"When in Rome": Identifying social norms using coordination games

Previous research in economics, social psychology, and sociology has produced compelling evidence that social norms influence behavior. In this paper we apply the Krupka and Weber (2013) norm elicitation procedure and present U.S. and non-U.S. born subjects with two scenarios for which tipping and punctuality norms are known to vary across countries. We elicit shared beliefs by having subjects match appropriateness ratings of different actions (such as arriving late or on time) to another randomly selected participant from the same university or to a participant who is born in the same country. We also elicit personal beliefs without the matching task. We test whether the responses from the coordination task can be interpreted as social norms by comparing responses from the coordination game with actual social norms (as identified using independent materials such as tipping guides for travelers). We compare responses elicited with the matching tasks to those elicited without the matching task to test whether the coordination device itself is essential for identifying social norms. We find that appropriateness ratings for different actions vary with the reference group in the matching task. Further, the ratings obtained from the matching task vary in a manner consistent with the actual social norms of that reference group. Thus, we find that shared beliefs correspond more closely to externally validated social norms compared to personal beliefs. Second, we highlight the importance that reference groups (for the coordination task) can play.