Presidential elections are relatively easy to analyze. The economic fundamentals provide an early handicap of the outcome, and by the end the countless polls are so accurate that you often need to try to get presidential elections wrong, as hopeful partisans do. A special congressional election with two unusual candidates—in today's case, disgraced former governor Mark Sanford (R-Appalachian Trail) and businesswoman Elizabeth Colbert Busch, Stephen Colbert's sister—is far more challenging. With less than 24 hours to go, analysts only have a few public opinion surveys, knowledge of the district, and a handful of reports about “momentum.” All considered, it’s anyone’s race, but one could argue that Sanford has the slightest advantage.
Two weeks ago Colbert Busch, the Democrat nominee, was riding high. Sanford’s ex-wife had accused him of trespassing, and the RNC withdrew its support for his campaign. With national Democrats still engaged and Colbert Busch’s fundraising efforts bolstered by her famous brother, Colbert Busch outspent Sanford on the air. A PPP survey showed Colbert Busch up 9 points, 50-41. As The Guardian’s Harry Enten pointed out, candidates with 9-point leads don’t lose often—even if it is just one poll in an unpredictable special election. Those developments shifted the national media’s discussion of the race, and the Cook and Rothenberg Political Reports classified the election as “Lean Democrat”—although Kyle Kondik of Larry Sabato’s Crystal Ball continued to characterize the race as a “toss-up.”
It was somewhat surprising that Colbert Busch was leading by such a wide margin, since South Carolina’s first congressional district voted for Romney by 18 points. But history makes it clear that deeply flawed candidates, like Mark Sanford, can lose all-but-safe districts. A study by Nicholas Chad-Long found that moral scandals cost incumbent House Representatives an average of 9 percentage points. Exactly how that applies to Sanford, who isn’t an incumbent, is hard to say—especially since his transgressions went beyond infidelity—but it was conceivable that Sanford’s credibility had deteriorated to the point that Republican partisans were unwilling to turn out on his behalf.
In retrospect, it’s unclear whether Colbert Busch ever had a large lead. The PPP survey that showed her up by 9 points assumed an electorate that voted for Romney by just 5 points, even though the district supported Romney by 18 points in November. In a low-turnout special election, it’s certainly possible for the electorate to differ quite a bit from the November electorate, but a 13-point gap was a little tough to swallow. That’s particularly true in South Carolina’s first, where nearly half of all Obama voters were black, and so an electorate that voted for Romney by just 5 points probably requires relatively high black turnout. Indeed, the PPP poll found that blacks represented 18 percent of the electorate, up from 12 percent in their first general election survey. But African American turnout is usually low in midterm elections—and perhaps especially low in a special election—making it tougher to accept PPP’s finding.
Sunday, PPP released a new survey with a different electorate and a different result. Now, Sanford leads by 1 point, 47-46, a 10-point reversal from PPP’s survey two weeks ago. The shift isn’t due to changes in preference, but mainly the composition of the electorate, which voted for Romney by 13 points, up 8 points from two weeks ago. Black turnout is back to just 13 percent of the electorate, a decline of 5 points from their prior poll. On balance, the composition of the electorate looks a lot like their first survey—which had Colbert Busch up 47-45 in late March—and bears little resemblance to their second. As an aside, these types of fluctuations are especially frustrating when they come from PPP, which employs an unorthodox approach to determining the composition of the electorate.1
PPP has a long and solid record. The other pollster of this race does not. Red Racing Horses, a conservative website, commissioned a survey showing a dead-heat, 47-47, but the data analysis, survey design, and weighting was “the sole responsibility of Red Racing Horses,” which, in their own words, is “run by a team of 6 volunteer hobbyists.” Perhaps as a result, they found that women represented 60 percent of the electorate. Red Racing Horses also bought their call-list from Gravis Marketing, which was only spared the wrath afforded to Gallup and Rasmussen because they were never credible enough to merit significant attention.
It’s too bad that there are only two surveys, but at least they agree that the race is close. Similarly, various news organizations have reported that Democratic and Republican private polls show a tight contest, with neither candidate building a lead outside of the margin of error. The Cook Political Report’s Dave Wasserman has moved the race back to a toss-up. Kyle Kondik at Larry Sabato’s Crystal Ball now says the race "leans Republican," while The Rothenberg Political Report now characterizes the race as “toss-up, tilt Democratic.” I give very little weight to media reports about “momentum” based on anonymous sources or on-the-ground observations, since I have no way of distinguishing reasonable claims from ridiculous assertions, but it is a data-point on the side of Sanford, even if a minor one.
Based on the available evidence, there’s not much cause to believe either side has a clear advantage. But to the extent that one can squint at a close race and search for tie breakers, most clues to tilt Sanford’s way: the most recent survey shows him ahead, if narrowly; there is a tenuous case for “momentum;” and most importantly, he’s fighting on friendly territory. The district voted for Romney by 18 points and, consequently, most undecided voters are Republican-leaners. According to PPP, undecided voters supported Romney by a 34-point margin, just 3 percent are black, and 47 percent identify as Republican. Many of these voters might not even show up at the polls, but, if they do, they seem likely to lean toward Sanford—perhaps especially since Sanford has increasingly tied Colbert Busch to the national Democratic Party. The best argument on behalf of Colbert Busch is the possibility that low turnout will dilute the district’s heavy Republican lean. But low turnout might not cut so clearly against Sanford, since as many as half of the district’s Obama voters might have been African American. On the other hand, a strong black turnout could take the race out of Sanford’s reach, since they represent about 20 percent of the population.
It’s important to emphasize that the case for a narrow Sanford advantage is somewhat tenuous. But the one thing we know is that this a tough district for Democrats. Only three House Democrats hail from more conservative districts. Although a deeply flawed candidate can blow a district as safe as South Carolina’s first, Sanford is competitive and recent surveys show that voters don’t despise him, even if they dislike him. But no one should be surprised if Colbert Busch prevails, either. It’s a true toss-up.
Follow me on Twitter @electionate
Rather than survey all adults or registered voters and weight respondents to match census or voter registration targets, PPP asks non-voters to hang-up the phone. With only self-described “voters” left on the line, PPP can’t and doesn’t weight their sample to match an authoritative source; instead, they weight to pre-conceived “target ranges.” Exactly how PPP comes up with these “target ranges” is a mystery, and how they decide where to weight within the “target range” is mysterious, as well. But it’s surprising that PPP’s “target range” for African Americans is, apparently, anywhere from 12 percent to 18 percent of the electorate. That’s an unacceptably wide range, especially in a racially polarized state like South Carolina, given that PPP doesn’t really have any reliable means, as far as I’m aware, for deciding where to weight within the “target range.”