Among the immediate measures to combat a fourth wave of coronavirus cases, wearing masks should be considered the least controversial. Unlike other public health measures such as lockdowns or capacity restrictions on businesses and organizations, requiring masks does not impede economic activity. Study after study has documented that masks can effectively diminish the spread of covid-19, the disease caused by the virus. Yet President Biden’s recent plea for people to wear masks — at least until the ongoing vaccination drive results in herd immunity — is likely to be ignored by many Americans, thanks to the nation’s extreme political polarization.

Our research found that the single most important predictor of local mask use last year in the United States was not covid-19 severity. Nor was it demographic characteristics, religious affiliation or social capital, or even local policies such as mask mandates. No — the strongest and most consistent local factor influencing mask use was what percentage of people in an area voted for Donald Trump in 2016. In areas where mostly Democrats reside, people tended to be likelier to wear masks. But the more people in an area who voted for Trump in 2016, the less likely it was that residents there would wear masks. And while mandates significantly increase mask use in communities that vote heavily Democratic, they have little additional effect in moderate or Republican places.

Mask use — or mask refusal — has become a symbol of a struggle between individual autonomy and social responsibility. Wearing or not wearing a mask becomes a cheap and effective way to demonstrate your political attitude. During the 2020 presidential campaign, both candidates tried to underscore the difference. Biden canceled rallies, held virtual events and excoriated Trump for his refusal to wear a mask in public. Trump, on the other hand, celebrated Amy Coney Barrett’s nomination to the Supreme Court with a large gathering in the Rose Garden. The event, emblematic of the Trump administration’s handling of the pandemic, ended up being a superspreader event.

Our survey data, courtesy of Dynata, included 250,000 responses from across the United States. Participants were asked to estimate how often they wear a mask when outside and around other people. These responses were then aggregated to calculate the frequency of local mask use. To measure partisanship, we used vote shares at the Zip code level from the 2016 election. The survey data allow us to show that partisanship is one of the most important predictors of local mask use. This gap in mask use between Republican and Democratic areas remains significant even where local governments mandated masks. Specifically, we found that an 18 percent swing in votes in favor of Trump in 2016 correlated with a decrease in mask use of 13 percent last year.

We tried to account for as many economic and demographic characteristics as we could find data on — including indicators of local mask mandates, social capital, covid-19 severity and comorbidity patterns. But we kept obtaining the same result: Partisanship not only matters, it swamps other factors by a large margin. The estimated effect of the Republican vote share on mask use is strongly negative and significant, regardless of anything else.

Other studies have documented additional aspects of the partisan divide over the pandemic in 2020. The gap between Democrats and Republicans manifests itself in differential risk perceptions, interest in covid-19 and social distancing behavior, all of which are higher for Democrats. But the use of masks is, in a sense, the most puzzling of all, as it appears to constitute a purely symbolic act of political belonging. Compared with lockdowns, masks impinge negligibly on economic and civil rights, so there seemed little reason to expect them to become the emblems of political affiliation. Yet often in politics, the most pointless quarrels can wield the most irresistible power.

Of course, the role of partisanship might be reversed if the partisan leader is right. We estimated the effect of Trump’s unexpected mask use at the Walter Reed National Military Medical Center on July 11 and his explicit endorsement of masks on July 20. After these events, there was an immediate and significant positive change in online discourse: Twitter users expressed more interest in mask use, as well as more positive sentiments toward the idea of wearing masks.

We can only speculate how much more Republican voters would have worn masks last year if Trump had consistently sent a pro-mask message and how many lives that would have saved. The coronavirus wave of the fall severely affected Republican strongholds in Iowa and Wisconsin, at a moment when it was long clear which tools were most effective in slowing the spread. Masks, if not a panacea, could have helped. Instead, after being hospitalized at Walter Reed with covid-19, Trump defiantly removed his mask and tweeted that no one should be afraid of the virus.

Now the partisan divide is haunting the Biden administration. Persuading the “Republican half” of the country to continue some attempt to mitigate the spread of the virus is paramount in buying time until enough people are vaccinated. A recent Pew survey found that 73 percent of Republicans think Trump was either a good or great president. If he were to come out in favor of wearing masks, it might help — but of course, if he’d been willing to do that, the past year would have been very different.

Read more: