Not in Kansas anymore

by Henry on December 12, 2005

I’d somehow missed this fascinating paper by Larry Bartels which Ezra Klein links to today. It uses NES data to argue that the thesis of Thomas Franks’ What’s the Matter with Kansas is completely wrong. Poor white voters have become more likely to vote Democrat over the last few decades. While they’re less likely to identify with the Democratic party than they used to be, the decline in Democratic party ID has been less marked among poor white Democrats than among richer ones, and is entirely attributable to losses in the South in the post-Civil Rights era. Nor, if you look at the preponderance of evidence, is there good reason to believe that poor white voters are more interested in cultural than in economic issues; if anything the opposite seems to be true.

Of course, Bartels’ argument isn’t only discomfiting to Franks; it also undermines the self-justifying claims of right wing pundits who consider themselves, against all the odds, to be populists. The one part of Bartels’ paper that I disagree with is its conclusion, which implies that mistaken Democratic angst over the party’s appeal to poor white voters is what motivates arguments over whether the Democrats need to fundamentally rethink their political message. If I understand Bartels rightly, he’s suggesting that the Democrats don’t need to change what they’re doing. I don’t think that’s true, and indeed it seems to me that some of Bartels’ earlier empirical findings point in the opposite direction. If, as Bartels has previously argued, the general public has a difficult time in connecting public policy with economic inequality, Democrats are likely to succeed to the extent that they can draw these connections in their rhetoric, and show how inequality affects not only the working class but the middle class too. That said, this paper seems to me to be a lovely example of how political scientists and other social scientists should be speaking to broader public debates, by using their expertise to examine whether the fundamental assumptions underlying these debates are fundamentally right or wrong. And it describes Peter Beinart’s arguments as “fatuous.” What more could you ask for?

{ 2 trackbacks }

Crooked Timber » » World Values Survey
12.14.05 at 2:45 pm
Crooked Timber » » Class dismissed
12.22.05 at 9:59 pm



Louis Proyect 12.12.05 at 8:54 pm

My take on Frank and Bartels:


coturnix 12.13.05 at 3:20 am

It’s nice to have some data to go together with my instinctive refusal to believe Frank’s hypothesis. I’ll take a closer look at the numbers tomorrow…


Doug 12.13.05 at 3:30 am

What more could you ask for?

Belle knows.


abb1 12.13.05 at 4:23 am

I strongly disagree that the Democrats don’t need to change what they’re doing.

However, I was a bit surprised to see this poll a coupla days ago:

CBS News/New York Times Poll. Dec. 2-6, 2005. N=1,155 adults nationwide. MoE ± 3.

“Regardless of how you usually vote, do you think the Republican Party or the Democratic Party comes closer to sharing your moral values?”

Republican Party 41%
Democratic Party 43%
Both (vol.) 3%
Neither (vol.) 5%
Unsure 8%


Andrew Gelman 12.13.05 at 9:27 am

Income is highly correlated with Republican voting in poor states, not so much in rich states. See here for details. We can understand the state average income effect as one of context. The Mississippi electorate is more Republican than that of Connecticut; so much so that the richest segment of Connecticutians is only barely more likely to vote Republican than the poorest Mississippians. In poor states, rich people are very different from poor people in their political preferences. But in rich states, they are not.


john emerson 12.13.05 at 9:32 am

Many of the conservative Christians I know are financially above the mean, and some of them are quite prosperous. Many are succesful in technical areas, but avoided the “liberal arts BS” during their educational careers, partly because it was not practical and partly because they wanted to continue to think the way they always had.

It’s hard to exaggerate the degree to which the “new right” is a rejection of the humanities, as taught in American universities 1950-2005. In the US (in contrast to Europe) Undergrad education as not really a good primary venue for political action and education, since many students don’t really respect their professors very much. Envious non-college people end up identifying liberalism with “the elite”, and plenty of college people ignore or resist it. You just end up with a splinter faction of militant students and hangers-on within a generally reactionary world. Historically, student movements have mostly failed.


Grand Moff Texan 12.13.05 at 10:25 am

Franks’ model works better in the South, not the midwest.


GKurtz 12.13.05 at 10:53 am

I lent my copy of WTMWK to someone, so I can’t verify this by quoting Frank, but my recollection is that he never makes the argument Henry & others are attributing to him. It’s been a few months since I read the book, but as far as I remember Frank never makes a social-scientific claim about voting patterns. His thesis in this book is not the sort of thing that’s easily subjected to soc-sci tests: he argues that right-wing rage (a phenomenon that certainly exists) is best understood as a sort of displaced class politics, in which economic class disappears and is substituted by a cultural divide that doesn’t neatly correspond to economic class lines. Frank makes some loose statements about voting patterns in order to set the context for his argument about ideology, but the point is that his argument is about ideology and not about voting patterns. (Or at least that’s how I remember it.)

