Monday, October 11, 2004
A paradox? Really?
There's an article on the paradox of voting in Slate. Assume people only vote to affect an election's outcome. Voting is costly (time, energy used going to the poll), and there is a payoff only when you break a tie. In a large election, ties are very very unlikely. So why do people vote?
Lots of clever and not-so-clever reasons for "rational" policy-motivated voting have been advanced, but I think the problem is the assumption that people only vote to influence the outcome.
I think actual voters can be broken down as follows:
Group 1: People who see the reason to vote as influencing the election outcome, and still rationally decide to vote because they are uncertain how close the election will be.
Size of group: Vanishingly small
Group 2: People who see the reason to vote as influencing the election outcome, and still vote because they misunderstand basic probability.
Size of group: A distinct minority
Group 3: People who vote because they would feel ashamed if they didn't (pressure from social networks) and/or who would like to tell other people how they voted (self-expression).
Size of group: Most voters
Political scientists have spent a lot of time trying to rationalize Group 1. Partly, this is because social scientists love the counter-intuitive, and a paper that showed it was rational to vote in a 100 million person election in order to influence the outcome would be as counter-intuitive as they come. But there is another reason: political science's economics envy. We wish we could build a general equilibrium model with voters voting for politicians based on policy, and politicians setting policy to win voters, all at the same time. We need predictable (perhaps, but not necessarily rational) policy-motivated voters to make this work.
A general equilibrium model would be very cool. But if voters are almost all Group 3 and 2, we're not modelling them right. Perhaps a behavioral economics/political science fusion might save the day?
Lots of clever and not-so-clever reasons for "rational" policy-motivated voting have been advanced, but I think the problem is the assumption that people only vote to influence the outcome.
I think actual voters can be broken down as follows:
Group 1: People who see the reason to vote as influencing the election outcome, and still rationally decide to vote because they are uncertain how close the election will be.
Size of group: Vanishingly small
Group 2: People who see the reason to vote as influencing the election outcome, and still vote because they misunderstand basic probability.
Size of group: A distinct minority
Group 3: People who vote because they would feel ashamed if they didn't (pressure from social networks) and/or who would like to tell other people how they voted (self-expression).
Size of group: Most voters
Political scientists have spent a lot of time trying to rationalize Group 1. Partly, this is because social scientists love the counter-intuitive, and a paper that showed it was rational to vote in a 100 million person election in order to influence the outcome would be as counter-intuitive as they come. But there is another reason: political science's economics envy. We wish we could build a general equilibrium model with voters voting for politicians based on policy, and politicians setting policy to win voters, all at the same time. We need predictable (perhaps, but not necessarily rational) policy-motivated voters to make this work.
A general equilibrium model would be very cool. But if voters are almost all Group 3 and 2, we're not modelling them right. Perhaps a behavioral economics/political science fusion might save the day?
