Of all the columnists occupying the choicest real estate at the paper of record, Thomas B. Edsall is closest in temperament and approach to the Policy Tensor. What I find particularly attractive about Edsall’s approach is that he pays close attention to contemporary scholarship. What Martin Wolf is to global economics and finance, Edsall is to US politics and society.
Donald Trump’s wholly unexpected triumph in 2016 is the main explanandum of a vast political science literature that has emerged in the three years since. Economic explanations predominated at the beginning. Since then, a different diagnosis has come to the fore that traces support for Trump to White racial prejudice. This diagnosis has achieved a nearly hegemonic position among political scientists and Democratic elites more generally. Diana C. Mutz’s paper “Status threat, not economic hardship, explains the 2016 presidential vote” spelled out the diagnosis.
How is it that the same American public that elected an African American to two terms as US President subsequently elected a president known to have publicly made what many consider to be racist and sexist statements?
A possible explanation is dominant group status threat. … For the first time since Europeans arrived in this country, white Americans are being told that they will soon be a minority race. The declining white share of the national population is unlikely to change white Americans’ status as the most economically well-off racial group, but symbolically, it threatens some whites’ sense of dominance over social and political priorities. Furthermore, when confronted with evidence of racial progress, whites feel threatened and experience lower levels of self-worth relative to a control group. They also perceive greater antiwhite bias as a means of regaining those lost feelings of self-worth.
Edsall’s latest dispatch was prompted by the publication of a new paper that tracks secular partisan realignment. Kitschelt and Rehm argue that education and income furnish good axes to track the realignment.
First, we show that the center-left party, in the United States at least, is being abandoned by lower-education/higher-income voters as much as by the working class (lower-education/lower-income). Second, we show that lower-income voters are divided sharply into two groups. One consists of highly educated people whose numbers and support for the Democratic Party are growing. Its members have become the core of center-left politics. The other group of low-income voters consists of those with lower education levels. Its members have drifted toward right-wing politics on the basis of appeals to authoritarian conceptions of social governance, racism, and xenophobia. But they have also become a “swing” group, up for grabs by either party, given their redistributive economic policy preferences.
Edsall spells out the import of these results for 2020.
The 2020 election will be fought over the current loss of certainty — the absolute lack of consensus — on the issue of “race.” Fear, anger and resentment are rampant. Democrats are convinced of the justness of the liberal, humanistic, enlightenment tradition of expanding rights for racial and ethnic minorities. Republicans, less so.
The salience of racial prejudice can hardly be doubted with this President in office. But are we not mistaking symptoms for causes? Is heightened anxiety over racial status the ultimate driver of the Trump phenomena? More pointedly: Is Trump’s base simply Hillary’s infamous “deplorables”? Or do we perceive it to be so because of the hold of Boasian antiracism on our minds? It is at least worth exploring the idea that support for Trump may be driven by real grievances.
What is common to this political science literature is that their empirical strategy relies on survey data. These are pretty large samples so the problem is not sample size. Rather the problem is that such surveys de-situate people. Each individual appears as an independent subject, grappling with socioeconomic and political trends. That’s fine as far as it goes. But what it leaves out is the spatial correlations due to the fact that people are members of situated communities.
We cannot afford to ignore geography because of two facts. First, the electoral college vote is of great consequence to party competition at the national level. Recall that the Trump coalition only prevailed because sparsely-populated regions in the interior are over-represented in the electoral college. Indeed, since the densely-populated regions on the coasts are Democratic strongholds, the electoral college system systematically discriminates against them. The result is that Democrats have a near lock on the popular vote, while the Republicans have a systematic advantage in electoral college votes. We should not be looking at nationally-representative, that is, population-weighted samples. Rather, we should be weighting by electoral college votes; at least in as much as we care about electoral outcomes and their drivers.
Second, the United States has increasingly become regionally polarized since the 1960s. It is possible, nay, likely, that people are angry, fearful, and resentful, not because their personal circumstances have changed for the worse, but because they see their communities falling apart and see no one in Washington paying any attention to it. As my democratic socialist friend, Ted Fertik mentioned:
Is your community suffering?—Was really the question Trump was speaking to.
So we must build geography right into the analysis. Once we start looking at electoral college-weighted, county-level correlates of the Trump swing—Trump’s vote share less Romney’s vote share—a very different pattern emerges. The three strongest predictors of the Trump swing are college graduation rate, population growth rate, and growth in deaths due to drug overdoses in 2003-2017.
