What factors influence the choice of electoral systems? How do the rules to change the rules affect this choice? Can we predict what coalitions will select which voting rules? In this paper we employ a laboratory experiment that tests a special case of a model of electoral rules selection (Benoit 2004), one in which there is no existing rule in place. In the experiment, groups of subjects must select a voting rule (Plurality, Runoff, or Borda) to be used in an election that ultimately determines their earnings. We collect data on subjects’ choices and negotiation processes in a computerized chat. We compare this data across two treatments which vary the level of agreement required, either majority or unanimity, to select the voting rule. We find that the negotiation process and subsequent choice of voting rule depends on the meta-level agreement threshold rule. The Plurality rule is selected more often when a majority is needed whereas the Borda count rule is selected more often when unanimous agreement is required. We also find support for testable implications of the special case in Benoit’s model: When only a majority is required, the model accurately predicts the coalitions that form along with their selected voting rule. When unanimity is required, negotiations and choices are more likely to focus on fairness, equity, and maximizing total payoffs. In addition, negotiations and subsequent choices under the unanimity rule are similar to subject behavior when payoff-uncertainty is introduced.