On the interpretation of giving in dictator games when the recipient is a charity

In experimental dictator games, subjects typically give substantial portions of their endowment to their partner, which has been interpreted as evidence that individuals are altruistic. Two influential studies find that when the choice set is expanded to allow subjects to take money away from an anonymous partner, they no longer give money. This suggests that when partners are anonymous, the seemingly altruistic behavior observed in standard dictator games is an artifact of the limited choice set. A vast literature employs dictator games in a different important context where the recipient is not anonymous: charitable giving. In this study, we examine whether generosity vanishes upon allowing subjects to take money when dictators are partnered with a charity instead of an anonymous partner. While we replicate the result that subjects are substantially less generous when the option to take money away from an anonymous partner is added to their choice set, we find that subjects remain generous when the take options are added when the partner is a charity. Regardless of whether subjects are able to take money from the charity or not, subjects give positive amounts just as often, the mean and median donation amounts are similar, and the modal choice is to give the entire endowment to the charity. These results suggest that dictator games remain a useful tool in studies of charitable giving.