Was the 2016 election actually a political realignment?

Did 2016 represent a realignment? That’s certainly an overused and somewhat taboo word these days. But to listen to media coverage today, you might get the impression that 2016 was a substantial break from voting patterns of the past. I’d like to explore this question a bit here and bring some voting data into the discussion.

To look at the state-level votes (as shown in the scatter plot below), we don’t see much of a shift in voting patterns between 2012 and 2016. How a state voted in 2012 tells us a great deal about how it voted in 2016. The vote in 2012 explains about 90 percent of the vote in 2016. (This is not as strong a correlation as between 2008 and 2012, but it’s close.) The red line is a marker for 2012 — if a state is above that line, Hillary Clinton did better there than Barack Obama did four years earlier. She clearly did a bit worse in most states, though not dramatically so.

But we see more substantial variation within several states. The map below (made with Maptile in Stata) shows the rise in the Republican vote share between 2012 and 2016 at the county level. Redder counties saw a larger shift toward the Republican ticket.

Maptile

What the map suggests is that the shift toward the GOP wasn’t uniform. It was concentrated in the Upper Midwest and rural areas of the Northeast. Monroe County, Ohio, moved 21 points in the Republican direction between 2012 and 2016. Howard County, Iowa, shifted 22 points toward the GOP.

As several political observers have noted, this wasn’t just a geographical shift. It was the counties with less educated and whiter populations that tended to shift the most in the Republican direction.

Below is a scatter plot of the county-level vote in Michigan. As with the scatter plot above, the red line shows how a county would have performed if Clinton had received the same vote share Obama received four years earlier. I have divided up Michigan’s 83 counties based on education: Hollow dots indicate counties where at least 25 percent of the population has a bachelor’s degree; the other counties have solid dots.

What’s striking is that the Democratic vote was almost unchanged in the high-education counties, while it dropped substantially in the low-education ones. Again, this doesn’t look like a realignment, in the sense that the more Democratic counties in 2012 still produced the more Democratic votes in 2016. Rather, there’s an intercept shift for the low-education counties. We see this pattern in almost every state, although it is more concentrated in the Upper Midwest.

This effect doesn’t look that big at the state level, since the counties that saw the largest shifts tend to be ones with smaller populations. But the distinction between these different types of counties is profound, and one we haven’t seen in other recent elections.

I used multivariate regression to predict the Republican vote increase in every state based on the county-level percentage of white residents and whether the county was a high-education one or not. (Data source: US Census.) The map below shows an estimate of the impact of the education variable on the Republican vote increase, controlling for each county’s racial composition. I refer to this as the education vote gap — the difference in the GOP vote share increase between the high and low-education populations. The redder states are those where there was a bigger gap between high and low-education counties’ vote shifts. In Michigan, for example, there was an 8.3-point gap — the typical low-education county moved more than 8 points in the Republican direction than the typical high-education county did. It was a similar gap in Iowa. In Wisconsin, it was about a 7-point gap.

Again, we see that this education vote gap was most concentrated in the Upper Midwest and the Northeast.

Now, these graphs help us see how the population shifted, although they don’t really help us understand just why we saw such a split from past voting patterns and why it was regionally concentrated. Was there something peculiar to Donald Trump that just animated lower-education whites in the Midwest? Or conversely, was there something about Hillary Clinton that caused those voters to defect from Democrats? If so, why just there, and will that same effect be around in 2020?

As Julia Azari notes, some recent qualitative literature helps illuminate the people and political beliefs of this region, but it doesn’t make it clear why just them and why now. We’re clearly in a period where race more closely correlates with party than it used to, and, as exit polls show, whites in some parts of the country are starting to vote more like whites in the South. In Georgia, for example, whites preferred Trump to Clinton by a 75-20 margin. Ohio whites went for Trump 62-33, a dramatic Republican shift from their 2012 voting patterns. Yet California’s whites preferred Hillary Clinton by a 5-point margin last year. Again, the racial polarization is not uniform. Even if we limit it to non-college whites, the gap in the vote was substantially larger in the Midwest than it was in the West.

To some extent, this may be a result of the decline of organized labor, which was strong in the Midwest compared to other regions and mobilized Democratic voting even among relatively conservative white laborers in its heyday. As labor lost its strength, perhaps this cleared a path for these downscale whites to sort into the Republican coalition. Or this may just be part of the trend of coastal whites developing different cultural preferences from those in the nation’s interior and the Trump campaign successfully exploiting those differences. Perhaps we’ll see this trend continue in other regions in the years to come, or maybe it was a temporary phenomenon. But our efforts to understand this shift are only beginning.

Will you help keep Vox free for all?

At Vox, we believe that clarity is power, and that power shouldn’t only be available to those who can afford to pay. That’s why we keep our work free. Millions rely on Vox’s clear, high-quality journalism to understand the forces shaping today’s world. Support our mission and help keep Vox free for all by making a financial contribution to Vox today.

One-Time Monthly Annual

$5/month

$10/month

$25/month

$50/month

Other

$

Yes, I'll give $5/month

Yes, I'll give $5/month

We accept credit card, Apple Pay, and Google Pay. You can also contribute via