Alabama’s long and strange special election for U.S. Senate comes to a close on Tuesday. Democrat Doug Jones and Republican Roy Moore face off in their bids to fill the seat vacated by Attorney General Jeff Sessions for the remainder of the term. Moore is the favorite, according to the polls, but Jones is just a normal polling error away from winning.
Let’s talk about where the race stands, what to keep an eye on after the polls close at 8 p.m. Eastern time and how today’s results could affect the national political landscape.
1. The race has been volatile
President Trump won Alabama by 28 percentage points in 2016. Alabama is a very red state, so this race really shouldn’t be close. Moore, however, was a divisive figure in Alabama politics before this campaign. He was already significantly underperforming Trump’s margin in surveys before The Washington Post first reported, on Nov. 9, that Moore had been accused of sexual misconduct. Since that Post report, the polls have swung to Jones and back to Moore again, and now they’re a bit muddled.
|POLLSTER||DAYS SINCE ACCUSATIONS||DOUG JONES||ROY MOORE||MOORE MARGIN|
The polling swings have roughly tracked with voter interest in the Moore scandals. According to Google Trends data, the volume of Alabamians’ searches for Moore dropped off about a week after the Post’s Nov. 9 story. That lines up well with when Jones’s polling surge stopped and Moore began to recover.
Moore was also probably helped by President Trump giving him a quasi and then a full-throated endorsement. Surveys taken before and after Trump got involved show that Moore’s margin over Jones among Trump supporters jumped by 10 to 15 percentage points. Given Trump’s national unpopularity, that jump might not have been a big deal in many states, but it’s a huge deal in Alabama, where Trump is still fairly popular.
All told, an average of polls conducted over the final 21 days of the campaign has Moore up by a few points, 49 percent to 46 percent. But there’s a big spread in results, everything from Moore +9 points to Jones +10 points. These divergent results make sense because …
2. This is a tough race to poll
Senate polling conducted within three weeks of an election has historically been decent, though by no means perfect. Since 1998, Senate polling averages that use at least 10 polls (like the Alabama race) have been off by an average of 3.5 percentage points. In other words, an average polling error in his favor could hand Jones the win.
There’s also reason to believe that there’s a higher-than-usual chance for error in Alabama. Perhaps because of the lower turnout usually associated with elections that are held on days other than the traditional November Election Day in a midterm or presidential year, Senate polls for these races have had an average error that’s nearly a point higher than the average error for those November races.
Indeed, no pollster really knows what turnout is going to look like in Alabama. In addition to the fact that this is a special election occurring two weeks before Christmas in an off-year, there hasn’t been a major competitive statewide general election in Alabama since 2002. SurveyMonkey, The Washington Post and YouGov all have Moore doing better among likely voters than registered voters. Fox News shows the opposite. SurveyMonkey has also demonstrated how weighting for different factors (including past voting patterns) can make a big difference in who a poll shows as being ahead. (My colleague Nate Silver wrote a more in-depth story about what’s going on with pollsters’ assumptions.)
Additionally, the way pollsters are conducting their surveys seems to be having an impact on their findings in Alabama. Only three pollsters who meet FiveThirtyEight’s gold standard1 have conducted polls in Alabama since Nov. 9: Fox News, Monmouth University and The Washington Post. Fox News gave Jones his largest lead of the campaign. Monmouth and The Washington Post surveys were more favorable to Jones than the average. Traditionally, gold-standard pollsters have been more accurate than other pollsters.
It’s also possible, though, that the non-gold-standard polls will be more predictive in this campaign. To meet the FiveThirtyEight gold standard, one thing a pollster must do is use a live interviewer to conduct its surveys. It’s plausible that some voters may not want to admit to another person that they plan to cast a ballot for someone accused of child molestation. They may feel more comfortable saying they are voting for Moore to a recorded voice or to a computer, which is how the vast majority of polls in Alabama have been conducted.
