Messaging matters: Investigating differences in WhatsApp communication patterns across different relationship constellations

A substantial proportion of everyday communication in close, interpersonal relationships happens through mobile instant messaging (MIM) services such as WhatsApp. So far, only few studies have focused on relationship-level differences with respect to chatting behavior. Our study seeks to address this gap by combining survey data (N = 357) with donated dyadic WhatsApp chat logs (N = 142) to investigate differences in chatting behavior across different relationship constellations defined by gender combinations, relationship types, and levels of interpersonal closeness. Results show that behavioral metrics representing messaging frequency, expressiveness, reciprocity, and timing substantially differ for different relationship constellations. Using Elastic Net and Random Forest models, the gender combination of chatters and relationship type, but not interpersonal closeness, could be predicted from chatting behavior with significantly better performance than a naïve baseline model. We contextualize our findings regarding the literature on interpersonal relationship and communication research, discuss limitations of this study, and provide recommendations for future work. • We combine relationship-level survey data with anonymized dyadic WhatsApp chats • We compute 18 metrics for chatting frequency, expressiveness, reciprocity & timing • We predict gender combination and relationship type of dyads from chat metrics • We discuss variable importance for gender combinations and relationship types • Results highlight potentials of chat log data for social relationship research