Reciprocity in a two-part dictator game

We conduct a dictator game experiment in which recipients in an initial game become dictators in a second game. When the subjects paired remain the same, the amount sent back is strongly correlated with the amount received despite the fact that the interaction is anonymous and is known to be one-time and zero-sum in nature. When the initial recipient is instead paired with a third subject, a less significant and lower-valued correlation between amounts received and sent is exhibited. Intelligence and personality test results, gender, and other characteristics also help to predict sending behavior and degree of reciprocity.