The Emergent Partisan Gap in Social Distancing
By Bennett Fleming-Wood, Yonatan Margalit, and Brian Schaffner, Tufts University
In the wake of over 2 million coronavirus cases and almost 120,000 deaths caused by coronavirus, attitudes towards the pandemic have become increasingly divided across party lines. Donald Trump and many other Republican elected officials have minimized the impact of the virus and called for a quick reopening of the country. Early research found that Americans were swayed by the reactions of political elites early on in the crisis: Republicans searched for hand sanitizer less when Republican leaders minimized the virus and more once Trump declared a national emergency. We know people are influenced by their party leaders in their attitudes towards coronavirus. But, to what extent has this influenced the degree to which people engage in social distancing behaviors?
Early research conducted during the first week of the emergency declaration found that partisanship was a reliable predictor of social distancing behavior and a recently published study using GPS data from smartphones also found stark partisan differences in social distancing behavior. But a lingering question is whether these partisan gaps are really driven by partisanship or if they are actually the consequence of a reality in which Democrats and Republicans have been experiencing the coronavirus pandemic differently. Indeed, during the first few months of the pandemic, Democratic areas were much more likely to have been hit hard by coronavirus outbreaks compared to Republican areas. Once we account for these differing experiences, do Republicans still engage in less social distancing behaviors than Democrats?
The Data
Data for Progress has been tracking self-reported social distancing behaviors since March as part of the Coronavirus tracking poll project. These surveys have asked respondents on a near weekly basis whether they had done any of the following during the previous two weeks:
Wore a face mask when in public places
Refrained from touching objects in public
Improved personal hygiene
Avoided crowded public places
Avoided physical contact with people
We also ask respondents whether they have been avoiding socializing in person with people from outside their households. The graph below plots the extent to which Democrats and Republicans report engaging in these social distancing behaviors during the course of this polling.
A first important point to make from this graph is that while partisan gaps have emerged, there are fairly high rates of compliance with social distancing across people from both parties. One striking example from the graph comes from mask wearing. Fewer than one-in-four Americans reported wearing masks in late March. But then the CDC reversed its position on mask-wearing on April 3rd and by the end of April, 70% of Republicans and 80% of Democrats reported that they were wearing masks. While we focus here on the partisan gaps in social distancing, it is worth keeping in mind that a remarkable percentage of both parties have significantly altered their daily behaviors in a relatively short period of time.
But partisanship does appear to shape the extent to which Americans are now abiding by social distancing guidelines. Starting in late-March, when our first survey went into the field, there were few partisan gaps in any of the social distancing behaviors we tracked. However, these gaps began to emerge starting in May. And by early June, Democrats were 15 percent more likely to report avoiding crowded public places and 16 percent more likely to avoid physical contact with others.
In order to more easily analyze the data, we combined the six behaviors shown in the graph above into a single measure of social distancing behavior. People with higher values on this scale report engaging in more social distancing behaviors, while people with lower values are not social distancing as much.
The graph below shows how Democrats and Republicans were distributed along this scale in surveys we fielded during March, April, and May/June. Note that in March, both Democrats and Republicans exhibited fairly similar levels of social distancing, though Democrats were slightly more likely to report social distancing. In April, social distancing hit its peak. This is seen by the large spike near the top of the scale and this happens largely because mask wearing became commonplace during April. During April we also see few, if any, partisan differences in social distancing.
Finally, in May and into early June, Democrats continue with high levels of social distancing behavior, but Republicans begin to deviate significantly from that pattern. In particular, many fewer Republicans are near the top of the distancing scale in May and June and an increasing number are reporting especially low levels of social distancing.
Testing alternative explanations
After Trump declared a national emergency on March 13th, Democrats and Republicans were on equal footing for social distancing behaviors. However, gaps emerged in April and May as Republicans leaders minimized the virus and called for the country to be re-opened. But is the gap we see in the graphs above really caused by partisanship? After all, there are many confounding factors that could account for the partisan gap that we detect. Republicans are more likely to live in rural areas, which were not as hard hit as urban areas and where social distancing may not need to be as strict. Red states were also more likely to end their stay at home orders earlier than blue states. And age may cut in the opposite direction, with older Americans (who are more likely to be Republicans) being the most at risk when they ignore social distancing guidelines. A study conducted in March by political scientists Shana Kushner Gadarian, Sara Wallace Goodman, and Thomas Pepinsky found that partisanship was still a key predictor after accounting for many other variables. However, that study was conducted early during the pandemic, before many Republican politicians acknowledged the seriousness of the threat. As we show above, the month of April witnessed broad bipartisan adoption of social distancing behaviors and the partisan gap only began to re-emerge in May.
In order to determine whether the more recent partisan gap is real, we examined several alternative explanations for the partisan gap.
Accounting for when states re-opened
One factor that might be driving the partisan differences that began to emerge in May is the fact that many of the states that re-opened earlier are predominantly red states. We used the New York Times and CNN coverage of state re-openings to divide respondents into groups based on when their home states began to re-open. The states that began reopening earliest – in April – do not show a clear trend in social distancing behaviors.
More importantly, the partisan gaps are evident among all three groups. As with the broad patterns we showed above, there are few differences in social distancing behaviors between Democrats and Republicans in April, but they begin to emerge in May. Importantly, the partisan gap that develops in May can be found even in states that did not start to re-open until June. This indicates that the emergent partisan gap was not simply a reaction to state re-openings. Republicans were beginning to drop in their social distancing behaviors during May even when they lived in states that had not begun re-opening yet.