We define the fixed-effect of a predictor as the product of its interquartile range and the slope coefficient. The interpretation is straightforward: The fixed-effect captures how much the response (here, Trump swing in an otherwise average county) moves when we move the predictor from its 25th percentile to its 75th percentile. Defined in this manner, fixed-effects allow us to compare the gradients of different predictors.
College Graduation Rate is a very strong predictor of Trump swing. The fixed-effect in a simple regression model with electoral college weights is -3.3%. That’s larger than the overall electoral-college weighted Trump swing, +3.1%. Counties that swung to Trump have a less educated population than the national average.
Population growth is also a strong predictor of the Trump swing. The fixed-effect in a simple regression model with electoral college weights is -2.8%. The fixed-effect of Net Migration Rate is also high: -1.6%. This suggests that Trump strongholds are places that are bleeding people. College-educated people are known to be much more mobile than people without a college degree. This suggests that the pattern we see in college graduation rates is due to college-graduates leaving, or not returning, to these places. This interpretation is strengthened when we observe that the college graduation rate is correlated with both population growth (r=0.456, p<0.0001) and net migration rate (r=0.396, p<0.0001).
But the most striking correlate of the Trump swing is growth in deaths due to drug overdose. The fixed-effect of this variable is a remarkable +3.4%. That’s higher than the fixed-effect of college graduation rate, population growth rate, and net migration rate.
Indeed, in a three-factor model, growth in deaths due to overdose emerges as the strongest predictor of the Trump swing. Although the fixed-effects of all three conditioners have the same order of magnitude.
The results are robust to a variety of controls. And no other conditioner comes even close in explaining the Trump swing.
Did counties where Whites are declining as a percentage of the population swing to Trump? That would be the implication of the racial resentment thesis—at least in as much as prejudice is driven by locally-visible demography. The answer is no. Change in the percentage of population that calls itself White (“DeltaWhite”) is positively correlated with the Trump swing. Although statistically significant, the fixed-effect is an order of magnitude smaller than that of our three main predictors. It is not that important a conditioner of the Trump swing.
What about income? Is it true, as the political scientists report, that Trump counties are actually richer? It is true that once we control for our three main factors, median income is positively correlated with the Trump swing. That’s congruent with results known from the surveys.
But the statement must be qualified. Without controlling for other factors, median income is negatively correlated with the Trump swing (fixed-effect = -2.6%). The interpretation of these correlations is straightforward: Counties that swung to Trump are poorer than those that did not, although this income differential is more than fully accounted for by variation in population growth and college graduation rates. Recall that median income is highly correlated with both college graduation rate (r=0.67, p<0.0001) and population growth (r=0.50, p<0.0001). So Trump country is poorer, of course, but not as poor as we would otherwise expect it to be, given the education and demographic differentials.
These results should disabuse us of the notion that Trump’s election had little to do with people getting left behind—I drop the quotation marks on purpose. Trump is in the White House because large parts of the country are in serious trouble. People can see the decline of their communities with their own eyes. What is pissing them off is that coastal elites keep ignoring their trauma and focus their attention on creating a more inclusive country.
But what does this have to do with racism? More pointedly: Why does the breakdown of elite-mass relations, now manifest in the Trump insurgency, exhibit the symptoms that it does? Why do people in Trump country, whose trauma is real enough, blame immigrants and minorities? Part of the answer is that people in Trump country regard Boasian antiracism as the hegemonic ideology of coastal elites—as indeed it is. Of course, they don’t call it that; they call it political correctness instead. Resentment of coastal elites, although driven by all-too-real decline of situated communities, is thus expressed as a wholesale rejection of the hated elites’ self-congratulatory worldview.
What I have argued here is that Democrats, including elite political scientists, have misdiagnosed the catastrophe of 2016. If I am right, unless Democrats wake up to reality fairly soon, Trump is going to win again.
Postscript. Unweighted OLS changes nothing.
Post-postscript. Highlights from the next dispatch.
In effect, Trump is a message from Flyover Country for elites. Are American elites listening? Democrats in particular need to pay attention. It is Democrats who repaired elite-mass relations through the 20th century and thereby re-stabilized the system. They must do it again. In order to do so, they must abandon the idea that racism is the key to 2016. It is not. Widespread despair is the key to 2016.