Six important things political science tells us about presidential elections

At long last, we’ve arrived at the finish line of this bruising, maddening, exhausting presidential race. As we anxiously await the election returns in the hours before the polls close, I thought it would be interesting to look at what the political science research says about presidential elections. Is this year’s race confirming or contradicting those findings? In what ways, and to what extent?

Before I dive into the research, a caveat: when researchers study presidential elections, a major dilemma they struggle with is that their unit of analysis is not terribly large. Statisticians refer to it as “the small N problem.” The issue, in a nutshell, is that since the dawn of the republic, we have had only 56 presidential elections (the 57th will occur tonight). With such a small number of observations, there’s a much greater chance that a relationship that appears to be causal might actually be the result of random variation. Moreover, those 56 elections have taken place over an exceedingly long time horizon, and many of the confounding variables that influence elections may well have had different effects at different points in time. The wonderful site XKCD has brilliantly illustrated the presidential small N problem here.

In short, when it comes to presidential elections, it can be difficult to sort out cause and effect.

Nevertheless, the research on presidential elections has come up with some well-documented findings that seem to be holding up pretty well this election year. Here are a few of them:

1. As Marty Cohen, David Karol, et al. demonstrate in their book The Party Decides, presidential candidates are, for the most part, chosen by party leaders and other unelected insiders. “Invisible primaries” occur well before a single primary vote is castle, and they produce clear frontrunners. Thus, the party tends to exert much more control over selecting the nominee than ordinary voters do. This finding certainly holds for 2012. Barack Obama ran unopposed and Mitt Romney, who was the choice of the G.O.P. party establishment from the get-go, won the Republican nomination by a decisive margin.

2. During the course of a presidential election, much ink is spilled and much bandwidth is eaten up on articles and blog posts about “independents” and “undecided voters.” But going back at least as far the seminal book The American Voter, published in 1960 and written by Angus Campbell et al., studies have shown that the vast majority of voters cast their votes on the basis of party identification. While it may be true that rates of partisan political identification have declined somewhat in recent decades, Campbell’s basic findings still seem to be holding up in 2012. An ABC News/Washington Post poll posted yesterday showed that 95% of Republicans supported Mitt Romney and 91% of Democrats were for Barack Obama. About two-thirds of those polled identified with one of the major parties, and only one-third identified as independent.

Moreover, as Keith et al.’s ground-breaking book The Myth of the Independent Voter showed, most self-identified independents actually do have strong partisan preferences and vote accordingly. In addition, the authors showed that truly independent voters — those without strong partisan leanings — were the least knowledgeable about the candidates they were voting for. More recent studies, such as this Pew poll from August, confirm these findings.

As for the “undecided” voter, one insight from the poli sci research, such as Paul F. Lazarsfeld’s classic book The People’s Choice, is that, due to partisan political identification, very few people change their voting preferences during a presidential election. This seems to be especially true of 2012, an election in which there seems to have been exceedingly few voters who are genuinely undecided.

3. Okay, if most people vote according to partisan political preferences and are therefore unlikely to change their vote, are there any events taking place during the campaign that tend to influence the undecideds? For example, might a successful convention or a strong political debate give tend to give a candidate a significant, and lasting, boost at the polls? The answer appears to be yes to the former, no to the latter. In their book, The Timeline of Presidential Elections, Robert S. Erikson and Christopher Wlezien found that while party conventions can have a real and lasting effect on the presidential campaign, debates rarely do. This year, since Barack Obama got a bigger convention bounce than Mitt Romney, that may bode well for him. By the same token, it would seem that Obama’s poor performance in the first debate probably did not hurt him much. We shall see. If Romney does end up winning after all, it would certainly call Erikson’s and Wlezien’s findings into question.

4. What about political advertising — does that make a difference? This year we’ve seen that, in spite of all that Citizens United money going to Romney, he is, as of today, running behind Obama in national opinion polls and in most of the swing state polls. The research shows that political advertising can significantly impact an election, but only under certain circumstances: when the candidates are unknowns, and when one candidate greatly outspends the other. Moreover, the impact of a political ad tends to be short-lived. So, even though Romney has much more Citizens United-related money than Obama, it looks like its impact has been minimal, because Obama is nothing if not rather well-known.

5. What, then, is the single most important deciding factor in an election? This one is easy — it’s the economy, stupid! A number of studies show that even a modest rate of economic growth makes it very likely that an incumbent president will be re-elected. We are experiencing modest growth right now (the emphasis is on “modest”!), which is why most of the poli sci forecasting models show Barack Obama defeating Mitt Romney, albeit by a narrow margin. If Romney does indeed prevail, a major rethinking of these forecasting models will be in order.

6) The research points to one more factor that can make a difference, and though it’s hard to measure it precisely, it is well worth a mention. That factor is the get-out-the-vote effort. As Donald Green and Alan Gerber’s important book Get Out the Vote documents, GOTV efforts significantly increase turn-out. Person-to-person contact is most effective, but phone calls, direct mail, and the like can also make a real difference.

So if at some point, during this long, hard slog of an election season, you’ve canvassed door-to-door, attended a phone banking party, volunteered to drive people to the polls, or even just did some old-fashioned envelope stuffing, give yourself a pat on the back. You’ve shown good citizenship, and chances are, the work you did had a real-world impact. Now you can feel free to kick back, get together with some friends, and experience the drama and nail-biting excitement of watching the final round of a tough, fierce, closely matched political race!

Kathleen Geier

Kathleen Geier is a writer and public policy researcher who lives in Chicago. She blogs at Inequality Matters. Find her on Twitter: @Kathy_Gee