Investigating spousal influence using moment-to-moment affect data from marital conflict

Gottman and colleagues proposed using a dynamical systems model to study dyadic interaction in marriage. In this model, each spouse's affect in each 6-s window is described as a function of an uninfluenced linear steady state and a nonlinear influence function of the partner's affect in the previous window. Recently, an alternative parameter estimation procedure for the equations of marriage was introduced, which is based on threshold autoregressive models. We apply this estimation procedure to data from a study of couples (N = 124) and newlyweds (N = 130) to compare different forms of spousal influence using the Bayesian information criterion. Although results show some statistically significant evidence for influence, this is only slightly greater than what would be expected by random association. One model of influence does not fit all couples. This suggests that for many people initial state and emotional inertia dictate the outcome of the conflict discussion far more than the moment-to-moment affect of the spouse. This latter finding is in conflict with most models of couples' interaction, which suggest that the outcome of conflict discussions are determined by the nature of the couples' mutual influence processes.