The Handbook of Conflict Resolution (3rd ed)

by Peter T Coleman

  The scientific climate in the early 1970s in psychological research suggested that the best measures of personality had been relatively unsuccessful in predicting and understanding individual human behavior, accounting for at most 9 percent of the variance (Mischel, 1968). Therefore, the prevailing climate asked, “What chance did psychology have of understanding relationships that involved two people? Wouldn’t one merely square the error and totally fail in predicting and understanding a relationship?” And yet, in hindsight, it is precisely within naturally occurring social organizations that clearly discernible behavioral patterns exist in highly social species. After all, in studying the bee, the Nobel laureate Karl von Frisch discovered the social dance of bees only by observing the hive (von Frisch, 1967). Had he studied only one bee in a laboratory, he would have probably concluded that bees suicidally dash their brains out on glass trying to get out of the window. This turned out to be as true of families of humans as it is of bees in a hive. John and Julie Gottman then took the research out of the ivory tower, created theory and interventions based on it, studied these, then created the Gottman Institute in 1996 to serve couples and their needs through providing research-based workshops and products, and to train clinicians in their research-based methods worldwide.

  STAGE 1: THE DISCOVERY OF RELIABLE PATTERNS OF INTERACTION

  Research on married couples began in 1938 with the publication of a classic book by Louis Terman (Terman, Buttenweiser, Ferguson, Johnson, and Wilson, 1938). Terman and his colleagues interviewed couples and gave them questionnaires. Fortunately, decades of excellent sociological research on couples later produced reliable and valid questionnaire measures of marital satisfaction and happiness. Sociologists were interested in various social factors that affected happiness and the longitudinal course of marriages, but they were uninterested in behavior.

  When John Gottman began his research in 1972 at Indiana University, he was the first to attempt to discover reliable patterns in the data. In this research with couples, he wanted to see if there were patterns of behavior or sequences of interactions that could discriminate happy from unhappy couples. It was not at all clear that these patterns existed.

  Gottman later teamed up with Robert Levenson in 1979. Together they used the newly available technology of home videotape to sample natural interactions—couples talking about the history of their relationship and how they thought about relationships and their own and their parents’ relationships. They also observed couples talking about how their day went (after having been apart for at least eight hours), performing tasks like the NASA moon shot consensus decision-making task (a task that tests decision-making abilities), talking about areas of conflict and trying to resolve them, and so on. It became clear that even highly distressed couples could do very well when they worked together on a standard lab task as long as the task was not personal. For example, on the NASA moon shot task, the couple’s score when they worked on the task together exceeded the best individual partner’s score regardless of the couple’s marital satisfaction. This meant that most couples were not deficient in decision-making skills, regardless of the condition of their marriage. Yet, when some couples, the unhappily married ones, tried to talk about their own conflicts, their conflict resolution skills evaporated into thin air.

  However, laboratory tasks that induced real marital conflict artificially, like a task in which each partner got a different version of some other couple’s travails and had to decide which partner was more at fault, easily discriminated happy from unhappy couples. But it was not actually a useful task because the task induced conflict in unhappy couples, and positive affects, like laughter, in happy couples. In other words, unhappy couples took the task to heart, but happy couples didn’t take the task seriously. Therefore, the task was not ecologically general or clinically useful because it failed to show how happily married couples dealt with their own very real conflicts. So the Gottman lab decided to study real conflicts in both happily and unhappily married couples.

  It was clear in this research that conflict was real and present in all couples, regardless of marital happiness. However, in unhappily married conflict interactions, most conversations began with negative affect, blaming the partner for the problem, while happily married couples were far more likely to begin gently and sometimes, even with positive affects like affection and humor. Also, Gottman and his colleagues quickly discovered that the way the conversation began in the first three minutes, regardless of marital satisfaction, determined how it would continue through the rest of the conversation—in 96 percent of the cases.

  Even in happily married couples, Gottman found that there were always some issues that created high levels of negative affect. Happily married couples initiated some discussions the way unhappily married couples did, and when that happened, the same sequences were observed as in unhappily married couples. However, Gottman determined that happily married couples could repair negativity far more easily and could rebound more easily when asked to talk about a positive topic than could unhappily married couples.

  For unhappily married couples, conflict often became pervasive in the relationship (transferring to our positive discussion topic), and eventually these couples began avoiding one another, leading parallel lives filled with loneliness. Levenson and Gottman charted this deterioration over time and subsequently it was called the Emotional Distance and Loneliness Cascade.

  The Gottman lab discovered stable sequences of interaction in a study of university students (using sequence analysis of Gottman’s observational coding system that scored videotapes and was called the couples’ interaction scoring system, CISS). Later, a graduate student of Gottmans, Mary Ellen Rubin, repeated the same experiment with couples in rural Indiana for her dissertation. Amazingly, the CISS numbers in the two studies discriminated happily from unhappily married couples, differing only in the second decimal place. This was the first hint that replication was possible.

