ALABAMA has seldom been of much interest to pollsters. No Democrat has held statewide office since 2010, or been elected to the Senate since 1990. Expecting another snoozer, pollsters largely ignored the special election to replace Jeff Sessions, now the attorney-general, to be held on December 12th. That changed after the Republican nominee, Roy Moore, a Bible-thumping former judge, was accused of offences ranging from sexual harassment to assault decades ago by eight women (many of whom were teenagers at the time). Mr Moore, who has strenuously denied the allegations, saw his polls plummet as prominent Republicans withdrew their endorsements and optimistic Democrats sent a flood of money to his opponent. Since then Mr Moore’s numbers have stabilised. Nearly 90% of Republicans in the state approve of the president, and support for Mr Moore in the primaries in Alabama’s 67 counties was strongly correlated with the Trump vote in 2016.
Still, Doug Jones, the Democratic candidate, has a chance of victory. An average of publicly available polls gives Mr Moore a lead of only 2.3 percentage points. But this estimate suffers from three big sources of uncertainty. The first is that polling for Senate races is especially unreliable: The Economist’s review of historic polling for such races from 1998 to 2014 suggests an average error of six percentage points. Second, many pollsters do not construct rigorously representative samples in Alabama, relying instead on lower-quality methodologies (like focusing on voters with landlines and ignoring those with mobiles). Third, all polls implicitly forecast which voters will show up on election day—something especially hard to do in a low-turnout special election.
According to conventional wisdom, Mr Jones needs a high turnout among black voters in order to win. They typically make up a quarter of the electorate and heavily tilt towards the Democrats. He is spending the last days of the campaign attending fish fries at black churches, emphasising his prosecution of two KKK members who bombed a church in Birmingham and appearing with John Lewis, a civil-rights icon. The effectiveness of such efforts remains uncertain: after decades in the political doldrums, the Democrats in Alabama are no longer good at mobilising voters in statewide races. According to a knowledgable source, the campaign hopes to push black turnout to 27% of the total to be in with a chance. Evidence from special elections earlier this year is not encouraging: black voters have made up a smaller percentage of voters than usual, according to numbers crunched at The Economist’s request by 0ptimus, a Republican-leaning data firm.
Instead, the key to a Jones victory is turnout among whites, which itself depends on whether the scandals around Mr Moore keep people away from the polls. Among white voters, who favour him by a whopping 35 points, roughly one-third seem to be persuadable by Mr Jones. A recent survey showed that 71% of Republicans believed the allegations against him to be false (compared with 37% of Republicans nationally). The same poll shows the mind-addling effects of partisanship: nearly one-quarter of white evangelicals in Alabama believes it is legitimate to defend sex with minors on Biblical grounds. The Jones campaign is running advertisements encouraging voters to spoil their ballots, as Richard Shelby, Alabama’s senior Republican senator, has already declared he will. Faced with the choice between a Democrat and someone accused of sexually assaulting teenagers, some might at least compromise by staying at home.