We’re now firmly in the analysis phase of the 2018 midterm elections, an election cycle in which Democrats slightly overperformed expectations in the House while seeing a net loss in the Senate, winning two-thirds of the contested Senate seats. One of the central questions for the party centers on the missed opportunities, particularly the races in Florida and Texas, where they got close — but not close enough — to victory. Why? What went wrong?
The New York Times’s Jonathan Martin offered an analysis of that question in an article published over the weekend. His argument overlaps with a central question within the Democratic Party itself: How far left is too far? Were the losses in Texas and Florida (and in Georgia’s gubernatorial race) in part a function of a too-progressive message that turned off rural white voters?
This is not a new question within the party, but the context in which it’s being asked has shifted dramatically. Members of the Democratic Party are much more likely to identify as liberal than they used to, with the self-identified moderate arm of the party shrinking as a result.
At the same time, the party is hoping to serve as a refuge for moderates turned off by President Trump’s embrace of the far right. How should the party thread that needle? And what didn’t work in the South?
Martin’s examination includes one specific argument.
The campaigns of Stacey Abrams in Georgia, Andrew Gillum in Florida and Beto O’Rourke in Texas may have electrified black and progressive white voters — just as Ms. Hyde-Smith’s comments may energize Mississippians to support Mr. Espy — but they had an equal and opposite effect as well: in rural county after rural county, this trio of next-generation Democrats performed worse than President Barack Obama did in 2012.
What’s nice about that claim is that we can evaluate it. But let’s not look at it solely in the context of Barack Obama’s performance in 2012. Lets look at how counties have voted in those three states in the past four elections and see what patterns emerge. Here, we’ve broken counties in each state into four groups — the most rural quarter to the least rural quarter of counties — and averaged the results in statewide races in each year.
What we see is that, in the most rural (darkest red) counties, the trend to wider margins for Republican candidates has been across the board since 2012. The 2014 gubernatorial election, pitting Gov. Rick Scott (R) against former governor (and now Rep.) Charlie Crist (D) showed a much smaller Republican margin in the most rural counties, perhaps a function of Crist having switched parties. But in each state, the results in 2016 were also more heavily Republican in more rural states than they had been four years prior.
Is that a function of how progressive Hillary Clinton was in 2016 relative to Obama in 2012? Is the steady support Republicans see in the most rural counties in that state a function of equivalently progressive candidates appearing on the statewide ballot? In some heavily rural counties, O’Rourke did worse than Obama had six years earlier. But so did Clinton. Is she as progressive as O’Rourke — or Gillum in Florida?
There’s actually a better way to consider the question Martin’s getting at. In Florida, there were two statewide candidates on the ballot, Sen. Bill Nelson (D) and Gillum. If we chart the difference in support for each candidate in every county against the density of the rural population in the counties, we see that more heavily rural counties were indeed more heavily supportive of Nelson than the unquestionably more progressive Gillum.
But the question is how Democratic candidates might decide how to approach elections. And to address that question, it’s important to add another dimension of data to the chart above: the density of the vote in each county.
Rural counties, by definition, are home to fewer voters.
Is shifting statewide rhetoric to appeal to more conservative voters in more rural places worth the effort? By our analysis, Gillum might have had about 20,000 more votes had the most rural half of counties in the state voted as favorably for him as they did for Nelson.
He lost by 34,000.
To our point at the outset of this article, there were differences between Nelson and Gillum that didn’t come down to rhetoric. Nelson is an incumbent. Gillum is black. The reason that Nelson did better in more rural counties isn’t simply a function of policy positions — but now we’re fumbling around in the gray area.
We have to come back, though, to the fact that not many people live in rural areas. Less than 10 percent of the votes cast in Florida’s Senate race came from counties that are among the most-rural half of counties in the state. The flip side to the question of moderating rhetoric is how candidates fare by amplifying their partisan bona fides.
One of the candidates making that argument most vociferously in 2016 was the guy who beat O’Rourke, Sen. Ted Cruz (R-Tex.). His pitch to the Republican electorate was that running hard-right would energize indifferent voters, while more moderate Republicans would show up to vote the party line anyway. It actually worked as a strategy — but for his primary opponent that year, Trump.
In Martin’s article, he notes that turnout was down in heavily Democratic Miami-Dade County in particular, relative to the rest of the state. A few days after the election, we noted that turnout in that county — which backed Clinton by nearly 30 points — was down relative to 2016, more so than any other county. Could a more forceful message to Democrats have boosted turnout there?
If turnout had been the same as it was in 2016 (which happened in a number of counties in the state) and Gillum captured the same percentage of the vote, he would have gained more than 39,000 votes — enough to win.
Again, this debate about how progressive the party should be is central to the Democratic debate with 2020 looming. There’s little evidence, though, that targeting campaigns to pick up a few thousand more votes in more rural areas would be worth it.
Perhaps you noticed that in the less-rural parts of the states on the line graph above, in Georgia in particular, the political movement had been toward the Democrats. A lot more people live in the least rural counties than the most rural ones. In Florida, the quarter of counties that have the lowest density of residents in rural areas are home to 69 percent of the population.
That shift in less-rural areas helps explain why all of the elections we’re talking about were closer than the 2016 election results. In Texas, Trump won by nine points; Cruz won by four. In Georgia, Trump won by five points; Republican Brian Kemp was elected governor by three. In Florida, Trump won by a little over a point; both Gillum and Nelson lost by less than half a point. In less close races, it’s hard to see why shifting messaging to pick up a few thousand rural votes is worth it, especially if it puts at risk the overwhelming margins in more populous areas.
It’s worth noting that we’ve written about this issue before, back in March when O’Rourke and other more progressive candidates won the Democratic primaries. That article was, in part, a response to a tweet from Martin getting at a similar question about the need for Democrats to be moderate.
Texas’s 7th District was one of the districts that flipped to the Democrats this year. The 23rd District race was close enough that the Associated Press originally called it for the Republican but then, in a rare move, withdrew that call as the lead bounced around.
Why did the Democrats do so well in those districts? Those candidates earned 114 percent and 95 percent of the Democrats' vote totals in 2016, respectively — in a midterm election. Those vote totals were likely in part because turnout was up statewide, a trend that has been attributed, among other things, to the energy O’Rourke’s campaign infused among Democrats.
But it’s something of a gray area.