Abstract
► In Dictator Game experiments, donations to charity sometimes depend on participant type. ► When donors have only one potential recipient, behavior does not vary significantly with recipient or participant type. ► When donors have multiple potential recipients, behavior does vary significantly with recipient and participant type. ► The results can be interpreted within a self-image model.
We conduct two simple experiments in which student participants are invited to give some of the money that they have earned to an international development charity for use in one of two African countries. In the between-groups experiment, participants are given the opportunity to donate to one country only. They are matched randomly with a country, and are given information about why that country might be poor; the information varies between the two treatments. In the within-group experiment, participants are given the opportunity to donate to either or both of the countries, and are given all of the information. Analysis of the results indicates that the effect of the difference in experimental design on the decisions made depends partly on observable participant characteristics. The results can be interpreted in terms of a model incorporating self-image.