Of all the different angles we can take in analyzing the 2016 presidential vote and the election of Donald Trump, one of the more interesting is to drill down to the level of our 435 Congressional Districts. This allows us to see not only the areas where the 2016 presidential outcome was substantially different from the 2012 presidential outcome, but also the areas where there was fairly extreme ticket splitting. That is, which parts of the country are still more than comfortable sending Democrats to Congress, even as they swung dramatically last year toward Donald Trump? What do those districts look like, and what can we learn from them heading into 2018 and beyond? Fortunately, Kyle Kondik at Sabato’s Crystal Ball (using data compiled by Daily Kos Elections) has done the legwork to identify these districts here. There’s no better place to start than Pennsylvania’s Congressional District 17 (PA-17), where voters re-elected Democrat Matt Cartwright by a large margin (54%-46%), while at the same time giving Donald Trump 53.4% of the vote, a substantial difference from Mitt Romney’s 43.3% in 2012.
How does that happen? As we start looking for an answer, it’s helpful to look at another district for a side-by-side comparison, preferably one in the same part of the country, to possibly identify which aspects of the 2016 turnout in PA-17 were unique and therefore may explain the shift toward President Trump. In this case, we’ll look at the PA-15, which runs more or less parallel directly to the southeast of PA-17. Congressional District 15 is represented by Republican Charlie Dent, who has been in the news recently as a key moderate in the debate over the Obamacare replacement bill. The Republican share of the presidential vote was nowhere near as volatile in PA-15, as Trump received 52% of the vote compared to Romney’s 51%.
One possible explanation for the results in PA-17 is that the composition of the 2016 electorate was different enough from the 2012 electorate to account for the change, meaning that either Republican turnout was higher, Democratic turnout was lower, or some combination of the two. As a basic first step toward determining whether this was in fact the case, we can place 2016 voters into one of three groups: Those who also voted in the district in 2012, those who were new to the district (meaning, not registered at the time of the 2012 election) and those who were not new to the district but did not vote in 2012. Below are tables showing the distribution of those three groups for both PA-15 and PA-17.
So at first glance, there is not a huge difference between the two. In each district, around 70% of those who voted in the 2016 election also voted within the district in 2012. If there existed a groundswell of newly registered Trump voters in PA-17, or conversely a group of would-be Clinton voters who stayed home, we would expect to see a difference in these numbers in comparison to PA-15. That difference simply isn’t there. However, these numbers obviously don’t tell the full turnout story. We can drill down further by looking at turnout by party registration. Below is a comparison of 2012 turnout by party and 2016 turnout by party, for PA-15:
This is exactly what we would expect in a district where there was no significant change in the results from 2012 to 2016. There was a slight uptick in the Republican and Independent percentages of the electorate, with a corresponding drop in the Democratic percentage, so it’s not surprising that Congressman Dent improved upon his 2012 performance and President Trump slightly improved upon Mitt Romney’s performance. But there is really nothing interesting here.
By contrast, take a look at the same comparison for PA-17:
Here, we can detect the Trump shift. Even though, as noted above, we know that 70% of those who voted in 2016 also voted in 2012, the Democratic percentage of the electorate drops by a significant 4.5 points. Most strikingly, the raw number of Democrats who voted barely changed from 2012, despite the fact that 26,000 more voters cast ballots in 2016 than in 2012. Nearly 20,000 of those additional votes come from Republicans. So what is going on?
What we appear to have is a textbook case of that oft-discussed group, the key to understanding the election, the cause of countless sleepless nights for Democratic politicians and strategists around the country…The Obama-Trump voters. And what’s interesting about PA-17 in particular is that we absolutely could have seen this coming. Because not only were these Obama-Trump voters changing their preferences from the confines of the voting booth, quite a few of them were actually changing their party registration from Democrat to Republican in the intervening years between 2012 and 2016:
This explains why, even with increased turnout across the board, there didn’t appear to be any growth in raw votes cast by Democrats. Some of those 2012 Democrats still voted in 2016, they’re just no longer Democrats. So they cancel out any Democratic growth through new registrants or new Democrats moving into the District. Of course one thing this data doesn’t show is how many voters maintained their Democratic affiliation and yet still voted for President Trump, though we can assume that that number is significant. There’s no other way to explain how, in a district where 52% of those who voted were registered Democrats, Hillary Clinton only received 43% of the vote.
The other important lesson that we can learn simply from the mere existence of these Obama-Trump voters is that four years is a long time in politics. It’s not as if they all voted for President Obama one day and then President Trump the next, though that was often the tone of the reporting in the days and weeks following the election. Clearly it is possible for the basic makeup of a Congressional District to change significantly in a four-year span. We can see this just by looking at registration by party for PA-17, rather than turnout:
First off, the bump for No Affiliation/Other voters is not surprising. In fact, it’s the trend across the country for younger voters to choose to remain unaffiliated from either of the two major parties. However, the registration shift toward Republicans is important because, as noted above, it means that anyone paying attention leading up to the election could have seen this coming.
Still, it’s important not to read too much into this trend and what it could mean further down the ballot, at least not yet. Voters in PA-17, and those in similar districts around the country, are not ready to give their votes to every politician on a ballot with an R next to their name. Don’t forget that the incumbent Congressman is Democrat Matt Cartwright, who, far from being a moderate, actually won the seat after defeating a moderate in the 2012 Democratic Primary. And yet, back he went to Washington even after sharing the ballot with President Trump.
So as the NRCC targets Cartwright in 2018, an opportunity that they appear to be relishing, they would do well to take a look at the last three Republicans to challenge him. In 2016, former race car driver and small businessman Matt Connolly described himself in a debate opening statement as a Constitutional conservative. Going back to 2014, former Schuylkill County coroner David Moylan, holds conservative views on social issues like abortion and the 2nd Amendment. And his 2012 opponent, Laureen Cummings, is the founder of the Scranton Tea Party. Suffice to say, President Trump is a different kind of Republican. While any of Cartwright’s last three Republican opponents may have fared well in a different district, the fact remains that PA-17 will likely be difficult for a traditional Republican candidate or a Tea Party-backed candidate to win. Which is perhaps the most important lesson for any political observer to take from the Obama-Trump voter phenomenon: Candidates still matter.