BEHAVIORAL SOCIAL CHOICE: CONSENSUS AMONG CONSENSUS METHODS
Like many disciplines, the Decision Sciences are split between theorists and empiricists. Rational choice theory develops mathematical models of how rational decision-makers must behave; behavioral decision research investigates how people actually make decisions. Professor Regenwetter believes these two approaches can be merged fruitfully in the domain of social choice theory (i.e., the mathematical theory of voting). During his Center appointment, he will continue his exploration of (a) the rational theory of how voting procedures work in principle and (b) empirical research on what happens in real elections.
Existing literature on social choice theory is primarily focused on analyzing and comparing mathematical properties of different voting methods. Much of that literature is commonly interpreted as showing that “democratic decision-making” is a myth, because there does not exist a universally rational method for aggregating preferences. The most famous result is Kenneth Arrow’s Impossibility Theorem. In lay terms, this theorem shows that it is impossible for any voting procedure to simultaneously satisfy a certain list of important desirable properties on all conceivable preference distributions.
Professor Regenwetter challenged some of this established wisdom in his coauthored book, Behavioral Social Choice: Probabilistic Models, Statistical Inference, and Applications (Cambridge University Press, 2006), demonstrating that real world elections circumvent the most famous voting paradoxes of the theoretical literature. A crucial reason suggested is that the real world does not encounter all conceivable preference distributions. Instead, real electorates display preference distributions with important patterns that are ignored by the Impossibility Theorem.
As an Associate, Professor Regenwetter will work on a second book on this topic and challenge the Impossibility Theorem more directly. His careful examination of real world survey and ballot data suggests that competing voting procedures may agree with each other to a surprising degree. In other words, it appears that the preference distributions of real electorates may avoid many of those “conceivable” distributions on which Arrow’s impossibility result hinges.