Crowds cheer as Hillary Clinton delivers her keynote address at the Democratic National Convention in Philadelphia on July 28, 2016. (Michael Robinson Chavez/The Washington Post)

In July 2020, the Democratic National Convention will go where it has never gone before: Wisconsin.

Funny, because — to hear most people tell it — neither has Hillary Clinton. And maybe that’s the point.

What’s the “Wisconsin narrative”?

Clinton’s failure to visit Wisconsin in 2016 has become a convenient shorthand for explaining why she lost to Donald Trump in that year’s presidential election. White House adviser Kellyanne Conway cited it when taunting Clinton on Twitter. Political commentators have done the same at the Conservative Political Action Conference, or CPAC. Late-night comedians have made this a punchline.

The logic of the “Wisconsin narrative” is simple. Trump defeated Clinton in the electoral college because he won Wisconsin, as well as Michigan and Pennsylvania, by less than one percentage point. But Clinton never even visited Wisconsin during the campaign. She took victory there — as in the election, generally — for granted. No wonder she lost!

But as I explain in a recently published article, this criticism rests on two uncertain assumptions: first, that campaign visits increase a candidate’s vote share, in general; second, that Clinton’s campaign visits, in particular, did so in 2016. Of course, we cannot replay the election to see whether Clinton would have won had she visited Wisconsin. But testing these two assumptions can give us a better idea of whether it would have made a difference.

What if Hillary Clinton had gone to Wisconsin?

My research uses an original database of presidential campaign visits in 2016 to estimate their effect on vote share, at the county level. I estimate these effects across all major battleground states, as well as within individual states, and for each of the presidential (Clinton and Trump) and vice presidential (Mike Pence and Tim Kaine) candidates.

What do I find?

First, across battleground states in general, I find no evidence that campaign visits had a statistically significant effect on county-level vote share — not even for Trump!

But within specific battleground states — Florida, Ohio, North Carolina, Pennsylvania, Michigan and Wisconsin — there is some evidence that campaign visits influenced voters. Or at least there are two examples. The first is Pence, whose visits to Ohio apparently won votes for the Republican ticket.

The second is Clinton. But only in Pennsylvania. There, Clinton’s visits are associated with a statistically significant increase in county-level vote share of about 1.2 percentage points per visit.

These results suggest that Clinton’s absence from Wisconsin — in and of itself — did not cost her a victory in that state. After all, her campaign visits in general and in all but one state had no discernible effect on local voters. Nor do I find evidence that the candidates who did visit Wisconsin — Trump (5 visits), Pence (4) and Kaine (6) — earned votes by doing so.

But it is possible that Clinton might have won votes by visiting Wisconsin, since she appears to have done so in Pennsylvania. For that matter, perhaps Clinton would have won Pennsylvania if she had visited there more often.

The problem with this last point is that Clinton did visit Pennsylvania. A lot. Eighteen times, to be exact. The only state that she visited more often was Florida, with 23 visits — and she lost there, too. So if the argument is that Clinton lost Wisconsin because she didn’t go there during the campaign, then how do we explain Pennsylvania?

Moreover, Clinton did not lose the election because she lost in Wisconsin. To win a majority in the electoral college, she also would have had to win in Pennsylvania and Michigan (which she visited five times). Any serious attempt to explain Clinton’s loss to Trump in 2016, then, also must account for her performance in those states — which campaign visits, alone, cannot explain.

How the 2020 convention in Milwaukee could help Democrats win Wisconsin

So does the “Wisconsin narrative” explain the Democratic Party’s choice to hold its 2020 national convention in Milwaukee? We cannot know for sure. But probably.

At least, this is how nearly everyone is interpreting that decision, judging by recent media coverage. In the words of Wisconsin-based political commentator Charlie Sykes, “the DNC’s decision to come to Milwaukee was … [about] a commitment to voters that they will show up this time.”

My research suggests that this is an overreaction to the ubiquitous Wisconsin narrative. Nevertheless, it may pay off for Democrats.

Why? Because — according to research by political scientists Matthew D. Atkinson, Christopher B. Mann, Santiago Olivella, Arthur M. Simon and Joseph E. Uscinski — national party conventions can win a party votes in a state, or at least in the host county. Of course, parties often lose the state in which they hold their convention. This was the case for Democrats in Pennsylvania, for instance, in 2016.

But, according to that study, parties do tend to win a higher vote share in the convention’s host county when (and here’s the caveat) the county in question already tends to vote for that party in presidential elections. Essentially, this is because intense media coverage of the event tends to strengthen local voters’ partisan predispositions. And since Milwaukee County voted 65 percent for Clinton in 2016, it seems likely that holding the Democratic convention there in 2020 will gain some votes in Wisconsin for the party’s presidential nominee — whoever she or he may be.

How much of an effect will this have in 2020? The research suggests that Democrats could gain about five percentage points in Milwaukee County because of the convention. In a close election, that may be enough to tip the balance in Wisconsin. But, then again, what are the chances that Wisconsin, alone, will decide the 2020 election? It didn’t in 2016.

Still, the Wisconsin narrative persists. More than that, it seems to be setting the narrative for 2020.

correction: Hillary Clinton won 65 percent of the vote in Milwaukee County in 2016. An earlier version of this article gave an incorrect figure. We regret the error.

Read more:

Mia Love is right. The loss of black Republicans in Congress really could hurt the party with black voters.

Secret groups of Democratic women are organizing deep in Republican territory.

People used to joke about ‘Democrats in disarray.’ They’re not joking now.

How the tea party paved the way for Donald Trump.

Christopher J. Devine (@ProfDevine) is an assistant professor of political science at the University of Dayton.