Predicting Committee Behavior in Majority Rule Voting Experiments

In past experiments, committees voting under majority rule have often failed to choose the Condorcet alternative (the core). Since this failure of theory might be due to flaws in experimental design, we developed a voting procedure in which the unique element in the core is also the unique outcome when a Nash equilibrium in undominated strategies for the whole game induces a Nash equilibrium in undominated strategies for every subgame. Nevertheless, our committees frequently choose other alternatives. To explain these results, we formulated a new theory that takes account of threshold effects and identifies a "selection set" predicted to contain the committee's choice. The revised theory performs well both in our experiments and in past experiments.