Laboratory experiments have become a popular tool to learn about preferences and behavior. Classic laboratory games ordinarily have two characteristics. First, the players are anonymous, and second, they are strangers. This study subverts these characteristics to explore behavior when subjects play against identified members of their own social network. We play an ultimatum game in highly connected, rural villages in Sierra Leone. We randomly assign players to either the classic game or a modified game with revealed identity and compare behavior across these variations. We find that classic games do not translate well to behavior in the local context. Revealing the identity of the receiver leads to an increase in offers, supporting the idea that people care about revealed attributes, and that their average belief about one’s personal reputation exceeds the average belief about the reputation of the status quo. Variation in sizes of offers made increases in the version of the game with the most information revealed as would be expected if social relationships can be both good and bad.
