Room Effects

We present clean evidence of a direct social context effect on behavior in a laboratory experiment: the gender composition of the room significantly alters the risk decisions of subjects even when the actions or presence of others are neither payoff nor information relevant. Our design is such that subjects do not know the decisions of others, nor can they be inferred. We find that women become more risk taking as the proportion of men in the group increases. This is most consistent with women imitating the expected behavior of others in the session. Our results imply that aggregate behavior is not a simple extrapolation of individual preferences. Groups might have more extreme behavior than the average individual. JEL codes: C91, D81, J16