When it’s working, American democracy is a peace process—its institutions were devised to settle radical disagreements nonviolently. On rare occasions, though, the institutions have failed. Is this another such moment? Bright Line Watch, a group of political scientists monitoring threats to democratic practices, has conducted a survey on the robustness of America’s democracy in the face of what it calls “nightmare scenarios.” Election Day is less than a week away, and Donald Trump has indicated that he’ll respect the result if he wins—but he has also been waging an unprecedented campaign to gin up belief in voter fraud. Should he lose in anything but a blowout, he seems poised to try to discredit the electoral process by casting doubt on the mechanics of voting and counting and by manipulating the vote in the Electoral College through spurious legal challenges, among other strategies. Which tactics seem most likely to damage democratic processes, given America’s institutional soft spots? More important, will Americans keep the peace, even as the President riles them?

Earlier this month, the Bright Line Watch team—led by Gretchen Helmke, of the University of Rochester, Susan C. Stokes, of the University of Chicago, and my colleagues John M. Carey and Brendan Nyhan, of Dartmouth College—posed such questions to some ten thousand academic political scientists in the United States. More than seven hundred, “across a diverse range of subfields,” responded, offering opinions about the vulnerabilities of America’s democratic institutions. The team simultaneously commissioned a YouGov survey of twenty-seven hundred citizens “selected and weighted to be representative of the U.S. adult population.” The survey asked them not the diagnostic questions posed to the political scientists but, to complement those, whether and in what ways they thought the vote could prove fraudulent. The team then divided those citizens into self-identified Trump “approvers” and “disapprovers” (the survey did not require respondents to identify as registered Republicans or Democrats, but it did establish which parties they felt “closer” to). The team examined the answers to determine how seriously Trump’s claims are believed—and, more important, what both groups might be prepared to do about it. “We asked a battery of questions regarding the ways fraud might be perpetrated,” Carey told me. “What’s fascinating, and cautionary, is the degree to which the President's supporters and his opponents are—no surprise—living in completely disconnected information ecosystems.”

The political scientists were asked to rate the likelihood of twenty-eight electoral scenarios in which things could go wrong, intentionally or not, from polling places being closed to the House of Representatives deciding the winner. Interestingly, many potential threats didn’t particularly disturb them: Russian hacking, for example, or a coronavirus-related emergency order that could curtail voting in various states, or “faithless electors” deciding the outcome in the Electoral College. The academics apparently think America’s electoral institutions are rooted enough that inherent weaknesses in law and formal decision-making procedures are not likely to be a source of major problems.

What does worry them (and, in fact, many others in the public arena) is the potential for bad faith deriving from party tribalism, which can roil supporters as the election proceeds: long lines at the polls on Election Day, notionally produced by efforts at voter suppression, or the likelihood that Trump will attack the “blue shift”—the tendency of returns in narrow state contests to favor Democrats in the hours and days after the polls close, largely because of the lag time in processing urban and mail-in ballots. They are most concerned, it seems, that false social-media claims will incite rage against any result that does not go Trump’s way, and they fear that, if Trump loses, he will, indeed, refuse to concede and may also encourage violence during voting and ballot counting.

On that ground, the YouGov survey of citizens is hardly reassuring. Both sides seem to agree on just one positive point: that poll workers at the local level, in towns and counties, can be counted on to tally votes in a trustworthy manner. Around eighty per cent think that their own votes will be counted fairly, a kind of proxy for a belief that their neighbors will be fair. Moreover, around three-quarters of respondents on both sides have “confidence that votes will be counted as intended” at the state level, although that confidence is shaken when the other side controls both houses and the governorship of the state. (Trumpists are somewhat more skeptical than anti-Trumpists on this score, even though Democrats face full Republican control in the key battleground states of Florida, Georgia, Ohio, and Iowa.)