Approaches to modeling interpersonal complementarity in intensive longitudinal data

Contemporary integrative interpersonal theory (CIIT) posits that successful social interactions are characterized by complementarity: correspondence in interpersonal warmth and reciprocity in interpersonal dominance. Interactions with high complementarity evoke more positive affect and less negative affect. Modeling complementarity is challenging because it requires capturing the interpersonal behavior of individuals along the two dimensions of warmth and dominance. This study compares three approaches—statistical interaction, multilevel response surface analysis, and Euclidean distance—for modeling complementarity across four datasets. The approaches varied in the consistency of findings and proportion of variance explained. Findings suggest the Euclidean approach for parsimony and theoretical coherence, whereas multilevel response surface analysis is preferable for comprehensively modeling the interplay of self and other on the interpersonal dimensions.