Skip to content

Stop me before I aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

Stephen Olivier points me to this horrible, horrible news article by Jonathan Haidt, “Why working-class people vote conservative”:

Across the world, blue-collar voters ally themselves with the political right . . . Why on Earth would a working-class person ever vote for a conservative candidate? This question has obsessed the American left since Ronald Reagan first captured the votes of so many union members, farmers, urban Catholics and other relatively powerless people – the so-called “Reagan Democrats”. . . .

Sorry, but no no no no no. Where to start?

Here’s the difference between upper-income and lower-income votes in presidential elections:

Ronald Reagan did about 20 percentage points better among voters in the upper third of income, compared to voters in the lower third. The relation between income and voting since 1980 is about the same as it was in the 1940s.

Oh yeah, Haidt said something about “across the world.” How bout this:

It varies. In most countries where we have the data, richer voters are more conservative, but it varies. So that’s interesting and worth studying. I’m not saying there’s nothing there, it’s just that Haidt’s whack-it-with-a-2×4 approach isn’t helping.

I agree with Haidt that economic policy is not the only driver of political preferences. I don’t agree with his claim that “blue-collar voters ally themselves with the political right.” And, yes, I realize that “blue collar” isn’t the same as “lower third of income”—but that’s part of the point! If someone is blue-collar and makes $100,000 a year, maybe it’s not such a surprise that he is voting for a party that supports upper-income tax cuts. Conversely, a non-blue-collar-worker making $30,000 might have a different perspective. Before getting into the psychologizing, Haidt would do well to get a better grip on the facts that he’s trying to explain.

This is one reason I’m such a big fan of descriptive research. Causal inference is fine, it’s great, it’s even in the title of this blog, but you’re gonna get into trouble if you try to come up with theories to explain purported facts that aren’t actually true.

How does this sort of thing get published?

Look, nobody’s perfect. I’m interested in psychology and have lots of friends who are psychology researchers, but if I tried to write something for the topic for a general audience, I’d probably make some mistakes. Similarly, it’s not Haidt’s job to be more knowledgeable than Tucker Carlson about U.S. politics. (From that standpoint, I’m more disturbed by the errors of David Runciman, who has political science in his job title but still has managed to botch his writings on American politics.) Perhaps Haidt read Thomas Frank’s book and it seemed convincing, he doesn’t keep up with scholarly debates on U.S. political science, so he didn’t know where to look.

B-b-b-but . . . Jonathan Haidt is not just some TV talking head. He teaches psychology at the University of Virginia! I’m sure he’s too busy to read up on the American politics literature, but doesn’t he have some colleagues across the quad whom he could talk with about this stuff?

I think it hasn’t helped Haidt to get this sort of uncritical press treatment. At some point it’s natural for him to start believing the hype and then just spouting off on whatever.

In any case, this sort of thing continues to bug me, that 4 years after our Red State Blue State book came out, that this sort of basic mistake could find its way into a major newspaper.

P.S. Commenter Alan T. makes a good point. Whatever misconceptions Haidt may have about voting patterns, his research on the psychology of variation in political attitudes might be valuable. I imagine it has some useful overlap with the work of John Jost. My irritation above is not intended to be a disparagement of Haidt’s research, only an expression of my frustration that he did not slow down and check the facts before changing gears and moving from research-mode to pundit-mode.

P.P.S. Haidt responds (and very politely, which I appreciate, considering the tone of the title and very first sentence of my post):

You are correct that I was too sweeping in my claim. Main error was not to specify that I was talking only about the white working class, and that I was talking especially about the last 8 years. It’s true that the poorer half of the population do generally vote Democratic. And you’re right that the claims about blue-collar Republicans were greatly overblown in What’s the Matter with Kansas. But if we look at whites only, in the last few elections, then there has been a shift, i believe. I was going by this article by Edsall.

In quick response:

1. I agree that, even as the general pattern of income and voting in America has remained roughly stable for most of the past seventy years, the social and geographic composition of the Democratic and Republican voting blocs have changed a lot. We have a graph of this in Red State Blue State (also I’ve blogged it once or twice) showing professionals moving toward the Democrats, business owners moving Republican, etc. It’s perfectly reasonable to focus on whites with less than college education, as long as we don’t confuse matters by conflating that with “working class” more generally. I have a similar problem with the term “blue collar” as it excludes many low-income working people; it’s a term with an emotional affect that I think can mislead.

2. Whether we’re talking about 30% of a group or 40% or 60%, it’s still interesting to learn about the motivations behind people’s vote choices. Most low-income Americans vote for Democrats and it’s worth understanding that preference; it’s also worth understanding the preferences of the Republican minority. If you’re particularly interested in conservative attitudes, I can see why you’d want to isolate various Republican-leaning white voters. You might want to write your next column on rich whites, as they are very strongly Republican in their voting.

