dysfunctional
(Bateson et al 1956); 9. need complementarity is functional (Winch 1958); 10.
healthy marriage is not possible unless neuroses in one’s primary family are
resolved (Scharff & Scharff 1991); 11. most marital conflict is projection
(Meissner 1978); first the marriage needs to become “conscious” (Hendrix
1988); 12. marriages start off happy, but over time reinforcement erosion oc-
curs and that is the source of marital dysfunction (Jacobson & Margolin 1979);
13. only equalitarian marriage is functional (Schwartz 1994).
The questions we must ask are, first, are any of these contentions support-
able, and second, are these results fundamental or epiphenomenal? It will be
important to consider this latter question when discussing the construction of
theory. As to the first question, the two primary approaches taken to date in-
volve finding the correlates of marital satisfaction and the predictors of long-
term stability and satisfaction.
The Criterion of Correlates with Marital Satisfaction
Gottman (1979) defined a dominance structure as asymmetry in predictability
of one partner’s behavior compared with the other’s and used time-series
analysis to assess its existence. Distressed couples had significantly greater
asymmetry in predictability, with husbands being dominant. A study of 122
societies by Gray (1984) showed that female power is related to more positive
sexual relations; however, a review by Gray & Burks (1983) concluded that
 MARITAL PROCESSES 183

wife-dominant marriages are the least happy. However, because of the diffi-
culty of operationalizing dominance and the lack of agreement among various
measures of power and dominance (Rushe 1996), it cannot be concluded that
the lack of a dominance structure is dysfunctional. The “demand-withdraw”
pattern or a “pursuer-distancer” pattern as characteristic of unhappy marriage
has been replicated a number of times (Gottman & Levenson 1988, Heavey et
al 1995). In this pattern, it is the wife who raises and pursues the issues and the
husband who attempts to avoid the discussion and tends to withdraw. A few re-
sults must qualify this conclusion. First, the pattern is also, to some degree,
characteristic of happy marriages. Second, the pattern depends to some degree
on whether the issue is the husband’s or the wife’s (Christensen & Heavey
1990). To date, there has been no research on the hypothesis that not being able
to change each other’s behavior is dysfunctional or that a good marriage is
characterized by acceptance in which spouses accept one another as they are
and do not try to get behavior change. The hypothesis that poor problem solv-
ing is dysfunctional has yet to be tested independent of other processes of com-
munication, such as negative affect. In therapeutic interventions, the two as-
pects of interaction are also confounded. Mindreading is a frequent way that
couples begin discussing an event or probing feelings (Gottman 1979). There
is no evidence that mindreading by itself is dysfunctional. However, a poten-
tially related process of negative trait attributions to one’s spouse is character-
istic of unhappily married couples. The simple leap from mindreading to nega-
tive trait attributions is manifested in phrases such as “You always” or “You
never.” The genesis of the transformation from simple, specific complaints to
these global complaints is unknown and unexplored, but Fincham & Bradbury
(1992) offered a clue when they reported that attributions of responsibility
were correlated with the amount of anger displayed by wives during a
problem-solving interaction and the amount of whining by both husbands and
wives. There is no evidence that not metacommunicating is dysfunctional;
Gottman (1979) found that the amount of metacommunication was the same
for happily and unhappily married couples. However, sequential analyses
showed that happily married couples used short chains of metacommunication
often with agreement in the chain, whereas unhappily married couples used re-
ciprocated metacommunication with metacommunication. This latter se-
quence shows that metacommunication in distressed couples was an ineffec-
tive repair technique. There has been little support for the other hypotheses of
what is dysfunctional in ailing marriages.
The Longitudinal Criterion: Divorce Prediction Research
Gottman (1994) reported that there were three types of stable couples. One
type of stable couple, called “volatile,” was very much like Raush’s bickering
184 GOTTMAN

couple, another type of stable couple was very much like Raush’s conflict
avoiding couples, while a third type of stable couple, called “validators,” was
like Raush’s harmonious couples. That is, all three types of couples Raush et al
had identified turned out to be stable. The three types of couples differed most
dramatically in the amount and timing of persuasion attempts: Volatile couples
had the most persuasion attempts and began them almost immediately; valida-
tors waited to begin their persuasion attempts until the middle of the conversa-
tion; and conflict avoiders avoided all persuasion attempts. The three types of
couples also differed in how emotionally expressive they were, with volatile
couples highest, validators intermediate, and avoiders lowest in emotionality.
The three stable types of couples differed from couples on a trajectory toward
divorce in many ways. First, using a balance theory of marriage, Gottman
(1994) reported that the ratio of positive to negative codes during the conflict
discussion was about 5.0 for the three types of stable marriages, while it was
0.8 for the unstable marriage. Second, couples headed for divorce were high on
four behaviors that Gottman (1994) called the “Four Horsemen of the Apoca-
lypse”; they are criticism, defensiveness, contempt, and stonewalling (or lis-
tener withdrawal). Consistent with the demand-withdraw pattern, women
were significantly more likely than men to criticize, whereas men were more
likely than women to stonewall.

