by making the measurement domain sufficiently rich. To some extent one always pursues this course in designing a study. However, within the context of programmatic research, another approach is possible, and that is the one I prefer from an aesthetic sense. This approach is to expand the theory only when forced to by the data. To some degree this is the well known criterion of parsimony. However, one person's parsimony is another's clutter, so building a theory is a matter of taste. The issue of how to build a theory is not unlike the issue of how to best build a candy store. Successful candy store owners will often tell you that you should be forced to expand by business itself. Don't start big and then try to cut the inventory.
My goal is to build a certain kind of theory, one that will hold as fairly robust with respect to cultural and secular variation. In other words, the appeal to the interface between emotional and biological processes is not accidental. I believe that there are laws of close relationships in our species; we are fundamentally social animals, as are bees, although considerably more complex. Nonetheless, the goal of this new theory is to discover these laws.
The notable gap in this work, in my view, is that it emerges only from the study of conflict. That is fine, because one of the phenomena in this family field that has held up to replication might be called the Primacy of Negative over Positive Affect. Variables related to negative affect are the ones that do the work for us in predicting and understanding variation. However, I believe that we will be led to the more positive aspects of family life through an understanding of the limitations of our current variables.
We do not understand conflict very well. It clearly occurs as a common phenomenon of close relationships, and it undoubtedly has many functions. In a marriage the negotiation of disagreement may preserve intimacy, assuming it is handled well. In one of Sam Vuchinich's tapes (personal communication, 1984) I was struck by the fact that we were seeing a slice of a father's issue with what he sees as his daughter's irresponsibility. He started the fight at dinner. It was not resolved at dinner. Perhaps this argument is enormously prosocial in a long-term developmental sense. Perhaps when this daughter is out on her own she will recall some of her father's words and think that the Old Man wasn't so dumb after all. The longitudinal time course of the resolution of this issue may be of the order of decades. Clearly our brief glimpses of family life are but a beginning. Hopefully the candy story will grow, but it should be forced to grow by the business itself.
This research was supported by NIMH Research Career Development Award K00257 and by PHS Grant MH29910.