A duo of columns in the New York Times a few days ago purported to puncture the soundness of Barack Obama’s sociology in his much-discussed "bitter" comment and his later clarifications.
First it was Princeton professor Larry Bartells, who wants to tell you that he as-much-as debunked Obama’s theory years ago in reaction to the book "What’s the Matter With Kansas" by one of Obama’s influences in this thinking, Thomas Frank. Bartels claims:
"Small-town people of modest means and limited education are not fixated on cultural issues. Rather, it is affluent, college-educated people living in cities and suburbs who are most exercised by guns and religion. In contemporary American politics, social issues are the opiate of the elites."
This didn't seem right to me, but then it was (supposed) economist Paul Krugman with a 100% redundant echo of the Bartells column the day before.
It’s true that Americans who attend church regularly are more likely to vote Republican. But contrary to the stereotype, this relationship is weak at low incomes but strong among high-income voters. That is, to the extent that religion helps the G.O.P., it’s not by convincing the working class to vote against its own interests, but by producing supermajorities among the evangelical affluent.
When I was shown the Bartells column by a conservative friend, I studied it hard for a while and was almost convinced that Obama and Frank have been mistaken. But then I looked more closely and eventually thought "Ooh, a great opportunity to get my sociology nerd on."
Here goes:
Upon close study, Mr. Bartels' demographic analysis is indeed shaky. The basis for his argument is the difference between the correlation of small-town, blue-collar voters' stance on each of several "cultural" issues like abortion with likelihood of voting republican (5%) vs. that of affluent, urban voters (58%). He points out that the two populations are of similar size, but he refuses to include either the actual Republican/Democrat split in either of the subpopulations or the proportion of abortion/choice belief, for example, in cities or towns.
Unlike Krugman, who purposely ignored the urban/small town split which Bartells used in his comparison, my intuition told me that that factor is far more important than income to both Bartell’s (faulty) logic, and Obama/Frank’s insight. In a typical city, 40% vote republican and 51% support some restrictions on abortion, according to Gallup. In a small town it's 60% republican and 65% favoring the same abortion restrictions. Clearly abortion is not overriding enough an issue in either setting to keep all anti-choice voters from voting for democrats, so in both settings it is the republican voters that are more homogenous than the democrats, who are somewhat split on the issue. Importantly, urban republicans are a smaller relative proportion of the anti-choice population in their cities, so it would seem that the urban republicans are more consistently anti-choice than rural republicans, kind of confirming Bartells' (misleading) observation about small town conservatism, but it hints at a very different reason why. The gap between Bartel's thinking and Obama/Frank's is that Bartel's argument only considers one vote-driving issue at a time.
My counter hypothesis to the Bartels theory is that cosmopolitan social conservatives, being in the minority in their communities, are just far more homogeneous in all of their views than small-town, low-income conservatives. In a small town, blue-collar conservatives meet others like them everywhere they go and are more likely to disagree with their fellow conservatives on some issues. In cities, educated conservatives are isolated and are far more orthodox in their conservatism. Admittedly I do not have the data needed to confirm my theory. To properly test this, we would need to know how many small-town working class pro-choice republicans are also pro-gun rights and how many of the republican voters among their rich "demographic opposites" in the cities are (just survey Fox News HQ?). If the GOP-COSMO-LIFE group is more consistently pro-gun than the GOP-RURAL-LIFE group, then I may be on to something.
If so, it is due to the fact that small-town republican voters aren't as strongly inclined to "cling" to their beliefs on abortion because if they don’t happen to be pro-life they can still find comfort in and vote republican on the basis of immigration or guns or illegal immigration or creationism in school. The Frank/Obama theory is that all these different 5% correlations add up to a substantial group of people voting against their economic interests for many different reasons. This theory depends on the important principal in statistics of covariance, which Bartels has conveniently left out of his discussion.
Bartels also talks extensively in his earlier writing (and Krugman parrots him) about how the number of democrats in small towns hasn’t actually decreased in the last several decades, but that’s not what either Obama or Frank claimed. What they are saying is that there should have been ever increasing democratic—hell, forget party, just say "populist"—support in small towns than there has been. If voters had cast their support based on where their economic interests lie, we would have had more than just two Democratic Presidents in the last 40 years.
Update:
Razor-sharp reader carolita points out in the comments that Bartels' income divider for small town/working class people is pretty ludicrous. Bartels splits America at the income level of $60,000, but according to the U.S. Census:
The wage difference between rural and urban or metro economics is significant, with rural wage earners bringing in a median household income of $32,837 in 2000, compared to the metro median income of $44,984 the same year. According to the USDA, more than 33% of households in agriculturally based communities have annual incomes below $15,000.
As I responded below, this data shoots a gaping hole in Bartels' case. Bartels' small town/working class group making under $60k swallows up almost the entire typical town. So, Bartels 5% vs. 58% comparison between town/poor and city/rich is even more subject to the effect of isolation-bred orthodoxy because those high-rise dwelling rich folk are not even close to the majority in the cities that they inhabit.
*note: this is my virgin diary and is cross-posted at my personal blog.
Your criticism is quite welcome so don't hold back, hear?