White educational polarization

Yes, I should have a post about Ezra Klein, “Shorism”, and popularism.  But for now, you get a related post with some nice data analysis from Alan Abramowitz on the issue of Democrats and working class white voters, “Can Democrats Win Back the White Working Class?”  It’s pretty much an anti-popularist take, without ever mentioning it.


— One of the defining features of American politics is the realignment of white, college-educated voters toward Democrats and that of white voters without a degree toward Republicans.

— There are competing views on how or whether Democrats can perform better among white non-college voters.

— Appealing to the economic interests of white non-college voters may not be enough for Democrats to win back their support…

There appear to be two major explanations for the political realignment of the white working class, and they have different implications for Democrats’ chances of a political comeback with this group. One school of thought, perhaps best represented by progressive scholar Ruy Teixeira, blames Democratic decline largely on the party’s prioritization of cultural and racial justice issues over traditional bread-and-butter economic issues. According to this theory, Democrats have failed to address economic problems such as the decline of manufacturing jobs and unfair trade competition that have led to growing economic insecurity among white working class voters. At the same time, many of these voters have been turned off by the Democrats’ increasingly liberal positions on issues such as gay rights, affirmative action, and immigration.

A second school of thought, represented by scholars such as Michael Tesler of the University of California, Irvine and John Sides of Vanderbilt University, argues that economic discontent has little to do with the flight of white working class voters from the Democrats. In their view, the main factor behind the shifting party allegiance of these voters is the success of Republican leaders like Donald Trump in appealing to the racial resentments and grievances of non-college white voters.

These two schools of thought have different implications for the ability of Democratic candidates to win back support from white working class voters. If economic discontent is the main driver of the shift to the GOP, Democrats could potentially win a larger share of the white working class vote by emphasizing concrete actions and policies to address these concerns while perhaps playing down liberal positions on cultural and racial issues. On the other hand, if racial resentment and grievances are the main drivers of white working class flight from the Democrats, paying more attention to the economic concerns of these voters might not be very effective. Moreover, downplaying or abandoning liberal positions on cultural and racial issues would potentially risk alienating voting blocs that make up key components of the party’s current electoral coalition including Blacks, Latinos, and college-educated whites.

In this article, I use evidence from the 2020 American National Election Study to examine the effects of various political attitudes on the candidate preferences of college and non-college white voters in the 2020 presidential election. In line with the arguments of racial resentment theorists, I find that economic insecurity had very little impact on white voter decision-making in 2020. However, I find that the rejection of the Democratic Party by white working class voters goes beyond racial resentment alone. Instead, I find that support for Donald Trump among white working class voters reflected conservative views across a wide range of policy issues including social welfare issues, cultural issues, racial justice issues, gun control, immigration, and climate change. In other words, the rejection of the Democratic Party by white working class voters is fundamentally ideological. This fact makes it very unlikely that Democrats will be able to win back large numbers of white working class voters by appealing to their economic self-interest…

Explaining the class divide on ideology

One important question raised by these findings is why non-college whites now hold much more conservative views across the board than white college graduates. In order to address this question, I conducted a regression analysis of conservative ideology among white voters. As predictors, in addition to education, I included two variables that I expected to have strong effects on ideology: party identification and racial resentment. I also included a variety of control variables including family income, sex, evangelical identification, and economic insecurity. The results are displayed in Table 4.

Table 4: Regression analysis of conservatism among white voters in 2020


Source: 2020 American National Election Study


The results in Table 4 show that the regression equation explains over 80% of the variation in conservative ideology. Racial resentment and party identification are by far the strongest predictors of conservative ideology. Evangelical identification has a significant impact as well, but its effect is not nearly as strong as the effects of racial resentment and party ID. Family income has almost no effect on ideology and economic insecurity has a negative effect, which means that greater insecurity is associated with less conservative policy preferences.

These findings indicate that while ideology was by far the most important predictor of candidate preference among white voters in 2020, ideology was itself largely explained by feelings of racial resentment. Conservative policy preferences among white working class voters on a wide range of issues were closely connected to their racial attitudes and specifically to their belief that white people have been losing ground in American society because of unfair advantages enjoyed by Blacks and other nonwhite groups.


