I’m a political scientist, but I don’t study elections. At least not real elections. So I disappoint folks—particularly reporters—who think that, as a political scientist, I should have a keen interest in, and insights into, presidential races. But just because I’m more interested in the votes that took place at the constitutional convention in 1787 than I am in the most recent poll doesn’t mean that I’m entirely oblivious to the presidential election. So let me serve up a few thoughts prompted by the election.
I don’t recall an election in which polling has received quite as much attention as it has during this election. Traditional state-by-state polling suggests a solid Obama victory in the Electoral College—which is where it counts, constitutionally—yet several relatively sober Republican and conservative commentators suggest that state-pollsters are using flawed assumptions of voter turnout. National polls have marginally moved back in Obama’s direction, but there are large discrepancies across polls.
Here’s the twist. The field of statistics is a fairly well-developed discipline. Given a characteristic in a population that we want to learn about, we know that sampling that population will give us an estimate of how many members of the population share that characteristic. And we know that our estimate gets better the larger the sample. We can take a sample and, depending on the sample size, provide as accurate an estimate of, say, the number of brown-eyed people in the population as we want to.
The difference between estimating the number of brown-eyed people, however, and estimating the vote for a presidential candidate is that the population that we’re trying to sample for an election doesn’t yet exist at the time we’re actually trying to sample the population.
In order to make a reasonable prediction of the election outcome, pollsters need not only to know the candidate preference of the people in their sample, they also need to know who will turn out to vote for an election that has yet to occur. That introduces a real twist to the estimation problem that cannot be remedied by increasing the sample size of the survey.
On top of this, many seem to think that turnout is a function of whether potential voters believe that their candidate has a chance to win. Thus partisan commentators have an incentive to spin stories about polls being wrong.
Here’s a general prediction: Whichever candidate says, “The only poll that counts is the one on election day,” that’s the candidate who will lose. As for today: I suppose if someone held a gun to my head and forced me to wager, I’d place my money on an Obama victory. But I have no reason to dispute those who claim that turnout models based on turnout in previous elections undercount turnout for Romney. So I’ll come home tonight after prison ministry and tune in to see. After all, the only poll that counts is today’s.
As I noted above, if a commentator states in a partisan context that turnout will decrease if people believe in advance that their candidate will not win, no one gets upset with what the commentator statement implies about voters. Yet articulate the underlying assumption as a full-blown theory, and it upsets lots of folks. It’s called the “paradox of voting,” and it suggests that, for most of us for most of the time, voting is irrational. That “political science” lesson clashes with the “civics” lessons that most of us have had drilled into us since we were children.
The critical move in the theoretical argument is to claim that your vote matters only if it’s the pivotal vote that elects your candidate. The thing is, the cost of voting, even though very small, is a constant. (Call it “C.”) You assume that cost whether your candidate wins or not. But your actions create that benefit (“B”) only if you are the voter who puts your candidate over the top. If your candidate wins by two or more votes, then you receive the benefit without assuming any cost. And if your candidate loses by two or more votes, then you receive a payoff of zero whether or not you vote. Therefore it is worth your time to vote only if your vote makes the difference in the election.
Consider an election in which 100,000 people vote (excluding you). That would be a small election. Think of an election in a modestly sized town (let alone a state or national election). Even placing a fairly high value on the benefits of your candidate winning, and placing a fairly low valuation on your time, the probability (“p”) that your vote is the pivotal vote is extremely small. So the expected benefits of voting for you will hardly ever outweigh the cost of voting (or pB – C < 0 for most people’s ps, Bs and Cs). If you don’t vote because your candidate is behind in polling, by the same logic you should never (or almost never) vote at all.
To be sure, we can play with the benefits side of the equation: voting can be an act of solidarity, a perceived obligation of citizenship, valued for its expressive sake, etc. And we talk about voter turnout all the time as if the cost of voting affects the propensity of registered voters to turnout at the polls. Nonetheless, forming the decision to vote as a sterile equation, and talking about the irrationality of voting, tends to jar students, particularly when it comes from political scientists.
There is a strong civil religion in the U.S. and its sacrament is voting. Even to appear to criticize voting is to attack that civil religion. Some years back a faculty member presented the bare-bones voting calculus to a large introductory section of U.S. politics. A student interpreted the presentation as the professor telling them “that people shouldn’t vote.” The University president at the time even made an inquiry after Rush Limbaugh reported on the controversy.
And me? I voted last week. After growing up in Nebraska, where most offices below the level of state executive offices are non-partisan and every office down to dog catcher is elected, I enjoy pushing the straight-party ticket button. Makes me feel like a rebel.
James R. Rogers is department head and associate professor of political science at Texas A&M University. He leads the “New Man” prison ministry at the Hamilton Unit in Bryan, Texas, and serves on the Board of Directors for the Texas District of the Lutheran Church-Missouri Synod. His previous “On the Square” articles can be found here.
Become a fan of First Things on Facebook, subscribe to First Things via RSS, and follow First Things on Twitter.