To look at how white voters cast their ballots this year, and to determine whether any underlying patterns exist, I gathered the exit poll data compiled by CNN. The resulting spreadsheet is available here. Nationwide, Obama won 43% of the white vote. When we look at the vote preference based on race, we see that Obama won a majority of the white vote in 18 states, plus the
Conversely, if we look at the McCain vote among whites, we see a strong geographical dimension to his support. Nationwide, McCain received 55% of the white vote. The map left highlights those states where his white support was the highest--over 10% above his national average. Thus, in a cluster of mostly southern states the Republican nominee got 2/3 or more of the white vote.
Another way we might approach these numbers is to go to those states with the largest black populations relative to the state as a whole. Here we get to the phenomenon described by Schaller—how does Obama do among whites in the “blackest” states?? At left is a chart showing his performance among whites in the 15 states with the most significant black electorates. Overall, he won 8 of these states. However, with the exception of
What happens when we look at the inverse of these states—i.e. the “least black” ones?? Does McCain do as well as he did in the
From this, can we conclude that white southerners voted against Obama because he was black?? Not necessarily. An alternative hypothesis is that these voters voted against Obama not because he’s black, but because he’s a Democrat. To get a better sense of what happened in November we need to go back further in time. Let’s look at 2004 when you had two white candidates on the ballot. If the results among white voters are quite similar to what we saw in 2008, our conclusions might need a bit more nuance. By looking at 2004 we can also get a sense of what changes might be taking place across the electorate and whether there is a geographic pattern. Thus, I also gathered the statewide CNN exit polls from the Bush/Kerry contest. What do we see?? Nationwide there wasn’t much difference between the support Obama and Kerry received among whites. In fact, Obama did better, getting 43% of the white vote to Kerry’s 41%.
Looking at the results state by state, we see that in all but 7 states, Obama did better among white voters than Kerry. In 8 states the performance of the two Democrats was identical. In the remaining 35 states, plus DC, Obama did better among whites than Kerry. If you look at the 18 states where Obama’s gains were the greatest (5% or more), he won 13. Four of these states voted Republican in 2004 (
Whereas Obama won 8 of the blackest states, Kerry only won 5. In these states, Kerry received below his national average among white voters (41%) in 8. Aided partially by increased support among African American voters (and also changes in turnout among whites and Republicans, plus greater support among Hispanics), Obama was able to outperform Kerry electorally. Narrowing further in on the white vote (and getting at the Schaller Effect) we see that in these states, Obama’s performance was worse than Kerry’s in four (
So where does that leave us? With the low level of support Obama received among white voters in some states--relative to John Kerry and in an extremely favorable electoral environment--its hard to discount the assertion that race played some role in the vote. While some of these states have been heavily Republican, we saw Obama outperform Kerry among whites in equally Republican states in other parts of the country. We've also seen many of these states elect Democrats to other offices. The fact that white voters behaved differently in various parts of the country suggests that race is an issue that cuts across the electorate in a multitude of ways. Because race has been intertwined with politics in some places, and completely absent in others, we shouldn't make sweeping assertions about how important this cleavage is. Rather, we should approach the issue with nuance, an appreciation for historical context, and a willingness to explore multiple explanations for the outcomes we've observed. Given that we've had only one election to test our hypotheses, we should hold off on definitive statements until there's more data to parse.