Family research has been hindered by a reliance on either empirical relationships or abstract theoretical claims. There has been no effective way to bridge the gap between statistical relationships and theoretical propositions. The application of linear structural equation models helps to bridge this gap. Empirical propositions are not developed. Rather, the statistical relationships between imperfectly measured variables are used to estimate the true effects between theoretical propositions. In this sense theory and method are brought together. This approach to theory construction allows for surplus meaning in theoretical constructs and emphasizes systemic meaning for these concepts. The approach is especially valuable for subjective orientations such as symbolic interaction. It is applied here to an example of family research. (Author/CM)