WHAT WE HAVE LEARNED: THE PSYCHOLOGY OF

MARRIAGE
The View from Observing
Only a few patterns seem to be consistently characteristic of ailing marriages.
The first is negative affect reciprocity, which this review has suggested is a re-
sult of the failure of repair processes. Second is the demand-withdraw pattern.
Third is a greater amount of negative than positive behaviors. Fourth is the
presence of particular forms of negativity, namely criticism, contempt, defen-
siveness, and stonewalling.
The View from Cognitive Psychology
A thorough review of this productive area in the study of marriage is not possi-
ble here (see Fincham et al 1990). There is a universal phenomenon in mar-
riage that has to do with how spouses in happy and unhappy marriages think
about positive and negative actions of their partner. In a happy marriage, if one
partner does something negative, the other partner tends to think that the nega-
tivity is fleeting and situational. For example, the thought might be something
like, “Oh, well, he’s in a bad mood. He’s been under a lot of stress lately and
 MARITAL PROCESSES 185

needs more sleep.” So the negativity is viewed as unstable, and the cause is
viewed as situational. In an unhappy marriage, however, the same behavior is
likely to be interpreted as stable and internal to the partner. The accompanying
thought might be something like, “He is inconsiderate and selfish. That’s the
way he is. That’s why he did that.” On the other hand, in a happy marriage, if
someone does something positive, the behavior is likely to be interpreted as
stable and internal to the partner. The accompanying thought might be some-
thing like, “He is a considerate and loving person. That’s the way he is. That’s
why he did that.” But in an unhappy marriage, the same positive behavior is
likely to be seen as fleeting and situational hence as unstable. The accompany-
ing thought might be, “Oh, well, he’s nice because he’s been successful this
week at work. It won’t last and it doesn’t mean much.” Holtzworth-Munroe &
Jacobson (1985) used indirect probes to investigate when a couple might spon-
taneously search for causes of events and what they conclude when they do
search for causes. They found evidence that distressed couples engaged in
more attributional activity than nondistressed couples, and that attributional
thoughts primarily surrounded events with negative impact. Nondistressed
couples engaged in relationship-enhancing attributions while distressed cou-
ples engaged in distress-maintaining attributions. Distress-maintaining attri-
butions maximize the impact of negativity and minimize the impact of positiv-
ity of the partner’s behavior.
Moreover, there was an important gender difference. Distressed husbands
generated more attributions than nondistressed husbands, but the two groups
of wives did not differ. They suggested that, normally, males may not engage
in much attributional activity but that they outstrip women once relationship
conflict develops. Relationship-enhancing attributions were responses to posi-
tive partner behavior in both groups of couples. Relationship-enhancing at-
tributions minimize the impact of negative and maximize the impact of posi-
tive behaviors of the partner. In an experimental study by Jacobson et al
(1985b), distressed and nondistressed couples were randomly instructed to
“act positive” or to “act negative.” They found that distressed couples were
likely to attribute their partner’s negative behavior to internal factors,
whereas nondistressed couples were likely to attribute their partner’s positive
behavior to internal factors. Thus, these attributions, once established, make
change less likely to occur. Behaviors that should disconfirm the attribu-
tional sets tend to get ignored, whereas behaviors that confirm the attribu-
tional set receive attention. Attributional processes may tap the way couples
think in general about the marital interaction as it unfolds in time. For exam-
ple, Berley & Jacobson (1984) noted that Watzlawick et al (1967) were refer-
ring to attributional processes when they discussed the punctuation fallacy.
The punctuation fallacy is that each spouse views himself or herself as the vic-
186 GOTTMAN

tim of the partner’s behavior, which is seen as the causal stimulus. Attribu-
tions and general thought patterns about negative behaviors may thus be theo-
retically useful in providing a link between the immediate patterns of activity
seen in behavioral interaction and physiological response and more long-
lasting and more global patterns that span longer time periods. It might be the
case that these more stable aspects of the marriage are better for predicting
long-term outcomes such as divorce than can be obtained from behavioral ob-
servation.
The content dimensions of negative attributions that have been studied in-
clude locus (partner, self, relationship, or outside events), stability (e.g. due
to partner’s trait, or a state that is situationally determined), globality (how
many areas of the marriage are affected), intentionality (negative intent—self-
ish versus unselfish motivation), controllability, volition, and responsibility
(e.g. blameworthiness). For attributions about negative events, all of the stud-
ies reviewed supported differences between happily and unhappily married
couples on the two dimensions of globality and selfish versus unselfish moti-
vation. It is likely that these attributional phenomena are what make the self-
report measurement of any aspects of the quality of the marriage so strongly
related. It is also what becomes problematic in attaching any specificity to the
measurement of marital satisfaction or marital quality (see Fincham & Brad-
bury 1987).
Another important cognitive dimension, called “sentiment override,” was
introduced by Weiss (1980). Weiss suggested that reactions during marital
interaction may be determined by a global dimension of affection or disaffec-
tion rather than by the immediately preceding valence of the stimulus. Notar-
ius et al (1989) evaluated the validity of this hypothesis in a remarkably crea-
tive study in which they employed a sequential stream of behavior and cogni-
tions to operationalize a number of hypotheses linking behavior and cogni-
tion. They found that distressed wives were more negative, were more likely
to evaluate their partner’s neutral and negative messages as negative (suggest-
ing the operation of a negative sentiment override), and, given a negative
evaluation of their partner’s antecedent message, were more likely to offer a
negative reply than were all other spouses. Vanzetti et al (1992) reported that
distressed couples have more negative and less positive expectations. They
measured “relational efficacy,” which is a shared belief that a couple can solve
its problems, and found that couples high in relational efficacy choose
relationship-enhancing attributions more often than do low-efficacy couples.
Low-efficacy marriages showed strong preferences for distress-maintaining
attributions.
To assess a larger cognitive unit than attributions, Buehlman et al (1992)
coded interviews with couples to assess the shared beliefs and narratives of
 MARITAL PROCESSES 187

couples about the history of their marriage and their philosophy of marriage. A
few simple variables (such as the husband’s fondness for his wife) were able to
predict divorce or marital stability over a three-year period with a great degree
of accuracy (100% for the divorcing couples, and 94% overall).

