A multivariate hierarchical model for studying psychological change within married couples.

This article illustrates new statistical methods for the study of psychological change in married couples. The design involves time-series data on each partner. The analysis combines longitudinal methods for studies of individual change with cross-sectional methods for the study of matched pairs. Each person is viewed as changing over time as a function of an individual growth curve or change function. As in previous studies of individual change, a person's trajectory depends on time-invarian t personal background characteristics and time-varying changes in the environment. However, unlike typical studies of individual change, a person's changing psychological profile depends, in part, on the influence of that person's partner. These methods apply directly to other types of longitudinal studies on families (e.g., studies that use teacher and parent reports of a child's social behavior). The methodology is flexible in allowing randomly missing data, varying spacing of time points, unbalanced designs, and time-varying and time-invariant covariates. The study of psychological change of married partners involves two basic features that have implications for conceptual and statistical models. First, each individual changes over time. Second, the change trajectories of partners are likely to be related. At minimum, the relational character of change in couples must not be ignored; it may, in fact, be of central interest in many studies. To represent this process, a pair of change functions is first specified for each couple in a Level 1 or within-couple model. At Level 2,