Gender- and frame-specific audience effects in dictator games

We study dictator allocations using a 2×2$2\times 2$ experimental design that varies the level of anonymity and the choice set, allowing observation of audience effects in both give and take frames. Changes in the distribution of responses across treatment cells allow us to distinguish among alternative motives as elaborated in recent theory. We observe significant audience effects that vary by both frame and gender. The pattern of responses suggests that heterogeneous concerns for reputation and self-signaling across gender give rise to the contextual effects associated with the give and take frames that have previously been observed in the literature.