Either way, there’s a lot of reason to be uncertain about how predictive the Alabama polls will be.
3. Jones must win big in the cities — really big
No Democrat has won a statewide race in Alabama since 2008. During that losing streak, Democrats have tended to do their best in two types of places: counties where a large share of the population is black and cities.
But to win the Alabama Senate race, Jones will have to do even better than a typical Democrat — after all, Democrats usually lose in Alabama. He’ll probably need to run up even larger margins than normal in the black belt and win handily in counties with significant population centers, such as Jefferson (Birmingham), Madison (Huntsville), Mobile and Montgomery. Beyond that, it’s difficult to pinpoint how well Jones will need to do in each county to win — we don’t have a string of competitive statewide elections to get a sense of the baseline.
So instead of one baseline, let’s use three that each share some characteristic with today’s election: a federal race, a race with Moore on the ballot, and a combo of the two. Below, I’ve created a table of the three county benchmarks, estimating how well Jones needs to do according to each to win the statewide Senate race. The first is based on the 2016 presidential election results. The second is based on the vote share margin between the 2012 Democratic candidate for state Supreme Court chief justice and Moore, who won that race. And the third is an average of the first two.2 To take one example, if Jones is winning Madison County (average benchmark Jones +9.5 points) by more than 9.5 points, that’s a good sign for his chances of winning statewide.
||SHARE OF 2016 VOTE▲▼
||2012 SUPREME COURT▲▼
Broadly speaking, the three benchmarks are similar. However, there are some differences. For one, the 2016 benchmarks show that Jones needs to do a lot better in heavily black counties like Bullock, Dallas and Macon than the 2012 benchmarks show. The 2012 benchmarks have Moore underperforming the usual Republican vote share in his home county of Etowah and the surrounding Cherokee and DeKalb counties, so if you go by those, Jones needs to do better in those areas. The 2012 benchmarks also have Jones needing to make up a disproportionate amount of ground in counties that are on the periphery of statistical areas (as defined by the federal government) such as Jackson and Lawrence, which are in the Chattanooga, Tennessee, and Huntsville, Alabama, statistical areas respectively.
It’s not clear which of these provide the best baseline. On the one hand, 2016 was a federal election, as is Tuesday’s election. On the other, Moore is a unique candidate, so it could be best to look at an election in which he was a candidate. For now, the best bet is probably to look at some average of the two.
4. This election has national implications
There will be a lot of talk about what a Moore win or loss says about the Republican Party. But the most immediate potential effect of the election would be from a Moore defeat, which would: (i) make Republican efforts to pass legislation in the current Congress more difficult and (ii) make Democratic efforts to take over the Senate in 2018 easier.
Republicans currently have 52 seats in the U.S. Senate. A Jones win puts that majority at just 51 seats. That means that Republicans could afford to lose only one GOP vote on legislation3 — two would be enough to sink any bill that is universally opposed by Democrats. Given all the trouble that Republicans have had passing legislation with a two-seat margin for error, it’s likely that they would struggle even more with a one-seat cushion.
Meanwhile, in 2018, Democrats hope to take control of the Senate, but they have limited opportunities. There are only two Republican-held seats up for election in states where Democrats are normally competitive: Arizona and Nevada. If Republicans have 52 seats heading into next year, Democrats will have to win those two seats and at least one additional GOP-held seat to take control of the Senate in 2019. If, however, Jones wins, the path to a Democratic-controlled Senate in 2019 gets a lot easier.
Finally, no matter who wins, Alabama’s Senate race will almost certainly be another special election this year in which Democrats outperform the presidential lean of the district or state. Although there are obviously some particular circumstances in Alabama, that Trump’s approval rating (even if still relatively high in Alabama) is below his percentage of the vote in last year’s election hasn’t helped Moore’s cause. Moore’s likely very poor performance versus Trump’s 2016 showing adds to the evidence that the national environment favors the Democrats heading into 2018.