This week, President Trump quietly disbanded his much-ballyhooed voting commission, originally charged with identifying the “millions” of illegal votes he alleged had been cast for his 2016 opponent Hillary Clinton. The effort had not identified any such fraud. Trump charged on Twitter that that was because states’ attorneys general were hiding the evidence:

While Trump’s attitude may strike some as a refusal to acknowledge facts, it’s not far from that of his base, as we found in a recent survey.

Many Democrats believe that claims of widespread voter fraud are just a ploy for passing strict voter-identification laws — laws that, according to analyses by the U.S. Government Accountability Office, tend to reduce turnout among voting groups likely to cast ballots for Democrats, like younger, less affluent or African American citizens. Such voter-ID laws have been challenged in a number of lawsuits, with mixed results in court. Recent political science research finds that Republican-controlled state legislatures tend to enact such laws in states where elections are becoming more competitive.

That’s the partisan frame. Republican elites tend to allege voter fraud and advocate for strict voter-identification laws; Democratic elites tend to dispute those allegations and argue that voter ID is a ploy for voter suppression. So what about their rank and file? In a study published last month in Public Opinion Quarterly, I found that partisans among the general public see voter ID in very different ways.

According to one comprehensive analysis, there were only about 31 allegations of voter fraud out of 1 billion votes cast over the past 15 years. At the same time, a majority of both Republicans and Democrats support voter-identification laws, although Republicans support such laws far more than do Democrats. Would learning about the nearly nonexistent rate of fraud reduce citizens’ support for voter-ID laws?

Here’s how I did my research

To test this possibility, I fielded a survey experiment in 2016 on a sample of more than 1,000 adult citizens across the U.S. The sample, collected by the Qualtrics survey company, was selected to be nationally representative in terms of race, age and geographic region. Early in the survey, I used the canonical seven-point party identification scale to find out which respondents were Democrats (46.51 percent), Republicans (30.94 percent) and independents (22.55 percent). All respondents were later randomly assigned to one of four conditions: a “baseline” condition that gave basic information about voter-ID laws; a “fraud” condition that gave information about the 31 alleged cases out of 1 billion votes between 2000 and 2014; or a condition containing the fraud information and also information that either Republican or Democratic voters would be disproportionately harmed by voter-ID laws in future elections.

Respondents were then asked whether they believed voters should or should not “be required to show an official photo identification before they are allowed to vote on Election Day.”

What do ordinary Republicans and Democrats think when told there’s little evidence of voter fraud?

The results were counterintuitive. After learning that voter fraud is extremely rare, Democrats were no less likely to support voter ID laws. Republicans actually became more likely to support such laws, as did independents. Simply by being prompted to think about voter fraud, Republicans became even more concerned about it — even if the prompt was informing them that such fraud is extremely rare.

The second part of the experiment told respondents that fewer voters of one party — some respondents were told “Democrats” and others “Republicans” — were likely to vote if a state required photo ID to cast a ballot. Once again, partisans responded very differently. Republicans supported voter-ID laws at roughly the same rate, whether told that Republicans’ or Democrats’ turnout would drop. But Democrats were significantly less likely to support voter-ID laws when told that Democrats would likely lose votes — and more likely to support them when told that Republicans would lose votes.

To see what else might be at work, I also analyzed Pew survey data to examine how much partisans’ support for voter ID changes according to how strongly voters favor their own party. Do those who are most passionate about beating the other team on Election Day feel differently about voter ID compared with their less passionate counterparts?

The answer differs by party. I find that, even among those Republicans who only weakly support the party and theoretically should be least motivated to want to disadvantage the opposing team, approximately 90 percent support voter ID. By contrast, Democrats most committed to their party are far less likely to support voter-ID laws than Democrats less emotionally invested in winning elections.

Republicans and Democrats see this issue very differently 

In short, Democrats seem most concerned about losing elections, while Republicans seem most concerned about voter fraud. Interestingly, that’s how Democratic and Republican elites most often frame the voter-ID issue. Democrats speak about preventing Republicans from gaining advantages in elections; Republicans talk about preventing election fraud. It’s possible that partisans are concerned only about the aspects of voter ID that elites have told them are important.

I don’t think that’s the whole story, though — because the results don’t really depend on partisans’ level of political interest and because, as noted above, most Democrats actually support a photo-identification requirement, unlike most Democratic elites. We still don’t know what’s behind these stances.

But we do know that many citizens support such laws, and not just because they want their party to win. And we know that showing the rarity of voter fraud doesn’t change minds. Concerns about voter fraud appear to be real, in other words, even if the fraud itself is not.

John V. Kane is an assistant professor at New York University’s Center for Global Affairs. His research focuses on partisanship, political behavior and political psychology.