About this Blog
About The Author
Email Me

RealClearPolitics HorseRaceBlog

By Jay Cost

« Obama's Success in Central Pennsylvania | HorseRaceBlog Home Page | At Princeton This Week »

Questions without Answers

There is a continuing conversation about whether Barack Obama can win working class whites in November. Some, such as John Judis, have argued that perhaps he cannot. Like McGovern, he will cede those voters to the Republicans. Others have argued that disaffection with Bush and the GOP will overcome any problems Obama might have with these voters.

So, who is right? I do not know. In fact, I don't think the question can be answered.

Before we get into this, we should probably be more specific. After all, it is wrong to assert that Obama cannot win the white working class. He did exactly that in Wisconsin. His problem has been the white working class in certain geographical reasons. The following graph, provided by Sean Oxendine, tells the tale.


Obama's weakest performances among whites have been in Appalachia, which is traced in solid black. Oxendine has put counties that Obama won in green, counties that Clinton won in blue. Note the expanse of deep, dark blue that moves from Mississippi to New York. This is where Obama has had his greatest problems. This is why Clinton will not drop out next week, even if she loses Indiana. West Virginia comes the week after, and Kentucky the week after that. She's bound to win both, and candidates do not drop out immediately prior to impending victories.

In all likelihood, weak general election performances in Kentucky, Tennessee, and West Virginia will not cost Obama the presidency - though he would be the first Democrat ever to win the White House having lost all three. The trouble comes with southern Ohio and western Pennsylvania.

To appreciate Obama's potential dilemma, consider the following chart. It examines Clinton's primary victories in the counties of metropolitan Pittsburgh compared to the general election performance of Republican candidates dating back to 1972.

Clinton and the GOP in Pittsburgh.gif

The first thing you'll notice is that the term "Reagan Democrat" is a bit of a misnomer here. In fact, Reagan barely improved over Ford, and his landslide reelection was nothing compared to Nixon's. Aside from Nixon, George W. Bush has performed best in metropolitan Pittsburgh.

Next, notice the wide variability in vote returns. Outside Allegheny County, where the African American vote provides a solid base for the Democrats, we see twenty-point swings in these counties.

Finally, turn to the primary results from last week. Obama did very poorly across the entire metropolitan area. Allegheny County was the sole exception. Interestingly, Fayette County is the poorest county in all of Pennsylvania, and no Pennsylvania county supported Clinton more strongly.

This is what John Judis is worried about. Even if you allocate most of Clinton's voters to Obama in November, he still might face trouble in metropolitan Pittsburgh. It might be that metro Pittsburgh Democrats, by swinging so heavily to Clinton, have indicated that the region as a whole is unimpressed by Obama and will therefore back McCain. Even if strong Democrats support Obama - weak Democrats, Independents, and persuadable Republicans might not.

What happens if counties like Fayette swing against Obama? It depends on how big the swing is. Would it be on the order of 1984, 2004, or (worst case scenario) 1972? Depending on the size of the swing, Pennsylvania could become trouble for him. After all, Obama had problems not just in metro Pittsburgh, but also in the northeast. He lost Lackawanna County by 48 points, and Luzerne County by 50 points. Kerry won both counties in 2004. Obama should pull a big victory out of Philadelphia, but trouble in the northeast and southwest would put more pressure on metro Philly to perform.

Ohio could also be trouble. Bill Clinton won the state in 1996 largely because he won the same counties that gave Obama just 30% of the primary vote. These are the counties that hug the Ohio River Valley in the second, sixth, and eighteenth congressional districts. Again, Obama might be able to improve on Kerry in the suburbs of Cincinnati, Cleveland, and Columbus - but trouble in the southeast would put more pressure on these areas.

Obama might also have trouble in Missouri. Oxendine's map does not indicate it, but he was very weak in southern Missouri, especially in the seventh and eighth districts. Many of these counties split their votes in 1996, which is one reason why Bill Clinton won the state twelve years ago.

