It took a little while, but I finally got enough time to do a bit of an election analysis. I tried to do a bit of a deeper dive than most of the post-election analyses. I looked at 2016 to get an idea of how voters are reacting to the Trump administration, 2014 for the nearest apples to apples comparison to midterm results and the Democratic wave in 2006 to get an idea of how the electorate is changing long term.
Overall, I found that a number of the narratives that emerged about the midterms are flawed when you look into them. Namely, the prevailing notion that the results were driven by affluent suburbanites and defecting Never Republicans doesn’t hold up. Suburbanites deserve some credit, but affluent voters don’t, and Never Trump Republicans certainly don’t.
On the other hand, other factors which may have been more significant aren’t getting overlooked. Low income voters had a much larger role in driving the results than they’re getting credit for, as did turnout and shifts among young voters, non-white voters and more progressive elements of the electorate. Moreover, there was a notable leftward shift among independents.
Before we get into the results, I do need to caveat everything by saying that I’m working from exit polls, which can sometimes be off in terms of estimating voter turnout. This isn’t so much of a problem because they’re usually accurate enough at indicating the swing in voter preferences of different demographics, which is what I’m mainly focusing on. They also indicate the margins in different voter blocks well enough, while other estimates which might be more accurate don’t. So while the exit polls aren’t perfect, they’re the still best we have.
Demographic Shifts
Racism Didn’t Really Help Republicans, But Gerrymandering and Voter Suppression Did
On election night, when it looked like Republicans were poised to over performing and might even retain the house, a lot of people concluded that the Republican strategy of stoking white ethnic anxieties was to thank. By ginning up fear over immigration and making dog whistles in places like Florida and Georgia Republicans had apparently gotten their base to turn out and salvaged a hopeless situation.
But now that the dust has settled a little bit, it’s looking more and more like that isn’t the case. For all their pandering to white nationalism, Republicans performed significantly worse with white voters. Numerous candidates who went hard for overt racism and xenophobia went down in flames, as they have tended to do historically, while others badly under performed. Hell, even Steve King, the candidate most associated with white nationalism, very nearly lost his incredibly safe seat in the whitest, oldest, most conservative corner of Iowa.
Of course, Republicans did still win in Georgia and Florida, but as FiveThirtyEight pointed out Abrams and Gillum still significantly over performed in states that tend to have a conservative bias. The fact that Kemp and DeSantis Barely won in states that they’d usually be favored in anyways doesn’t suggest they had a particularly good strategy.
In fact, the strategy probably backfired. As noted, it probably did turn off a number of white voters. But it was also probably encouraged a strong backlash among non-white voters. Perhaps more than twice as many Hispanic voters came out in 2018 as had in 2014. Moreover, they also voted for Democrats at much higher rates. Asian voters in particularly went from splitting the vote almost 50-50 in 2014 to voting for Democrats by a 3-to-1 margin.
Mind you, Democrats still could have done better. Latino turnout in Arizona, Georgia and Florida still lagged significantly, and better outreach would have helped turn those races from nail biters to solid wins. But still, the performance really was ground breaking for a midterm.
But the downside is that the more tangible aspects of this strategy, namely disenfranchisement through gerrymandering and voter suppression, were quite effective. Once again, states like North Carolina saw significant shifts towards Democratic candidates, but still didn’t see them pick up a single seat. Moreover, throughout the election countless stories of naked attempts to preempt voting in place like Georgia came out. The effect of these efforts likely was enough to tilt the races in many races with narrow margins.
Hopefully, down ballot wins will help counteract this in future cycles as Democrats can start to take apart the vote suppressing apparatus. And the fact that Democrats were still able to do as well as they did means this isn’t an insurmountable issue. But it is still a significant one we’re going to be dealing with for some time to come, so we need to be ready to challenge it wherever we can.
Yes Suburban Voters Helped Win The House, But The Most Impressive Gains Were Actually Elsewhere
One of the most common takes after the election is that it was won primarily on the basis of suburban voters shifting decisively to the Democrats.
