A sober, quantitative analysis of a completely bonkers primary.
On Tuesday, West Virginia Republicans will head to the polls to nominate a candidate for the state’s upcoming Senate race. This is a high-stakes contest—Sen. Joe Manchin is one of the few remaining moderate red-state Democrats in the upper chamber, and West Virginia Republicans have diligently worked to grow a bench of candidates that could take on a strong incumbent like Manchin.
But there’s at least one speed bump—Republican primary voters might decide to nominate a convict.
Don Blankenship, a coal magnate who recently spent a year behind bars for violating mining standards (the case was connected to the Upper Big Branch Disaster, where 29 men lost their lives) and has recently made headlines for using terms like “ China People ” and calling the Republican Senate Majority Leader “ Cocaine Mitch ”, has a real shot at winning the GOP nomination.
There’s a obviously a lot going on in this race. But I’m going to stick to the electoral analysis—the current state of the Republican primary, what usually happens when a problematic candidate like Blankenship gets a major party nomination, and what the outlook for the general election is.
The Basics: Nobody Knows Who is Going to Win the Primary
There are three major candidates in the race. Two of them—Congressman Evan Jenkins from the 3 rd district (the southern third of the state, known for its coal production) and Patrick Morrisey (the state’s attorney general and a favorite of ideological conservatives)—both fit the mold of traditional GOP Senate candidates. Both would likely run solid campaigns attempting to tie Democratic Sen. Joe Manchin to the national Democratic party and supporting President Trump’s agenda. The third candidate is Blankenship, QED.
The polling suggests that any of these three could win the nomination.
On Monday, John McCormack published the results from two internal tracking polls from rival campaigns. Both showed Blankenship ahead by narrow margins. Fox News released the only public poll of the race, which showed Jenkins in the lead with Morrisey close behind and Blankenship with a fighting chance.
There’s no way to come up with a clear prediction based on these numbers. Most of our data comes from internal polls (which can have real issues) and the polls we have disagree with each other. Maybe most importantly, the margins on these polls are thin. Jenkins leads Morrisey by 4 points in the Fox poll while the internal polls show a 1- and 3-point lead for Blankenship. None of those results translate into a safe lead for any candidate.
How Much Worse Would Blankenship Be Compared to a Replacement-level Republican?
If Blankenship does win the nomination, he’ll face Manchin in the general election. Nobody knows with certainty how Blankenship will perform in a general election ( problematic candidates have won before). But it’s not crazy to think that his baggage might give him worse odds than a replacement-level Republican. So it’s worth sifting through the data to get a sense of how far below average he might perform.
I got a read on this by feeding some data (state-level presidential approval, whether an incumbent was running, whether one or of the major candidates was widely considered to be ineffective in some way, etc.) into a simple model (one I worked on with Sean Trende at RealClearPolitics) and asking it to predict the results of competitive senate races from 2004 through 2016 (2008 is omitted because George W. Bush’s approval rating was low and partisanship gave GOP candidates a higher ceiling than his numbers suggested).
The model came up with some interesting outputs. When one party ran a bad candidate, it estimated that the other party got roughly 3 extra percentage points tacked onto their final vote total (with some error around that estimate).
In a close election, 3 points is a lot. In 2012, Republican Dean Heller won his Senate seat by about 1 point over Democratic candidate Shelley Berkley. Berkley wasn’t an especially great candidate, and Berkley’s weakness helped Heller win despite Barack Obama’s simultaneous 7-point statewide victory. Similarly, in 2010 Democrat Michael Bennet won a Colorado senate seat by 1.7 points, partially because Republicans nominated Ken Buck, who also had some issues (though Buck recovered from that loss and eventually made a comeback in the House).
It’s possible to imagine a case where Blankenship’s baggage gives Manchin a similar boost and allows him to win a race he otherwise might lose.
But many Republicans fear that Blankenship would be even worse than your run-of-the-mill bad candidate. And that’s not a crazy fear—we’ve seen other candidates perform a lot worse than Buck or Berkley.
In December 2017, Republican Roy Moore lost a special Senate election in Alabama to Democrat Doug Jones. A brief refresher on Moore—he was removed from the state supreme court twice, underperformed Mitt Romney by a huge margin in 2012, managed to win the 2017 senate special election primary in Alabama against Luther Strange ( who had his own issues) and then was credibly accused of improper sexual conduct with teenage girls while he was a man in his 30s. Republicans generally haven’t fared well in House and Senate special elections in the Trump era, but Moore’s performance was the worst of the bunch.
The Moore data point—an almost 30 point underperformance—stands out. A more generic Republican might have underperformed Trump by a significant margin (maybe 10 or 20 points), but Moore’s underperformance went about 10 points beyond that and allowed Democrats to take a Senate seat in Alabama.
And Moore isn’t the only candidate to recently underperform the president by double digits. When Joe Arpaio (the infamous sheriff who was recently charged with contempt of court, pardoned by the president and has vocally questioned the veracity of President Obama’s birth certificate) ran for another term as Maricopa County in 2016, he lost to his Democratic opponent, Paul Penzone, by 11 points. Trump simultaneously won Maricopa County by 3 points .
I could go on, but the point here isn’t to hang a specific number on a Blankenship candidacy, but just to demonstrate that candidates matter and that when you nominate someone with Blankenship’s baggage, you put your party at risk of anything from a small underperformance to a colossal underperformance.
So What Would a Bad Republican Performance Look Like?
We can get an idea about this by looking at past election results
This .gif shows the county-wide two-party presidential vote in West Virginia from 1988 to 2016 with red and blue indicating higher Republican or Democratic county-wide vote shares (you might have seen versions of it here or here). The story here is simple: As Democrats moved to the left on culture and coal (think of the difference between Bill and Hillary Clinton) and the GOP moved right (or the difference between George H.
