The Working White Working Class Really Is Leaving the Democrats
[See the end of this post for how the debate/discussion has played out… it has been quite civil and productive]
I published an essay in The Guardian two weeks ago offering one reason why most white working class people in the USA vote Republican. Two excerpts:
politics at the national level is more like religion than it is like shopping. It’s more about a moral vision that unifies a nation and calls it to greatness than it is about self-interest or specific policies. In most countries, the right tends to see that more clearly than the left….
In focusing so much on the needy, the left often fails to address – and sometimes violates – other moral needs, hopes and concerns. When working-class people vote conservative, as most do in the US, they are not voting against their self-interest; they are voting for their moral interest. They are voting for the party that serves to them a more satisfying moral cuisine.
I had been inspired by a New York Times column by Tom Edsall, on how Obama’s re-election team had largely written off the white working class vote:
In the United States, Ruy Teixeira noted, “the Republican Party has become the party of the white working class,” while in Europe, many working-class voters who had been the core of Social Democratic parties have moved over to far right parties, especially those with anti-immigration platforms.
My essay was strongly condemned by two worthy critics. First, Andrew Gelman said that I had gotten the basic facts wrong. The working class does NOT vote Republican, he said; it votes Democrat, and has for over 70 years. There’s been no change. And it votes for the Left in most (but not all) countries.
I made a careless mistake in not specifying in my excerpt above that I was talking only about the WHITE working class. That’s the subject of the big debate, ever since Thomas Frank’s book “What’s the Matter with Kansas.” Of course African Americans and Latinos vote heavily Democratic, and make up much of the working class. But even when we focus only on whites, Gelman says I (and therefore Edsall and Teixeira) still got it wrong.
Second, George Monbiot published a rebuttal in The Guardian, saying that whatever the merits of The Righteous Mind, I had stumbled “stupidly and disastrously” in applying my ideas to politics. He pointed to an analysis by political scientist Larry Bartels that refuted Thomas Frank’s claims about the very existence of a shift away from the Democrats of white working class voters. Bartels’ analysis, like Gelman’s seems to show that there has been essentially no change in the allegiance of the white working class to the Democratic Party in the last 70 years. Rather, Monbiot argues, the working class (in Britain and the US) has become less likely to vote at all, because it sees little difference between the economic programs of the two major parties. It would therefore be foolish for the Democrats or the Labour party to try to appeal to working-class voters by “triangulating” on moral issues when what is needed is a stronger shift to the LEFT, to draw them back to the ballot box by offering them a better deal on economic issues.
But was I really wrong in claiming that the white working class has moved away from the Democrats? It all depends on how you define “working class.” All of the authors in question grant that there is no gold standard. You can use education, income, job classification, self-identification, or some combination of those criteria. They all intercorrelate, but different criteria sometimes lead to different conclusions.
Bartels and Gelman focus on income. They operationalize “white working class” as whites in the bottom third of the national income distribution, for any given year of analysis. Gelman’s very interesting book Red State, Blue State, Rich State, Poor State explains the paradox that richer STATES in the USA vote more Democratic, yet within almost all states, richer INDIVIDUALS are more likely to vote Republican. Point taken: there is a robust tendency for people to become more conservative and/or Republican as they get richer (when you hold everything else constant). I had not known this.
Bartels focuses on Thomas Frank’s claim that there has been a “backlash” against the Democrats – Frank claims that the white working class has changed its allegiance in recent years. In chart after chart, based on ANES data from the 1950s through 2004, he shows that there’s been no change, no shift to the right, no abandonment of the Democrats by whites in the bottom third of the income distribution. Once again, there’s no reason for the Democrats to change strategy or reinvent the party because there is no problem to be addressed.
But things look different if you define class based on education. In a paper by Ruy Teixeira and Alan Abramowitz, titled “The decline of the white working class and the rise of a mass upper middle class,” the authors note that taking whites in the bottom third of the total national income distribution (not the white distribution, which can’t be computed from ANES data) gives you an odd sample: the majority of them are not working. It’s mostly students, retired people, homemakers, and people who are unemployed or on disability. In 2004, only 39% of white voters in the bottom third were currently employed (compared to 73% of white voters in the middle third of income). So this is a group that is more dependent on government programs. Bartels’ analyses do indeed show that this group is still and has always been reliably Democratic.
Teixeira and Abramowitz argue that education is a better way to identify the working class. (For one thing, it correlates more closely with people’s own descriptions of their class than does current family income). They focus on whites who have not completed a 4 year college degree. (They also look at other ways of slicing the data, including an index of all the major predictors). Using the lack of a college degree as the criterion, they show that the white working class has indeed shifted over to voting for Republican presidents with Nixon (70% in 1972) and never really returned to the Democrats. Clinton drew about as many of them as did his two Republican opponents, but Gore lost them by 17 points, Kerry lost them by 23 points. Obama lost them by 18 points, and the gap seems to be growing. Nate Cohn analyzes several recent Pew and Qunnipiac polls and concludes: “over the last four years, Obama’s already tepid support among white voters without a college degree has collapsed.” (See also this Gallup poll).
I wanted to examine these relationships in more detail, particularly the difference between employed and non-employed whites, so I downloaded the ANES cumulative data file, limited my analysis to whites, and then created 4 groups based on ANES variables vcf0140a (education level = 6 or 7) and vcf0116 (work status = 1):
Group 1 = no college, no job
Group 2 = no college, job
Group 3 = college, no job
Group 4 = college, job
The crux of the debate, therefore, is what has happened to group 2, in comparison to the others. The Democrats have been the party of the working man (and woman) since FDR’s New Deal coalition. So has THEIR allegiance changed in recent decades? The most direct measure for us to look at is ANES question vcf0302: “Generally speaking, do you usually think of yourself as a Republican, a Democrat, an Independent, or what?” Look at the red line below, which plots the declining percentage of people in group 1 who said “Democrat” in response to that question.
