Last year, about a month before the runoff in the heavily watched special election for Georgia’s 6th Congressional District, another race in the same corner of the state quietly slipped by with little notice from national media. This race was for Georgia’s 32nd state Senate District, which lies within the boundaries of the 6th, though the Senate seat is even more conservative than the House seat.
In the end, the Democratic candidate, Christine Triebsch, wound up losing, but what was important is how she lost. Triebsch was defeated by a 57-43 margin, a 14-point spread that was identical to Hillary Clinton’s 54-40 loss in 2016—and 22 points better than the 67-31 thumping Barack Obama had endured four years earlier.
We sat up and took notice at Daily Kos Elections, even though the Republican won that state Senate special election by a healthy margin. Why? Because if the same margin changes when compared to the last two presidential elections also showed up in the congressional special election, that pointed to a virtual tie. And indeed, the margin changes in the House race were similar, coming in just 3 points under Clinton’s and 19 points better than Obama’s, or in other words, about 3 points weaker in both cases than the shift in the Senate race. And indeed, Democrat Jon Ossoff lost by 3.6 points, 51.8 to 48.2.
What we didn’t pay much attention to at the time was a similarly situated special election in South Carolina’s 84th state House District, where the Democrat improved on Clinton’s margin by 13 points and Obama’s margin by 8. A few weeks later, in the special election for South Carolina’s 5th Congressional District, the Democrat improved on Clinton’s margin by 15 points and Obama’s by 8. While separated geographically, the two districts were similar demographically—more on that below.
This year, in January 2018, we saw a special election in Pennsylvania state House District 35, where the Democrat over-performed Clinton’s margin by a whopping 29 points and Obama’s by 19! Looking ahead to the special election in an adjacent district, Pennsylvania’s 18th Congressional District, where Clinton lost by 20 and Obama lost by 17, it certainly gives us cause for optimism that Democrat Conor Lamb has a real chance of winning. It’s also consistent with recent polling showing a single-digit race.
But, as it turns out, PA HD-35 and PA-18 are different enough demographically that the comparison might be a little off. Fortunately, a part of HD-35, White Oak, is more similar to PA-18. But how did they vote?
White Oak is one little town that voted in a landslide for Trump and then voted in a landslide to send a Democrat to the Pennsylvania House.
Trump won White Oak by 14 points; Romney won by 12. But Democrat Austin Davis won the town by a 26-point margin. That’s a massive 40- and 37-point shift, respectively, in a white, working class/middle class area. Towards the Democrat.
PA-18 is wealthier, whiter, and more educated as a whole than PA HD-35. These demographic differences, of course, are important variables when it comes to voting behavior. Below is a map showing the two districts and the racial demographics:
You can compare the demographics with Matthew Isbell’s results maps here. White Oak is the big red block on the right in the 2016 results maps, turning a pretty mix of blue for the 2018 special.
Here are the numbers for each of the districts discussed:
The biggest differences between the pairs of districts in Georgia and South Carolina is turnout. Indeed, turnout was about double in the congressional district races! Nonetheless, the margin shifts are consistent in both.
Still, the PA-18 special could easily have turnout that is quadruple the turnout of the PA HD-35 special. For such a discrepancy, we have one more natural experiment this year to compare to: Virginia’s House District 85. In a January 2017 special election, Cheryl Turpin underperformed Clinton and Obama by about 5 points and lost. But in the November rematch, with four times the turnout, Turpin over-performed both Clinton and Obama by about 2 — and won. Of course, these two elections are separated in time by more than the other pairs. However, we’ve also seen in additional repeat elections that a party’s over-performance is not related to turnout.
With only 767 votes total in White Oak for the recent Pennsylvania special, these results are simply another set of tea leaves to read. In other words, we wouldn’t use this to predict a massive 40-point shift in the PA-18 race! Rather, it shows that given the demographics of the district, Democratic over-performance is a little more likely rather than less likely, and a shift large enough for a Democratic win is well within the realm of possibility. There’s every reason to believe it when the polls show that this area, like most of the rest of the country, is shifting back to Democrats, and plenty of reason to think that Conor Lamb can pull out a win.