Inside a ‘mega-study’ on election denial, polarization, and violence—and how to stop it

Across major races for the upcoming midterm elections, Republican voters have put forward nominees who openly embrace election conspiracies.

Doug Mastriano, the GOP nominee for governor of Pennsylvania, said that if he’s in power, he’d have the ability to decertify any election result “with the stroke of a pen.” Kari Lake, GOP nominee for Arizona governor, declared even before her results came in that there’d been evidence of fraud in her race.

“If we don’t win, there’s some cheating going on,” Lake preempted. (When she won, she simply said her supporters had “outvoted the fraud.”)

The trend is widespread. According to a FiveThirtyEight study, at least 120 Republican midterm candidates don’t accept the results of the 2020 election, representing 49% of Republicans on the ballot for Senate, House, governor, attorney general, and secretary of state. Only 37 have fully accepted Joe Biden’s presidential victory.

Stanford’s Polarization and Social Change Lab wanted to find out why there’s such widespread support for these antidemocratic attitudes. The Strengthening Democracy Challenge, which researchers described as a “mega-experiment,” was the largest of its kind, recruiting more than 32,000 “partisan” participants who resembled the American population with regard to age, gender, and ethnicity—and, critically, identified as either Democrat or Republican.

In addition to examining antidemocratic sentiment, the study looked at animosity among Democrats and Republicans and support for violence against the opposing party. Crucially, it also looked at things that might help reduce negative sentiments between the parties by testing a set of 25 online interventions (with mixed results).

Similar to how public opinion polls are conducted, the study asked participants to answer a set of online questions. To gauge antidemocratic attitudes, researchers asked whether their political party should reduce voting locations in areas that support the other party, or if their party should not accept election results if they lose.

To assess views on political violence, they asked if it’s fair to send threatening messages to other party leaders, or if using violence for political goals is justified. For political animosity, they were asked to rate their feelings about members of the opposing party, from 0 (very cold) to 100 (very warm).

But before receiving these questions, respondents were randomly assigned to one of 25 interventions that were designed to reduce antidemocratic views, partisan dislike, and support for political violence. All less than eight minutes long, these activities included short films, quizzes, readings, essay-style questions, chatbots, and a guided meditation.

Researchers found that the most effective interventions were those that corrected misperceptions about the opposition. In general, members of both parties wrongly assumed that members of the other party took more extreme policy positions and that they disagreed on more than they actually did. That tracked with an older study by the lab that found members of both parties overestimated rivals’ support for political violence by up to 442%, which is “orders of magnitude wrong,” says Robb Willer, professor of sociology at Stanford, who led the study.

The theory is that showing people that they actually agree on a number of things diminishes the feeling that the other side is “out to get” them, Willer says, noting, “You can ratchet these things down by just giving people accurate information.”

An especially effective intervention was a short film that showed the two Utah gubernatorial candidates from 2020, a Democrat and a Republican, agreeing to honor the election result no matter who won. But this was only one of three interventions that had a significant effect on all three areas of the study: reducing partisan animosity, support for political violence, and antidemocratic attitudes.

Other interventions didn’t work as expected: For instance, a quiz that asked participants to imagine being in the shoes of someone who grew up differently and would therefore have different views on issues like guns and abortion had a meaningful effect only on reducing partisan animosity.

In fact, the majority of strategies worked to reduce partisan animosity. “We came out of that thinking, we have really powerful tools for intervening on this right now,” Willer says.

But the two other areas—antidemocratic attitudes and political violence—proved much harder to crack. One intervention was a film about democratic collapse and violence in other countries, including Zimbabwe and Turkey, which worked to significantly reduce antidemocratic attitudes and partisan animosity among all who watched it. But it also produced a “backfire effect,” in that it actually increased the will for political violence among Republicans.

In recent years, the right has engaged in more antidemocratic actions and political violence than the left. By and large, Republicans have been the ones denying election results and attempting to suppress voters; they were behind the Capitol insurrection; and they have engaged in a number of acts of political violence, including hate crimes, armed demonstrations, and white supremacist events. According to the Anti-Defamation League, right-wing groups caused 75% of political violence in the past decade, versus 4% from left-wing groups.

But many studies have found that latent potential for violence is just as high among Democrats. The difference, Willer says, is that Republican leaders and candidates—like Mastriano and Lake—are the ones currently capitalizing on the appetite for these positions.

“When you have a situation where two groups hate the other side, it’s easy to get away with stuff,” he says. Misinformation is also to blame, as biased media sources place audiences in different “information environments.” This means that many people don’t even view certain acts as undemocratic; for them, their party is not overturning an election but rather taking back one that was stolen.

Given the current climate, these interventions could have real-world applications. The Utah video, for example, could be replicated in other states; political funders or state parties could encourage candidates to participate in such a film as a condition of receiving campaign donations.

One of the most efficient ways to disseminate the digital interventions to a wide audience could be for social media companies like Facebook to promote them. They could also be promoted by various grassroots groups whose work focuses on bridging the gap between Democrats and Republicans, such as Living Room Conversations and the Listen First Coalition. The Stanford lab is holding a conference in September to inform such nonprofits about how best to implement offerings like these.

The strategies will also need further study. “We found a few things that work [but] they’re not transformative,” Willer says.

One critical takeaway was that each area is fundamentally different, even if they seem related. It makes sense that partisan animosity may be a prerequisite to political violence—but while 23 interventions worked to reduce partisan animosity, only 5 affected attitudes toward political violence.

“Our study really shows that you cannot just treat dislike in order to treat violence,” says Jan Voelkel, a Stanford PhD student who was involved in the research. “You really need to think directly about how to treat violence.”