P.P.P.S. Some cognitive psychology research suggests how all this confusion could’ve happened.


  1. Jonathan says:

    No hat tip to Jonathan either! I posted it in the last Haidt comments (I am being completely vain.)

    My favorite excerpt:
    “Across many kinds of surveys, in the UK as well as in the USA, we find that people who self-identify as being on the left score higher on questions about care/harm. For example, how much would someone have to pay you to kick a dog in the head? Nobody wants to do this, but liberals say they would require more money than conservatives to cause harm to an innocent creature.”

    • Andrew says:


      I recall your comment but I’d thought you were joking! “For example, how much would someone have to pay you to kick a dog in the head? Nobody wants to do this, but liberals say they would require more money than conservatives to cause harm to an innocent creature.” That sounds like a parody of social psychology research, I couldn’t imagine it was a real quote.

      • Chris says:

        Unfortunately, in the social psych/social cognition literature on moral judgment, that’s a pretty typical example. I blame the Trolley/Footbridge problems. They started people down the path of increasingly absurd stimuli.

        • Jonathan says:

          What IRB won’t approve the actual killing of the animals with a mock weapon (that is infuriating-sarcasm)!

  2. Mark says:

    As a social psychologist who does work on politics…

    …It is Haidt’s job to know about U.S. politics. He has written peer-reviewed journal articles on the topic and recently published a book. He also often gives talks about politics, served as representative of “political psychology” for a grad student “mentor lunch” at a recent social psychology conference, and appears in many news outlets as an expert on politics. If you are a social psychologist (or really any expert) and have clear connections with your other social science brothers and sisters, you should be familiar with their research literature.

    • Andrew says:


      The problem, perhaps, is that Haidt is so invested in the idea that everybody else in his field is biased, that he is inclined to an uncritical, unmoderated view of all his uninformed attitudes. In my own work, I’m all too aware of my deficiencies.

  3. Anonymous says:

    I had a discussion with him about 7 years ago in which he said he had never listened to Rush Limbaugh or read the National Review because he found the idea unsettling or something to that nature. I found it odd at the time, as the talk he gave was on the moral bases of politics — why liberals think X and conservatives think Y. Perhaps this has changed since then.

  4. Max Hayward says:


    I think that you have interpreted the ambiguous claim: “blue-collar voters ally themselves with the political right?” as meaning: “all/the majority of blue-collar voters ally themselves with the political right?” (or, indeed, “a greater proportion of blue-collar workers than rich people vote conservative?” – I have NO idea what part of the text you get that question from).

    As I read it, it is intended to mean: “some/many blue-collar voters ally themselves with the political right.”

    Working out why this occurs is an important and troubling question, to which Haidt is offering an answer. Or did I miss something?

    Remember: charitable interpretation is the first step on the path of constructive criticism.


    • Andrew says:


      Republicans have been winning a good share (40% or more) of low-income voters for a long time. This is not new with Reagan, despite what Haidt claims.

    • Jeremy Fox says:

      Max, your reading seems “charitable” to the point of putting more accurate words in Haidt’s mouth.

      Haidt says “Reagan first captured the votes of so many union members, farmers…” Your reading ignores Haidt’s claim that something changed with Reagan. (And on the contrary, Andrew’s data show that nothing changed with Reagan.)

      Haidt says “blue collar voters ally themselves with the political right.” Interpreting this as meaning “some” blue collar voters ally themselves with the political right seems to me to beyond charitable–it seems like special pleading. “Some” people in every group vote for candidates of every political stripe.

      You say that Haidt is trying to work out why blue collar workers ally themselves with the political right. But Andrew’s data show that that amounts to trying to explain a non-existent fact.

      And how is it not constructive criticism for Andrew to provide data?

      I think you are reading Haidt more than charitably, and reading Andrew less than charitably.

      By the way, I say this as an ecologist who has never heard of Haidt. I’m merely an intelligent, distinterested party here. I find Andrew’s reading of Haidt perfectly fair and natural, and his data to be perfectly on point.

      As Andrew and other commenters note, Haidt is not some inexperienced student from whom mistakes are only to be expected. He is a prominent academic who frequently offers public commentary on politics. Surely it is not too much to expect him to familiarize himself with basic data on voting patterns before commenting on the reasons for those patterns? And surely it is not uncharitable to assume that he is perfectly capable of choosing his words with reasonable precision? So that it is perfectly fair to criticize the most natural reading of his words, rather than stretching so as to read his words “charitably”?

    • Emily says:

      Yes. This is how I read it as well. And the question “Why on Earth would a working-class person ever vote for a conservative candidate” is far more consistent with this explanation. Otherwise it would have been “Why do most working-class people vote for conservative candidates,” or “Why do more working-class people than rich people vote for conservative candidates?”

