Behavioural data are important for group research and are widely used to inform and validate relevant theoretical propositions. There is more debate about the role of such data in illuminating the psychological processes that underpin group behaviour itself. This article focuses on this debate and outlines that behavioral data are limited because (a) they have difficulty distinguishing between interactions between individuals-as-individuals and individuals-as-group members, and (b) when the psychological group is properly understood it becomes clear that much of group research concerns individuals-as-group members and the emergence of a shared social identity (along with shifts in self-definition from I to we). The implication is that behavioural data per se are limited in being able to infer the very psychological processes that are central to group research. In conclusion, examples are outlined where in combination with a social identity perspective (and related work) behavioural data have been informative in advancing group research.