Testing models of information avoidance with binary choice dictator games

• In a dictator game experiment, subjects can avoid seeing a receiver’s payoffs.
• The data are used to test several models relating to information avoidance.
• Even those who sacrifice to help the receiver in a stark choice avoid information.
• This allows them to make self-serving choices.
• They are more likely to avoid seeing payoffs when it is unlikely to hurt a receiver.

Standard social choice experiments generally force subjects to make decisions about giving money to another person, but the ability to avoid information outside of the lab could lead to less altruistic or fair behavior than such experiments tend to suggest. I expand on the design of Dana, Weber, and Kuang (2007) to better study information avoidance in an experimental setting. Subjects are given the chance to avoid information about a recipient’s payoffs in a dictator game. I vary the probability that a dictator’s payoffs will be aligned with the recipient’s in order to assess the role of beliefs on avoidance and test contradictory models. The within-subjects approach shows that even people who are generous in a stark choice will make self-serving decisions when they can avoid knowing the recipient’s outcome. People avoid information more often when the self-serving choice is unlikely to hurt the recipient, which supports Rabin’s model (1995) of moral rules and moral preferences.