The View from Psychophysiology

A recent and productive approach to studying marriages has been a social psy-
chophysiological procedure. Beginning with Kaplan et al (1964), simultane-
ous psychophysiological recording was taken of two conversing individuals.
The initial finding was that the galvanic skin responses of the two interacting
people were correlated only when they disliked each other. Levenson & Gott-
man (1983) later extended this finding to the construct of physiological link-
age, i.e. predictability of one person’s physiology (across channels) from the
other’s (controlling for autocorrelation), and reported greater physiological
linkage for unhappily married compared with happily married couples. The
linkage variable accounted for over 60% of the variance in marital satisfaction.
Levenson & Gottman (1985) later reported that measures of physiological
arousal in cardiovascular channels and in skin conductance were able to pre-
dict drops in marital satisfaction over three years, controlling for the initial
level of marital satisfaction. The pattern was later used by Gottman (1990) to
suggest that diffuse physiological arousal, that is, arousal in more than one
physiological channel (but not necessarily more than one physiological sys-
tem, e.g. increased heart rate as well as contractility, or blood velocity) would
be associated with decreased information processing capability and a reliance
on overlearned patterns of behavior and cognition, particularly those associ-
ated with fight or flight. Brown & Smith (1992) studied 45 married couples
and found that husbands attempting to persuade their wives showed the great-
est increase in systolic blood pressure before and during the discussion. In
males, physiological effects were accompanied by increased anger and a hos-
tile and coldly assertive interpersonal style. Although wives showed behavior
patterns that were similar to husbands to some degree, they displayed neither
elevated systolic blood pressure nor anger. In addition, Fan2) found that both
the size of the systolic blood pressure responses of husbands during marital
conflict and their recovery times exceeded those of their wives. Malarkey et al
(1994) simultaneously studied the secretion of stress-related hormones in five
samples of blood taken during the conflict interactions of 90 newlywed cou-
ples. Hostile behavior (coded with the MICS) correlated with decreased levels
of prolactin and increases in epinephrine, norepinephrine, ACTH, and growth
hormone, but not cortisol.
188 GOTTMAN

Physiological approaches have added something important theoretically.

The general contribution they provide is in directing the organization and
search for patterns of behavior and cognition within balance theories, in which
positivity and negativity are in a state of dynamic balance around a steady
state, or set point (for a mathematical model, see Cook et al (1995). In the
body, many systems are in a state of dynamic homeostatic balance around a
steady state through the action of opponent processes (e.g. the regulation of the
heart’s rate through the parasympathetic and sympathetic branches of the auto-
nomic nervous system).
Three additional concepts are discussed here. First, negative affect reci-
procity may exist as a function of spouses’ inability to soothe themselves and
each other. Second, this variable of soothing may be the basis for the large re-
lapse effect in marital therapy. In therapy the therapist plays the role of soother
instead of the spouses, and when therapy ends the spouses are unable to soothe
each other, and old patterns of behavior and cognition reassert themselves.
Third, Gottman & Levenson found that the husband’s stonewalling (with-
drawal as a listener) was related to his physiological arousal (reported in Gott-
man 1994). In addition, Levenson et al (1995), in a study of older long-term
marriages, reported that the husband’s but not the wife’s physiological arousal
was related to his self-report of feeling negative (in a video recall paradigm);
they speculated that the husband’s withdrawal in the demand-withdraw pattern
may be related to his physiological arousal, since it is known that males are
more aware of their own physiology than women. Thus, physiological meas-
ures suggest a biological basis for the gender effect in the demand-withdraw
pattern. This observation is not intended to mean that such a biological basis is
unrelated to socialization.
The View from Intervention
A major contribution by psychologists to the study of marriage was the idea of
systematically evaluating marital therapy. Unfortunately, as is typical of psy-
chotherapy in general, new therapies emerged in this field largely from specu-
lative writings by therapists instead of from careful empirical work. There
have been a number of reviews of the marital therapy literature to date (e.g.
Baucom & Hoffman 1986, Bray & Jouriles 1997, Dunn & Schwebel 1995,
Hahlweg & Markman 1983, Jacobson & Addis 1993, Pinsof & Wynne 1995,
Prince & Jacobson 1997). Their major conclusions are summarized here.
Meta-analytic studies of marital therapy outcome appear at first blush to pres-
ent a more optimistic picture than the one presented here, but meta-analytic
studies tend to ask a very global question about “effectiveness” relative to a
control group, which obscures issues of which therapies are successful on
which outcome measures (e.g. only self-report?), issues of clinical versus sta-
 MARITAL PROCESSES 189

tistical significance, and issues of longitudinal follow-up and relapse. In addi-

tion, meta-analyses tend to lump good and bad studies together and examine
effect sizes; as is well known, studies with the best control groups necessarily
have the smallest effect sizes because they attempt to control everything ex-
cept what they suppose to be the active ingredients of the intervention.
The most frequently evaluated marital intervention is behavioral. Baucom &
Hoffman (1986) noted that the skills usually taught by behavioral marital ther-
apy (BMT) are communication and problem solving, and contingency contract-
ing (based on the quid pro quo assumption). They distinguished these commu-
nication skills as problem solving in nature, as opposed to communication skill
programs oriented toward the expression of emotions and listening skills. Bau-
com & Hoffman concluded that (a) couples receiving BMT (compared with a
wait-list control group) improve significantly in negative communication and
self-reports of problems (Jacobson’s program was the only one to report im-
provements in positive communication); (b) BMT is superior to nonspecific
and attention control groups; and (c) there are no major differences in the ef-
fectiveness of two components of BMT, nor in their order of administration.
However, Jacobson et al (1985b, 1987) reported some evidence that the com-
munication/problem-solving (CO) training was superior to the behavior ex-
change (contingency contracting) (BE) condition. Upon two-year follow-up,
couples in the CO condition were most likely to be happily married and least
likely to be separated or divorced. They also noted, however, that while statis-
tically significant changes were obtained by BMT compared with a waiting list
(and other) control group, “60–65% of the couples either remained somewhat
distressed or failed to change during treatment” (p. 605). In a more recent re-
view, Jacobson & Addis (1993) reached similar conclusions to those of Bau-
com & Hoffman (1986). They estimated that about 50% of couples cannot be
considered successes. They are considerably more pessimistic than Baucom &
Hoffman about the long-term effectiveness of BMT. Recently, Snyder et al
(1991) reported that the insight-oriented and BMT groups were equivalent at
termination. At four-year follow-up, however, they found that the divorce rate
was 38% for the BMT group and only 3% for the insight-oriented group. Ja-
cobson (1993), however, challenged the meaning of the Snyder et al results; in
a comparison of treatment manuals for the two groups, he concluded that their
BMT manual was 10 years out of date and that the insight-oriented manual was
far closer to current BMT treatments. Regardless of how one labels the inter-
vention, the Snyder et al four-year follow-up results are quite encouraging.
Summary
First, if one requires replicated effects, treatment gains are not generally main-
tained over time. Second, it does not seem to matter very much what one does
190 GOTTMAN