The deep political divide between college and non-college white voters in recent elections reflects a deep ideological divide between these two groups. Non-college white voters are now far more conservative than college educated white voters on a wide range of issues including cultural issues but also social welfare issues, immigration, racial justice, gun control, and climate change. This class divide appears to have little or nothing to do with economic self-interest and everything to do with the diverging racial attitudes of these two groups.

These findings indicate that efforts by Democratic leaders to win back the support of white working class voters who have been voting for Republican candidates in recent years by appealing to their economic interests or shifting to the right on issues like immigration and gay rights are unlikely to bear much fruit. Moreover, tacking to the right to win votes from a shrinking population of white working class voters might turn off large numbers of college educated white voters with liberal views on these issues.


Think rational

Fairly recently, Scott Alexander reviewed Julia Galef’s Scout Mindset.  As they are both prominent members of the (un)official rationalist community, it’s no surprise that Alexander gave a very positive review.  Of course, I’m a big fan of the book myself.  But, if you’re not going to read it (and let’s be honest, you’re probably not) at least due yourself the favor and read Alexander’s thorough review/summary (how Alexander writes so damn much will never not amaze me).  And, if you don’t read it, here’s an extended excerpt with some really good practical advice on thinking more rationally:

The book divides learning Scout Mindset into an intellectual half (Part II) and an emotional half (Part III – V). The intellectual half emphasizes probabilistic thinking and thought experiments.

You’ve probably heard the probabilistic (aka Bayesian) side of things before. Instead of thinking “I’m sure global warming is fake!”, try to think in terms of probabilities (“I think there’s a 90% chance global warming is fake.”) Instead of thinking in terms of changing your mind (“Should I surrender my belief, and switch to my enemy’s belief that global warming is true”), think in terms of updating your probabilities (“Now I’m only 70% sure that global warming is fake”). This mindset makes it easier to remember that it’s not a question of winning or losing, but a question of being as accurate as possible. Someone who updates from 90% to 70% is no more or less wrong or embarrassing than someone who updates from 60% to 40%.

(this comes up again in the last part of the book, the part on how to be emotionally okay with changing your mind. “Probability update” is less emotionally devastating than “I said X, but actually ~X, so I was dead wrong.”)

Not sure how sure you are? The book contains a fun probability calibration exercise. I won’t violate its copyright, but you can find a very similar automated test here

My results on the quiz above. See if you can get closer to the line than I did!

But you probably already knew all of this. One of the genuinely new ideas in Scout Mindset is its endorsement of various counterfactual “tests”. The idea is, imagine yourself considering a similar question, under circumstances that would bias you the opposite direction. If you stick with your opinion, it’s probably honest; if you’d change your opinion in the counterfactual, you probably had it because of bias.

So for example, if a Republican politician is stuck in some scandal, a Republican partisan might stand by him because “there’s no indisputable evidence” or “everyone in politics does stuff like that” or “just because someone did one thing wrong doesn’t mean we should fire them”. But before feeling too sure, the partisan should imagine how they would feel if a Democrat committed exactly the same scandal. If they notice they’d feel outraged, then their pro-Republican bias is influencing their decision-making. If they’d let the Democrat off too, then they might be working off consistent principles.

I try to use this test when I remember. I talk a good talk about free speech, and “don’t cancel other people for discussing policies you don’t like, they have a right to their opinion and you should debate it instead”. But a while back I read an article on Harvard hosted a conference on “the risks of home schooling”, with an obvious eye towards seeing whether they could get home schooling regulated or banned. My first twenty thoughts were something like “is there some way to get revenge on Harvard for being the sorts of people who associate with causes like this?”, plus anger that the administration was probably going to pretend it was neutral on this issue and just “encouraging debate”. Then by my twenty-first thought I remembered this is exactly the sort of thing I was supposed to be against, and grudgingly decided to be more understanding and sympathetic of everyone in the future.

Or: sometimes pundits will, for example, make fun of excessively woke people by saying something like “in a world with millions of people in poverty and thousands of heavily-armed nuclear missiles, you’re really choosing to focus on whether someone said something slightly silly about gender?” Then they do that again. Then they do that again. Then you realize these pundits’ entire brand is making fun of people who say silly things (in a woke direction) about gender, even though there are millions of people in poverty and thousands of nuclear missiles. So they ought to at least be able to appreciate how strong the temptation can be. As Horace puts it, “why do you laugh? Change the name, and the joke’s on you!”

