On Election Day 2004, Donald Trump — then a private citizen and a registered Democrat — went to vote. In accordance with his status as a television star, “Access Hollywood” tagged along, with host Billy Bush watching as Trump tried repeatedly to figure out where he was supposed to cast his ballot. The result was a brief segment in which Trump ultimately cast a provisional ballot, yielding the second-most politically important snippet of tape involving Bush and Trump.

The point is that Trump clearly didn’t vote in person very often. As president, he still doesn’t, voting by absentee ballot even when it’s not necessary. He’s not alone. White House press secretary Kayleigh McEnany also generally votes by absentee ballot, as does Trump’s senior adviser Kellyanne Conway.

Normally such details are unimportant, if not expected. But Trump, McEnany and Conway have all spent the past week or two disparaging voting by mail, bolstering the president’s insistence that mail-in ballots are rife with fraud. They aren’t.

The White House has tried to differentiate between their requested absentee ballots and ballots sent by states to voters to reduce in-person voting in the middle of a global pandemic. Ultimately, though, states have mechanisms for establishing the validity of both ballots, reducing the utility of the distinction.

Of course, it’s not only Trump, McEnany and Conway who vote absentee. Washington Post analysis of current voter registration data from L2 Political shows that more than 9.2 million currently registered Republicans cast absentee ballots four years ago. Rules governing absentee ballots vary by state, but in many states, absentee ballots made up a significant portion of the vote. And, given that most Republicans voted for Trump, that means that absentee ballots probably made up a big chunk of Trump’s own votes in the presidential election of 2016.

We can make a rough comparison of the actual votes cast for Trump and Democrat Hillary Clinton in each state with the number of absentee ballots cast by likely Democrats and Republicans. The results look like this.

Data weren’t available for every state, but we’ve highlighted three particularly interesting states for which it was: Michigan, Utah and Wisconsin. There’s some imprecision here; since Michigan and Wisconsin don’t have partisan registration, L2 uses sophisticated modeling to determine the likelihood of a voter’s partisanship. But in those states and Utah, the margin by which Trump won is actually smaller than the difference in the absentee votes likely cast by Democrats and Republicans.

In other words: These are states in which absentee votes could theoretically have given Trump a win.

Let’s look at Michigan and Wisconsin. Those states were two of the three that were central to Trump’s electoral college victory, flipping from blue in 2012 to red in 2016 and giving Trump the White House.

In Michigan, Trump’s margin of victory was nearly 11,000 votes. Currently registered voters in the state who are likely Republicans cast more than 45,000 more absentee votes than likely Democrats. Assume all of those voters voted on party lines and remove them? Clinton wins Michigan.

The numbers in Wisconsin are smaller, but the math is similar. Trump won by about 23,000 votes, with likely Republicans casting about 50,000 more votes than likely Democrats by absentee.

There are a ton of caveats here. Again, it’s currently registered voters, meaning that the actual figures from 2016 will vary from these. (They would be higher.) It’s also not the case that every Democrat voted for Clinton or every Republican for Trump. Eliminating the absentee vote would also eliminate absentee votes from third-party and absentee voters, which mix up the math. Nor is this enough to have cost Trump the White House; flip Michigan, Wisconsin and Utah and he still prevails by a single-digit electoral-vote margin.

But it does serve as a reminder that, for many Republicans, casting ballots by mail is standard procedure — and was important for Trump in past elections. After all, it’s a lot easier to cast a ballot by mail than to try to figure out where you’re supposed to vote while the cameras are rolling.

Lenny Bronner contributed to this report.