in treatment. In general, the effect sizes are roughly the same, regardless of the
exact nature of the intervention. This latter fact is remarkable when “disman-
tling” studies done in BMT are considered. Rather than having identified an
active ingredient of BMT, they suggest the conclusion that any of the parts
equals the whole. Furthermore, it appears possible that all parts may be as ef-
fective as contingency contracting, an approach based on an erroneous as-
sumption about what makes marriages work. If this remarkable conclusion
were true, almost all marital treatment effects to date would be due to nonspe-
cifics such as trust in the therapist, hope, the existence of a structured program,
all of which could be considered to be placebo effects.

SUMMARY: AN INTEGRATIVE MODEL OF HOW

MARRIAGES MAY DYSFUNCTION AND FUNCTION
The Stable Phenomena of Marital Process and Outcomes
To date, research on marriage by psychologists has been remarkably produc-
tive. We can identify seven consistent patterns across laboratories. These pat-
terns are (a) greater negative affect reciprocity in unhappy couples, which may
be related to the failure of repair; (b) lower ratios of positivity to negativity in
unhappy couples and couples headed for divorce (this includes a greater cli-
mate of agreement in happily married couples); (c) less positive sentiment
override in unhappy couples; (d) the presence of criticism, defensiveness, con-
tempt, and stonewalling in couples headed for divorce; (e) greater evidence of
the wife demand–husband withdraw pattern in unhappy couples (though it is
probably also there to some extent in happily married couples); (f) negative
and lasting attributions about the partner and more negative narratives about
the marriage and partner in unhappy couples; and (g) greater physiological
arousal in unhappy couples. What is needed at this juncture is a theoretical
model of how these various patterns may fit together. The search for a theory
needs to be guided by two questions, the question of what is dysfunctional in
marriages (i.e. how the seven negative patterns are related and the ontogeny of
the seven negative patterns in ailing marriages), and the question of what is
functional, that is, what couples whose marriages are doing well are doing
(thinking, feeling, etc) differently. Both questions are necessary—identifying
negative dysfunctional patterns does not imply that one has also simultane-
ously identified positive functional patterns.
What Is Dysfunctional in Unhappy, Unstable Marriages?
Gottman (1994) proposed a theory of dysfunction based on longitudinal re-
search and the correlates of marital satisfaction. Unstable marriages are likely
 MARITAL PROCESSES 191

to be unable to work out over time one of three stable adaptations Gottman dis-
covered: volatile, validating, conflict avoiding. In their attempts to accomplish
a stable adaptation, some couples fall into a pattern that does not maintain the
balance of positivity to negativity at a high level. Spouses tend to formulate
their complaints as criticisms and contempt; they tend to respond defensively
and eventually withdraw from each other. At the core of this formulation is a
balance theory, an ecology of marital behaviors in which a ratio of positivity to
negativity that is highly tilted toward positivity needs to be maintained. This
ratio is suggested as the quantity that needs to be regulated at a high level, ap-
proximately 5.0. Related to marital interaction is the subtext of how the inter-
action is perceived by each partner. If the ratio of 5.0 is significantly violated,
the perception of well-being is replaced by one of distress, which is some com-
bination of hurt and anger (fight, or the “righteous indignation” perception)
and/or hurt and perceived attack (flight, or the “innocent victim” perception).
The length of time in a state of distress, in turn, determines feeling flooded by
one’s partner’s negative affect, with accompanying diffuse physiological
arousal (DPA), negative subjective affective states, and negative attributions
of the partner. Flooding begins the Distance and Isolation Cascade, which en-
tails perceiving one’s marital problems as severe, as better worked out alone
rather than with the spouse, arranging one’s lives so that they are more in paral-
lel than they used to be, and loneliness within the marriage. Eventually, even
one’s perception of the entire relationship is affected, and the couple’s narra-
tives change. In the oral history interview, people (particularly husbands) ex-
press disappointment with the marriage, declare little fondness for the partner,
and present themselves as separate entities who do not see the past the same
way or share a common philosophy of marriage. They also tend to see their
lives as chaotic and out of control, and they see all the marital conflict as point-
less and empty. This grim formulation is balanced by the reverse results that
couples whose marriages are stable use positive affect and persuasion in very
different ways—ways that buffer them from the physiological stresses of DPA
and from the perception of their partner’s negative emotions as horrible, dis-
gusting, terrifying, overwhelming, disorganizing, and impossible to predict.
An Intergrative Account: the Bank Account Model
What is the etiology of these dysfunctional patterns, which are predictive of un-
happiness and divorce? This review introduces a new theory, called the Bank
Account Model (BAM). The first premise of the BAM is that the answer to the
question of the ontogeny of the seven negative patterns in ailing marriages lies
in asserting that psychologists have been looking in the wrong place. They
have studied almost exclusively the resolution of conflict, and BAM suggests
that the seven dysfunctional patterns reviewed here reflect the endpoint of the
192 GOTTMAN