Some other counterfactual tests like this you can try:

Status Quo Test: If you’re defending the status quo, imagine that the opposite was the status quo. Would you be tempted to switch to what you have now? For example, I sometimes feel tempted to defend American measurements – the inch, the mile, Fahrenheit, etc. But if America was already metric, and somebody proposed we should go to inches and miles, everyone would think they were crazy. So my attraction to US measurements is probably just because I’m used to them, not because they’re actually better.

(sometimes this is be fine: I don’t like having a boring WASPy name like “Scott”, but I don’t bother changing it. If I had a cool ethnically-appropriate name like “Menachem”, would I change it to “Scott”? No. But “the transaction costs for changing are too high so I’m not going to do it” is a totally reasonable justification for status quo bias)

Conformity Test: Imagine that some common and universally-agreed idea was unusual; would you still want to do it? If not, you might be motivated by conformity bias. Suppose only 5% of people got married or had kids; would you still want to be one of the 5%? Suppose almost everyone started a business after high school, and going to college instead was considered a weird contrarian choice – would you take it anyway?

Again, sometimes this is fine. Doing the same thing as everyone else earns you friends, and is usually good evidence that you’re not making a terrible error. But it’s at least worth being aware of. Julia writes:

When I was a kid, I idolized my cousin Shoshana, who was two years older than me…during a family camping trip one summer….as we sat in her tent, listening to the latest album on her cassette player, Shoshana said “Ooh, this next song is my favorite!” After the song was over, she turned to me and asked me what I thought. I replied enthusiastically “Yeah, it’s so good, I think it’s my favorite too.”

“Well, guess what?” she replied. “That’s not my favorite song. It’s my least favorite song. I just wanted to see if you would copy me.”

It’s possible my cousin Shoshana crossed paths with Barack Obama at some point, because he used a similar trick on his advisors when he was president. It was essentially a “yes man” test: If someone expressed agreement with a view of his, Obama would pretend he had changed his mind and no longer held that view. Then he would ask them to explain to him why they believed it to be true. “Every leader has strengths and weakness, and one of my strengths is a good BS detector” Obama said.

The Selective Skeptic Test: How credible would you consider the same evidence if it supported the other side?

A meta-analysis of ninety careful studies, published by a prestigious psychology professor, shows that there is no such thing as telepathy, p < -10^10. Does that put the final nail in the coffin? Does it close the debate? Is anyone who tries to pick holes in it just a sore loser? Does it mean that anyone who keeps believing in telepathy after this is a “science denier”?

In the real world, a study meeting that description shows there is such a thing as telepathy. Hopefully you left yourself some room to say that you think the study is wrong.

This is another one with some subtlety. By Bayes’ Rule, you should believe evidence for plausible things more than you believe evidence for implausible things. If my friend says she saw a coyote out in the California hills, I believe her; if she says she saw a polar bear, I am doubtful. I think the best you can do here is understand that, a giant meta-analysis proving telepathy is false doesn’t force a believer to change her mind any more than a giant meta-analysis proving it’s true forces you to change yours.

A lot of the best rationalists I know instinctively apply these tests to everything they think. One technique for cultivating this practice (not the book’s recommendation) is to go on Twitter, where the adage is “there’s always an old tweet”. Argue that people who say racist things should be cancelled, and someone will dig up your old racist tweet and make you defend why you shouldn’t face the same consequences. Argue that it’s disgraceful when the other party uses extreme violent language about their outgroup, and someone will dig up an old tweet where you used even more extreme language about yours. Demand that the Republican senator resign for sexual misconduct, and someone will find the old tweet where you said the Democratic senator should tough it out. Eventually, if you want to maintain any dignity at all, you learn to double-check whether your beliefs are consistent with one another or with what you’d believe in vaguely similar situations.

Scout Mindset says: why not try the same thing, even when you’re not on Twitter, just to determine what’s true?

When I read Galef’s book, I did not actually try the included probability calibration.  But, since Alexander included a link to an online version, I went for it.  And, pretty, pretty pleased with how it turned out, if I don’t say so myself.  Here’s my results below.  I’d love to know how you did via comment or email.   


%d bloggers like this: