Cool story bro.
This assumes that each vote is independent, which is a great simplifying assumption but laughably false. If you assume that whatever factors cause you to vote or note only apply to you, then the claim is plausible, although you would have to throw out everything people think they know about social science. A small price to pay, I’m sure.
Do you mean that if nobody else votes, your vote will count? As one of the approximately 40% of the population that doesn’t vote (hey, I don’t even live in a swing state), I am never surprised to see the next day that huge numbers of people acted differently from me and showed up at the polls.
Isn’t the premise of a social contract that citizens are going to make individually suboptimal decisions in support of an institution (or put another way, citizens weight the persistence of the institution in their own utility)? It seems like this kind of reasoning would regard almost every social contract as stupid (regardless of how the cost-benefit in this particular case is estimated).
In case anyone is wondering:
Steve Levitt: “Well, one good indicator of a person who’s not so smart is if they vote in a presidential election because they think their vote might actually decide which candidate wins.”
Andrew has a paper which argues that it is rational for people to vote in presidential elections. The required ingredient is that people care about strangers. Even if a single vote has a miniscule chance of deciding an election, the results of the election affect so many people that it is still worth it to vote.
Few can be as smart an Economist, let alone Levitt, but in this case I don’t think the calculation in Gelman’s paper quite captures the probability as seen by an individual voter. If anything, Levitt’s remark looks even worse with a more careful calculation. Presumably an individual voter would calculate:
P(“my vote is decisive”| “what I know about the election”)
But an individual voter doesn’t know beforehand who else is voting. Knowing this can radically change the given probability. A more realistic calculation might involve something like: “the fraction F of the eligible population who votes”. The above probability becomes:
P(“my vote is decisive’)=Sum_F P(“my vote is decisive”|F)P(F)
The prior on F can be affected in funny ways as my background knowledge evolves. Suppose I do this calculation initially and discover that P(“my vote is decisive”) is practically zero. Then if I believe other people are making the same calculation, I may suspect many (like Levitt) will skip the election. That in turn would cause me to shift the probability mass for the prior P(F) to smaller values of F. The effect would be to increase P(“my vote is decisive”)!
The net result is that I may be motivated to vote not because P(“my vote is decisive”) is large, but rather to insure that my P(F) is accurate and consequently the probability that my vote is decisive is low. In other words, an intelligent agent could rationally decide to vote precisely to ensure that their vote isn’t decisive (and to force everyone else into similar behavior)!
Incidentally, I have some first hand experience with this. During the first major election in Iraq, the Sunnis boycotted the elected. Well, most of them did anyway. One party, the Iraqi Islamic Party, didn’t boycott the election and took all the seats.
Later on, during the Awakening Movement, the Sunni tribesmen joined the Marines and began killing Al Queda in earnest. The fact that the Iraqi Islamic Party held all the political seats was bitterly regretted by most Sunnis and caused endless problems for us Marines until there was another election.
Just to clarify: because of the game theoretic nature of the problem, the best a rational agent can hope for is to ensure that the candidate who would win if everyone voted is the same as the one who will actually win. An individual voter might like their candidate to win no matter what, but they can’t bring that about in equilibrium with their vote. They can however increase the chance that the actual winner is the one who deserves to win (by decreasing the probability that their vote is decisive)
I think this does reflect how many people to think of it. The 2000 election motivated many people to vote just to ensure nothing like the 2000 election happens again.
In any other branch of science, if the data do not fit the theory then the scientist has to think of a better theory. It’s a shame when economists decide that the data are wrong (People are irrational!) and so is any scientist that disagrees with the theory (Those guys aren’t smart!)
I agree—but, as I’ve discussed before in this space, there are two modes of reasoning for conventional (whoops, I mean “rogue”) economists of the Levitt variety. One mode of reasoning is, as you say, to figure out the optimal course of action and then to declare that everyone who disagrees is irrational; the other is to declare that everyone is rational as a starting point. The usual “counterintuitive” way to proceed seems to be to take any behavior that seems reasonable or moral and prove it’s irrational, then to turn around and take any behavior that seems foolish or antisocial and assume it’s rational, sometimes taking an indignant populist view in doing so. Thus we can smugly laugh at silly upper-middle-class middle-aged people who vote at a 90% rate while soberly respecting the rationality of felons and drug addicts. In disagreeing with this attitude, I’m not trying to claim that all upper-middle-class Americans are rational or that felons and drug addicts are not (and, of course, there’s some overlap among these categories), but rather that mainstream (sorry, I mean “maverick”) economists seem to me all to eager to mock and disparage commonplace social behavior and to “counterintuitively” accept the rationality of all sorts of behaviors that most people would consider antisocial, mistaken, or even stupid.
A kindred spirit?
The median vote in 2000 was Sandra Day O’Connor. So much for Levitt.
There is an article in the American Economic Review (top econ journal) that empirically makes a case for voting to be worth one`s while, but I don’t recall the title or authors (!). Wish I could find the link.
I’m still struggling with the idea that the aim of voting is to cast the decisive vote in the election – why?
In team sports the aim is that your team wins, not that you make the winning score…is that irrational too?
Conditional on deciding to vote, yes, I agree that it makes sense to vote for your preferred candidate. The question I was addressing was whether it makes sense to go and vote in the first place. I do not say that “the” aim of voting is to cast a decisive vote, but it is “an” aim—despite Levitt’s claim that it’s “not so smart” to think that way.