A Behavioral Theory of Elections

4/14/2011 - Why do people vote? Why do voters form parties? When do candidates adopt centrist platforms in general elections?

Political scientists have debated these questions for generations, based largely on the assumption of “rational choice,” as used in traditional economic models. In this rational-choice tradition, voters and candidates make fully rational choices that weigh costs against benefits in line with personal preferences which, in the case of voters, may include policy motivations or the desire to see a particular candidate win but also include the cost of turning out to vote.

However, the new book A Behavioral Theory of Elections attempts to move beyond rational choice toward a real-world model of elections — one that’s rooted in behavioral science and combines insights from psychology with models from game theory and stochastic processes.

Authored by Daniel Diermeier, the IBM Professor of Regulation and Competitive Practice, along with Jonathan Bendor, David A. Siegel and Michael M. Ting, the book offers several important insights, including:

• Voters and candidates perform no heroic feats of computation. Instead, they adapt their choices through trial and error. 
• Real-world decision makers don’t optimize as much as they “satisfice.” That is, they tend to seek satisfactory solutions, rather than optimal ones. 
• “Satisficing” behavior by winners and “searching” behavior by losers leads to a string of winning platforms and government policies that tend toward central policy positions but never fully converge. 
• The rational-choice model suffers from the paradox of voter turnout: Truly rational voters would realize that one vote doesn’t matter and should free-ride, leading to extremely low turnout. However, turnout rates in U.S. presidential elections are consistently above 50 percent. 
• In the behavioral model, higher voter turnout is no paradox. Voters whose candidates win tend to be satisfied with the outcome and will tend to “do the same thing,” i.e., continue to vote. Non-voters whose preferred candidate loses are more likely to be dissatisfied, and to “try something different,” i.e., vote, the next time around. Both these actions push voter turnout well above the extremely low rates predicted by the rational-choice model. 
• Even naïve voters quickly learn to develop persistent party affiliations.

A Behavioral Theory of Elections has earned praise from several prominent political scientists for its novel insights and innovative methodology. Diermeier and his colleagues also developed computational models that illustrate the theory and can be found on the book’s website.

The authors hope readers will use those models to explore other questions and possibilities. Meanwhile, Diermeier and his colleagues plan to expand on their model to address other unanswered questions, such as the effects of generational change on party affiliation and voter turnout.

Diermeier is also the author of the forthcoming book Reputation Rules: Strategies for Building Your Company’s Most Valuable Asset (McGraw-Hill), which will be published in May.