This is true in many ways. Suburban districts did make up a majority of the Democrat’s gains, and Democrats did improve their performance in suburbs. Between 2014 and 2018 they went from losing the suburbs by 14 points in 2014 to a draw in 2018.
But Democrats improved their performance even more in the cities, where they went from winning by 14 points in 2014, to winning by 32 points in 2018. That, coupled with lagging turnout and notably improved performance in rural areas, was perhaps an even more significant factor in tilting the midterms.
All this becomes even clearer when you compare the results to 2016. Democrats only modestly improved on their margins in suburbs, but they made even more impressive gains in cities and made up a fair amount of ground in rural areas. And if you go back 2006, you see that there really isn’t a long term shift towards suburbs.
This is to say that the commonly repeated notion that the Democratic base is shifting towards the suburbs isn’t born out by the numbers. Those aren’t where the most notable shifts happened. Moreover from a long term perspective, Democrats only maintained, and modestly built on their suburban gains in 2016, which was itself only recreating the results Democrats had enjoyed in 2006 and 2008. And anyone arguing that the future of the Democratic party is in the suburbs should beware that the results may be just as ephemeral this time around as well.
On the other hand, the yawning urban-rural gap is still very real. This geographic narrowing of the Democrat’s base is problematic insofar as urban districts are easier to gerrymander. Broadening the base, while at the same time addressing the structural iniquities of the current electoral system, are clearly needed. However, as the most recent elections have demonstrated there can be an upside to all this. If Democrats can run up the score enough among urban voters that were broken up into predominantly suburban or rural districts they may still be able to overcome gerrymandering to win them anyways.
Affluent Voters Didn’t Actually Help Democrats That Much, Most of the Gains Came From Lower and Middle Class Voters
Part and parcel with the idea that the Democrats won on the basis of suburban voters is the idea that they won because of affluent suburban voters. This doesn’t actually seem to be the case though.
It is true that Democrats did also gain a substantial number of votes from high income voters who make more than $100,000 a year on, about 9 million on 2014. But this was offset by gains made by Republicans. In the end Democrats only netted about 2.1 million votes in the demographic on 2014. And compared to 2016 they actually did worse with affluent voters.
By contrast, low income working class voters came back to the Democrats in substantial numbers. Among low income voters making less than $50k a year, more than 11 million more voted for Democrats than in 2014. The Democrat’s margin among these voters increased by nearly 6.5 million votes. Gains among middle income voters were similarly impressive, with democrats netting 4.5 million votes on 2014.
If Democrats have anyone to thank for their wins, it’s lower and middle class voters. All this is to say the Democratic base is still very much the working class, and there isn’t really a trend towards more affluent voters.
This isn’t to say things were entirely great with lower class voters though. Democrats under performed with the working class in states like Ohio, Indiana, Missouri and North Dakota, and this was in large part of the reason they lost they key races there. By contrast, rust belt states where they were able to cruise to victory, like Minnesota, Wisconsin, Michigan and Pennsylvania, are all states where the working class came out and went left. This is clear enough if you compare races relative to a hypothetical scenario where low income voters came out at rates proportional to their share of the population and voted for Democrats at the average rate of 60%.
So to sum up, Democrats made up a lot of lost ground with the working class, but they still have quite a bit of work to do. This may be in part because the economy is still doing well, and parts of the Midwest still see things like Trump’s trade policies in a positive light. However things may look very different in 2020 if there’s a recession (and there probably will be).
Democrats Made Less Progress With Low Education “Working Class” Voters, But that’s Really Just Saying They’re Not Doing Well with Older, Rural Voters
In recent years, the media has tended to define “the working class” as basically meaning people with less than a college education. I really do not care for, this, and I’ve gone on great lengths in the past explaining why it’s bad analysis that terribly muddies the waters of class politics in a counterproductive way. But let’s consider it anyways.
From this vantage point the Democrats made gains in “working class”, but they weren’t as pronounced. They gained nearly 13 million voters with less than a college education on 2014, and went from losing the demographic by 3 in 2014 and 5.5 million votes in 2016 to running about even. That’s a respectable amount, but it’s much less than the 13.5 million lower and middle class votes we alluded to earlier.
