Structural flaws in the bridge from basic research on marriage to interventions for couples

In an article that received widespread media attention, Gottman, Coan, Carrere, and Swanson (1998) reported on a longitudinal study of newlywed couples and recommended major changes in the practice of marital therapy and education. Most notably, they called for the abandonment of interventions that promote active listening, they concluded that anger was not detrimental to marital outcomes, and they advocated an interaction pattern wherein wives raise issues more gently and husbands more readily accept wives' influence. Because of several methodological and conceptual shortcomings, the recommendations of Gottman et al. appear to us to be premature at best. We raise methodological concerns about the nonrandom selection of research subjects, failure to control for factors that may have differentiated couples initially, procedural issues regarding observational and physiological data, and ambiguity about statistical tests employed. We raise conceptual concerns about how the labeling of patterns affects the conclusions drawn and also about the use of correlational data to make strong causal inferences. We conclude that the article by Gottman et al. risks influencing couples and practitioners alike in a manner that, in our view exceeds the scope and methods of the underlying research.