Accounting for the urban/rural divide
One factor we might expect to be particularly important is whether individuals live in urban areas or not. Democrats are more likely to live in dense urban areas and dense urban areas have been hit harder by the pandemic. Does this partisan divide persist even when we divide our samples into groups based on whether they live in urban, suburban, or rural areas?
As with the previous analyses, the graph shows that respondents in all types of areas began the pandemic with no real partisan differences in social distancing. Still, the gap opened up in May in urban, suburban, and rural areas alike and by the beginning of June, rural Republicans were social distancing 13 points less than Democrats. Notably, the partisan gap appears to be smallest in the suburbs, where suburban Republicans are now social distancing 6 points less than Democrats. By contrast, we might expect the partisan gaps in social distancing to shrink in dense urban areas where Americans are most at risk, but urban Democrats were social distancing 14 points more than urban Republicans by the height of the pandemic in the beginning of May. Thus, the partisan gap in social distancing is not simply a function of the urban-rural divide.
Accounting for age
Next, we examined partisan gaps by age groups. We might expect age to matter for social distancing since older Americans are at considerably higher risk of dying from coronavirus than those who are younger and middle-aged. While Millenials and Gen Z respondents had only 4 point partisan gaps in March and April, younger Democrats were social distancing 12 points more than young Republicans by the beginning of June. Gen X followed a similar pattern with small partisan gaps throughout March and early April. However, Gen X Democrats were social distancing 18 points more than Gen X Republicans by early May. It makes sense for large gaps to emerge for younger respondents who are less vulnerable to coronavirus. Younger Republicans are able to social distance less than younger Democrats without facing massive consequences. However, we found that the partisan gap persisted even amongst older generations.
Despite the risk for higher age groups, we find that even among Boomers and Silent Generation respondents Republicans socially distance less than Democrats. Boomers and the Silent generation also started out with small partisan gaps in March and April, but still reached a 10 point difference between Democrats and Republicans in early June.
Accounting for personal experiences with coronavirus illness
Finally, we examined the possibility that personally knowing someone who has at some point been infected with coronavirus might affect the partisan gap. Our polls showed that during the first month of the pandemic, Democrats were significantly more likely to know someone infected compared to Republicans. For example, in April, 27% of Democrats said they knew a family member, friend, or co-worker who had been infected with the coronavirus, while just 21% of Republicans said the same. In the surveys fielded in May and June the difference was somewhat smaller – 33% of Democrats and 29% of Republicans reported knowing someone who had been infected. If knowing someone with coronavirus makes people more likely to take social distancing seriously, then the fact that Democrats are more likely to have this experience compared to Republicans may be part of the explanation for the partisan gap in social distancing.
Interestingly, people who know someone who has been infected with coronavirus do not appear to engage in significantly more social distancing behaviors than those who don’t know anyone with the virus. Furthermore, the partisan gap is relatively consistent among those who know someone who has been infected and those who don’t. Thus, while Democrats are more likely to know someone who has been infected, it does not appear to bear on the size of the partisan gap in social distancing.
Accounting for all factors simultaneously
So far, we show that when accounting for each of these alternative explanations individually, the partisan gaps that opened up in May still persist. However, as a stronger test of the partisan gap in social distancing, we concluded our analysis by estimating a regression model that allowed us to control for all of these factors simultaneously, as well as several other variables such as sex, race, and education. Specifically, we estimated a model that included all of the survey respondents interviewed in May and June (when the partisan gap re-emerged). The dependent variable in the model was our social distancing scale and the independent variables were party, age, sex, race, education, and knowing someone who had been infected with coronavirus. Additionally, we controlled for where a respondent lived by controlling for their state of residence and whether they lived in an urban, suburban, or rural area. (As an even stronger test, we estimated a second model that controlled for the zip code where the respondent lived and we still found similar patterns.)
Our model uncovered a number of noteworthy results For example, there is a gender gap in social distancing – controlling for the other factors, women are about 5 points higher in their social distancing behavior compared to men. Education is also important, as people with college degrees engage in more social distancing than those with lower levels of educational attainment. Additionally, as we saw in the graph above, older Americans are considerably more likely to adhere to social distancing guidelines than their younger counterparts.
But even after accounting for all of these traits, as well as where each respondent lives, we continue to find that partisanship is a powerful predictor of how much social distancing people have been doing since the beginning of May. Specifically, Republicans report levels of social distancing that are 14 points lower than that of Democrats even after controlling for all these factors.
Conclusion
In a relatively short period of time, the coronavirus pandemic has significantly altered how Americans engage with the world on a daily basis. The vast majority of Americans now regularly avoid crowded places, interact only occasionally with others, and frequently wear masks in public, behaviors that would have been difficult to imagine just a few months ago. Broadly speaking, it is impressive how many Americans have complied with social distancing norms.
While acknowledging the impressively broad compliance with social distancing guidelines, we also note the troubling partisan gap with this compliance that has emerged since the beginning of May. Republicans are now reporting less social distancing behaviors than Democrats, and this emergent partisan gap exists across generations, and even in urban areas and among people who know someone that has been infected. As the coronavirus case count in America shows no sign of subsiding, partisanship may very well be a big part of why the virus continues to spread and is now increasingly affecting Republican areas.
Brian F. Schaffner (@b_schaffner) is the Newhouse Professor of Civic Studies at Tisch College and the Department of Political Science at Tufts University.
Bennett Fleming-Wood (@bennettfw36) is a rising senior majoring in political science at Tufts University.
Yonatan Margalit (@tunadahuman) is a rising junior majoring in political science at Tufts University.