When you break up the non-college white population by employment status, as Teixeira and Abramowitz suggest, you see what they are saying: Both of the non-college groups slope downward, but group 2 (the red line) goes down more steeply. White people who have a job but no college degree have been leaving the Democratic Party. Are they becoming Republicans, or just getting disengaged, as Monbiot suggested (about the British working class)?
They are moving toward the Republican Party. The red line slopes upwards. That first sharp uptick appears to confirm the reality of the “Reagan Democrats.” And, interestingly, the purple line slopes upward too. That’s group 4, people with a job and a college degree. In other words, the Republicans are increasingly becoming the party of white people who currently hold jobs. This seems like a dangerous situation for the Democrats. Should they do anything to address it? Or are my critics correct that there no problem here, no trend, no loss of the white working class?
Bartels and Gelman are far more skilled at this sort of analysis than am I. Most likely I have missed something. I welcome their corrections, which I’ll post or link to below. Furthermore, Monbiot, Bartels and Gelman are probably all correct when they say that economic concerns played a stronger role in recent electoral shifts than the sorts of moral/cultural issues that I and Thomas Frank were talking about. I should not have suggested that concerns about national greatness and such things were the MAJOR drivers of change. I should have more modestly said: “look, here are some moral misunderstandings that are probably contributing to the ongoing alienation of many white working class voters from left-wing parties in the US and UK.” I thank my critics for pointing out this serious error in my initial essay, and for doing it in a way that displayed passionate disagreement about ideas without personal animosity or insult.
But the bottom line is that I (and Edsall, Teixeira, Abramowitz, and Frank) seem to have been correct in our basic claim that the white working class is leaving the Democratic Party. Or, at least, it depends how you define class, and it depends on several moderator variables, including employment status. (I assume the trends would be even more dramatic if we exclude unionized and public-sector employees.)
I’m not saying the Democrats must or can recapture the working white working class. It’s a shrinking demographic, many winning coalitions are possible, and I know little about electoral strategy. But the Democrats have been trying to figure out what and whom they stand for, and they’ve been trying to find their narrative, for a while now. (See Matt Bai.) If the Democrats want to be the party of the American working man and woman, they should first figure out whether they are in fact losing the working white working class, and if so, why. Moral psychology may offer part of the explanation.
6/17/12: Edsall posted a followup column, backing up his earlier column with analyses of exit polls in congressional voting, showing that swings in the white working class vote predict the fortunes of the Democratic party; swings in other groups are much less predictive. So its bad news for the Dems when this group (“canaries in the coal mine”) turns against them.
6/18/12: Larry Bartels responds here, mostly critical, saying that the trend I show here in party ID is mostly due to the South; it’s very weak outside the south and doesn’t show up in presidential voting.
6/19/12: Political scientist Chris Johnston responds here. He does not disagree with Bartels’ quantitative analyses, but he is supportive of my overall argument. He shows that the moral values items on the ANES do predict voting as well or better than does income, and this happens even among Latino voters.
6/19/12: Nate Cohn comments here, agreeing with Bartels overall but noting that the pattern of changes among white working class support for the Democrats is complex–down in rural areas, up in suburbs–and that since 2008 things are way down mostly OUTSIDE the South.
6/20/12: Andy Gelman responds here, trying to reconcile it all: “In short: Republicans continue to do about 20 percentage points better among upper-income voters compared to lower-income, but the compositions of these coalitions have changed over time. As has been noted, low-education white workers have moved toward the Republican party over the past few decades, and at the same time there have been compositional changes so that this group represents a much smaller share of the electorate.” “Lower-class whites (especially in the south) may well be trending Republican, but upper-class whites are even more strongly in the Republican camp, and it’s worth understanding their motivations as well.”
6/25/12: Edsall wrote a followup NYT column covering this whole debate, and drawing in commentary and graphs from Abramowitz supporting my basic claims (which had been based on Abramowitz’s earlier analyses), and expanding the discussion to include occupation-type (i.e., “blue collar” vs. “white collar.”). My only disagreement with Edsall’s column is that he describes the tone of the argument as “furious.” Perhaps there was furious argument in the past, before I stepped into this “minefield” (Bartels’ term). But the discussion and debate that followed my initial Guardian essay has seemed to me to be very civil. Not just in the dueling blog posts, but in the email discussions among me, Gelman, Bartels, Abramowitz, and Edsall. It has not exactly been a warm discussion among friends, but neither has there been any anger or ad-hominem argumentation. It’s just researchers and a journalist trying to hash out the conflicting signals that emerge from multiple datasets to try to figure out a basic factual question: have the Democrats really been losing the white working class, or is it a statistical illusion that has been too heartily embraced by the press? After all this discussion, I think the answer is still yes–at least for the working white working class. (Gelman has convinced me that the poorest slice of the white electorate has not shifted away). But I see now that the issue is more complicated than I thought when I first stepped into the minefield.
Note #1: The ANES dataset has about 1100 white respondents in each election year (every even year) going back to 1948, but some of the items we need for analysis only begin in 1972, so I start there. The graphs look spiky when you plot every single election, so I grouped elections by decade (e.g, 1980 – 1988), and averaged together the 5 elections in each decade.