After last week’s loss in Georgia’s 6th Congressional District, Democrats still need to gain 24 seats to take back the House in 2018. The question is how they get there. They could:
- Go after traditional swing districts, many of which are in the “Rust Belt,” where Barack Obama performed adequately in 2012, but where Hillary Clinton got crushed by Donald Trump. Let’s call this the “Obama path.”
- Go after “new” swing districts, many of which are in the “Sun Belt,” where Clinton did historically well for a Democrat, but where Obama had a harder time against Mitt Romney. Georgia 6 is one of these districts. We’ll call this the “Clinton path.”
- Compete everywhere.
Option No. 3 might stretch resources more, but special election results so far this year suggest that Democrats probably need to compete in the Clinton path and the Obama path to really have a chance to take back the House.
Since Trump was sworn in as president, there have been 23 special elections1 for the U.S. House and state legislatures. In previous years, the average result in both of these types of special elections has been at least somewhat predictive of the following midterms.2 So if the Obama path were better for Democrats in 2018, we’d expect the special election results to look more like the 2012 map. If the Clinton path were better, they’d look more like the 2016 results.
So which map have the special elections followed? A multiple linear regression in which we try to explain the 2017 results by the district’s partisan lean in 2012 and its lean in 20163 reveals that the special election results so far this year track more closely to the 2012 lean. The analysis suggests that the best starting point to explain special election results in the Trump era is a weighted average in which the 2012 lean is weighted close to twice as much as the 2016 lean.
In essence, if you’re interested in where the country is politically — is Wisconsin now red because Trump carried it? Or is it still blue because Obama won it in 2012? — you’re best off with a political map that looks mostly like 2012’s.
To be clear, the 2016 results aren’t useless in understanding a district’s lean. Both the 2012 and 2016 partisan leans have the same basic correlation with this year’s special election results. Rather, the special elections we’ve had this year simply suggest that Trump has not remade the political map.
Indeed, compare the 2017 results to the 2012 results district by district, or compare the 2017 results to the 2016 results district by district, and you see the same general shift among voters. There’s been an average shift of 14 percentage points in favor of the Democrats using the 2012 presidential election lean and a shift of 15 points using the 2016 baseline.
|SWING IN SPECIAL ELECTION RESULT COMPARED TO …|
|DATE||DISTRICT||2012 DEM. LEAN||2016 DEM. LEAN|
|6/20||Georgia 6th Congressional District||+23||0|
|6/20||South Carolina 5th Congressional District||+12||+17|
|6/20||South Carolina House District 48||+9||+5|
|6/20||South Carolina House District 70||+11||+15|
|6/15||Tennessee House District 95||+30||+14|
|5/30||South Carolina House District 84||+13||+15|
|5/23||New Hampshire House District Carroll 6||+20||+13|
|5/23||New Hampshire House District Hillsborough 44||-3||0|
|5/23||New York Assembly District 9||+32||+41|
|5/23||New York Senate District 30||+1||+1|
|5/16||Georgia Senate District 32||+26||+2|
|5/9||Oklahoma House District 28||+39||+50|
|4/29||Louisiana Senate District 2*||+58||+63|
|4/25||Connecticut House District 68||-30||-21|
|4/11||Kansas 4th Congressional District||+23||+23|
|4/4||California 34th Congressional District*||+22||+16|
|2/28||Connecticut House District 115||-7||+12|
|2/28||Connecticut Senate District 2||-20||-20|
|2/28||Connecticut Senate District 32||+7||+11|
|2/25||Delaware Senate District 10||+3||+6|
|2/14||Minnesota House District 32B||+10||+25|
|1/31||Iowa House District 89||+22||+36|
If we look at all the special elections, it’s interesting to note that Democrats this year have been outperforming their 2016 baseline most in the districts where Clinton underperformed Obama and least in the districts where Clinton outperformed Obama. The latter group tends to include more highly educated, suburban areas that traditionally voted Republican. These places may be growing bluer, but perhaps Clinton did as well as could be hoped for by a Democratic candidate for the time being. In the districts where Clinton didn’t do as well as Obama had, it’s possible that traditional Democratic voters who voted for Trump in 2016 may not be voting for Republicans in 2017.
|WHERE CLINTON DID WORSE THAN OBAMA*||DEM. SWING FROM 2016|
|Connecticut House District 68||-21|
|New Hampshire House District Hillsborough 44||0|
|Delaware Senate District 10||+6|
|Connecticut Senate District 32||+11|
|Connecticut House District 115||+12|
|South Carolina House District 84||+15|
|South Carolina House District 70||+15|
|South Carolina 5th Congressional District||+17|
|Minnesota House District 32B||+25|
|Iowa House District 89||+36|
|New York Assembly District 9||+41|
|Oklahoma House District 28||+50|
|Louisiana Senate District 2**||+63|
|WHERE CLINTON DID BETTER THAN OBAMA*||DEM. SWING FROM 2016|
|Connecticut Senate District 2||-20|
|Georgia 6th Congressional District||0|
|Georgia Senate District 32||+2|
|South Carolina House District 48||+5|
|New Hampshire House District Carroll 6||+13|
|Tennessee House District 95||+14|
|California 34th Congressional District**||+16|
|Kansas 4th Congressional District||+23|
In the eight special-election districts where the district became more Democratic compared with the nation as a whole between 2012 and 2016, the shift to the Democrats in 2017 has averaged just 7 percentage points from the 2016 partisan lean. Georgia 6 is one of those eight. In the 14 districts where Clinton underperformed Obama, the special election margin shifted 21 points, on average, toward the Democrats.4 The South Carolina 5th Congressional District, where Democrat Archie Parnell nearly pulled off a stunning upset last week, is one of those.
So what to make of this data?
For starters, it suggests that Democrats should not give up on areas where Clinton did significantly worse than Obama did. Democrats gained a seat in a special state legislative election in New York where Clinton underperformed Obama by 9 points compared with the national vote, for example. Many of the areas where Clinton did worse than Obama and where there was a large swing in the 2017 special elections, like South Carolina 5, have a low percentage of college-educated voters. So perhaps Democrats should continue to try to compete in traditionally competitive districts with fewer college-educated voters — voters that have traditionally gone Democratic, but went for Trump. These voters may have cast a ballot for Trump, but they’ll apparently still pull the Democratic lever, at least when Trump isn’t on the ballot.
Inversely, it suggests that Democrats would be silly to focus only on or mostly on the Sun Belt suburbs in 2018. These are the areas with a lot of college-educated voters (like Georgia 6) that have traditionally gone Republican, but where Trump is not as popular as other Republicans are. It’s not that Democrats can’t win in those places. But in the areas where there was a very large shift toward Clinton in 2016 and where Republican incumbents are still running (i.e., have an incumbency advantage), it may be a bit much to expect many Democratic candidates to do better than Clinton did.
More broadly, the results of this year’s special elections suggest — as have some commentators — that Democrats would benefit from a broad House playing field in 2018. When the electoral map is in flux, like it is now, what the partisan baseline is for an individual district — how Democratic is it, compared with the nation as a whole? — isn’t going to be entirely clear. And because Democrats need 24 seats to take back the House, they really can’t afford to leave any stone unturned.