PREDICTING THE LONGITUDINAL COURSE OF MARRIAGES*
I am very honored to receive this award from the American Association of Marriage and Family Therapists. I wish to acknowledge those students and colleagues who have contributed to my work through the years. In particular I want to mention Cliff’Notari-ous, Howard Markman, Lowell Krokoff, and Lynn Fainsilber. They have been terrific partners in an exciting adventure. I would also like to acknowledge the contribution of my wife, Dr. Julie Schwartz Gottman, who adds spice to every aspect of my life.
For the past ten years of my life, I have been blessed by a collaboration with Robert Levenson, and the work I will refer to today is a product of this fruitful partnership. Today I will be talking about new analyses of research just completed a few weeks ago, all as yet unpublished.
The current estimate by Andrew Cherlin of the probability that a marriage will end in divorce in its lifetime is 50%. This is an alarmingly high rate and there is no question that the rate has shown dramatic increases in our century. Furthermore, it has been shown that marital dissolution is clearly a clinical issue of serious consequences in terms of the mental and physical health of spouses as well as their children. Despite these facts, it is remarkable that research has been quite unsuccessful at predicting which marriages will separate or divorce. In fact, if you go to the library you will discover that in the nearly 2,000 studies that have been published to date with the words “divorce” or “marital separation” in the title, only three studies have been longitudinal prospective studies. This fact alone was surprising to me. In the vast majority of the studies researchers were interested in the effects of divorce or separation, not in the prediction of divorce or separation.
Furthermore, these three studies have not been very successful in their attempts at prediction. The correlations have been weak, accounting for about 4% of the variation in separation or divorce. Also, the pattem of results has not been very informative. None of these three studies has observed the interaction of couples. So, to summarize, research has not even tried to specify which marital interaction processes are antecedents of the dissolution of marital relationships. Without this ability to predict, we will never understand why marriages fall apart or work well. We need the prediction data to be able to fashion a science of how families function.
Since 1984 Bob and I have been struggling with the problem of building an empiri-cally-based theory of how marriages change. Our questions are simple. How do marriages change over time? Are there any particular interaction pattems that characterize those couples whose marriages deteriorate or improve over time?
The philosophy of our research has been to describe marital interaction and how it is perceived by the couple with as much precision as possible. Therefore, we built a
laboratory that involved collecting videotapes of couples interacting in which we also obtained physiological data synchronized to the video time code. In other words, couples came to our lab at the end of the day and they were interviewed to identify a major area of disagreement in their marriage. They were asked to discuss this area and try to come to some resolution. While they talked we videotaped them and simultaneously collected physiological data including their respective heart rates, blood velocity and the amplitude of each quantity of blood that their hearts ejected on each beat, skin conductance
and gross motor movement.
In the first study we did, we found a result that surprised us. The result was that physiological arousal, particularly of the husband, predicted the longitudinal deterioration of marital satisfaction. Couples whose hearts beat faster, whose blood flowed faster, who sweated more and moved more during marital interaction or even when they were just silent but anticipating marital conflict, had marriages that deteriorated in satisfaction over the course of three years. Also, couples who were physiologically calmer had marriages that improved over time. The strength of the prediction in this longitudinal study was amazing to us. For example, the higher the husband’s heart rate during interaction, controlling marital satisfaction at Time One, the greater the deterioration in marital satisfaction over three years. The correlation was 0.92. This meant that using only physiological data we could predict, with over 95% accuracy, which couples’ marriages would improve and which would deteriorate in the next three years.
Our next question was what marital interaction patterns were related to the deterioration of marital satisfaction over time? Both in this study and in an additional longitudinal study we used the videotapes to analyze the behaviors of the couples. We found that some patterns of marital conflict were beneficial to the marriage in the long run even if they were upsetting at the time. In contrast, when wives were only agreeable an compliant the marriage would deteriorate over time. We also identified a destructive pattern of the husband’s withdrawal as a listener, a pattern we call stonewalling. Stonewalling is a behavior pattern in which the listener presents a stone wall to the speaker, not moving the face very much, avoiding eye contact or using what I call a monitoring eye contact pattern, holding the neck rigid and not using the usual listener responses such as head nods or brief vocalizations that tell the speaker that the listener is tracking. When husbands stonewalled, marital satisfaction decreased over time.
Interestingly enough, we found that deterioration in marital satisfaction didn’t adequately predict separation or divorce. Basically, this was because many unhappily married couples stayed married. There was a missing link between the deterioration of marital satisfaction and later separation and divorce.
To identify this link we designed another study. In this study, we observed 79 couples at Time One and were able to follow up 92.4% (73) of these couples four years later. In addition to physiology and videotapes, we obtained several questionnaires: (1) A Parallel Lives Scale(sample item: My partner and I live pretty separate lives); (2) The Extent to Which the Husband Does Housework, a questionnaire filled out by the wife on the extent to which the husband did household chores; (3) Loneliness (sample item: Sometimes I feel so lonely it hurts); (4) Chronicity of Problems, based on subjective estimates of how long a set of issues in the marriage had been problems; (5) Escalation, which assessed the extent to which the partner’s negative emotions were perceived as aversive, irrational, unexpected, and overwhelming; and, (6) The Avoidance of Conflict, which assesses the extent to which the subject believes that negative feelings and problems were best worked out alone rather than by talking things over.
We coded the behaviors of the couples on the videotapes using these observational coding systems, one of which examined only facial expressions of emotion.
For the facial coding system only the video portion of the videotape was employed.
We measured the rate of particular facial expressions of emotion: happiness, surprise, anger, sad-miserable smiles. Miserable smiles involve raising the lip comers without the eye involvement that sometimes creates wrinkles in the comers of our eyes. The miserable smile is the kind we are likely to make when a photographer instructs us to smile and the smile looks unreal or pasted on, or when we are trYi!!.g to put on a happy face.
What did we find? We found the missing link in predicting separation and divorce.
Briefly, the results were that marital separation was predicted by the following Time One variables. When the videotape’s sound was turned down and only the faces were observed, couples who were more likely to separate showed more of the following facial expressions: wife’s disgust (and this was the strongest correlation in the study, 0.51), husband’s fear, husband’s miserable smile, and wife’s miserable smile. Accompanying these facial expressions were the following behaviors: Husbands and wives were more defensive (they made more excuses, denied responsibility more), wives complained and criticized more, husbands disagreed more and both husbands and wives yes-butted more.
To predict divorce, one needed to add the husband’s stonewalling and the wife’s verbal expressions of contempt.
I would like to summarize these very detailed results for you. I think that, put very simply, couples who are likely to separate and divorce at Time Two were already separating emotionally at Time One. Gary Larson has a cartoon I wish I could show you. It has a drawing of several animals that nature has endowed with barbs and other dangerous protrusions. Then there is a drawing of a man dressed in a crazy costume, carrying a bazooka. Larson’s caption is “Nature’s way of saying ‘don’t touch’.” That is the way I understand these results. Couples who are more likely to dissolve their marriages were more defensive at Time One. I think that a part of defensiveness iii a fending off of contact. In particular it appears for those’ couples more likely to separate or divorce the wives are already more emotionally detached, more critical of their husbands and the husbands showed more fear facial expressions. The wives reported that their lives had already been arranged to be parallel, as if they were living side by side like two railroad tracks that run in the same direction but never touch. I think what we have discovered, quite simply, is that at Time One the wives had already disengaged and emotionally withdrawn from their husbands.
The first stage begins with marital conflict in which the husband becomes very physiologically aroused and stonewalls with his wife. Then, finally, emotionally withdraws from the conflict. Over time he becomes overwhelmed by his wife’s emotions and avoidant of any conflict with her.
The husband’s stonewalling is very aversive for the wife and leads to her physiological arousal. She responds by trying to re-engage her husband.
The second stage is marked by the withdrawal of the wife. She expresses criticism and disgust. Their lives become increasingly more parallel and he is fearful. In short, the husband’s withdrawal from hot marital interaction is an early precursor of the wife’s withdrawal. When both withdraw and are defensive, the marriage is on its way toward separation and divorce.
How powerful were these results? A regression analysis showed that with about six of these variables we could account for over 60% of the variance in predicting marital separation. This is a multiple correlation coefficient of 0.8. This means that if we wanted to predict marital separation four years later using our Time One data we would correctly classify about 90% of the couples as separated or not separated. These are extremely powerful results, about as strong as those studies, for example, that link blood cholesterol with cardiovascular disease.
To get some idea of the gain we obtained by this precise methodology, it is interesting to compare these results to how well the predictions would be based on Time One measure of marital satisfaction. The correlations are about -0.23. This means that Time One marital satisfaction would correctly classify about 26% of the couples.
There were some other interesting results in these analyses. First, the husband’s stonewalling predicted his loneliness, which predicted the deterioration of his physical health over four years. So, the stonewaller pays a high price. We also discovered that men who did housework were far healthier four years later than those who did not.
Men who did housework were less overwhelmed by their wives’ emotions, less avoidant of conflict and had lower heart rates during marital conflict than men who did no housework.
Why might these gender differences exist? Why do men stonewall significantly more than women? Bob Levenson and I proposed that these differences are related to gender differences in physiological response to conflict. Briefly, our hypothesis is that it takes men longer than women to recover from physiological arousal. There is a lot of evidence to suggest that this hypothesis is valid.
Let me tell you why this is an interesting hypothesis. Suppose it is the case that it takes men longer than women to recover from physiological arousal, and that this arousal is aversive. It is logical that males will then become more avoidant of conditions that lead to this aversive, extended arousal. We already know that hot marital interaction is physiologically arousing and so we can deduce that stonewalling should be more likely for males than for females.
What has been left out of this grim picture is the pathway by which marriages improve over time. Interestingly enough, hot marital conflict by itself is not destructive for a marriage if it also includes positive affects such as affection, humor, positive problem-solving, agreement, assent, empathy, and active non-defensive listening. Positive affects reduce physiological arousal.
As far as we know right now, the ratio of these positive to negative affects needs to exceed 10 to 1 for the marriage to be on a trajectory of increasing marital satisfaction.
This brings me to our current series of studies. These expand on our previous work in several ways. First, we are beginning short-term longitudinal studies at different stages of the family life cycle, starting with newlyweds and moving to young parents to couples in their 40s and couples in their 60s. Second, in collaboration with Hans Ocks and Janice Kiecolt-Glaser and Ron Glaser we are including endocrine and immunological assessment to map out the relationships between emotion and health.
Third, we are building an apartment laboratory in which newlywed couples will be observed continuously for a 35-hour weekend. We hope to be able to study the role of positive emotions more closely in this lab.
Fourth, with Neil Jacobson we are studying domestic violence.
Gender differences are clearly important in understanding how marriages function
and dysfunction. In the future we should be able to map out these relationships with,
hopefully, a great deal of precision.
Thank you again for this great honor.
*Reprinted with permission from AAMFT. Previously published as part of the 1989 AAMFT
John M. Gottman, PhD, is Professor, University of Washington, Department of Psychology
N125, Seattle, WA 98195.
Glass, G., Willson, V., & Gottman, J. M. Design and analysis of time–series experiments. Boulder,
CO: Colorado University Press.
Gottman, J. M. (1976). Distressed marital interaction: Analysis and interventions. Champaign, IL: Research Press.
Gottman, J. M. (1979). Marital interaction: Experimental investigations. New York: Academic
Gottman, J. M., Gonso, J., Notarius, C., & Markman, H. (1978). A couple’s guide to communication.
Champaign, IL: Research Press, 1976. Chapter 1 reprinted in G. C. Davison & J. M. Neale
(Eds.), Contemporary readings in psychopathology. New York: Wiley.
Gottman, J. M. (1982). Emotional responsiveness in marital conversations. Journal of Communica-
tion, 32, 108–120.
Gottman, J. M. (1978). Nonsequential data analysis techniques in observational research. In Carl
Haywood & Gene Sackett (Eds.), Observing behavior, Vol. 2 Data collection and analysis
methods. Baltimore, MD: University Park Press.
Gottman, J. M. (1980). The consistency of nonverbal affect and reciprocity in marital interaction.
Journal of Consulting and Clinical Psychology, 48, 711-717.
Gottman, J. M., & Krokoff, L. J. A rapid observational system for global coding of marital interac-
tion, Behavioral Assessment, in press.
Gottman, J. M., & Krokoff, L. J. The relationship between marital interaction and marital satisfac-
tion: A longitudinal view. Journal of Consulting and Clinical Psychology, in press.
Gottman, J. M., & Levenson, R. W. (1986). Assessing the role of emotion in marriage. Behavioral
Assessment, 8, 31-48.
Gottman, J. M., & Levenson, R. W. (1988). The social psychophysiology of marriage. In P. Noller
& M.A. Fitzpatrick (Eds.), Perspectives on marital interaction. Philadelphia: Multilingual
Gottman, J. M., & Levenson, R. W. (1984). Why marriages fail: Affective and physiological patterns
in marital interaction. In J. Masters (Ed.), Boundary areas in social and developmental
psychology. New York: Academic Press.
Gottman, J. M., & Porterfield, A. L. (1981). Communicative competence in the nonverbal behavior
of married couples. Journal of Marriage and the Family, 43, 817-824.
Gottman, J. M., & Markman, H. (1978). Experimental designs in psychotherapy research. In S.
Garfield & A. Bergin (Eds.), Handbook of psychotherapy and behavior change, New York: John
Wiley & Sons.
Gottman, J. M., Markman, H., & Notarius, C. (1977). The topography of marital confiict: A
sequential analysis of verbal and nonverbal behavior. Journal of Marriage and the Family,
Gottman, J. M., Notarius C., Markman, H., Bank, S., Yoppi, B., & Rubin, M. E. (1976). Behavior
exchange theory and marital decision making. Journal of Personality and Social Pathology,
Krokoff, L. J., & Gottman, J. M. (1982). The structural model of marital interaction. In H. H.
Blumberg & P. Hare (Eds.), Small groups, Sussex, England: John Wiley & Sons, Ltd.
Krokoff, L. J., Gottman, J. M., & Roy, A. K. (1988). Blue-collar marital interaction and a compas-
sionate philosophy of marriage. Journal of Personal and Social Relationships.
Levenson, R. W., & Gottman, J. M. (1983). Marital interaction: Psychological linkage and affective
exchange. Journal of Personality and Social Pathology, 45,587-59.
Levenson, R. W., & Gottman, J. M. (1985). Physiological and affective predictors of change in
relationship satisfaction. Journal of Personality and Social Psychology, 49, 85-94.
Mettetal, G., & Gottman, J. M. (1980). Reciprocity and dominance in marital interaction. In J.
Vincent (Ed.I, Advances in family intervention, assessment and theory. Greenwich, CT: JAl
Notarius, C., Markman, H., & Gottman, J. M. (1983). The couples interaction scoring system:
Clinical applications. In E. E. Filsinger (Ed.), A Sourcebook of marriage and family assessment.
Beverly Hills, CA: Sage Publications.