      • Andrew says:


        Sure, but in that case there’s no reason for the question to have “obsessed the American left since Ronald Reagan.” Working-class persons have been voting Republican since the very first Gallup polls in the 1930s. You can be working-class and still support conservative economic policies. No magic five-factor psychology model is needed to explain that.

  5. John says:

    You mean this Steven Runciman? He sounds like a wonderful, fascinating guy…

    (from Wikipedia)

    It is said that he was reading Latin and Greek by age five. In the course of his long life he would master an astonishing number of languages, so that, for example, when writing about the Middle East, he relied not only on accounts in Latin and Greek and the Western vernaculars, but consulted Arabic, Turkish, Persian, Hebrew, Syriac, Armenian and Georgian sources as well. A King’s Scholar at Eton College, he was an exact contemporary and close friend of George Orwell. While there, they both studied French under Aldous Huxley.

    In 1921 he entered Trinity College, Cambridge as a history scholar and studied under J.B. Bury, becoming, as Runciman later commented, “his first, and only, student”. At first the reclusive Bury tried to brush him off; then, when Runciman mentioned that he could read Russian, Bury gave him a stack of Bulgarian articles to edit, and so their relationship began. His work on the Byzantine Empire earned him a fellowship at Trinity in 1927.

    After receiving a large inheritance from his grandfather, Runciman resigned his fellowship in 1938 and began travelling widely. From 1942 to 1945 he was Professor of Byzantine Art and History at Istanbul University, in Turkey, where he began the research on the Crusades which would lead to his best known work, the History of the Crusades (three volumes appearing in 1951, 1952, and 1954). Most of Runciman’s historical works deal with Byzantium and her medieval neighbours between Sicily and Syria; one exception is The White Rajahs, published in 1960, which tells the story of Sarawak, an independent state founded on the northern coast of Borneo in 1841 by an Englishman James Brooke, and ruled by the Brooke family for more than a century.

    In his personal life, Runciman was an old-fashioned English eccentric, known, among other things, as an aesthete, raconteur, and enthusiast of the occult. According to Andrew Robinson, a history teacher at Eton, “he played piano duets with the last Emperor of China, told tarot cards for King Fuad of Egypt, narrowly missed being blown up by the Germans in the Pera Palace hotel in Istanbul and twice hit the jackpot on slot machines in Las Vegas”.

    He died in Radway, Warwickshire, while visiting relatives, aged 97. He was interred in Lockerbie, Dumfriesshire.

  6. Nameless says:

    When you’re objecting to Haidt by saying that richer voters tend to be more conservative, I think that you’re missing the picture by looking only at income and ignoring ethnic/minority aspects.

    Here in the States, there are two kinds of blue-collar workers: whites and non-whites. Non-whites vote overwhelmingly Democratic, regardless of income. They tend to be poorer than whites, which makes the overall party affiliation vs. income relationship appear skewed.

    The curious part, the one that Haidt tries to address, is that majority (non-Hispanic white) voters lean conservative, even when they are low-income (working class).

    In 2008, whites with incomes of less than $50,000 voted 47% Obama / 51% McCain. And that understates the magnitude of the effect, because many whites with these incomes are college students or otherwise single, younger people who aren’t working class and tend to vote Democratic. Looking by educational attainment, the ones with only a high school degree or less voted 40% Obama / 58% McCain, which is actually a bigger margin than among “rich whites” (income over $100,000, 43%/55%).

    Working-class whites are the core demographic of the Republican Party. The party wouldn’t exist if it didn’t get the majority of the vote in this demographic, because there are too few rich voters, it does not even attain a large margin among rich voters, and it generally fails miserably among minorities.

    As to the other claim, that something changed radically with Reagan, I can’t address that one because I don’t have the data.

  7. Josef Fruehwald says:

    Strangely, I heard him say something similar during a talk back in April, and wrote to him with a link to this post: He even wrote back to me, and mentioned the post in his reply, suggesting that he at least glanced at it. I guess the impact was insufficient to change his thinking.

  8. Anonymous says:

    I think it makes sense in that the working class may be struggling financially, but they may also hold conservative values. This is especially true for the working class in the U.S.

  9. Haidt has visitied UVA’s politics department for workshops a few times – where he would be challenged on statements like this one – but now that he’s moved to NYU, I guess anything goes!

  10. will says:

    Reagan won by large margins across the board, so of course his margins with any particular group would be more than usual.

  11. Thomas says:

    At least some of the language you object to is in the sub-headline. Are you sure that Haidt wrote it, or saw it prior to publication? My understanding is that writers don’t write their own headlines, and I’d guess they also don’t write the sub-headlines of the sort you object to here.

