Disclaimer. Don't rely on these old notes in lieu of reading the literature, but they can jog your memory. As a grad student long ago, my peers and I collaborated to write and exchange summaries of political science research. I posted them to a wiki-style website. "Wikisum" is now dead but archived here. I cannot vouch for these notes' accuracy, nor can I say who wrote them.
Fowler. 2006. Altruism and turnout. Journal of Politics 68 (August): 674-83.
The traditional calculus of voting assumes that citizens will turn out if p*B > C; that is, the benefits (B) of getting the best outcome (defined in terms of pure self-interest), weighted by the probability (p) that one's vote is decisive, must exceed the costs (C) of voting. Since p rapidly goes to zero, this creates the much-discussed paradox of non-voting.
In truth, though, we have no reason to define B purely in terms of self-interest. Suppose some people also care about others, not just self interest. Then the voting calculus looks like this: p*(B + a*S*N) > C. The p, B, and C terms are defined as before. The "a" represents an individual's level of altruism--how much he cares about others' welfare. The S represents the benefits to society (i.e. to the average citizen) of having the "best" candidate/policy elected. And the N term represents the size of the population. Since p is roughly 1/N, this equation reduces (more or less) to a*S > C. Fowler tests this theory experimentally and finds promising results.
Experimental economists have long known that people exhibit varying degrees of altruism. For example, a purely self-interested economic model of the "dictator game," in which participants are awarded a cash prize and given the opportunity to share this prize with an anonymous other, would predict that self-interested people would share none of the winnings; in experimental tests, though, participants, regularly share quite a bit of this prize (see Figure 1).
Fowler replicates these "dictator game" experiments, but rather than attributing this "sharing" to error, he uses it as a measure of each participant's level of "altruism." By sharing, participants reveal their preference (or lack of it) for helping others.
Previous studies have found that more partisan voters are more likely to think that a particular candidate/policy is more likely to benefit society more. Thus, partisanship correlates with a voter's perception of 'S'. Fowler uses a standard 7-point partisanship scale.
Note that the model predicts an interactive hypothesis: a*S > C. Thus, when voters are altruistic, an increase in partisanship will lead to an increase in turnout; when voters are not altruistic, it will not. Similarly, when voters are partisan, an increase in altruism will lead to an increase in turnout; when voters are moderate, it will not.
As expected, Fowler finds these results. See Figure 2. Note that only the left-hand graphs show significant relationships (as predicted). Interestingly, these variables work even when survey measurements of civic duty (the oft-maligned "D" term) are included in the analysis--and the "D" term variables are not significant. (Of course, the measure of "D" is suspect; see pg 682. He uses a standard measure, but it seems conceptually weak. To have high civic duty, a respondent must disagree that "If a person doesn't care how an election comes out he shouldn't vote in it.")
The experimental setting was not ideal, though it was probably good enough for preliminary research. Fowler administered a questionnaire and a dictator game to a couple hundred undergraduates. Turnout was self-reported. He controls for the other variables known to influence turnout.
A generally insightful study. But what about policies known (by individual voters) to hurt some groups while benefiting others? For example, strongly partisan Democrats are more likely to favor redistribution (e.g. progressive taxes, at the very least); surely they know that these policies hurt the wealthy. Do Democrats consider the net societal cost/benefit (as Fowler seems to imply), or do they consider the benefits only to those most like them? How would this distinction change the predictions?
Does Fowler's model give us a way to integrate a voter's sense of "justice" into the voting calculus? Could this explain why people are motivated to vote by abortion, gay marriage, torture, etc., even when they have no "vested interest" (in the overly narrow sense used by Ratner and Miller) in these issues? (Recall Ratner and Miller's argument: You don't have a self interest in abortion, gay marriage, or torture of enemy combatants unless you are actually female, gay, or an enemy combatant.)
Research by the same authors
Research on similar subjects