Relative importance of conflict topics for within-couple tests - The case of demand/withdraw interaction

Abstract

Researchers commonly employ observational methods, in which partners discuss topics of concern to them, to test gender differences and other within-couple differences in couple conflict behavior. We describe a previously-unidentified assumption upon which statistical tests in these observational studies are frequently reliant - whether each partner is more concerned or dissatisfied with the topic selected for them than the partner is. We term this the relative importance assumption and show that common procedures for selecting conflict discussion topics can lead to widespread violations of the assumption in empirical studies. Study 1 conducts a systematic review of the literature and finds that few existing studies ensure relative importance is met. Study 2 uses two empirical samples to estimate how often relative importance is violated when not ensured, finding it is violated in one-third of interaction tasks. Study 3 examines the potential consequences of violating the relative importance assumption when testing within-couple differences in observed behavior, focusing on gender differences in the demand/withdraw pattern. Results show that these tests were profoundly impacted by violations of relative importance. In light of these violations, we conduct a more rigorous test of demand/withdraw theories and clarify previously-inconsistent results in the literature. We recommend explicit consideration of relative importance for studies testing within-couple effects, provide methodological recommendations for selecting topics in future studies, and discuss implications for clinical practice.