Given that I’m from Michigan, you might think that the headline is a slam against our neighbors to the south. It’s true: Michiganders are much smarter than Ohioans — after all, we somehow managed to trade Toledo for the entire Upper Peninsula. Also, we’re much better-looking and we’re better at sports, possibly excluding football.
But really, the headline is referring to this analysis by The New Republic’s William Galston, in which he argues that President Obama is unlikely to win the election without winning Ohio.
In a literal sense, this is of course true. Ohio is close to the national average in terms of its propensity to vote for Democrats or Republicans. It’s also well-balanced demographically. If Mr. Obama appeals to the sorts of voters that he needs to appeal to in order to have a good election night over all, Ohio will probably come along for the ride. If he isn’t, it probably won’t. As an actuarial matter, the odds that a candidate wins the election without winning Ohio are probably something like 10 to 1 against.
The same is true, however, for almost all states that are close to the national median, including Colorado, Florida, Nevada, Pennsylvania and Virginia.
In terms of endgame tactics, the precise ordering of these states matters. If the popular vote is extremely close, for example, the fact that (for instance) Pennsylvania is even a point or two more Democratic than Ohio could have fairly profound effects on the optimal allocation of resources like campaign visits and advertising. It is, however, much too early to have that discussion. We know, more or less, which states are likely to be close and which ones aren’t. But we don’t know exactly how the states will line up, and we need to have a fairly precise understanding of that before the conversation about electoral tactics is all that salient.
Mr. Galston’s article, however, is not so concerned with these tactical questions. Instead, it is concerned with strategy: what sort of policies should Mr. Obama pursue to win another term? That is the right question to be asking at this point, and his piece has a lot of valuable insights.
Mr. Galston suggests that Mr. Obama should pursue a strategy to woo the white working-class — the sorts of voters that he says are in abundance in Ohio — instead of the somewhat more upscale voters that he says exist in Colorado. He regards this as tantamount to Mr. Obama’s appealing to the center of the electorate rather than to his base.
Where I think Mr. Galston’s thesis goes astray, however, is in suggesting that there is any one type of voter — or any one type of state — that will be key to Mr. Obama’s re-election prospects. As I noted last week, swing voters come in all different shapes and sizes (not all of them are white, by the way) and in all different states.
In last week’s article, we developed a statistical model to estimate the likelihood that a voter cast a ballot for Mr. Obama or John McCain based on all different sorts of demographic characteristics. The article concluded that, with some exceptions, like African-Americans, most voters aren’t easily classified on the political spectrum solely on the basis of their demographics.
That analysis deliberately ignored what we might think of as political variables — like whether the voter identified as a Democrat or a Republican, or whether they consider themselves conservative or liberal — things that are a matter of choice rather than birth. In the real world, of course, questions like how many liberals are there in Ohio, or how many Republicans are there in Florida, are relevant to the strategies that Mr. Obama and his Republican opponent might choose to pursue.
I’m going to re-run the numbers, then, with these characteristics included. I’ll also include a few demographic factors that we skipped initially, like different types of religious affiliation (e.g. Mormon or Jewish voters), and whether the voter is a gun-owner. And I’ll confine the analysis to those who are registered to vote.
This version of the analysis is more in line with what a campaign would conduct in order to do its voter targeting. Once we’ve accounted for these political variables, we’re not so much in the dark about whom a voter is likely to cast a ballot for. Sometimes, demographic variables can outweigh things like partisanship and ideology — but usually they don’t.
In this version of this analysis, 35 percent of registered voters can be considered to belong to the Republican base, which we’ll define as being at least 75 percent likely to have voted for Mr. McCain. Another 37 percent belongs to the Democratic base and were at least 75 percent likely to vote for Mr. Obama. That leaves only 28 percent of the country that we might describe as swing voters.
What’s lost in this version of the analysis is the question of why a voter has the preferences she does — why, for instance, she identifies as a conservative Republican despite having demographic characteristics that would point toward her being somewhere in the middle of the spectrum. Those are important questions and ones that both political scientists and pundits don’t spend enough time thinking about. We do know, however, that given that the voter identifies as both conservative and Republican, she’ll be very likely to vote against Mr. Obama.
These patterns, however, can differ somewhat from state to state. The data set that we’re using (from the Annenberg National Election Survey) is so large that we can examine these things directly.
Here, for instance, are the preferences for voters from California. There certainly are some Republican voters in California — about 31 percent qualify as Republican base voters. But 45 percent belong to the Democratic base. A Republican can win elections in California, but only if she wins essentially all the swing voters, or the Democratic base does not turn out.
