THE TWO previous papers in this series reported on the theoretical framework and method for evaluating the "Family Interaction Scales" and on the data analysis and research findings.^1,2 This paper will focus on some of the methodological issues of the research and on a discussion of the substantive findings. Details on the theory, method, and research findings were covered in the first two papers and will not be repeated here.

ReviewÌý The goal of the "Family Interaction Scales" project was primarily methodological: to develop and evaluate an instrument for investigating whole family interaction. We concluded from our findings that this instrument was a reliable and valid one, that is, the Scales had relatively high interand intra-rater reliability and the variables derived from the Scale categories yielded statistically significant differences among several groups of families.A series of hypotheses, expressed in terms of the variables, were written in advance of