Meanwhile, among higher educated voters Democrats also built on their gains among highly educated voters from 2016, gaining a total of 7 million votes on 2014 and netting almost 9 million votes on 2014 thanks in large part to collapsing support for Republicans. And unlike with income, there is a pretty pronounce long term shift to more highly educated voters away from less educated one.
So how do we reconcile the fact that the working class as defined by income is still the Democratic base while the working class as defined by education is increasing not?
Because the fact of the matter is, using education level as a basis for “working class” status heavily biases the terms towards people who are older and more rural, while also containing a number of people who are actually quite well off because they work in jobs that are managerial, petit bourgeois or involve extraction industries. Arguably the main reason why this version of “the working class” appears to be shifting Republican is because (duh) those are all groups that either always favored Republicans or have shifted Republican for other reasons.
At the same time, saying highly educated voters are shifting to Democrats is mainly just saying younger urban voters are shifting to democrats. Of course they are.
And if you’re skeptical that this is basically what’s driving the apparent shift in “working class” preferences, you can look at the most recent general election in Britain to see just how much age and other factors can obscure the picture. Much like the US, there were plenty of stories of lower class Conservatives and upper class Labourites. But if you look within age cohorts you actually find that lower class status invariably correlated to higher support for Labour. Something similar is almost certainly happening in the US.
So we can dismiss the idea that Republicans somehow understand the real heart of the “working class” or that Democrat’s attempts at economic populism don’t work. Those are flawed conclusions based on bad analysis.
Still, this isn’t an irrelevant shift. The Democrats is still the working poor, but who exactly those working poor are has shifted to workers in service jobs in urban centers. Many of them are even broke despite their education. Hence the recent shift in the Democrat’s working class politics towards issues that appeal towards those types of workers, like minimum wage, housing, free college and so forth.
The Age Gap Is Starting To Work In The Left’s Favor, Relatively Liberal Older Boomers Are Becoming Senior Citizens
As in past cycles, there was a big contrast in voting patterns between age groups, with the general rule being that the older the voter the older the more likely they are to vote Republican.
This last election saw that trend pushed to new extremes, with voters under 30 going for Democrats at a rate of 2-to-1. Moreover, as voters have aged up they’ve largely retained their voting preferences. Whereas back in 2006 Democrats really only won double digits among voters under 30, today they dominate voters under 40, and win among voters up to about 50.
One minor point of interest is that while Republicans continue to win among older voters, Democrats actually picked up a lot of votes over 2014 among voters over 65. By contrast, the shift left among voters in the 50-65 range was much more muted.
This kind of makes sense if you’ve been following how generational politics has unfolded in recent years. I made this image recently as a bit a joke, but it basically captures the spirit of what we tend to see. Namely, Older Baby Boomers are somewhat more liberal thanks to their formative experiences with the Vietnam war and Watergate. As they shift into the 65+ bracket they’ve moved it somewhat left.
By contrast, the Younger Boomers and Gen Xers raised during the era of 70s malaise and Reagan tend to more conservative. As such, you’ll notice that people in the 50-65 range tend to be the biggest supporter of things like Trump and the Tea Party.
This would be somewhat of a mixed blessing for Democrats. On the one hand, it would mean the key demographic of senior citizens would be less hostile for a decade or so. But on the other hand, people expecting to win out on demographic attrition might need to wait a bit longer than they think.
Base Mobilization and Party Switching
No, Trump Skeptical Republicans Did Not Swing The Midterms
With Democratic gains predominantly in suburban districts some, particularly Vox have attributed the Democrat’s performance in the midterms to Never Trump Republicans who supposedly finally came out against Trump.
There’s virtually no evidence to support this theory.
if you look at the exit polls, Republicans voted for Republican House candidates at a rate of 94%, the exact same as they had in 2014. Maybe you can cherry pick a few cases where there were substantial defections from Republicans, but on average they were about as loyal as they ever were. Never Trumpers could have easily tipped the balance in the Senate had they come out to support Heidikamp, Donelly, McCaskill or Nelson, but they didn’t. In all of those races, Republicans continued to support their own at rates of over 90%.
