Methodological advances to studying similarity effects in friendships

Similarity plays an important role in the emergence and quality of friendships. However, past work has commonly used (absolute) difference scores or profile correlations to study similarity effects which introduce substantial ambiguities. Here, we review a statistically more precise and detailed approach, namely polynomial regression and response surface analyses, that can be used examine and illustrate how similarities in people's characteristics (e.g., personality traits, human values, attitudes) relate to friendship outcomes. We discuss strengths and rarely discussed limitations of the approach, call on future research to use this approach in an exploratory fashion to provide more evidence on the nature and boundary conditions of similarity effects in order to generate testable hypotheses, and briefly highlight the role of complementarity.