The study Henry cites sounds interesting and significant — I’m not arguing with that claim. But it would be a shame if Frank’s argument about class and ideology got lost in the shuffle. This is the stuff he’s been writing about for years, in The Baffler and in his other books, and he’s pretty sharp about it.


abb1 12.13.05 at 12:23 pm

I agree with Louis that class is not defined by income, it’s a rather obvious point. A $100K/year software developer seeing her job transferred to $20/year job in India is as much a victim here as her house cleaner. I also liked Louis’ Democrats/Republicans ‘good cop/bad cop’ metaphor.

But I don’t think the ‘urban/rural’ aspect is emphasized enough in all this: I suspect that small town ‘petit bourgeois’ do benefit from Republican policies – low taxes, no social programs on the federal level; that’s just how it is.


Erik 12.13.05 at 12:47 pm


Erik 12.13.05 at 12:49 pm

Sorry, messedm up the link. The paper’s title is “What’s the matter with Connecticut” You can find it here:


Sam Rosenfeld 12.13.05 at 6:01 pm

See the response here. I suspect the truth is somewhere in between, and obviously what’s relevant in these discussions is what we mean by “working class.”


Complete Stinker 12.13.05 at 11:22 pm

We can negotiate what we mean by “working class,” haggle over details, find exceptions and produce counterexamples but the bottom line clear: this is class warfare and the lower classes are, and always have been, the bad guys.

Poverty, hardship, insecurity and ignorance make people conventional, uncritical, brutal, intolerant, selfish, narrow-minded, short-sighted and bad. Education, privilege, security and wealth make people good. Virtue is a luxury item.

The solution is simple even if the means to achieve it isn’t: win the class war. Suppress and disenfranchise the lower classes. And then obliterate the working class by fixing the system so that everyone is reasonably educated, secure and comfortably off–so that everyone is upper middle class.


Larry Bartels 12.15.05 at 6:06 pm

I was surprised by Gopoian and Whitehead’s demographic profile of whites in the bottom third of the income distribution, so I checked the NES data. All I can say is that their tabulations don’t look like mine. They claim that only 35% of low-income whites (in 2004, I assume) were actually working, while 43% were retired or disabled. I have 49% working (with another 6% temporarily laid off or unemployed) and 35% retired or disabled. (Weighting the data as I did in my paper reduces both those percentages slightly, while increasing the percentage of homemakers and students.) Whatever group they are looking at, it is not the group of low-income whites characterized as “working class” in my paper.

More generally, if social scientists have a “prevailing definition” of the term “working class” I missed the memo. Apparently the people we’re talking about did, too. Among whites in the bottom third of the income distribution in 2004, 55% called themselves “working class”; among Gopoian and Whitehead’s whites without college degrees, only 48% did. Further restricting the definition to people with “incomes that surround the median household income for the nation” makes me even more curious why we don’t just use a well-established term that would seem to fit these people comfortably: “middle class.”

Of course, analysts can use whatever labels they want as long as they are clear about their definitions. My focus on low-income whites was inspired primarily by Frank’s reference on the first page of his book to Democrats as “the party of workers, of the poor, of the weak and the victimized,” and by his subsequent insistence on considering class “in the material, economic sense, not in the tastes-and-values way our punditry defines class.” It also fits nicely with my own broader interest in the politics of economic inequality. It is very easy to think of significant government policies that distribute costs and benefits on the basis of income, but much harder to think of instances in which “who gets what” depends on whether they happen to have a college degree.

Finally, a brief comment on the broader debate in which Gopoian and Whitehead’s analysis is situated. It is certainly true that the Democratic Party has lost support among whites without college degrees. (As with Democratic fortunes more generally, most of that decline is directly attributable to the demise of the artificially Democratic Solid South of the Jim Crow era. But let’s ignore that elephant in the room – along with the growing proportion of the electorate that happens not to be white.) What should we conclude from that trend? Many observers seem to leap to the conclusion that the party needs to reconnect with “traditional values.” Whites without college degrees are, indeed, more conservative than better-educated whites are on social issues like abortion and gender roles. But they also attach much less weight to those issues in their voting behavior. In 2004, the statistical connection between social issue preferences and presidential votes was more than twice as strong among college-educated whites as among those without college degrees. (In contrast, the connection between economic issue preferences and presidential votes was equally strong among both groups.) If anyone has a magic formula for appealing to less-educated socially conservative whites while retaining the loyalty of better-educated – and apparently more attentive – socially liberal whites, I’m all for it. But in the real world of hard political trade-offs, it is by no means obvious that moving to the right on social issues would be a net vote winner for Democrats.


Andrew Leigh 12.15.05 at 8:15 pm

A new paper by Ed Glaeser and Bryce Ward also seems relevant to this debate. Discussion and trackback.

Comments on this entry are closed.