Has Roy Moore, the Alabama Republican Senate nominee accused of sexually touching a 14-year-old girl many years ago, profited from the charges that Minnesota Democratic Senator Al Franken sexually harassed a woman performer on a USO trip a decade ago? You could make a case that the answer is yes, from a brief examination of the poll results in Moore’s race against Democrat Doug Jones in the run up to the December 12 special election.

Real Clear Politics lists 16 polls conducted between September 8 and November 28. Four polls conducted between September 8 and October 1 showed Moore leading Jones on average by 48.5 to 38.7 percent. Moore led by a slightly lesser percentage, perhaps due to an increase in awareness of Jones, in three polls conducted between October 14 and 19.

No other reported poll conducted interviews between October 20 and November 9, the day that the Washington Post ran its well-sourced blockbuster story alleging that Moore while in his thirties sought to date high school students and had sexual contact with one girl when she was 14.

In six polls conducted between November 9 and 15, Moore’s lead disappeared, as Jones averaged 46.5 percent and Moore a statistically indistinguishable 46.3 percent. It looked like Jones had a serious, perhaps better than even chance of winning a Senate race in a state that Donald Trump carried 62 to 34 percent over Hillary Clinton.

No polls conducted interviews on November 16, when Los Angeles radio anchor Leeann Tweeden charged that Franken sexually harassed her on a USO trip in 2006. Those charges were accompanied by a photo of a smiling Franken holding the breasts of an apparently sleeping Tweeden.

The three Alabama polls conducted between November 20 and 28 show Moore ahead once again, by a 49.7 to 45.3 percent margin.

Summarizing, Moore led by about 7 percent until the Post story appeared, after which the race became a dead heat, and then Moore jumped to a 4 percent lead after charges appeared about Franken and, in the days following, multiple liberal media figures.

Working hypothesis: Some significant quantum of Alabama voters recoiled away from Moore after the Post story, but then at least some of them went back to supporting him after major Democratic officeholders—Franken and then Congressman John Conyers—clung to office in the face of less serious charges. One suspects that many asked: if their side won’t punish their wrongdoers, why should I punish my side’s wrongdoers? Partisan loyalties reasserted themselves, making Moore appear to be a more likely—though far from certain—winner.

Caveats are in order. Polls have error margins, and averages of results include polls with different interviewing techniques and sampling and weighting procedures. We may just be looking at statistical noise. All that said, my working hypothesis seems at least plausible.