A central problem in the area of research on the family is that of the relative inability of researchers to empirically test current theoretical models. Any theory regarding the influence of family characteristics and interaction patterns on developmental processes ought to provide for the confirmability of constructs and the replicability of associated effects. Both of these qualities are nearly absent from the activities of family researchers today. By Kuhn's (1962) standards, research on family processes is in the pre-scientific stage and may require certain methodological advances to overcome present problems. A sequence of methodological options that may be used to develop the kind of paradigm(s) that are appropriate for research on family processes includes (1) structuring a research situation to observe parent-child interaction years after intervention at an early age; (2) developing measures; (3) coding; (4) establishing reliability; (5) quantifying multivariate patterns of behavior; (6) treating patterns of behavior as dependent variables to examine the influence of treatment and demographic factors separately; and (7) using interaction patterns as independent variables in accounting for variability in numerous individual outcomes exhibited by the child. In general, researchers need to converge their efforts on utilizing common methods to test significant aspects of family process models. (Author/RH)