As the delta variant pushes U.S. covid-19 cases back upward, Americans are again sharply divided by party over such public health tools as mask and vaccine mandates.

After the past year’s fierce partisan battles over mask requirements and shutdown orders, many have speculated that the pandemic has made Americans even more polarized than they were before. In May 2020, Axios correspondent Bryan Walsh wrote: “Far from being the unifying force other catastrophes have been, the covid-19 pandemic is tearing a divided America — and world — further apart.” In the same month, Christian Science Monitor staffer Linda Feldman wrote: “Polarization, building for decades, was already intense before covid-19. Now it’s on steroids.”

Were these pundits correct? Did polarization increase during the onset of covid-19? Our research suggests that an important type of partisan polarization — affective polarization, or how strongly Americans dislike and distrust adherents of the “other” party relative to those of their own party — actually decreased during the first stages of the pandemic.

How we did our research

Social scientists use the term “polarization” to mean divisions between political groups. “Affective polarization” is a specific type of polarization that refers to how much Republicans and Democrats dislike each other, distrust each other or have negative feelings toward each other. That distrust and dislike between parties has been increasing quickly in the United States over the past several decades. Government gridlock, partisan discrimination in job hiring and avoidance of those from the other party in social situations have all been linked to affective polarization.

We used several large-scale surveys to examine trends in affective polarization in the United States during the initial stages of the coronavirus pandemic in 2020. Our analysis included data from the well-respected Democracy Fund Voter Study Group’s Nationscape and the American National Election Studies. These surveys asked people how “warmly” or “coolly” they felt toward or how favorable their views were of various parties and figures, such as the president and Republicans or Democrats in Congress. We converted all data sources to a common 0-100 temperature scale, with higher numbers indicating warmer feelings.

We measured affective polarization by computing the difference between one’s (presumably favorable) feelings toward one’s own party and one’s feelings toward the other party.

When the pandemic hit, polarization fell

We found that in the months immediately after the pandemic first hit, Americans actually felt closer to one another, regardless of party, as you can see in the figure below. Before March 2020, our measure of affective polarization stayed relatively constant at just under 50 degrees, meaning respondents felt almost 50 degrees more warmly toward their own party than toward the other party. Affective polarization then fell by more than 3 degrees by mid-May as Americans’ pandemic anxiety grew. That’s comparable to more than one-third of how much U.S. affective polarization rose during the increasingly antagonistic years between 1996 and 2016.

To ensure our findings were not a statistical fluke, we conducted five other analyses from distinct data that measured trends in affective polarization at roughly the same time the pandemic hit the United States: three from Nationscape (measuring related variables), one from the American National Election study and one from a previously published study. Three of the five other data sources showed a decline in polarization, while two showed no significant change.

But did the pandemic cause the drop in antagonism?

To explore whether the link between the pandemic and polarization is causal, we did an experiment. In September and October 2020, we asked part of a nationally representative sample of 1,503 adults, recruited via the survey firm Bovitz Inc., to read news excerpts about and reflect on the onset of the pandemic. We then asked them to report their feelings toward each party. By comparing their responses to those of others who were not assigned to reflect on the pandemic, we found that thinking about the pandemic reduced affective polarization measures by nearly 4 degrees.

Why? Political science research often finds that common national threats increase national unity — and the pandemic was an enemy to us all. Consistent with that theory, in April 2020, 90 percent of Americans told a YouGov survey that “we’re all in it together.” Only 63 percent had agreed with this sentiment in the same survey in 2018.

Polarization increased again after George Floyd’s murder

The pandemic was not the only politically-charged event in 2020. In June, a video circulated showing a Minneapolis police officer kneeling on George Floyd’s neck until he died. This prompted unprecedentedly broad global protests of racial injustice, a small but important share of which devolved into violent riots. While 86 percent of Democrats supported the protests, according to Kaiser Family Foundation polling, only 36 percent of Republicans did.

If you look back at the figure above, you can see that affective polarization increased after Floyd’s murder, undoing most of the 3-degree decline that followed the pandemic. Many of the other data sources we examined showed similar results.

In other words, while the pandemic may have been a common enemy, Floyd’s murder appears to have pushed Americans back into their opposing views, with dislike and distrust. U.S. history has been shaped by political conflict over race from the founding; today, racial views are sharply divided along partisan lines.

What can we learn?

As Americans try to heal partisan wounds, policies trying to ameliorate polarized behaviors may not fix the underlying divisions. While some events may bring Americans together, others will just as quickly divide the country with strong, even violent, distaste for the “other” point of view.

Levi Boxell (@levi_boxell) is a PhD candidate in economics at Stanford University and a National Science Foundation graduate research fellow.

Jacob Conway (@JacobCConway) is a PhD candidate in economics at Stanford University and a National Science Foundation graduate research fellow.

James N. Druckman is the Payson S. Wild Professor of Political Science and faculty fellow at Northwestern University, and author of the forthcoming Experimental Thinking (Cambridge University Press, 2022).

Matthew Gentzkow is the Landau Professor of Technology and the Economy at Stanford University.