Alternatively, here are voters from Utah. The graph looks a little bit more erratic because the sample size is smaller, but the overall pattern is clear: about 60 percent of voters belong to the Republican base, so generally a Democrat can’t win even under the best of circumstances.
Other states, however, may have a fair number of swing voters even though they aren’t usually thought of as swing states. Here, for instance, is the data for Massachusetts.
Only 15 percent of voters belong to the Republican base in Massachusetts. But the remaining voters aren’t necessarily Democrats. While 44 percent belong to the Democratic base, another 41 percent are swing voters. What happened to Martha Coakley in the Senate race last year? She got almost all of the base voters in her race against Scott Brown, but then almost nobody else, so she finished with 47 percent of the vote over all.
The closest thing to the opposite of Massachusetts is probably a prairie state like Nebraska. Republicans start out with a big advantage there, but there are enough swing voters — 32 percent, above the national average — that moderate Democrats like Ben Nelson can win on occasion.
Even among what we would think of as swing states, there are really two different categories. Here, for example, is the chart for North Carolina. North Carolina is a swing state, but it has relatively few swing voters (only about 25 percent of the population). Instead, it is competitive because the extremely Republican voters (like evangelical Christians) roughly balance out against the number of extremely Democratic ones (like African-Americans and highly educated white liberals in the Research Triangle).
Other swing states really do have a lot of swing voters. In Iowa, for example, 36 percent of voters fit this description:
So do 41 percent of voters in New Hampshire:
Because it has so many swing voters, New Hampshire tends to overshoot the national trend; if the country moves 2 percentage points to the right, New Hampshire might move 4, and vice versa.
Here are the figures for all states, excluding Hawaii and Alaska where Annenberg did not conduct any surveying. Keep in mind that the sample sizes are small for the smaller states, so there is a fairly high margin of error. Nevertheless, according to the Annenberg data, the states with the most swing voters are …. Rhode Island and North Dakota. They aren’t normally competitive, because they have a large number of Democratic and Republican base voters respectively and that is a lot of gravity to overcome. But this helps to explain why they do unorthodox things on occasion, like Rhode Island electing Lincoln Chafee (both as a Republican and an independent) or North Dakota electing Democratic senators like Byron Dorgan.
But what about Ohio? Actually, as you might expect, it’s very close to the national averages, with 36 percent Democratic base voters, 35 percent Republican base voters, and 29 percent swing voters.
On one level, this lends support to Mr. Galston’s hypothesis: Ohio really does look like the United States in microcosm. But there’s a subtle distinction. In being a good model for the United States as a whole, Ohio — like the rest of the United States — mostly consists of base voters. It doesn’t necessarily follow, then, that President Obama is more likely to win Ohio by pursuing swing voters instead of base voters. If he takes a course that appeals to his base, he’ll enliven African-Americans in Cleveland and college kids in Columbus, while enraging evangelicals in the Ohio River Valley and well-to-do businessmen in the Cincinnati suburbs.
Maybe that trade-off is worth it and maybe it isn’t; he might be better off ignoring his base and going for swing voters in Youngstown and Shaker Heights and Dayton instead.
But Ohio so closely resembles the rest of the country that there is almost no deliberate strategy a president could pursue in order to single it out. What working-class voters in Youngstown think about Mr. Obama will greatly resemble what working-class voters in Reno think. The college kids in Columbus aren’t all that different from the ones in Charlottesville, Va.
A president could conceivably pursue a strategy to win North Carolina or New Hampshire or New Mexico, since those states are more idiosyncratic. But there’s not much he can do to exploit any advantage in Ohio. This is why I say there’s nothing special about it — although one could argue that Ohio is special precisely because it is so uncannily average.
Nor does it follow that working-class voters are going to be especially important in 2012 just because there are a lot of them in Ohio. There are roughly as many swing voters, for instance, in upper-crust Colorado, the state that Mr. Galston contrasts Ohio against:
Yes, the swing voters in Colorado are different than the ones in Ohio. But, as we mentioned last week, demographics alone don’t do all that much to predict how someone will vote. In contrast to 2004, when the conversation was all about what “security moms” in the suburbs would do, the focus now seems to be on the white working class. But both groups contain their share of winnable votes. And there are about as many electoral votes in swing states with above-average incomes (for instance, Colorado, Virginia, New Hampshire and Minnesota) as there are in states with below-average ones like Ohio or New Mexico.
In short, analyses like these risk confusing cause and effect. It’s not so much that as Ohio goes, so goes the nation. It’s more that as the nation goes, so goes Ohio.