Non-selfish behaviour: Are social preferences or social norms revealed in distribution decisions?

People frequently behave non-selfishly in situations where they can reduce their own payoff to help others. It is typically assumed that such pro-social behaviour arises because people are motivated by a social preference. An alternative explanation is that they follow a social norm. We test with two survey experiments (N=2,408) which of these two explanations can better explain decisions people make in a simple distribution game under three different elicitation mechanisms. Unlike previous studies, we elicit preferences and perceived social norms directly for each subject. We find that i) norm-following better explains people’s distributive choices compared to social preferences and ii) lack of confidence in one’s social preference –itself explained by weaker social identification— predicts norm-following. Our findings imply that the Pareto criterion has weaker (than previously thought) foundations for welfare evaluations, but this effect may be attenuated in societies with stronger social identification. Perhaps unexpectedly, but unsurprisingly given i) above, we find that different mechanisms for eliciting social preferences have no effect on distribution decisions.