New coronavirus polling shows Americans are responding to the threat unevenly
American voters are treating the pandemic like a partisan issue
By Emily D. Bello-Pardo, Monika Nayak, and John Ray
As part of our efforts to track the 2020 election, YouGov Blue regularly fields surveys of the general electorate including items on topics of general interest. Without a doubt, coronavirus is the most pressing story. Here, we report on new polling showing how Americans are responding to what is now a global pandemic that has reached all fifty U.S. states and the District of Columbia.
We fielded a survey of 1,260 likely general election registered voters between March 13 and 17, 2020 that included a battery of questions about the coronavirus outbreak, including questions about how the government is handling the response, whether individuals are worried about the virus, and questions about behavioral changes.
In this piece, we explore the extent to which self-reported registered voters have changed their behaviors in response to the coronavirus outbreak. Perhaps unsurprisingly in the current political climate, we find that political affiliation clearly predicts which respondents are adjusting their behavior in response to the coronavirus pandemic. In short, we observed significant differences in American voters’ health behavior changes between voters of varied party identification and age groups, among other characteristics.
We asked voters the following question to measure seven independent behavioral adjustments:
Some people are going about their daily lives at the moment as normal, while others have had their routines change dramatically as COVID-19 has spread. How would you say that each of your following behaviors have changed over the last week as a result of the coronavirus?
Handwashing: “Amount of times you wash your hands every day”
Changed Travel: “Travel plans over the next month”
Working From Home: “Working from home / telecommuting”
Stockpiling Food: “Buying non-perishable food or increasing the amount of groceries you usually purchase”
Stockpiling Medicines: “Buying flu or cold medicine, or making sure to have enough prescription medicines to last three months”
Child And Elder Care: “Changing plans to account for caring for children, the elderly, or disabled family members whose typical daytime care, such as school and support programs, have closed”
Social Distancing: “Canceling existing social engagements or deciding against new social engagements you might otherwise have made, a practice known as social distancing”
Respondents were asked about each of these behaviors at a random order, and they could answer that their behavior “has changed dramatically”, “has changed somewhat”, “has only changed a little”, “has not changed at all”, or “not sure”.
In the following sections, we summarize some key results. Future posts will drill down further into Americans’ attitudes toward the coronavirus and other questions we included in this survey.
Some Americans are changing some behaviors, but a majority are not
Voters reported that, over the past week, they were particularly likely to start washing their hands more, to change their travel plans, and to engage in more social distancing. About 58 percent of voters reported that their handwashing activity had changed “somewhat” or “dramatically,” as did 49 percent of voters who changed their previous patterns of social distancing, and 40 percent who changed travel plans.
Slightly fewer voters reported they had begun stockpiling food (37 percent), reorienting their child and elder care (31 percent), and working from home (27 percent). Similarly, 25 percent of voters also reported they had begun stockpiling medicine. The following graph shows these overall differences:
Moreover, over half of our respondents reported that their working from home (56 percent), stockpiling medicines (56 percent) and child and elder care responsibilities (50 percent) had not changed at all as a result of the coronavirus outbreak. About 3 in 10 individuals reported that their social distancing patterns (31 percent) had not changed at all over the last week, and 2 in 10 individuals (22 percent) said that their handwashing had not changed either over the same period. These two are important indicators because public health authorities are asking Americans to practice social distancing and increased handwashing as crucial behaviors in the fight against COVID-19.
Overall, however, if we sum the total number of changes each voter in our sample reported making in their behavior in response to the coronavirus, we see some clear and consistent differences in these groups. The following chart shows the average number of possible behavior changes we asked about for voters overall, for voters broken out by age, and voters broken out by party identification. While the average voter changed between 2.5 and 2.8 of those behaviors “somewhat” or “dramatically,” voters in the 18–29 range changed somewhere between 3 and 4 behaviors. In contrast, voters in the 55–64 age range changed between about 2 and 2.5 behaviors. Democrats changed between 3 and 3.5 behaviors, compared to Republicans who changed around 2 behaviors. Additionally, voters who have children report changing their behaviors about as much as do Democrats and voters in the 18–29 age bracket. We explore this result in further detail below.
Democrats have changed their behavior more than Republicans
While contagion does not know party affiliation or beliefs, we find that response to contagion sure does. We found that 18 percent more Democrats than Republicans (68 percent to 50 percent) report that the coronavirus has changed the amount of handwashing they do “dramatically” or “somewhat.” By a 50–30 margin, 20 percent more Democrats have changed their travel arrangements than Republicans. In terms of “social distancing” — one of the most frequently recommended anti-contagion measures — 62 percent of Democrats have changed their behavior, compared to just 35 percent of Republicans.
Notably, there are even partisan differences in behavioral changes that, presumably, are not entirely up to individuals. As has been frequently reported, employers typically make recommendations for their workers to work from home on a company-wide, department-wide, or team-wide basis, suggesting that one individual’s partisan preferences are not necessarily major factors in determining whether this behavior can change. The implication here is somewhat obvious: Partisanship does not necessarily play a causal role in how people respond to coronavirus, and we don’t have enough data to assert that it does, but it is very likely that partisanship correlates highly with the type of work people do. As YouGov Blue polling has been used to demonstrate elsewhere, employment and partisanship interact. Here, we report partisan differences because they consistently emerge as significant predictors of behavior pretty much no matter how we wrangle the data, even if we don’t have data that would allow us to assert a causal relationship.