  12. Alan T. says:


    Yes, Haidt clearly gets an important demographic fact wrong. Pity he hasn’t read your book. His error is again clear near the end of his article: “When working-class people vote conservative, as most do in the US…”

    Nevertheless, I am avidly reading Haidt’s book “The Righteous Mind” because it provides the most plausible answer I have yet found to an important question: “Why doesn’t everyone share my political views, that seem so obvious and reasonable to me?” Other writers who address this issue (including Frank, Lakoff, and Hetherington & Weiler; I haven’t read Westen) seem to perceive only the trunk or the tail of the elephant, not the whole beast.

    If you or anyone else who reads this comment can suggest a writer who has a better answer to my question, please let me know. I don’t expect any answer to this question to reach scientific standards of proof, because causation is difficult to prove.

  13. noahpoah says:

    The difference between conservative vote share for two income groups doesn’t bear directly on claims of absolute vote share for any one of those groups. Conservative vote share for upper income people minus conservative vote share for lower income people can be, say, 20%, and this is consistent with baseline conservative vote share for lower income people being anywhere from 0% to 80%. The graphs in the post don’t support your claim from the comments that “Republicans have been winning a good share (40% or more) of low-income voters for a long time. This is not new with Reagan, despite what Haidt claims.”

    • Andrew says:


      Feel free to read our book for all the details. We have charts with absolute vote shares and also relative vote shares. Or if you don’t want to buy the book, you can look up the national votes on the internet and go from there.

  14. Jonathan Haidt says:

    you are correct that I was too sweeping in my claim. Main error was not to specify that I was talking only about the white working class, and that I was talking especially about the last 8 years. It’s true that the poorer half of the population do generally vote democratic. And you’re right that the claims about blue-collar republicans were greatly overblown in What’s the Matter with Kansas. But if we look at whites only, in the last few elections, then there has been a shift, i believe. I was going by this article by Edsall:

    but if you know that Edsall is wrong, please do tell me:
    jon haidt

  15. Dubi says:

    That comparison graphic – is that from red state blue state?

  16. […] « Stop me before I aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!… […]

  17. Gaurav says:


    Bartels makes the same point with great aplomb. If economic policies changed under Reagan, then isn’t it a bit surprising that poor support for R remained roughly constant. It would be nice if we could calibrate positions of both parties on economic issues and then see whether there is something amiss.

  18. Gerald Clore says:

    The political scientists scored points by correcting Haidt’s claim about the importance of Reagon in the timeline, but they came off otherwise looking a little silly — How could he be so misinformed, hasn’t he read all my blogs? One would have hoped that in their specialty they might see the forest as well as the trees. I gather, however, that despite being political scientists, they find nothing worth commenting on regarding the key question, which why might people who probably aren’t going to benefit economically from a Republican victory nevertheless vote Republican. Might the fact that a nontrivial number of people do that suggest that political allegiances are much less about economic gain and loss and more about moral reasoning than most of us would have assumed?

    • Andrew says:


      I don’t know what you’re talking about. Nobody has to read my blog or even my book. But if someone’s going to write about politics in a national newspaper, it makes sense for him to get the facts right.

  19. Corey says:

    Best SMCISS post title ever!

  20. […] Campaign Hey, That Famous ‘Skills Shortage’ You’ve Heard About? It’s a Myth Stop me before I aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!… (how Jonathan Haidt is factually challenged) Spending under Obama lowest in post WWII period (this […]

  21. […] Commenter Alan T. makes a good point. Whatever misconceptions Haidt may have about voting patterns, his research on the psychology of […]

  22. […] of this post to go to its ‘permalink’ version, which shows comments at the bottom. Gives a link to a useful debunking post by Andrew Gelman on the Statistical Modeling, Causal Inference, and […]

  23. […] 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; […]

  24. […] our discussions of psychologist Jonathan Haidt’s opinions about working-class voters (see here and here), a question arose on how to reconcile the analyses of Alan Abramowitz and Tom Edsall […]

  25. […] realizing that he’s trying to explain a phenomenon that does not exist.” Gelman included a link to his own blog with the intriguing headline, “Stop me before I aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa […]

  26. Critical Thinker says:

    I would suggest that you are making a critical mistake as well. It seems to me you are making no distinction between those who work and those who are poor, but do not work.

  27. Martin K says:

    Where do these charts come from ? It purports that French poor tend to vote more conservative than French rich, which is dubious, see for instance red: Hollande, blue: Sarkozy. Obviously there are some confounding variables here, notably age, but even so, I have always read that French rich tend to vote more conservative (French blue-collars tend to vote much more for far-right parties though).

    • Andrew says:


      The graphs are from the Red State Blue State book. Details on the data appear in the notes at the end of the book.