WikiSummary, the Social Science Summary Database

Geys: 'Rational' theories of voter turnout

From WikiSummary, the Free Social Science Summary Database


Geys. 2006. 'Rational' theories of voter turnout: A review. Political Studies Review 4:16-35.

In Brief

Downs (1957) long ago pointed out a paradox: If voters are rational, then rational choice predicts non-voting. In the ensuing decades, scholars have made several attempts to resolve this paradox. Geys reviews the primary attempts, arguing that recent attention to social groups and learning are most promising.

The Paradox

Downs's paradox is well known. The benefits of voting (B) are weighted by the probability (P) that one's vote will be decisive. But since voting is costly (takes time, mainly), then the probability if actually turning out, P*B - C, will be negative for almost any positive C. Thus, no voting.

Solutions to the Paradox

Though there have been several attempts to resolve this paradox, Geys argues that a successful theory should explain not only the 'level' of turnout, but also allow for

  • Strategic voting. Models that require sincere voting don't work.
  • The tendency of certain groups (more educated, wealthier, more partisan, older, and women) to vote more.
  • Higher turnout under PR and in close races.
  • Higher turnout for "first-order" elections (e.g. for Congress) than for "second-order) elections (e.g. for state or local elections, or for supernational European Parliament elections).

Consumption Benefits

Early on, Riker and Ordeshook proposed adding a "D" term: P*B - C + D. This D term would capture the benefit one gains from expressing himself, or from complying with the ethics of voting.


  • This implies that turnout is motivated (primarily) by D - C, since P*B is nearly zero. Why, then, are partisans more likely to turn out?
  • Unless we can explain 'why' some people have a higher D than others, then we've just got a tautological theory: People vote because they get benefits from voting.

Schuessler tried to patch some of these problems by bringing in social theory, "being" vs "doing." I don't vote for candidate Y to obtain candidate Y's victory ("doing"), but to be a candidate Y voter ("being").

The ethical voter

Self interest does not need to be egocentric; it can also be altruistic. Your utility function can include both egocentric and altruistic concerns, with some scalar determining how much weight you place on each set of concerns. Experimental evidence confirms that people are willing to incur personal costs to help others.

Fowler argues that "discriminating altruists"--i.e. those with altruistic preferences toward specific groups--are motivated to vote by altruism, since their B term is affected (i.e. it matters who wins). Those who care about everybody's utility equally ("unconditional altruists") will refrain from voting since there is no net gain from redistribution that will help everybody (zero-sum).

This argument, as well as others about "warm-glow" vs "pure" altruism, help us understand why people might be more interested in voting if they are embedded in a larger social group. More on this below.

Minimax regret

Ferejohn and Fiorina once proposed that voters vote in order to minimize their regret in a future worst-case scenario. If your candidate loses, that's bad; to minimize this probability, and your accompanying regret, you turn out.

Lots of problems here. First, voting must be sincere, not strategic. You always vote for your most preferred candidate. Second, this model doesn't explain why turnout goes up in close elections; apparently, voters should turn out no matter how close the election is. Third, this model overlooks an empirical problem: People don't act the same way when it comes to disaster insurance. People in earthquake-prone areas ought to minimize their regret in a worst-case scenario by purchasing earthquake insurance, yet few do.

Game theoretic approach

The above models all use decision theory. A game theoretic approach would have voters considering what they think everyone else will do. Namely: If everybody votes, then the probability that my vote matters is negligible. Thus, I should not vote. But if I think everybody else will think this way, then nobody will vote. Thus, I should turn out, since my vote will definitely be decisive. But everybody else will think this way, too. Thus, whether I vote depends on how I read everybody else's likely behavior.

Problems: Requires perfect information. Not realistic. Also, this analysis becomes less useful as the size of the electorate increases. Generally, these approaches don't work well.

Group-based models

Man is a social animal. Man should therefore act with his group to achieve group benefits. Problem: Free riding (see Olson 1965). Schram tries to skirt this problem by differentiating opinion leaders from ordinary voters. The opinion leaders (like Olson's entrepreneurs) have their own incentives to incite turnout; voters might want to build credibility or a reputation with these leaders, so they want to turn out. Thus, social pressure leads to turnout.

Alternatively, a social norm might be at play. Particularly in groups that interact frequently, groups might have social rewards for voting (i.e. furthering the group's ends) and social sanctions for abstention. This fits empirical evidence that more connected people vote more.

Referring back to the altruism theories, Geys points out that members of groups might vote if they include their group members' utility in their own personal utility fucntions.

Information models

People have limited information. As Matsusaka argues, people will be more confident about B if they have more information, so informed people will be more likely to vote.

Although this logic seems reasonable, it must assume some predisposition to vote. It can't explain why anybody votes in the first place.

Learning theory

People learn good strategies by observing what worked well for them in the past. If voting was good last time, I'll do it again. If not, I won't. [Doesn't this have the strange implication that, if my candidate loses this year, I'm less likely to turn out next year? And what about learning 'who' to vote for, not just 'whether' to vote?]

This is a D term argument. I learn either that voting (in general) is worth it, or I learn the opposite. Recall Geys's criticism that (early) D-term arguments failed because they did not explain where the D term came from. Learning theory seeks to right this wrong; the D term changes in response to our previous experience with voting.

One branch of this approach is the "habit" school: "Voting and abstention, in other words, are habit forming" (Gerber et al 2003).

Geys calls for more research to test the empirical implications of these theories.