We study group decision making in a two-step process. In the first step, group members decide by a random dictator mechanism upon the rule they will use in the second step of their decision process. In the second step, all group members then vote between two alternatives and the decision is implemented according to the rule chosen in the first step. One alternative implies zero payoffs for all group members, the other alternative can have positive and negative valuations for each different group member, where valuations are drawn independently. Selfish players should choose a rule in the first stage that implements their preferred choice for sure in the second stage. Inequality averse players should choose even for small positive valuations a rule that implements the alternative that yields zero payoffs for all and subjects that are concerned with maximizing efficiency should for small positive or negative valuations choose majority voting as the decision rule (or a rule between majority voting and one that implements their payoff-maximizing outcome for sure). We find that in the second stage group members almost always vote in favor of the alternative that maximizes their own payoff, whereas the rule choice is often inconsistent with selfish payoff maximization. Furthermore, the rule choice shows no evidence of inequality aversion, but is consistent with efficiency concerns.

http://www.creedexperiment.nl/creed//