Whatever else one might say about the tragicomic 2016 Republican primaries, nobody can deny that they’ve attracted a lot of attention. The debates have garnered remarkably high ratings—perhaps because so many people enjoy watching Donald Trump’s act, whether or not they’re considering voting Republican. But Republican turnout in the primaries and caucuses is also up a great deal, and has outpaced Democratic turnout considerably. So far, almost five million more voters have shown up for Republican primaries and caucuses than for Democratic contests. 

To Donald Trump, the lesson is clear: He’s capable of bringing large numbers of new voters into the Republican Party. At the March 3 debate in Detroit, the mogul declared that he was “very, very proud” because “millions and millions of people have come to the Republican Party over the last little while.” Trump, who has shallow roots in the GOP and is viewed unfavorably by upwards of 70 percent of the public (and “very unfavorably” by half), is a very weak presidential candidate by most conventional measures. But should Democrats worry that his ability to mobilize voters will overcome these liabilities, particularly given that likely Democratic nominee Hillary Clinton also is viewed unfavorably by a majority of voters? 

Democratic Party elites shouldn’t be high-fiving each other,” The Huffington Post’s Zach Carter wrote in late February. They should be very, very worried.” The social science, however, says they shouldn’t fret. As both Harry Enten of FiveThirtyEight and Barry Burden and Jordan Hsu of The Monkey Cage have recently explained, there is no evidence that turnout in the primaries is correlated with turnout in general elections. Turnout in primary elections is so low (compared to that of general elections), that it can increase substantially without indicating that the party is attracting significant numbers of new followers. As Enten points out, it’s not really surprising that the free-for-all Republican primaries have attracted a higher number of voters than the Democratic race, which despite Bernie Sanders’s unexpectedly strong challenge has always had a clear, well-known frontrunner in Hillary Clinton. 

Based on past patterns, then, Democrats have nothing to worry about. But, of course, if past patterns were an ironclad guarantee, then Donald Trump wouldn’t be the most likely Republican nominee in the first place. The best social-science analysis indicated that Trump’s lack of support among party elites would doom his chances at the presidency. If the 2016 campaign has taught observers any lessons, it’s not to be wary of history repeating itself, it’s not to expect history to repeat itself.

Trump’s ascendance should remind us of two things about the political science of elections. First, we’re dealing with a small “n” (i.e, a small sample), because the modern primary system is a relatively recent phenomenon. The small number of competitive primaries means that what appears to be a strong pattern tends to be less robust than it first appears. And, second, the political universe is dynamic rather than static. Historically, for example, Congress has been roughly as productive in times of divided government as it has in times of unified government. But it would be unwise to bet on this trend continuing, since more disciplined and polarized parties, combined with a steep decline of norms that historically have compelled congressional leaders to make deals with presidents from the other party, mean that divided government is very likely to be less productive than unified government going forward.

With those caveats, I think it’s likely that this is one case where the findings of social science will hold. Donald Trump will almost certainly be a very weak general-election candidate. Because his chief appeal is to voters who are already Republican-leaning, there’s little or no chance he can attract significant numbers of new voters to the Republican coalition. And those voters he does attract are likely to be outnumbered by the counter-mobilization against him. In fact, the anxious Democrats have it backward: It’s likely that Trump’s presence on the ballot will increase Democratic turnout, not Republican turnout, in November. 

The argument that Donald Trump can transform the Republican coalition runs something like this. With its relentless focus on upper-class tax cuts and reductions in middle-class entitlements—which even conservative voters mostly don’t want—the GOP has shackled itself at the national level to a declining base of voters. Trump, by saying he would leave Social Security alone and expressing skepticism about trade deals, has a populist message that could bring white working-class voters into the Republican fold. (Trump also supports a massive upper-class tax cut, although he hasn’t emphasized it.) Combine the populism with his unique ability to get across to people who don’t typically pay attention to politics, and Trump might be able to produce a Republican majority.