  In a series of research studies, Gottman developed new observational coding systems with Cliff Notarius, at the time Gottman’s student, and the lab applied new methods for analyzing sequences of interaction that were developed by Jim Sackett and Roger Bakeman. These sequences (described in Gottman, Notarius, and Markman, 1979) described the skillful conflict management patterns of happily married couples and how very different they were from the patterns of unhappy couples. One of the first discoveries was that during conflict discussions, the ratio of positive to negative interactions was 5/1 on average for happily married and stable couples and 0.8/1 on average for unhappily married and unstable couples. The laboratory then asked whether all negative interactions were equally corrosive. The answer was no. In particular, four behaviors turned out to be excellent predictors of divorce. Gottman called these “the four horsemen of the apocalypse”: criticism (expressing a complaint as a defect in one’s partner), defensiveness (counterattacking or acting like an innocent victim), contempt (insult, mockery, disrespect, acting superior), and stonewalling (listener withdrawal, no usual listener verbal or nonverbal responses).

  The Gottman lab also began using social exchange theory, an application of John von Neumann and Oskar Morgenstern’s (1949) book on game theory, which was followed by Thibaut and Kelley (1986) in their classic book, The Social Psychology of Groups. To operationalize exchange theory within interactions, Gottman built a device called a “talk table” in which people could interact and also rate after every turn at speech how positive or negative their intentions were and how positive or negative the impacts of the messages they received were. This was the first application of game theory to couples’ interaction, a theme the Gottman lab later returned to in studying trust and commitment. The lab used these methods to define reliable patterns of interaction and thought during conflict. Following a series of peer-reviewed journal articles, Gottman published these results in a series of scientific papers and a book (Gottman, 1979).

  In a randomized clinical trial attempting to apply thes
e early findings to change unhappy marriages, the Gottman lab found they could get large changes in marital satisfaction, but that these changes mostly relapsed within a year. However, one of Gottman’s students, Howard Markman, applied the same intervention (described in Gottman, Notarius, and Markman, 1979) to newlywed couples and consistently discovered that the same intervention as a preventive measure was effective at preventing marital discord and divorce. This was the first discovery of a general effect: Preventive effects with couples who are not yet unhappy are much larger and more stable than intervention effects with unhappy couples.

  STAGE 2: PREDICTION AND THE REPLICATION OF THE PREDICTION

  The second stage of the Gottman research program was attempting longitudinal prediction. Prediction in psychology means being able to predict important outcomes from the patterns observed. In repeated studies over time, the patterns and sequences Gottman observed were able to discriminate happy from unhappy couples.

  In 1979, another research breakthrough occurred. Robert W. Levenson and John Gottman teamed up to combine the study of emotion with psychophysiological measurement and a video recall method that gave them rating dial measures of how people felt during conflict. This was the new way of getting talk table numbers. The research also became longitudinal. Few predictions were made in the first study. They were interested in a measure of physiological linkage because a prior study showed that the skin conductance of two nurses was correlated only if they disliked one another. They thought this phenomenon might be linked to negative affect in couples as well. Indeed it was.

  They were also amazed that in their first study with thirty couples, they were able to predict the change in marital satisfaction (over a three-year period) almost perfectly using just their physiological measures. Time 1 was their first observation of the couples in their new laboratory, and Time 2 occurred three years later. The correlations between Time 1 and Time 2 were very high, with Time 2 marital satisfaction (from the .70s to the .90s) controlling for Time 1 marital satisfaction. They found that the more physiologically aroused couples were in all channels (heart rate, skin conductance, gross motor activity, and blood velocity), the more their marriages deteriorated in happiness over a three-year period, controlling the initial level of marital satisfaction.

  As predicted by exchange theory, the rating dial and observational coding of the couples’ interaction also predicted changes in relationship satisfaction. Levenson and Gottman had never seen such large correlations in their data before (correlations ranged from .7 to .9).

  In another study, they asked couples to first have an events-of-the-day reunion conversation in which couples talked about the events of their day, then next a conflict discussion, and finally a third conversation about a positive topic. What was surprising was that during the conflict discussion, the use of a harsh start-up (mostly by women) was predictable by the partner’s disinterest or irritability in the events-of-the-day discussion; the responses of men during the events-of-the-day conversation were especially important in this prediction of harsh start-up during conflict. It became clear at that point that to understand conflict, one had to also examine the quality of nonconflict conversations. This finding was the beginning of realizing that the quality of the couple’s friendship, especially as maintained (or not maintained) by men, was critical in understanding conflict. Furthermore, the ability to rebound from conflict to the positive conversation became an important marker of the emotion regulation ability of couples. As conflict persists without resolution, apparently it could come to pervade all of a couple’s life.

  Both Levenson and Gottman had discovered Paul Ekman and Wallace Friesen’s facial affect coding system (FACS; Ekman and Rosenberg, 2005) and began working with the Ekman laboratory. Gottman subsequently developed the specific affect coding system (SPAFF), which was an integration of FACS and earlier systems in the Gottman lab. The SPAFF directly coded affect using all channels of communication in a cultural informants system. Gottman also began applying time-series analysis to the analysis of interaction data.

