“Trump’s Big Gamble,” the New York Times headline reads, “Can He Pull Alabama Senator to Victory?”
It’s certainly possible, but is it likely?
According to public opinion polls, don’t count on it. Trump’s pick is trailing the opposition Roy Moore in the Alabama Senate GOP primary runoff.
According to the final week of polling, incumbent appointed Senator “Big Luther” Luther Strange, is falling short of the former Alabama Supreme Court Chief Justice by nearly 10 percentage points. To those who tuned in for the first round of the contest, this will not surprise; Moore came away with nearly 40% of the vote compared to Strange’s 32 points.
But there is something surprising about the race. The all-out war between Trump’s former chief strategist Stephen K. Bannon and the president has raised eyebrows among spectators. Bannon has brought an army of conservatives and far-right leaders to Alabama to shore up support for Judge Moore. At the same time, Trump and Senate Majority Leader Mitch McConnell have been pulling (and moving money) for Strange since the very beginning.
Even more peculiar is that Trump picked Strange in the first place. Ideologically, Trump and Moore are much closer aligned that him and Strange; both are populist politicians who built their careers off of countering the establishment and promising to restore the roots of American success. This conventional wisdom about Trump and Moore is borne out of the data, too. looking at the relationship between Trump’s 2016 primary performance and Moore’s share of the vote in round one of the Senate race, a pretty clear relationship emerges; where Trump did better, so did Moore.
To be fair, Moore’s vision for America has always been a tad more conservative than Trump’s — the judge famously denied to allow same-sex couples to marry after the Supreme Court struck down laws that forbade it in 2015 — and certainly more religious. Nevertheless Trump endorsed the establishment candidate in a largely outsider political environment of Republican primary elections. That is unlikely to pay off for him, at least electorally.
It could be that Trump will receive something in return from McConnel and the gang for supporting their candidate. In light of recent events, however, like the inability for the Grand Old Party to pass health care reform, whatever quid-pro-quo Trump gets probably won’t be worth the pain. That’s bad news for the president if the polls are right in hinting that his candidate will probably lose the race.
|Pollster||Date (%)||Moore (%)||Strange (%)||Lead (%)||Source: Various Pollsters|
|Strategy Research||Sep. 20||54.0||46.0||8.0|
|JMC Analytics||Sep. 27||47.0||39.0||8.0|
My average of polls taken in the final week of the race pegs the President’s candidate ten points behind the guy rolling with the alt-right. Of course, polling is not gospel; we can reasonably expect that the polls should err, maybe by quite a bit. In round one of the GOP primary, polling missed the mark by an average of 6%. Polls of other special elections this cycle — like that in Georgia’s sixth Congressional district, Montana’s At-Large district, and the like — have been off by similar margins.
Taking the average of those polling misfires (really, it’s the root-mean-square error), we can expect a margin of error of roughly ±14% on Tuesday. An error that large would be more than enough to elect Luther Strange instead of Roy Moore, but would only happen about 15% of the time. Under normal circumstances — as in, one where we have a large sample size of elections to calculate an error — I would leave this 15% chance alone. But we don’t have a lot of past elections for which to craft our assumptions about polls of Senate primaries.
We can use Bayes’ theorem to adjust for lack of knowledge. Often when statisticians don’t have enough information to make a thorough conclusion they use one of Bayes’ tools called an uninformative prior. Simply, we take the number of times we expect an event to happen plus one and divide it by the total number of occasions in our sample plus two Since we think Roy Moore may have an 85% shot at winning, that math works out to (8.5 + 1) / (10 + 2) = 0.79, or 79%. Instead of saying Moore has an 85% chance of winning based on polls, Bayes’ theorem tells us we can more reasonably expect that chance to be more like 79%.
But we shouldn’t stop there; though the math says he is the clear front-runner, Moore has other weaknesses that could increase the likelihood of a Luther Strange upset victory. Still, I wouldn’t put his chances anywhere above 30%. He is a true underdog.
In order to pull off a ten-point upset, Strange will have to look for big gains in Montgomery and Mobile counties. These are the counties in Alabama that contain the most opportune votes for Strange; they’re the largest counties that he didn’t win. He would still need to rack up huge margins in other urban areas, however, as the voters outside the cities voted for Roy Moore 46% to Strange’s 31% in the August round of the primary. Given the amount of money that the Republican Leadership Fund spent on his candidacy, an upset of this magnitude would not be unheard of. In fact, outside groups supporting Strange have outspent those on Moore’s side nearly eight-to-one in the last two months of the campaign. That’s an incredible spending margin for a losing campaign. If Roy Moore has a weakness it’s this.
Needless to say, Strange could win, but polls and votes say that his path to victory is pretty darn tough. Outside expenditures are his only real silver lining.
Alabama’s Senate Republican primary race is coming to a close, but not without a bang. If Roy Moore really does best Luther Strange, we’ll know a lot more about how deep the divide within the Republican party — the chasm between the establishment and the conservative right — is growing. Polls say we can expect that to happen even if the race goes against much of what we know about how parties choose candidates. Then again, if we’ve learned anything from the past two years of campaigning, it’s that what we thought we know about party nominations may, in fact, be pretty wrong.
Roy Moore has a ten point lead on Luther Strange in the race for Alabama’s Senate Republican nomination. We’ll see if that pans out on Tuesday evening. Polls close at 7:00 PM CDT.
As always, I’ll have live updates and projections of the race running in a Google spreadsheet. Commentary will flow through my Twitter profile and I encourage you to follow along and send me your comments. If polls are right, we won’t have a long night — but we’ll have fun anyways! Here’s the spreadsheet.
- R for Political Data Science Week 12: Do Voters Still Care About The Economy?
- R for Political Data Science Week 11: Is Beto O'Rourke the Media's Sweetheart?
- R for Political Data Science Week 10: What If Each State Allocated Their Electoral College Votes Proportionally?