As is typical for issues of this kind, Independent voters typically sit somewhere between Republicans and Democrats on many of these measures. However, as only about 12 percent of our voters identified as “true Independents” — that is, Independents who did not lean toward one party or the other — this is an imprecise estimate.
Younger voters have changed their habits more than older voters have
Even controlling for other factors like party identification, there are significant differences in how people have responded to the coronavirus depending on how old they are. Unfortunately, these differences do not necessarily accord with the recommendations from global and national health organizations, which emphasize the risk to older people stemming from coronavirus: Younger voters have changed their behavior more than have older voters across most behaviors we polled.
For example, about 56 percent of voters aged 18–29 say they’ve begun stockpiling food, compared to just 30 percent of voters over 65 who report they’ve done the same. While 62 percent of voters in the 18–29 range have changed their behavior with respect to social distancing “somewhat” or “dramatically,” only 47 percent of voters over 65 report they have done the same. While each group has significantly changed its behavior, the most dramatic changes have occurred among the youngest voters.
Some of the differences between younger and older voters are clearly functions of age itself — for example, it is not surprising that fewer voters over 65 report fewer changes in their approach to “elder care or childcare” than do voters in the 18–29 or 30–44 range. It should also not be surprising that fewer voters over 65 report they’ve begun “working from home” than other voters.
Indeed, these differences are not all significant. For example, there are no statistical differences by age in terms of how much voters wash their hands or change travel plans. Moreover, there are no differences between the age group 18–29 and the group 30–44 in terms of how much child and elder care they report doing or whether they report having begun stockpiling food, or how much social distancing they are engaging in, but there are differences between 18–29 and all other age groups across those two behaviors.
Putting it all together
We conducted a regression analysis to understand whether the differences we report above are statistically significant at conventional levels. In each of these regressions, we included typically relevant factors used in political data analysis: party identification, age, whether the respondent self-identifies as a “born again” Christian, whether the respondent has children, the type of place that the respondent lives in, race, gender, and education.
The following graph reports the results of each of these regressions for each of the seven behaviors we polled. When reading these graphs, the vertical axis shows each of the categories for each of the variables in the regression analysis (with the reference category omitted for each), and the horizontal axis represents the coefficient sizes. Each of the dots represents the specific estimate for each covariate, and the horizontal lines around the dot represent the 90 and 95 percent confidence intervals. Estimates below zero mean that that specific group reports less behavioral change than the reference category, while estimates above zero represent more behavioral changes. If one of these intervals crosses the vertical dashed line, that one particular estimate is not statistically significant.
In this graph, we can see that the relationships between party identification and behavioral change we discuss above are all statistically significant. That is, across all behaviors we polled, Republican and Independent voters report significantly less changes when compared to Democrats.
Similarly, there are clear and significant age differences across most, but not all behaviors we polled. For example, older people report less behavioral changes with regards to social distancing, stockpiling food, and stockpiling medicine than do those who are 18–29 years old. However, there are no significant age differences with regards to handwashing, and only those between 55–64 seem to have reported less changed travel.
Indeed, individuals who have children report having changed their travel behaviors and, unsurprisingly, their child and elder care more than those without children. And handwashing seems to have increased significantly among those who live in urban or suburban areas, and among Hispanic respondents.
The following chart reports all of the regressions we discuss above and can contain some further interesting insights for readers:
Younger voters, voters with children, and Democratic voters are most clearly responding to the current crisis. Other voters have taken less action, controlling for other factors.
Conclusions
Our data suggest that many voters have begun making substantial changes to their lives in response to coronavirus, but those changes are uneven across “familiar culprits” of political behavior like party identification and age, and these findings should be worrisome to those concerned about public health during this crisis. Parents, Democrats, and younger voters are driving the current response to coronavirus.
Methods: This survey is based on 1,260 interviews conducted by YouGov on the internet of registered voters likely to vote in the general election in November, 2020. The survey was fielded by YouGov Blue from March 13, 2020 to March 17, 2020, and it was weighted according to gender, age, race, education, region, and past presidential vote based on the American Community Survey and the Current Population Survey Registration and Voting Supplement, conducted by the U.S. Bureau of the Census. The weights range from 0.05 to 6.04 with a mean of 1 and a standard deviation of 0.58.
About YouGov Blue: YouGov Blue is a custom research division of YouGov that works with Democratic and progressive clients. Leveraging YouGov’s panels, methodology, and technology infrastructure, we are pioneering novel capabilities for progressive clients.
About The YouGov Blue Core: This survey data was fielded as part of the YouGov Blue Core. The YouGov Blue Core is a subscription polling service for the 2020 election, collecting over 15,000 survey responses each month. The Core tracks opinion for both the Democratic Primary and the General Election in 2020, conducting polls at the national and state levels, as well as among demographic subgroups of interest. For more information, please reach out to YouGov [dot] Blue [at] yougov [dot] com to see how the YouGov Blue Core can fit your needs.
