A structural equation model of dyadic relationships within the family system.

Ideas about the association between family interaction patterns and mental health have inspired a great deal of research on family relationships. Conventional statistical models (e.g., analysis of variance), however, are not suited for the analysis of the nonindependent observations produced by family systems; yet, these sources of nonindependence (e.g., reciprocity, partner effects) are very interesting to many family researchers. This article demonstrates how structural equation modeling enables the study of family processes in a way that is consistent with a systems perspective. An analysis of perceived coerciveness in 2-parent, 2-child families illustrates the method.