In all, these results have been so lopsided that people have begun to wonder about Obama. Can he hold the line in these places? If he can't, will that cost him these states, and therefore the presidency itself? Some people see this happening. Some do not. This is what has brought about this discussion. His poor showings in these places is a bit of a puzzle, and has induced a debate.

What I find interesting about the conversation over places like Fayette County is that people from there are not really participating in it, at least on their own terms. Their role in the discussion is mediated by the public opinion survey. They merely answer the questions the pollsters ask them.

So, if you think about it, the voters' ability to influence the conversation is really quite constrained. In a statewide survey of Pennsylvania, you might get three or four residents from Fayette who answer "yes" or "no" to a handful of questions. That's it. No amplification permitted.

This makes me wonder about these competing theories. Just how much do they depend on the theorists, and how much on the voters?

What is more, the candidate in question happens to be black. This further limits the utility of the polls. There has been a heated debate in political science for a quarter century over whether white America has become less racist, or whether its racism has become less transparent. Researchers have found that whites now tend to answer straightforward questions about race in a much less racist way. However, some researchers claim to detect racism in responses to seemingly unrelated questions. Others vehemently disagree. Putting aside the merits of the arguments pro and con, the point is that polls have failed to arbitrate the dispute. Researchers look at the same answers to the same questions, yet see two different opinions.

Those outside the academy have been struggling with similar difficulties as regards the "Bradley Effect." Like so-called "latent racism," the issue hinges on whether public opinion surveys give us a clear read on the thinking of the American voter.

Unfortunately, they probably do not. They are limited, imperfect tools for understanding him - especially on the issue of race. Above all, they tightly restrict his role - and therefore limit his ability to influence our conclusions about him. It is very possible for a researcher or pundit to approach the polls with erroneous preconceived notions, and have the data's ambiguity "validate" the false theory.

The problem is that we are exclusively dependent upon the polls for our understanding of public opinion. Nobody is really going "out there" to interact with average people, to find out what they are thinking. We have the polls, and that is all we have. So, we only know as much as they tell us. When they are silent, ambiguous, or misleading, we are ignorant.

They did not always monopolize our knowledge base. While studying for my qualifying exam in American politics several years ago, one of my favorite reads was a dusty old book called Political Ideology by Robert Lane. Unfortunately, it is no longer part of the canon. I only read it by mistake. My department gave me an outdated reading list to study from, and I read it before the error was corrected. I'm glad I did. It is a fascinating read, not so much for its conclusions, but for its methodology. Lane engaged in detailed, dialectical conversations with a group of fifteen men from "Eastport, USA." His study was certainly not wide, but it was very deep.

For better or worse, public opinion researchers have not embraced Lane's method. Today, research is dependent upon survey data. Academics develop abstract theories of how voters think, and test them via these surveys. While some academics interact with their subjects, this is by no means required for publication. The same goes for analysts outside the academy. Extended, dialectical conversations are not required for opining on the average voter. Instead, the latest readings from Gallup are.

There are good reasons to go with the public opinion survey over Lane's methodology. If I were only going to select one method, I'd select the former without thinking twice. Lane's methodology is simply too subjective. However, the objectivity of the survey comes at a great cost - namely, distance from our subjects. This distance gives the answers to the survey questions ambiguity, and thus an opportunity for subjectivity to weasel its way back in. Ideally, both methods should be employed. The public opinion survey, with its objective and quantifiable answers to specific questions, should be supplemented with extensive conversations with our subjects.

Practically speaking, this kind of methodological pluralism rarely happens - though there are exceptions. Instead, most researchers rely exclusively on the polls. This is a very unfortunate development. It is particularly debilitating when it comes to understanding Obama and Appalachian whites. As should be clear from the discussion of "latent racism" and the "Bradley Effect," polls are particularly unhelpful here. Race is a hard topic to explore from thirty thousand feet, which is essentially what the polls do.

What we need, then, is somebody like Robert Lane. We need an expert who is up-to-date on the latest scholarly research, and who has spent time soaking-and-poking in places like Fayette County to see whether people there are willing to vote for Obama. As far as I know, there is no such expert.

And so, we have no answers, just questions.

-Jay Cost