Losing is hard, particularly when you think you’re going to win. Democrats are viscerally aware of this in the wake of the 2016 election. Since then, though, Republicans have had to cope with that disappointment. Sure, they eked out a few narrow wins in places that usually vote Republican (like Georgia’s 6th Congressional District last year), but they’ve now also suffered two defeats in places that usually vote Republican by a lot.
The first was Roy Moore’s loss in the Alabama Senate special election in December. That one was easy to wave away: The guy was accused of sexually accosting underage girls! Moore was one of the most politically damaged major-party candidates in modern political history.
But then there was Tuesday night in Pennsylvania. Rick Saccone, the Republican candidate for the 18th Congressional District, was a state legislator and a veteran running in a district that backed President Donald Trump by nearly 20 points 17 months ago. And Saccone appears to have lost, too, albeit only by a few hundred votes.
Republicans have mastered the art of spinning overwhelmingly bad special election results by focusing on wins vs. underperformance in House races (treating as nonexistent a House race in California that the Democrats won), ignoring losses at the state level and treating Roy Moore (justifiably) as an outlier. Saccone is harder to dismiss.
One effort focuses on how moderate Democratic candidate Conor Lamb’s positions were. House Speaker Paul D. Ryan called Lamb a conservative, implying that Lamb wasn’t really a Democrat, so, therefore, this isn’t really a Democratic victory.
Someone needs to tell the Republican Party opposition research team about that.
CNN’s Jeff Zeleny spoke with people at the White House who pointed to one silver lining: Hey, at least the race was close. This, of course, ignores the fact that there was a 20-point shift to the Democrats from November 2016. The difference between Republicans doing 20 points worse and doing 25 points worse is a subtle one.
Many Republicans, including Ryan, have sought to isolate the president from the implications of that shift, insisting that his visit to the region Saturday helped turn a relatively easy Democratic win into a narrow one.
On CNN Tuesday night, former Trump campaign spokesman Jason Miller put it this way:
“Well, I think they’re going to be pointing to the fact of the big Trump bump that Saccone got on the home stretch,” Miller said. “The fact that they closed the five-point gap, and you talk about some of the other things the Trump folks did behind the scenes: There was a robo-call from the president yesterday.”
That argument was reiterated on Fox News’s “Fox and Friends” Wednesday morning.
“Last week, it looked like Mr. Lamb was going to win by six points,” host Steve Doocy said. “So something drew it closer together, if you believe in polls. Maybe it was the president’s visit and the visit from the Trump family.”
Maybe! Or maybe these are generous misreads of the polling.
So last week there was indeed a poll from Monmouth University that had Lamb up by six points. But the pollsters were very careful to qualify their findings.
Every pollster makes assumptions about who will turn out to vote in an election that is then used to weight the survey results. Talking to half Democrats and half Republicans may seem fair, but if three-quarters of voters in the district are usually Republican, an even partisan split means that your survey is too heavily Democratic. (Poll nerds: Yes, this is oversimplified to make the point.) In a special election in a weird political year, it can be hard to predict who’s going to turn out to vote.
So Monmouth included three results in its report. One was its model based on how Democratic voting has surged in past special elections since 2016. Another model looked at a lower predicted turnout, more in line with past midterm elections. A third looked at even higher turnout, akin to a presidential race.
The six-point lead was for the Democratic-surge scenario. In a lower-turnout model, Lamb only led by two points.
So what did turnout actually look like? In an email, Monmouth University Polling Institute director Patrick Murray explained that their low-turnout model would mean about 210,000 people voted and their surge model (the six-point one) would mean about 240,000 votes. As of this writing, about 230,000 votes were cast.
In other words, based on the turnout, Monmouth’s poll would have shown a smaller-than-six-point advantage for Lamb. And on top of that, the poll had a margin of error of about five percentage points.
Murray explained that their surge model incorporated assumptions based on those past special elections. Had Monmouth’s model used only individual voter history and data from the survey itself, they would have given Lamb a two-point lead.
But Murray also noted evidence that Trump helped shift things slightly to Saccone. In the more Republican part of the district, the results were close to what Monmouth modeled in its poll’s Democratic-surge scenario. In the more Democratic parts, including in Allegheny County, Lamb underperformed, relative to that scenario.
“Trump’s visit may have countered the extra Dem surge we postulated was possible,” Murray wrote, “but the outcome was pretty close to what we showed using a standard turnout model.” Trump might have had some impact - but he didn’t shift the race by five points.
Saccone didn’t run as strong a race as he could have: He was beaten on fundraising and outperformed on messaging. As a candidate, he was middle-of-the-road but hardly in Roy Moore territory. Even a narrow win for him would have been bad news for Republicans nationally, given how red the district was in 2016.
So the best spin at hand? Saccone at least didn’t get walloped, and maybe Trump’s visit helped tug the final results slightly back to the middle.
But there’s a flip side to that. Democratic enthusiasm in 2018 - enthusiasm which is largely responsible for tugging a 20-point Republican district to the left - is primarily a function of Trump’s unpopularity with everyone besides fervent Republicans. Monmouth showed that voters were split on Trump: 49 percent approved of his job performance, and 49 percent disapproved. Most Lamb backers disapproved of Trump strongly. That gets a little lost in defenses of how Trump aided Saccone at the end of the race.
Putting it another way: If Trump weren’t in the White House, Saccone quite probably wouldn’t have been trailing in polls in the first place.