This doesn’t mean that Ms. Warren or Mr. Sanders are doomed to do a few points worse than Mr. Biden. These figures are basically averages, and while Democratic Medicare for all opponents tended to fare better on average, not all did. It is even possible that Mr. Trump is so weak that Democrats could win even if they did alienate centrist voters — though Mr. Trump’s seemingly persistent advantage in the Electoral College may raise the burden on Democrats to appeal to more conservative areas.
And the midterms do not disprove an important argument about how a more left-leaning Democrat might win a presidential election. Many left-leaning Democrats argue, for instance, that they can mobilize a new coalition of young, nonwhite voters who typically sit on the sidelines of politics. These voters are generally unrepresented in the House battleground districts. Even if these kinds of voters were well represented, few House candidates could reasonably hope to mobilize a new coalition of irregular voters in their districts. The sort of less engaged voters who don’t even participate in presidential elections aren’t paying close attention to individual House campaigns.
But the House results are nonetheless a reminder that turnout isn’t everything. Last year, Mr. McCready outperformed Hillary Clinton’s 2016 result in the district by 11 points, even though the partisan composition of the electorate was essentially unchanged and the black share of the electorate was lower than it was in 2016. In our final poll of North Carolina’s Ninth, 52 percent of those surveyed approved of Mr. Trump, and just 41 percent disapproved. The very same respondents backed the Republican Mark Harris by just 0.5 points over Mr. McCready.
In her pursuit of the governorship in Georgia in 2018, Stacey Abrams improved by a modest 3.7 points over Mrs. Clinton’s result in the state, a smaller increase than Democrats nationwide, even though she succeeded in increasing the black share of the electorate. Elizabeth Warren, perhaps surprisingly, ran three points worse than Mrs. Clinton in her Massachusetts Senate victory in 2018.
The presidential election, of course, could be very different. Patterns in congressional voting might not hold up in a higher-turnout, higher-stakes election. The pivotal states could end up differing from the House battlegrounds or the nation in important ways that change the calculus. And Mr. Trump’s divisive style brings its own uncertainties.
It has been argued, for instance, that nonwhite and female candidates might be at a disadvantage against Mr. Trump among swing voters in the Rust Belt, even though there is little evidence that nonwhite or female candidates do worse than white or male candidates as a general proposition. According to our analysis, a candidate’s race and gender were essentially irrelevant to results in last year’s House elections.
Maybe presidential elections or Mr. Trump are different, and maybe the kind of election where race and gender matter more is one where ideology matters less. Maybe mass television appeal and superficial markers of strength, leadership or masculinity would matter more in a high-turnout election, when less educated and less political engaged voters who know nothing about Medicare for all will flock to the polls after a year of watching two candidates dominate television coverage. There is no comparison to a congressional race, so perhaps the congressional results should be taken with a grain of salt.
But there is a difference between not being sure about candidate “electability” and knowing nothing at all about it. The evidence for the advantage of nominating relatively moderate candidates exists all the way to the present, even possibly tonight when the polls close in North Carolina.