A possible solution to the paradox of voter turnout

The paradox of voter turnout is a major empirical puzzle that has been unresolved in rational choice theory. Why do rational actors contribute to the public good of electoral outcomes, especially since the likelihood that their vote will be decisive is nearly zero? I propose a possible solution to this paradox based on the stochastic learning model rather than the subjective expected utility maximization model. In the stochastic learning model, actors are conceived to be backward-looking adaptive learners, rather than forward-looking utility maximizers, and use the past correlations between their choices and collective action outcomes as a guide for their decision whether or not to vote. The stochastic learning model of calculus of voting can solve the paradox because now p ≅ .500 instead of p ≅ 0. The analyses of the 1972-74-76 panels of the American National Election Study largely support the hypotheses derived from the stochastic learning model