failure of three related processes. The first process is the couple’s ratio of fairly
low-level positivity to negativity in nonconflict interaction. Nonconflict inter-
action consists primarily of the mundane, everyday interactions of married
life, each of which holds the possibility of what might be called either “turning
toward” or “turning away” from one’s partner. An example follows: One
spouse is in the bathroom in the morning, in a hurry getting ready for the day,
when the partner comes into the bathroom and says, “I just had a disturbing
dream.” An example of turning away would be, “I don’t have time for this right
now,” while an example of turning toward would be, “I’m in a real hurry, but
tell me about your dream.” A greater balance of turning away compared with
turning toward implies that there will be many moments of what could be
called “unrequited interest and excitement,” in which one person’s interest and
excitement is not responded to by the partner, and many moments of “unre-
quited irritability,” in which one person’s low-intensity anger is not responded
to by the partner. This lack of responsiveness leads to the presence of the first
of the Four Horsemen, criticism. Criticism leads to the other horsemen.
The theory proposes that a greater proportion of turning toward compared
with turning away leads to positive sentiment override (Weiss 1980), whereas
a greater proportion of turning away compared with turning toward leads to
negative sentiment override. Physiological soothing of one’s partner using a
variety of positive affects (e.g. interest, affection, validation, empathy, humor)
during everyday stress reduction interactions (typically events-of-the-day dis-
cussions and errand talk) is central to contributing to positive sentiment over-
ride, and this is accomplished through the simple mechanism of escape condi-
tioning. The second process is the amount of cognitive room that couples allo-
cate for the relationship and for their spouse’s world. We call this the “love
map.” The husband’s love map is particularly predictive of the longitudinal
course of marriages. This process has to do with knowing one’s partner’s
world and continually updating that knowledge. The third process, which is
another contributor to positive sentiment override is the existence of what we
call the Fondness and Admiration System (tapped by the oral history inter-
view). Admiration is the antidote for contempt. Couples who are on a stable
and happy trajectory express spontaneous admiration and affection for their
partner much more than couples on the trajectory toward divorce. The Fond-
ness and Admiration System affects and is affected by both cognition and be-
havior.
The existence of either positive or negative sentiment override determines
the success or failure of repair processes (Gianino & Tronick 1988 discussed
these in mother-infant interaction) in conflict interaction. The success of repair
minimizes negative affect reciprocity. Positive sentiment override, however,
also leads to the presence of three other processes (in addition to successful re-
 MARITAL PROCESSES 193

pair) during conflict interaction. The first has been called “editing.” Part of
breaking the chain of negative-affect reciprocity, editing is assessed as a low-
ered conditional probability of becoming a negative-affect speaker after one
has been a negative-affect listener. There is a fair amount of evidence that hap-
pily married couples are more likely to edit than unhappily married couples
(Gottman 1979, Notarius et al 1989). Editing is responsible for decreasing the
probability of what the Patterson group has called “negative startup,” which
means moving from one’s partner’s neutral affect to one’s own negative affect.
It is related to the beginning of the demand-withdraw pattern, and it is more
characteristic of wives than husbands (Gottman 1994, 1996). The second pro-
cess is one we call “respectful influence.” Respectful influence involves using
positive affect in the service of the de-escalation of conflict while attempting
to influence one’s partner without using the Four Horsemen, and it also in-
volves accepting influence from one’s partner. Rushe (1996), in a detailed
analysis of persuasion and influence tactics during marital conflict resolution,
found that assertive means of persuasion during conflict resolution was the one
discriminator between happily married and distressed nonviolent couples.
Coan et al (1997) described rejecting influence as characteristic of violent cou-
ples. The third process is positive affect, which this theory hypothesizes is
used carefully to avoid one’s partner’s defensiveness and to reduce physiologi-
cal arousal. Our laboratory has evidence that among newlyweds the Time-1
positive affects of interest, affection, humor, and validation predict and dis-
criminate whether couples will eventually divorce, be stable and happily mar-
ried, or stable and unhappily married (in a five-year follow-up). These pro-
cesses affect the couple’s narratives, the stories they tell themselves and each
other about the marriage and the partner. The current evidence is that these nar-
ratives are quite powerful predictors of marital trajectory (Buehlman et al
1992). Narratives may guide cognition and attributions even when the couple
is not together.

Visit the Annual Reviews home page at

http://www.AnnualReviews.org.

Literature Cited
Bakeman R, Gottman J. 1986. Observing In- and GSEQ. New York: Cambridge Univ.
teraction: An Introduction to Sequential Press
Analysis. New York: Cambridge Univ. Bank L, Dishion T, Skinner M, Patterson GR.
Press 1990. Method variance in structural equa-
Bakeman R, Quera V. 1995. Analyzing Inter- tion modeling: living with “glop.” In De-
action: Sequential Analysis with SDIS pression and Aggression in Family Inter-
194 GOTTMAN

