It certainly would be nice to know if there’s a political price paid when a candidate allegedly assaults a reporter. I’m speaking not just as a member of the media here, though that status certainly does predispose me toward a higher level of sympathy for a reporter being allegedly thrown to the ground simply for asking a question. Generally speaking, regardless of victim, it seems as though such an allegation — especially an allegation corroborated by eyewitnesses, an audio recording and a subsequent apology from the politician — should influence how people vote.

In the case of Greg Gianforte, the Montana Republican accused of just such an assault the day before winning his election on Thursday, it’s very hard to suss out how his (alleged) actions affected the outcome.

Gianforte was lucky enough to (allegedly) assault the Guardian’s Ben Jacobs after the majority of ballots had been cast in early voting. For most voters, the assault had no effect on their vote — because they’d already voted. The timing of the (alleged) assault does offer one way in which we might measure the effects of it, though: By comparing the early vote [pre-(alleged)-assault] to the vote on Election Day itself, we might be able to determine how much the vote moved.

Unfortunately for us, though, the state doesn’t release data broken out by when the vote was cast. (A request for such information from the secretary of state in Montana wasn’t returned.) Fortunately for us, the good people at the Associated Press were able to determine early voting data for six of Montana’s 56 counties: Flathead, Gallatin, Lewis & Clark, Missoula, Sanders and Yellowstone.

In the early vote in those six counties, Democrat Rob Quist had a small, 0.5 percentage point lead. On Election Day? The lead grew to 5.8 points.

Boom. There was a price to be paid, after all.

Or … was there?

One problem with this analysis is that it’s impossible to tell how much of the change between the early vote and day-of voting was a function of the assault. There are long-standing differences between the sorts of voters who vote early and those who vote at their polling place that could also affect what happened.

What’s more, the shift wasn’t evenly distributed. In four of the six counties for which the AP had early-vote data, Gianforte did worse on Election Day. In two, he ended up doing better.

What’s more, these six counties were more heavily pro-Quist than the rest of the state. In these counties, Quist won by 1.8 percentage points. Across the state, he lost by more than six.

So what was the effect of the (alleged) assault? We don’t know. I’d be perfectly happy, though, if we didn’t run more experiments of this sort to try to determine an answer.