  Levenson and Gottman began attempting to replicate observations from the first study. The subsequent studies that they conducted in their two labs (some with colleagues Laura Carstensen, Lynn Katz, Sybil Carrere, and Neil Jacobson in the Gottman lab; Jacobson and Gottman, 2007) eventually spanned the entire life course, from a study following newlyweds through the transition to parenthood to a study of two groups of couples (one in their forties and one in their sixties) in the Levenson lab at University of California, Berkeley, on the transition through retirement. The study of couples in later life involved following couples for twenty years in Levenson’s Berkeley lab.

  The Gottman lab at the University of Illinois also studied the linkages between marital interaction, parenting, and children’s social development (with Lynn Katz) and later, at the University of Washington, studied these linkages with infants (with Alyson Shapiro). Gottman had begun studying families, at first examining children from age three longitudinally up to age fifteen. He developed the concept of meta-emotion, which is how people feel about emotion in general, specific emotions (like anger), emotional expression, and emotional understanding (Gottman, Katz, and Hooven, 1997). The idea of emotion coaching emerged from that research, a scientific validation of the work of child psychologist Haim Ginott (2003). In a study of newlyweds, Gottman began studying the transition to parenthood and learning how to do research on babies and parents (Gottman, 2004; Gottman and DeClaire, 1998).

  Gottman and Levenson discovered that couples interaction had enormous stability over time (about 80 percent stability in conflict discussions separated by three years). They also discovered that most relationship problems (69 percent) never get resolved but are perpetual problems based on personality differences between partners (reported in Gottman, 1999). In seven longitudinal studies, one with violent couples (with Neil Jacobson), the initial findings and predictions replicated. The researchers could predict whether a couple would divorce with an average of over 90 percent accuracy across studies using the ratio of positive to negative SPAFF codes, the four horsemen of the apocalypse (criticism, defensiveness, contempt, and stonewalling), physiology, the rating dial, and an interview they devised called the oral history interview, as coded by Kim Buehlman’s coding system. They could also predict whether stable couples would be happy or unhappy using measures of positive affect during conflict, which Jim Coan and Gottman discovered was used not randomly but to physiologically soothe the partner. They also discovered that men accepting influence from women was predictive of happy and stable marriages. Levenson discovered that humor was physiologically soothing and (with Anna Ruef) that empathy had a physiological substrate in a study using the rating dial.

  The Gottman-Levenson labs’ prediction of divorce was often misunderstood by laypeople who were not very acquainted with the mathematics of probability. Some critics, for example, claimed that a 90 percent divorce rate was not impressive since the national divorce rate was about 50 percent. They said, “If you guess that everyone will divorce, you will be right half the time.” However, the commonly reported 50 percent rate is an estimate of the chance of divorcing over a very long forty-year period. Our divorce predictions were over much shorter time periods, like six years. In six years, for example, seventeen newlywed couples divorced out of 130 newlywed couples, or only 13.1 percent. Guessing that each of these newlywed couples would divorce in 6 years would produce about an 87 percent error rate. A 90 percent correct prediction rate is like blindly picking correctly (by chance alone) 15 out of 17 red balls (the couples who will divorce) in an urn that also contains 113 white balls (the couples who do not divorce). The chance of that correct prediction is about 10–13. Hence, the prediction rate in the Gottman lab was probably not a chance event.

  Later, Jacobson and Gottman collaborated on a basic study of domestic violence with four groups of couples: (1) happily married, nonviolent, (2) unhappily married, nonviolent, (3) situationally violent, a
nd (4) characterologically violent (all men). They discovered a typology of battering that has mostly been replicated in the literature. Later this finding led to a successful treatment for situational domestic violence.

  In 1986 Gottman built an apartment laboratory at the University of Washington, in which his student Janice Driver spent a decade (first as a volunteer and then a doctoral student) discovering the basis of friendship and intimacy and its relation to conflict through a bids and turning coding system. With that work, Gottman and Driver discovered how couples create and maintain friendship and intimacy and how turning toward or away from a bid for emotional connection (during nonconflict interaction) was related to behavior during conflict, especially repair. Newlyweds who divorced six years after the wedding had turned toward bids 33 percent of the time, while newlyweds who stayed married six years after the wedding had turned toward bids 86 percent of the time. The idea of the friendship being an “emotional bank account” was verified. Friendship was related to repair of negativity and, also surprisingly, to the quality of sexual intimacy.

  When fourteen-year longitudinal data became available, Gottman and Levenson discovered a second dysfunctional pattern, emotional disengagement. It was marked by the absence of both high levels of negative affect and any level of positive affect during conflict (no interest, affection, humor, or empathy). Now they could predict not only if a couple would divorce, but approximately when. Couples who had the four horsemen divorced an average of 5.6 years after the wedding, while emotionally disengaged couples divorced an average of 16.2 years after the wedding. This was a very new finding.

  Levenson, Carstensen, and Gottman began studying marriage in later life with two groups of couples in the Bay Area, one in their forties and one in their sixties. Thanks to Levenson’s tenacity, this work has turned out to be a twenty-year longitudinal study that his lab is now finishing.