Another thought on voter turnout, courtesy of the Horserace weblog:
Since everybody is talking 'bout high turnout…
I thought I should answer this question: Does it favor Democrats this year?
Not necessarily! The principle behind the idea is that marginal voters (i.e. people who are not necessarily going to vote every year) tend to be marginal Democrats (i.e. people who generally, but not always and not commitedly, favor Democrats). Thus, the more people that come out to vote, the more marginal voters that are coming out to vote, the more Democratic votes there are.
But let’s think about this marginal voter for a second. This would be the voter whose expected utility from voting is lower than the expected utility of a committed Democrat or committed Republican.
Game theorists would argue that people vote based on a cost-benefit analysis. There are inherent costs to voting (information costs required to learn about the candidates, transportation costs, time lost costs) and these can be compared to the psychological benefits (i.e. fulfilling civic duty, following tradition). Now, in 2000 the benefit from voting was greater than the cost for voting for about 54% of eligible adults. In other words, about 46% of America decided that the costs were too great, 54% decided that they were not.
Suppose that these early turnout indications are true, that the Democratic and GOP faithful are coming out in droves upon droves.
What happens to the marginal voter? The probability that he will vote actually declines, as his time costs are going through the roof. He gets to the polls, sees a wait of up to two hours and decides, “Awww…forget it! I don’t like either of those bozos, anyway!”
This is what would be called by game theorists a “tipping model.” Predicting turnout is a very complicated matter – as a high turnout at time X might indicate a low turnout at time X+1 because people coming at time x+1 see the long wait as too costly. This, in turn, indicates a high turnout at X+2, lower turnout at x+3, etc. What happens is that the new arrivals at the polling place at each point in time calculate the cost of voting and compare it with the expected benefit from waiting.
In general, though, suppose that the GOP has boosted its GOTV effort to rough parity with the Democrats. Turnout during the day will rise and fall according to the above model, but in general it will be higher than in elections past, as there would be more committed/energized Republicans coming to the polls than there were in 2000. In other words the wait to vote at time 2000x<2004x,>
That means that, in general, the maginal voter becomes that much less likely to vote. It does not matter when he arrives, he faces greater costs relative to voting in 2000.
This is perhaps what we have seen in the ABC News/ Wa Po tracking poll, which has noted that the people who are “definitely going to vote” has fallen (outside the MOE) over the last week. In other words, the marginal voter has prudently calculated that there will be longer-than-normal lines and has decided that the costs simply outweigh the benefits.
If we presume that the “marginal voter” leans Democratic, and that the GOP has increased the number of its base supporters, higher turnout would actually benefit Republicans.
This general principle is, incidentally, why bad weather hurts turnout. It does not really affect the base of either party. They’ll come out to vote regardless the weather, wait, etc. Rather, it affects the marginal voter. Bad weather increases the cost of voting (i.e. you suffer the cost of getting wet, poor driving conditions, etc). As this marginal voter leans Democratic, we can say that bad weather helps Republicans.