The only major race where Never Trumpers might have significantly affected the results was in Arizona where Sinema got won 12% of Republican voters or so. That may have made up the margin. But it still represents a very minor shift, maybe 2% or so. Sinema could probably have just as easily gotten as many votes, if not more, had she invested in better Hispanic outreach.
So no, Trump Skeptical Republicans did not swing the midterms.
Then what did?
Republicans Got Wrecked in Turnout
On election night there was a conventional wisdom that formed that Trump and Republicans had managed to avert complete disaster by winding up their base and getting them to turn out while Democrat’s own turnout fell short of their high expectations. In hind sight this assessment was off the mark.
While it’s true that about 10 million more Republicans had turned out than had in 2014, that’s dwarfed by the 15.5 million addition Democrats and nearly 13 million more independents who turned out. Maybe Republican turnout was impressive compared to most midterms, but they were still clearly lagging.
All in all Democrats beat Republicans in strict partisan voting by something like 5 million votes. That’s slightly more than half margin of victory.
Independents Broke Hard Against Republicans
The other half was made up of Independents breaking sharply against Republicans. In 2014 Republicans won independents by 12 points. In 2018 they lost independents by 12 points. Moreover because of the surge in turnout Democrats got more than twice as many independent votes as they had in 2014.
All in all, the Democrats netted about 4 million votes from Independents, slightly less than the 5 million vote edge they had partisan voting but still very significant.
Why did Democrats win independents? A very large component of it was likely just generic protest voting. People are still by and large disappointed in the state of the country and since Republicans are in charge they’re getting most of the blame. A lot of it was due to alienation by Trump and Republicans over healthcare, immigration and various scandals. It’s possible that there may been contingent of nominally independent Republican leaning voters turned off by Trump specifically. Or conversely it may have been that there were a lot of nominally independent Democratic leaning voters who were mobilized this time around.
My sense is that it was probably 50% mobilized Democratic leaners and 50% alienated voters, but there’s no really way of knowing for certain.
The Democratic Base Is Turning Left In A Big Way
The midterms saw more than 10.5 million more voters who identified as “liberal” than 2014, and as a share of the electorate self-identified liberals continued to gain. Overall, the share of self identified liberals has risen from 20% of the electorate in 2006 to 27% in 2018.
This trend is even more impressive in shifting the makeup of the Democratic party. In 2006, liberals only represented 33% of Democratic voters compared to 54% who identified as moderate and 12% who identified as conservative. Today, liberals make up 46% of Democratic voters compared to 43.3% who identify as moderate and 11% who identify as conservative, and this trend will likely only continue.
This has been reflected in the party as well, as progressive groups have gained while conservative ones have fallen away. To put this in perspective, in 2009 the Progressive Caucus had 67 members in the House of Representatives compared to 54 in the Blue Dog Caucus. Come 2019 the Progressive Caucus will have 97 members while the Blue Dogs will only have 24.
This helps put a lot of the developments over the last year into perspective. Insurgent progressive candidates were frequently frustrated in both the primaries and elections, but they clearly have momentum. By contrast, attempts by the Democratic centrists to recreate their success in 2006, when they ran primarily moderates and veterans who they thought could win over Republicans, was met with a much more lukewarm reception than it was back then. Similarly, it helps explain why overtures towards bipartisanship after the election were met with a good deal more hostility this time around.
Much of this is due to the shifting ideology of the base to be sure, with young voters in particularly favoring a more ambitious leftist platform in the wake of the Great Recession and increasing Republican extremism. But a lot of it is also just a matter of self-identification. Democratic voters were arguably always in favor of the sort of ambitious platform progressive candidates nowadays are offering, but they also found the notion of coming together to overcome common problems emphasized in the rhetoric of politicians like Barack Obama appealing. However after the Obama years disenthralled people of any notion that Republicans were interested in compromising on anything Democrats are clearly shifting towards candidates who will use the power the have more aggressively to achieve left wing goals.