Lots of factors in the mix here but I think the likeliest explanation for the shift is the dullest one. Simply put, the story’s gotten a bit stale over the holiday. When an alleged scandal first breaks, it hits like a roundhouse; the public recoils from the subject of it instinctively but then begins to process it. For many people the process in a red state like Alabama is destined to mean lots of day-to-day contact with Moore fans making the case in conversation that the accusations are garbage. Populist conservative media is all-in for Moore too. Slowly, some (but maybe not all) of the fencesitters who were bothered by the stories about teen girls initially will come back to him.

There haven’t been any new allegations recently either, which is important. Timing matters with these things. Trump shook off a polling downturn after the “Access Hollywood” tape emerged a month before the election but Hillary couldn’t shake off the publication of James Comey’s letter 10 days before the polls opened. If WaPo really did have an electoral agenda in defeating Moore, it would have held the story about Leigh Corfman until a week before the vote. As it is:

Two weeks ago, after WaPo’s story was published, Change Research had Democrat Doug Jones up three. Since then nothing has happened to discredit Moore’s accusers, just the passage of time and the public’s interest moving on to other topics. Today: Moore 49, Jones 44.

What has changed? The largest difference is turnout: many Republicans who ten days ago said they might not vote, now say they plan to show up on Election Day and vote for Moore. In mid-November, 82% of those who voted for Donald Trump in 2016 said they would “definitely” vote on December 12; that number has climbed to 88%. Additionally, Moore has made some gains with his base: his 91–5 lead with them ten days ago has grown to a 93–4 edge. In mid-November, 10% of voters had planned to cast a write-in vote; that number has dropped to 7%…

Compared to ten days ago, fewer Republicans believe the allegations against Moore. While all voters believed the allegations by a 46–30 margin ten days ago, they now believe them by only 42–38. Among Trump voters, the split was 16–51 (believe-don’t believe) in the middle of the month, and it’s 9–63 now

As was true in our previous polling, those who think the allegations are false say that very little could make them change their minds. 2% said they might believe them if more accusers come out; 1% would believe them if the President says that he does. 97% say the accusations are garbage, and nothing would make them believe they are true.

Ninety-seven percent. Some of those people are exaggerating, of course; if Leigh Corfman produced a photo of her 14-year-old self in her underwear with Moore, some opinions would change. But 97 percent is a big number to start with when we’re talking about allegations of sexually assaulting a 16-year-old and molesting a 14-year-old. In hindsight, Trump should have said that he could shoot *or molest a kid* on Fifth Avenue and he wouldn’t lose any voters.

As I say, there are other factors. The Franken and Conyers scandals and Pelosi’s pitiful dithering about the latter have probably pushed some reluctant Moore voters firmly back into his camp. Trump’s tepid re-endorsement of Moore on grounds that the GOP agenda can’t survive another Democratic seat in the Senate moved some voters too. (But not many. Only 3.5 percent of Trump voters said his support for Moore made them more likely to support Moore.) In any election there’s a tangle of motives for why voters vote as they do: In the case some sincerely don’t believe Moore’s accusers, others think he may have sinned but surely has since atoned, others believe that the GOP is under no obligation to play by a loftier set of rules than the ones Democrats follow themselves.

But there’s another terrible possibility. I think John Podhoretz has hit on a dark truth here:

In JPod’s formulation: “I am so committed to my views that I am being tested and I will pass the test.” I haven’t articulated it that way but I’ve had the same sense that, paradoxically, the more accusers come forward, the *less* likely Republicans will be to desert Moore. That’s the opposite of logic but it makes sense from the nihilistic perspective Podhoretz describes. The more the evidence mounts, the more your partisan faith is being tested and the more devotion you must show. Since the last poll conducted by Change Research, outlets like AL.com and the NYT have run stories quoting local residents who remember Moore pursuing teenaged girls. It may be that those stories helped bind wary Republicans closer to him when, in a saner world, they should have done the opposite.

In lieu of an exit question, an intriguing catch by Will Saletan at Slate:

I’d forgotten all about that Moore soundbite. When he said it at the time, I assumed he was just blowing smoke. There would be no “investigations,” it’s just something he was telling supporters to encourage them to remain skeptical of the WaPo story. But what if Moore’s campaign was coordinating with Project Veritas on the busted WaPo sting? Sending a fake rape victim in to try to discredit a paper’s reporting on a woman who claims to have been molested by Moore as a child would be ultra-sleazy. And if JPod is right, Team Moore’s role in it would actually help him gain votes. He’s trying to defeat the pharisees who are challenging people’s Republican faith. They can’t waver.