We discuss the logic of research designs for dyadic interaction and present statistical models with parameters that are tied to psychologically relevant constructs. Building on Karl Pearson's classic nineteenth–century statistical analysis of within–organism similarity, we describe several approaches to indexing dyadic interdependence and provide graphical methods for visualizing dyadic data. We also describe several statistical and conceptual solutions to the ‘levels of analytic’ problem in analysing dyadic data. These analytic strategies allow the researcher to examine and measure psychological questions of interdependence and social influence. We provide illustrative data from casually interacting and romantic dyads.