Pennsylvania’s 18th Congressional District will host a special election on March 13, 2018 after its previous incumbent, Republican Tim Murphy, resigned due to a sex scandal. The candidates in this special election are Republican Rick Saccone, a state representative, and Democrat Conor Lamb, a former Assistant U.S. Attorney. A Libertarian candidate is also in the race. The 18th district, based in southwestern Pennsylvania, includes parts of four counties (Allegheny, Greene, Washington, and Westmoreland) and has voted solidly Republican in recent presidential elections.
But what exactly is the district like? Does it have any Democratic areas, or does the entire district vote for Republicans? Do the voters in the district tend to split their tickets? And most importantly, does Lamb have a chance to harness the energy of those opposed to Donald Trump (who has campaigned for Saccone) enough to pull off an upset victory? For answers to those questions, plus a bunch of maps, read on.
PA-18 takes in a swath of working-class rural areas, small towns, and middle- and upper-class suburbs of Pittsburgh. While many of those rural areas used to vote for Democrats due to the influence of unions, they have trended sharply Republican recently, and in 2016 Donald Trump won every single rural township not just in PA-18 but in all of southwestern Pennsylvania. But what about those suburbs?
As the map shows, with the sole exception of the city of Washington, Pennsylvania, all of Hillary Clinton’s areas of strength in PA-18 were in the suburban areas. Clinton only won an absolute majority in eight suburban municipalities, however one of them, Mount Lebanon (which was her best municipality in all of PA-18), is actually the second-most populous municipality in PA-18 (after Hempfield Township in Westmoreland County). Mount Lebanon is a long-established, wealthy, and well-educated suburb, and its neighboring suburbs of Scott and Castle Shannon, both of which also voted for Clinton, are middle-class inner-ring suburbs. Just to the south of Mount Lebanon are two other large suburban towns – Bethel Park and Upper St. Clair – that both voted for Trump but by fairly small margins. But as the map shows, as soon as one crosses the border into Washington County, support for Trump surges as the townships become exurban and then rural.
The pattern in Pittsburgh’s eastern suburbs and in Westmoreland County is similar. Pittsburgh’s inner-ring eastern suburbs, which are not in PA-18, are Democratic, but once one enters Westmoreland County and PA-18, the suburban townships there are strongly Republican. Democrats used to have some strength in Westmoreland’s small towns including Jeannette and Latrobe, but the working-class residents there have switched to voting for Republicans, first at the top of the ticket and now for most offices.
But let’s take a closer look at how the voting patterns of the people of PA-18 have changed over time.
The name of the game here is that most of the suburban townships and small inner suburbs got bluer, while the more working-class suburbs and practically all of the rural areas got redder. Districtwide, those shifts largely canceled each other out, since Trump only did 0.2 percent better than Romney. The aforementioned suburbs of Mount Lebanon and Upper St. Clair swung particularly hard toward Clinton, unsurprising since they are both wealthy and well-educated places. The more middle-class suburbs of Bethel Park, Scott, and Castle Shannon swung moderately toward Clinton. At the same time, every single township in rural Greene County, which used to be a Democratic stronghold, swung heavily toward Trump, leading to him winning the county by more than 40 percent.
But with all those massive swings to Trump in the rural areas of the district, can a Democrat win PA-18 anymore? Let’s look at the other statewide elections in 2016 to try to figure that out.
The closest that any Democrat came in 2016 to winning PA-18 was in the Auditor election, where incumbent Democrat Eugene DePasquale defeated Republican challenger John Brown by 5 percent statewide, but Brown narrowly edged out DePasquale by less than one percentage point in PA-18. How did DePasquale get so close to winning PA-18 at the same time that Clinton was being blown out there? The answer is that DePasquale did much better among rural voters. You’ll notice that Brown received more than 70 percent of the vote in only a single municipality in the district (Green Hills, population 29). DePasquale also improved on Clinton’s performance in the middle-class suburbs in the district – note that he won Bethel Park, South Park, Pleasant Hills, and Greensburg, all of which Clinton lost. And DePasquale achieved those improvements all while running roughly even with Clinton in the sorts of wealthy, well-educated suburbs where she did well across the country – places like Mount Lebanon and Bethel Park. (Notably, Democratic U.S. Senate candidate Katie McGinty lost mainly because she was unable to match Clinton’s performance in those well-educated suburbs.) And yet, even with those strong performances almost everywhere in PA-18, DePasquale still narrowly lost the district. Does that mean that Lamb can’t win? No, but it certainly won’t be easy for him to win.
You may say that DePasquale’s numbers are skewed in his favor because he was an incumbent. So let’s take a look at the Democrats’ second-strongest performance in PA-18 in 2016 – that of Joe Torsella, the Democratic candidate for Treasurer. The Treasurer election was an open seat, and Torsella won by about 6 percent statewide, but he lost PA-18 by just under 5 percent.
Torsella underperformed DePasquale by a small margin almost everywhere in the district – in the rural areas, in the middle-class suburbs, and in the wealthy suburbs. This means that Torsella still outperformed Clinton substantially in the rural areas and in many of the middle-class suburbs (including Bethel Park, Scott, and Hempfield), but underperformed Clinton in well-educated suburbs such as Mount Lebanon and Upper St. Clair. Torsella’s numbers were substantially better than either Clinton’s or McGinty’s, but the fact that he still lost the district by a not-especially-close margin underscores how difficult it will be for Lamb, who doesn’t have the incumbency advantage that DePasquale had, to win the seat.
The two other statewide elections that occurred in 2016 were the U.S. Senate election and the Attorney General Election. I do not believe that either of those elections will be especially predictive of the results of the special election, but I made maps of them anyway.
Does Lamb have a path to victory, and if so, what is it?
I believe that yes, Conor Lamb does have a path to victory in the special election. The path is narrow, but it exists. A good comparison to make is to the Alabama U.S. Senate special election last December, where every single thing had to go right for Doug Jones to be able to win, and sure enough, everything did go right for him. Lamb’s path to victory will require almost everything to go right with his campaign, and a little “help” from Saccone certainly wouldn’t hurt.
Based on the results and patterns from other special elections during the most recent Congress, I believe that Lamb’s most likely and realistic path to victory consists of the following:
1) Match Torsella’s performance in the rural areas and working-class towns
2) Match Clinton’s performance in the well-educated suburbs
3) Have high Democratic turnout in the most Democratic areas of the district
Importantly, all three of the above are necessary for Lamb to pull off the upset. Only achieving #1 would give Lamb approximately 47 percent of the two-party vote; achieving both #1 and #2 would give him 48-49 percent of the two-party vote. High Democratic turnout in the bluest areas of the district would be necessary for Lamb to pull off the victory.
I’d like to address one final point here. Why am I not predicting that Lamb will (or could) outperform Clinton in the well-educated towns where she already did very well? This final map helps explain why.
Towns colored in darker red experienced less ticket-splitting in 2016, while towns colored blue saw a larger proportion of their electorates split their tickets in 2016. As the map shows, the two large towns where Clinton improved the most over Obama – Mount Lebanon and Upper St. Clair – are both colored dark red and therefore saw a low rate of ticket-splitting. Therefore, I believe it is unrealistic to expect that even more of those voters will split their tickets for the March 2018 special election when they did not in 2016.
At this point, I’d say that Lamb has about a 25 percent chance of winning the election. The three points of Lamb’s path to victory that I outlined above are all realistic, and have all occurred with other Democrats in other special elections in the past year, but the fact that all three of them are necessary for the victory makes it somewhat unlikely that Lamb will actually be able to pull it off. I await with bated breath the results of the election, and will analyze the results in great detail afterwards.