The most obvious flaw in this argument is that there just aren’t that many “Reagan Democrats” who still vote for Democrats—people who Trump can sway to the GOP. White voters with some college or less already overwhelmingly vote Republican. And Trump’s sky-high unfavorable ratings put him at a disadvantage in terms of attracting educated professionals who aren’t already voting for the GOP. The most promising terrain for new white voters—and for flipping states won by President Obama—is in the Rust Belt. But to carry the Rust Belt states (with the exception of Ohio), Trump would have to win a much, much larger number of white voters than Mitt Romney did in 2012Since the white voters whose message Trump is most likely to appeal to are already voting Republican, this seems massively unlikely.

Perhaps Trump could succeed not so much by attracting new Republican voters as getting more Republican-leaning voters to the polls? This won’t be easy, either, as Republicans reliably turned out for Romney in 2012. It’s true, as purveyors of the “missing Republican voters” theory say, that the Republican coalition hasn’t been growing in recent years—so theoretically, at least, there’s room to add new voters. But any increase in white turnout is likely to be counterbalanced by the way Trump will mobilize minority voters, and particularly Hispanics, to the polls for Democrats. We can’t know for certain until the election happens, but it seems likely that Trump would will compel more Hispanic voters to turn out in opposition than Romney did. That will increase the number of new white voters Trump would need to attract. 

This aspect of a Trump nomination is particularly important, because it has effects that will extend beyond 2016. It’s worth remembering that George W. Bush lost Hispanic voters by only 18 points in 2004. Two cycles later, Romney and his “self-deportation” message lost them by 44. A Trump nomination would only accelerate this trend—and once people start identifying as party supporters, they become unlikely to change back. 

If Trump gets the nomination, “anti-immigrant and particularly anti-Mexican-immigrant rhetoric is very likely to mobilize opposition on the part of the Latino community in November,” said Hannah Walker, a doctoral fellow at the Washington Institute for the Study of Inequality and Race at the University of Washington. Scholars have documented several examples of similar counter-mobilizations in the Latino community. Most notably, mid-’90s California initiatives directed at immigrants and bilingual education and Mitt Romney’s “self-deportation” comments in 2012 seem to have mobilized Latino voters to turn out, and the passage of the anti-immigration “Sensenbrenner bill” in 2006 led to huge opposition rallies. It seems very likely that a Trump nomination would similarly motivate Latino voters to come to the polls in opposition.

There’s an additional problem for Trump and the Republicans: gender. Hillary Clinton has some serious liabilities as a candidate—the aforementioned poor favorability ratings, for starters, as well as being an ultra-establishment candidate in a year in which voters appear to be in an anti-establishment mood. But as the New Republic’s Elspeth Reeve has observed, one of Clinton’s strengths as a candidate is her ability to give blustering sexists enough rope to hang themselves. She did this to considerable effect in her 2000 Senate campaign against Rick Lazio, and Trump’s misogynist condescension dwarfs Lazio’s. It seems likely that some number of affluent, Republican-leaning women might be inclined to vote for Clinton after seeing more of Trump’s act. 

None of this is to say that Democrats should be overconfident as they look toward November. Trump would have to be a not merely bad, but positively catastrophic, candidate for the Democrats to have a significant chance of retaking the U.S. House; even winning the handful of seats needed to regaining the Senate majority (particularly crucial given the vacant Supreme Court seat) won’t be easy. Under the current partisan alignment, any major-party candidate can potentially win the presidency under the right conditions (although not only Trump, but Ted Cruz, is unattractive enough to test the limits of this). But the Democratic Party currently holds a structural advantage in the Electoral College, and nominating Trump would probably make it even harder for the Republicans to assemble a majority of voters.

Trump primarily appeals to a shrinking, already-Republican slice of the electorate—and he alienates the pool of voters that’s growing. Barring an economic or military crisis, it’s hard to see his path to the White House. There isn’t an obvious opportunity for him to bring a lot of new voters into the Republican fold, and he is likely to both increase the turnout of Hispanic voters and make them even more predominantly Democratic. For both 2016 and beyond, that doesn’t look like a winning Republican formula—no matter how many folks are turning out for the GOP primaries.