action, ed. GR Patterson, pp. 247–79. 1991. Becoming a family: marriage, par-
Hillsdale, NJ: Erlbaum enting and child development. In Family
Bateson G, Jackson DD, Haley J, Weakland J. Transitions, ed. PA Cowan, EM Hether-
1956. Toward a theory of schizophrenia. ington. Hillsdale, NJ: Erlbaum
Behav. Sci. 1:251–64 Cowan PA, Cowan CP. 1987. Couple’s rela-
Baucom DH, Hoffman JA. 1986. The effec- tionships, parenting styles and the child’s
tiveness of marital therapy: current status development at three. Presented at Meet.
and application to the clinical setting. In Soc. Res. Child Dev., Baltimore, MD
Clinical Handbook of Marital Therapy, ed. Cowan PA, Cowan CP. 1990. Becoming a
N Jacobson, A Gurman, pp. 597–620. New family: research and intervention. In Fam-
York: Guilford ily Research, ed. I Sigel, A Brody. Hills-
Bell RQ. 1968. A reinterpretation of the direc- dale, NJ: Erlbaum
tion of effects in studies of socialization. Cummings EM, Davies P. 1994. Children and
Psychol. Rev. 75:81–95 Marital Conflict: The Impact of Family
Berley RA, Jacobson NS. 1984. Causal attri- Dispute Resolution. New York: Guilford
butions in intimate relationships: toward a Dunn R, Schwebel AI. 1995. Meta-analytic re-
model of cognitive behavioral marital ther- view of marital therapy outcome research.
apy. In Advances in Cognitive-Behavioral J. Fam. Psychol. 9(1):58–68
Research and Therapy, ed. P Kendall, Easterbrooks MA. 1987. Early family devel-
3:2–90. New York: Academic opment: longitudinal impact of marital
Bloom B, Asher S, White S. 1978. Marital dis- quality. Presented at Meet. Soc. Res. Child
ruption as a stressor: a review and analysis. Dev., Baltimore, MD
Psychol. Bull. 85:867–94 Easterbrooks MA, Emde RA. 1988. Marital
Bray JH, Jouriles EN. 1997. Treatment of and parent-child relationships: the role of
marital conflict and prevention of divorce. affect in the family system. In Relation-
J. Marital Fam. Ther. In press ships Within Families: Mutual Influence,
Brown PC, Smith TW. 1992. Social influence, ed. RA Hinde, J Stevenson-Hinde, pp.
marriage, and the heart: cardiovascular con- 27–45. Oxford: Clarendon
sequences of interpersonal control in hus- Emery RE. 1982. Interparental conflict and the
bands and wives. Health Psychol. 11:88–96 children of discord and divorce. Psychol.
Buehlman K, Gottman JM, Katz L. 1992. How Bull. 92:310–30
a couple views their past predicts their fu- Emery RE. 1988. Marriage, Divorce, and
ture: predicting divorce from an oral his- Children’s Adjustment. Newbury Park,
tory interview. J. Fam. Psychol. 5: CA: Sage
295–318 Emery RE, O’Leary KD. 1982. Children’s
Burgess EW, Locke HJ, Thomes MM. 1971. perceptions of marital discord and behav-
The Family: From Traditional to Compan- ior problems of boys and girls. J. Abnorm.
ionship. New York: Van Nostrand Rein- Child Psychol. 10:11–24
hold Emery RE, O’Leary KD. 1984. Marital dis-
Burman B, Margolin G. 1992. Analysis of the cord and child behavior problems in a non-
association between marital relationships clinical sample. J. Abnorm. Child Psychol.
and health problems: an interactional per- 12:411–20
spective. Psychol. Bull. 112:39–63 Fankish CJ. 1992. Warning! Your marriage
Campbell DT, Fiske DW. 1959. Convergent may be hazardous to your health: spouse-
and discriminant validation by the multi- pair risk factors and cardiovascular reac-
trait-multimethod matrix. Psychol. Bull. tivity. Diss. Abstr. Int. 52:5532
56:81–105 Fincham FD, Bradbury TN. 1987. The assess-
Christensen A, Heavey CL. 1990. Gender and ment of marital quality: a reevaluation. J.
social structure in the demand/withdraw Marriage Fam. 49:797–809
pattern. J. Pers. Soc. Psychol. 59(1):73–81 Fincham FD, Bradbury TN. 1992. Assessing
Coan J, Gottman JM, Babcock J, Jacobson NS. attributions in marriage: the relationship
1997. Battering and the male rejection of attribution measure. J. Pers. Soc. Psychol.
influence from women. Aggress. Behav. In 62:457–68
press Fincham FD, Bradbury TN, Scott CK. 1990.
Cook J, Tyson R, White J, Rushe R, Gottman Cognition in marriage. In The Psychology
J, Murray J. 1995. The mathematics of of Marriage, ed. FD Fincham, TD Brad-
marital conflict: qualitative dynamic bury, pp. 118–49. New York: Guilford
mathematical modeling of marital interac- Fitzpatrick MA. 1988. Between Husbands and
tion. J. Fam. Psychol.9:110–30 Wives: Communication in Marriage. Bev-
Cowan CP, Cowan PA, Heming G, Miller NB. erly Hills, CA: Sage
 MARITAL PROCESSES 195

