Donald Trump is no longer on social media, but that doesn’t mean he can’t cause a stir with a brief message. On Wednesday, reporters received a release from Save America, his cash-rich political action committee. It included a warning for Republicans who didn’t demonstrate sufficient fealty to Trump’s central animating cause.

“If we don’t solve the Presidential Election Fraud of 2020 (which we have thoroughly and conclusively documented),” Trump’s statement read, “Republicans will not be voting in ‘22 or ‘24. It is the single most important thing for Republicans to do.”

It set off an immediate scramble. Here was the Republican Party’s guiding light offering an indirect threat against its candidates. If Republicans don’t “solve” election fraud, he suggests, he’ll just shrug at the fact that people don’t turn out to vote. It’s an impossible demand for a few reasons, of course, the two most prominent being there was no rampant fraud in the election and, therefore, there’s not anything specific that can be done about it.

There was an invocation that popped up over and over: Trump was threatening to do to Republicans nationally what he did in Georgia for the Senate runoff elections in January. At the time, twin Democratic victories were framed as being a function of Trump’s efforts to cast doubt on the presidential election results in the state. The Atlanta Journal-Constitution looked at county-level results and found that turnout declines in January were heavier in some counties that had voted more strongly for Trump in November.

That’s true. You can see it below: The more supportive a county was of Trump in November 2020, the bigger the downward shift in turnout in January 2021. The dots are generally arrayed from upper left to lower right. More Trump support, bigger drop-off.

But this isn’t the whole story. If we look at the shift by county from the 2018 gubernatorial race to the 2021 contest, there’s no such pattern. In other words, the increase in votes cast in the runoff election was pretty uniformly distributed relative to 2018 vote.

In fact, analysis of data on registered voters from the political data firm L2 suggests that there wasn’t a correlation between how a county voted in 2020 and the difference in turnout among Democrats and Republicans in 2021. There are a lot of caveats here — it’s only people still registered, and it’s modeled partisanship, meaning that it’s imputed from other data about voters — but if there was a big tamping down of support among Republicans in 2021, we would probably see it there, too.

The implication, then, may be one that we’ve encountered in the past. It wasn’t the 2021 election that was exceptional, it was the 2020 one. And the differentiating factor was whether Trump was on the ballot.

If you don’t think that the GOP is concerned about turning Trump voters out to vote, you need only look at the gubernatorial race in Virginia. There, Republican nominee Glenn Youngkin is running a two-track campaign, presenting himself to mainstream Virginia voters as a traditional Republican but carefully reaching out to Trump’s base through more-muted channels. One of those efforts, an appearance on former Trump adviser Seb Gorka’s radio show, was elevated by the anti-Trump group the Lincoln Project in an ad this week. But, otherwise, Youngkin’s been “dancing on the edge of a razor,” in the evocative phrasing of a Trump campaign staffer, balancing between the GOP’s two worlds.

Why? Obviously because he very much needs those Trump-supporting voters to transfer their enthusiasm to him. It’s not clear how much of the polling errors in the 2016 and 2020 presidential contests was a function of failing to capture support for Trump, but it’s hard not to notice that the polling in 2018 was right on the mark. There wasn’t a surge of polling-resistant Trump voters who came out that year, presumably helping to keep the polls from fritzing. If Youngkin benefits from energized Trump voters, his position is strengthened greatly.

To date, though, Trump has shown little to no ability to shift his base’s enthusiasm to other candidates in general elections. His party lost in 2018 and in special elections — such as the Alabama Senate race — during his presidency. That was clearly in part because Trump motivated so many voters to come out against his endorsed candidates. One study published last year found that Trump’s vocal endorsement could have prompted enough of a backlash to cost Republicans a number of seats in 2018. If, for example, a number of independent voters came out to vote for Trump in 2020 but not in the Georgia runoff race, that would explain the presidential-election-correlated drop-off in votes cast and the lack of a connection between Democratic and Republican drop-off seen in the L2 data.

Last month, The Washington Post and the Schar School of Policy and Government at George Mason University co-sponsored a poll in Virginia evaluating the state of play around the gubernatorial race. One thing the poll revealed was that Trump’s top-line assertion is wrong: There is no difference in enthusiasm to vote between Republicans and Republican-leaning independents who think the 2020 election was stolen by fraud and those who don’t. There was also no difference seen between those concerned about their vote being counted and those who weren’t.

In other words, there’s no evidence for Trump’s insistence that Republicans who are depressed about 2020 will depress 2022 turnout.

A natural conclusion, then, is that we’re getting Trump’s threat backward. Perhaps it’s not that Trump is threatening to withhold the enthusiasm of his base unless the party does the impossible thing he demands. Instead, he’s using the failure to do the impossible thing as the rationalization for why he won’t be able to turn out his base.

The Youngkin race in Virginia will be a significant test of this thesis. If Youngkin beats the polls and wins with unusually high Republican turnout (as opposed to winning with a measurable decline in Democratic voting), we might learn that there is a Trump voting bloc that can be energized for candidates who aren’t named Trump. But if that base doesn’t turn up, Trump has his excuse ready to go.

Emily Guskin and Lenny Bronner contributed to this report.