Eyes on 2018: PA-18 Special Election

Donald Trump won Pennsylvania’s 18th district by 20 percent in 2016; now, less than two years later, a Democrat has a real chance of defeating a Trump Republican there. Rather than embracing the spotlight, Democrat Conor Lamb hopes he can coast to victory by merely avoiding major mistakes.

Though lower-profile than the special elections in Alabama and Georgia’s sixth district, many eyes are on tomorrow’s election. While an imperfect indicator, the results of this race will shed light on the fate of Democrats up for election in red states. And this race’s unpredictability only adds to its excitement.

The District

Pennsylvania’s 18th district has more registered Democrats than Republicans, but these numbers are deceptive. Despite historical union strength, the district has been red for years, with Republican Tim Murphy running uncontested in 2014 and 2016.  Murphy’s resignation, however, has thrown things into flux.

The district is made up primarily of three different counties: Allegheny, Washington, and Westmoreland. As one additional county, Greene County, makes up less than 2 percent of eligible voters in the district, it is easier to think of it merely as an extension of Washington County, which it borders.

Allegheny County is generally considered a swing county, mostly made up of Pittsburgh suburbs. Washington and Westmoreland Counties feel much more like traditional Trump territory, though some signs suggest that support for the President may be waning in these areas.

It should also be noted that this special election is essentially for a yearlong rental of a seat, as the state is set to change the confines of the district significantly before the next election.

The Candidates

Conor Lamb, 33, is a former federal prosecutor and Marine, running as a moderate. He has publicly expressed personal opposition to abortion and is decidedly moderate on immigration. His path to victory requires heavy union support and large margins in Allegheny County, the county which contains both the most voters and the most liberal voters.

Rick Saccone, 60, is a four-term state representative and former Air Force special agent who proudly considers himself “Trump before Trump was Trump.” He hopes to rally Trump’s base of working-class whites in Westmoreland and Washington Counties to boost him to victory. The race is his to lose, but he seems to be trying his best to do just that. Many Republican operatives have openly insulted him, and polling has narrowed as more voters have become familiar with him.

While Lamb has outraised Saccone significantly, Saccone has been boosted by massive spending by Republican groups which actually shift the funding advantage slightly to his side.

The State of the Race

Pennsylvania’s 18th district has not been competitive for years, and special elections are notoriously difficult to predict. As a result, all polling for the race should be taken with several grains of salt.  Nonetheless, there are five polls taken within the last month from at least semi-reputable pollsters:

  • 3/8-3/11, Monmouth: Lamb (D) +6
  • 3/1-3/5, Gravis: Saccone (R) +3
  • 3/1-3/3, Emerson: Lamb (D) +3
  • 2/13-2/15, Gravis: Saccone (R) +6
  • 2/12-2/14, Monmouth: Saccone (R) +3

Over the course of the last few years, I have developed a system of predicting election outcomes based on polling, demographic data, and other factors which vary from race to race. A key component of this system is a weighted polling average which weighs polls based on timeliness (the more recent the better) and quality (some pollsters are more reputable than others). Using the five polls listed above, the weighted polling average for this race is Lamb +3.4.

Polling is, however, just one part of the equation, and demographics are a bit more favorable to Saccone. The district is over 94 percent white, so Lamb cannot afford to lose much more than half of white voters. Its unemployment rate is 6.7 percent, considerably higher than the national average. And 75.9 percent of occupied homes in the district are owned, rather than rented, numbers typically only seen in Republican strongholds. Average ages and many other data points do not clearly favor one side over the other.


Using polling, and demographics, and a bit of tweaking, I developed a model for the race. A table below makes use of the model to predict the outcome of tomorrow night’s election.






This table assumes that counties will make up a similar percentage of the electorate to their representation in the 2016 general election, just with Allegheny County turnout slightly higher and Westmoreland County turnout slightly lower. Somewhat counterintuitively, then, this model predicts that results will more closely mimic 2012 than 2016. This is a result of a general lack of enthusiasm on the part of Republican voters going into tomorrow evening.

Whereas Republicans were on average more enthusiastic about the 2016 race than Democrats, the reverse was true in 2012. All five polls assessed above indicated more excitement among Democratic voters, more closely mirroring 2012. Many Trump voters will likely stay home, and a considerable number will likely flip, so Lamb has more room to grow in Republican-leaning counties than in Allegheny. This is especially true given his embrace of the blue dog label.

The table’s third row outlines the model’s predictions of margins by county, reflecting changes in turnout from 2016. The second row, meanwhile, shows the margins in each county most reflective of a tied race.

Though few polls have included the race’s libertarian candidate, he is unlikely to garner much support. The model thus predicts a 1.3 percent victory for Lamb, with election results as follows:

  • Conor Lamb (D), 50.2 percent
  • Rick Saccone (R), 48.9 percent
  • Drew Miller (L), 0.9 percent

Lamb’s margin of victory in this model is quite slim, and his chance of winning stands at 62 percent. The model also estimates with 95 percent confidence that the race will fall somewhere between Saccone +6.3 percent and Lamb +6.1 percent. If that seems like a big margin, that’s because it is. Different elements of the race suggest very different outcomes, and it is quite difficult to determine what will matter by the time the night is over.

Previous: Eyes on 2018: Indiana Senate Race

Next: The Ives of March: How Bruce Rauner was nearly taken down by “fringe candidate” Jeanne Ives