14 - Couples, Gender, and Time: Comments on Method

Summary

The difficulties of conducting research in the social sciences are compounded when one studies longitudinal married couples. To begin with, married couples are not randomly paired. They are similar to one another on a host of variables. Second, there are all sorts of missing data. One or both members do not show up for one of the sessions; if enough time passes about half the couples will separate. Third, the partners influence each other. Fourth, the meaning of variables changes over the course of the relationship. Despite these and many other difficulties, many researchers still attempt to study marriage over time.

This chapter provides a checklist of things I would like to see in the analysis of longitudinal data on heterosexual couples, with particular emphasis on effects involving couples, gender, and time. I indicate how these three types of effects were treated in the seven empirical chapters in this volume that report on the same variables assessed at least twice. Because I am simply trying to illustrate how these effects have been handled and how they might be handled differently in the future, I discuss only the major analyses, in these chapters; basically, I concentrate on analyses presented in tables and figures.