2014 年 30 巻 1 号 p. 56-67
This note gives a primitive idea on coalition formation in a weighted voting experiment. The experimental protocol requires that the roles of subjects be partially fixed throughout each session and that each subject be in favor of only one proposal including his or her own at a time. In two four-player weighted voting games with no veto player, the orders of observed frequencies of minimal winning coalitions (MWCs) were the same as those of predicted ones derived from the following presumptions. (1) Each voter prefers a MWC to another MWC when his or her relative voting weight in the MWC is larger than that in another MWC. (2) The probability of each MWC occurring depends on a score the MWC makes in the Borda count, given such individual preferences of voters over MWCs. Tensions among MWCs in the bargaining process affected the resource allocation among the members within each MWC, which violated Gamson's law in some MWCs. Nevertheless, the above prediction was robust to small violations of Gamson's law.