Forehand R, Brody G, Long N, Slotkin J, Fau- Hendrix H. 1988. Getting the Love You Want.
ber R. 1986. Divorce/divorce potential and New York: Holt
interparental conflict: the relationship to Hetherington EM. 1988. Coping with family
early adolescent social and cognitive func- transitions: winners, losers and survivors.
tioning. J. Adolesc. Res. 1:389–97 Child Dev. 60:1–14
Friedman HS, Tucker JS, Schwartz JE, Tomil- Hetherington EM, Clingempeel WG. 1992.
son KC. 1995. Psychosocial and behav- Coping with marital transitions. Monogr.
ioral predictors of longevity: the aging and Soc. Res. Child Dev. 57(227):1–242
death of the “Termites.” Am. Psychol. 50: Hetherington EM, Cox M, Cox R. 1978. The
69–78 aftermath of divorce. In Mother-Child,
Gianino A, Tronick EZ. 1988. The mutual Father-Child Relations, ed. JH Stevens, M
regulation model: the infant’s self and in- Matthews, pp.117–45. Washington, DC:
teractive regulation coping and defense. In Natl. Assoc. Educ. Young Children
Stress and Coping, ed. T Field, P McCabe, Hetherington EM, Cox M, Cox R. 1982. Ef-
N Schnieiderman, pp. 47–68. Hillsdale, fects of divorce on parents and children. In
NJ: Erlbaum Nontraditional Families, ed. M Lamb, pp.
Glenn ND, Kramer KB. 1985. The psycho- 233–88. Hillsdale, NJ: Erlbaum
logical well-being of adult children of di- Heyman RE, Eddy JM, Weiss RL, Vivian D.
vorce. J. Marriage Fam. 40:269–82 1995. Factor analysis of the Marital Inter-
Gottman JM. 1979. Marital Interaction: Em- action Coding System (MICS). J. Fam.
pirical Investigations. New York: Aca- Psychol. 9:209–15
demic Holtzworth-Munroe A, Jacobson NS. 1985.
Gottman JM. 1990. How marriages change. In Causal attributions of married couples:
Depression and Aggression in Family In- When do they search for causes? What do
teraction, ed. GR Patterson, pp. 110–35. they conclude when they do? J. Pers. Soc.
Hillsdale, NJ: Erlbaum Psychol. 48:1398–412
Gottman JM. 1994. What Predicts Divorce? Howes P, Markman HJ. 1989. Marital quality
Hillsdale, NJ: Erlbaum and child functioning: a longitudinal in-
Gottman JM, ed. 1996. What Predicts Di- vestigation. Child Dev. 60:1044–51
vorce: The Measures. Hillsdale, NJ: Erl- Jacob T. 1987. Family interaction and psycho-
baum pathology: historical overview. In Family
Gottman JM, Katz L. 1989. Effects of marital Interaction and Psychopathology, ed. T
discord on young children’s peer interac- Jacob, pp. 3–24. New York: Plenum
tion and health. Dev. Psychol. 25: 373–81 Jacobson NS. 1989. The maintenance of treat-
Gottman JM, Levenson RW. 1988. The social ment gains following social learning-
psychophysiology of marriage. In Per- based marital therapy. Behav. Ther. 20:
spectives on Marital Interaction, ed. P 325–36
Noller, MA Fitzpatrick, pp. 182–200. Jacobson NS, Addis ME. 1993. Research on
Clevedon, Engl: Multilingual Matters couple therapy: What do we know? Where
Gottman JM, Roy AK. 1990. Sequential are we going? J. Consult. Clin. Psychol.
Analysis: A Guide for Behavioral Re- 61(1):85–93
searchers. New York: Cambridge Univ. Jacobson NS, Christensen A. 1997. Integrated
Press Cognitive Behavioral Marital Therapy.
Gray JP. 1984. The influence of female power New York: Guilford. In press
in marriage on sexual behaviors and atti- Jacobson NS, Follette VM, Follette WC,
tudes: a holocultural study. Arch. Sex. Be- Holtzworth-Munroe A, Katt JL, Schma-
hav. 13:223–31 ling KB. 1985a. A component analysis of
Gray LB, Burks N. 1983. Power and satisfac- behavioral marital therapy: 1-year follow-
tion in marriage: a review and critique. up. Behav. Res. Ther. 23:549–55
Psychol. Bull. 93:513–38 Jacobson NS, Margolin G. 1979. Marital
Hahlweg K, Markman HJ. 1983. Effective- Therapy. New York: Brunner/Mazel
ness of behavioral marital therapy: em- Jacobson NS, McDonald DW, Follette WC,
pirical status of behavioral techniques in Berley RA. 1985b. Attributional processes
preventing and alleviating marital dis- in distressed and nondistressed married
tress. J. Consult. Clin. Psychol. 56: couples. Cogn. Ther. Res. 9:35–50
440–47 Jacobson NS, Schmaling K, Holtzworth-
Heavey CL, Christensen A, Malamuth NM. Munroe A. 1987. Component analysis of
1995. The longitudinal impact of demand behavioral marital therapy: 2-year fol-
and withdrawal during marital conflict. J. lowup and prediction of relapse. J. Marital
Consult. Clin. Psychol. 63:797–801 Fam. Ther. 13:187–95
196 GOTTMAN

Kaplan HB, Burch NR, Bloom SW. 1964. Martin TC, Bumpass L. 1989. Recent trends
Physiological covariation in small peer in marital disruption. Demography 26:
groups. In Psychological Approaches to 37–51
Social Behavior, ed. PH Liederman, D Meissner WW. 1978. The conceptualization
Shapiro, pp.21–43. Stanford, CA: Stanford of marriage and family dynamics from a
Univ. Press psychoanalytic perspective. In Marriage
Katz LF, Gottman JM. 1991a. Marital discord and Marital Therapy, ed. TJ Paolino, BS
and child outcomes: a social psycho- McCrady, pp. 25–88. New York: Brun-
physiological approach. In The Develop- ner/Mazel
ment of Emotion Regulation and Disregu- Mischel W. 1968. Personality and Assess-
lation, ed. K Dodge, J Garber, pp.129–58. ment. New York: Wiley
New York: Cambridge Univ. Press Murstein BI, Cerreto M, MacDonald MG.
Katz LF, Gottman JM. 1991b. Marital interac- 1977. A theory and investigation of the ef-
tion processes and preschool children’s fect of exchange-orientation on marriage
peer interactions and emotional develop- and friendship. J. Marriage Fam. 39:
ment. Presented at Meet. Soc. Res. Child 543–48
Dev. Notarius CI, Benson PR, Sloane D, Vanzetti
Kelly LE, Conley JJ. 1987. Personality and NA. 1989. Exploring the interface between
compatibility: a prospective analysis of perception and behavior: an analysis of
marital stability and marital satisfaction. J. marital interaction in distressed and non-
Pers. Soc. Psychol. 52:27–40 distressed couples. Behav. Assess. 11:
Kolb JM, Straus MA. 1974. Marital power and 39–64
marital happiness in relation to problem- Nye FI. 1988. Fifty years of family research
solving. J. Marriage Fam. 36:756–66 1937–1987. J. Marriage Fam. 50:305–16
Kulka RA, Weingarten H. 1979. The long- Patterson GR. 1982. Coercive Family Process.
term effects of parental divorce on adult Eugene, OR: Castalia
adjustment. J. Soc. Issues 35:50–78 Peterson JL, Zill N. 1986. Marital disruption,
Kurdek LA. 1993. Predicting marital dissolu- parent-child relationships, and behavior
tion: a 5-year prospective longitudinal problems in children. J. Marriage Fam.
study of newlywed couples. J. Pers. Soc. 48:295–307
Psychol. 64:221–42 Pinsof WM, Wynne LC. 1995. The effective-
Lederer WJ, Jackson DD. 1968. The Mirages ness of marital and family therapy: an em-
of Marriage. New York: Norton pirical overview, conclusions and recom-
Levenson RW, Carstensen LL, Gottman JM. mendations. J. Marital Fam. Ther.
1994. The influence of age and gender on 21(4):339–623
affect, physiology and their interactions: A Porter B, O’Leary KD. 1980. Marital discord
study of long-term marriages. J. Pers. Soc. and childhood behavior problems. J. Ab-
Psychol. 67:56–68 norm. Child Psychol. 8:287–95
Levenson RW, Gottman JM. 1983. Marital in- Prince SE, Jacobson NS. 1997. A review and
teraction: physiological linkage and affec- evaluation of marital and family therapies
tive exchange. J. Pers. Soc. Psychol. 45: for affective disorders. J. Marital Fam.
587–97 Ther. 21:377–402
Levenson RW, Gottman JM. 1985. Physio- Raush HL, Barry WA, Hertel RK, Swain MA.
logical and affective predictors of change 1974. Communication, Conflict, and Mar-
in relationship satisfaction. J. Pers. Soc. riage. San Francisco: Jossey-Bass
Psychol. 49:85–94 Reid JB. 1970. Reliability assessment of ob-
Lewis M, Rosenblum LA, eds. 1974. The Ef- servational data: a possible methodolgical
fect of the Infant on Its Caregiver. New problem. Child Dev. 41:1143–50
York: Wiley Revenstorf D, Vogel B, Wegener C, Halweg
Maccoby EE. 1990. Gender and relationships: K, Schindler L. 1980. Escalation phenom-
a developmental account. Am. Psychol. ena in interaction sequences: an empirical
45(4):513–20 comparison of distressed and nondis-
Malarkey WB, Kiecolt-Glaser JK, Pearl D, tressed couples. Behav. Anal. Modif. 2:
Glaser R. 1994. Hostile behavior during 97–ll6
marital conflict alters pituitary and adrenal Rushe R. 1996. Tactics of power and influence
hormones. Psychosom. Med. 56:41–51 in violent marriages. PhD thesis. Univ.
Margolin G, Wampold BE. 1981. Sequential Wash., Seattle
analysis of conflict and accord in dis- Rutter M. 1971. Parent-child separation: psy-
tressed and nondistressed marital partners. chological effects on the children. J. Child
J. Consult. Clin. Psychol. 49:554–67 Psychol. Psychiatry 12:233–60
 MARITAL PROCESSES 197

Rutter M. 1979. Protective factors in chil- communication patterns in high and low
dren’s responses to stresss and disadvan- marital adjustment groups. Hum. Com-
tage. In Primary Prevention in Psychopa- mun. Res. 9:306–l9
thology: Vol. 3. Social Competence in Vanzetti NA, Notarius CI, NeeSmith D. 1992.
Children, ed. MW Kent, JE Rold, pp. Specific and generalized expectancies in
3–27. Hanover, NH: Univ. Press New Eng- marital interaction. J. Fam. Psychol. 6:
land 171–83
Schaap C. 1982. Communication and Adjust- Vincent JP, Friedman LC, Nugent J, Messerly
ment in Marriage. Netherlands: Swets L. 1979. Deman characteristics in observa-
Zeitlinger tions of marital interaction. J. Consult.
Scharff DE, Scharff SS. 1991. Object Rela- Clin. Psychol. 47:557–66
tions Couple Therapy. Northvale, NJ: Ar- Watzlawick P, Beavin JH, Jackson DD. 1967.
onson Pragmatics of Human Communication.
Schwartz P. 1994. Peer Marriage. New York: New York: Norton
The Free Press Weiss RL. 1980. Strategic behavioral and
Shannon CE, Weaver W. 1949. The Mathe- marital therapy: toward a model for assess-
matical Theory of Communciation. Ur- ment and intervention. In Advances in
bana: Univ. Ill. Press Family Intervention, Assessment and The-
Shaw DS, Emery RE. 1987. Parental conflict ory, ed. JP Vincent, 1:229–71. Greenwich,
and other correlates of the adjustment of CT: JAI Press
school-age children whose parents have Weiss RL, Hops H, Patterson GR. 1973. A
separated. J. Abnorm. Child Psychol. 15: framework for conceptualizing marital
269–81 conflict, a technique for altering it, some
Snyder DK, Wills RM, Grady FA. 1991. data for evaluating it. In Behavior Change,
Long-term effectiveness of behavioral ver- ed. LA Hamerlynck, LC Handy, EJ Mash.
sus insight-oriented marital therapy: a 4- Champaign, IL: Research Press
year followup study. J. Consult. Clin. Psy- Whitehead LB. 1979. Sex differences in chil-
chol. 59(1):138–41 dren’s responses to family stress. J. Child
Terman LM, Buttenweiser P, Ferguson LW, Psychol. Psychiatry 20:247–54
Johnson WB, Wilson DP. 1938. Psycho- Wiggins JS. 1973. Personality and Prediction.
logical Factors in Marital Happiness. Reading, MA: Addison-Wesley
New York: McGraw-Hill Winch RF. 1958. Mate Selection: The Study of
Thibaut JW, Kelley HH. 1959. The Social Psy- Complimentary Needs. New York: Harper
chology of Groups. New York: Wiley Brothers
Ting-Toomey S. 1982. An analysis of verbal