Carnegie Mellon University


Report 1: Baseline nationally representative panel survey

This report includes basic findings from phase 1 of a longitudinal panel. The full results can be found in the accompanying scientific report.

Measures: The survey includes 3,974 responses to a 15-minute survey about responses to the coronavirus pandemic, focusing on three protective behaviors hand washing, respiratory hygiene, and wearing masks. Using a detailed methodology, we also obtained a high-quality measure of the degree of exposure survey respondents report to people beyond their household. Finally, we ask demographic questions including age, race, sex, and county of residence.

Respondents report that in the previous 7 days a median of 3 people came into their home and were within 6 feet of distance (interquartile range 1 to 6 people). Outside of the house, they report coming into such too-close contact in the past 7 days with a median of 2 people (interquartile range 0 to 9 people). The sum of these two counts (truncated at 100 and log-transformed for statistical analysis) provides a measure of potential exposure to the novel coronavirus. This measure is strongly predicted by self-reported social distancing behavior measured with independent psychometric qualities. Thus, our survey results indicate  that adherence to social distancing recommendation is linked to amount of contact with others that could result in spread of the virus.

Analyses: In the analysis presented here, we predict what kinds of people are experiencing more potential exposure to the novel coronavirus, as a first pass at how demographics and beliefs predict vulnerability. We then proceed to investigate why we find the differences that we do, to better understand the underlying psychology that is leading some people to expose themselves to more risk of contracting the novel coronavirus.

Results: We find that older people have less exposure; in particular, elderly respondents (age 70 and older) have less exposure than those under 70. We find that women tend to have lower exposure, as reflected by their reported increased efforts to protect themselves. In contrast, those who are employed report more exposure. Although some of these people may be working from home, they may have less control over their exposures. Finally, we find that trust in government is a robust predictor of exposure. Trust in science is only weakly related to exposure, and even that relationship disappears when controlling for other variables. However, having more trust that the policies of social distancing, shutting down restaurant dine-in services, and self-isolation of those with symptoms are optimal for dealing with the virus is a strong predictor of reduced exposure.  In contrast, expressing more support for policies reflecting skepticism about the impact of the virus was a very strong predictor of reduced anxiety about Covid-19, and also of increased potential exposure, even controlling for political party affiliation and other demographic factors.

Conclusions. Where Americans choose to place their trust appears to be strongly associated with behaviors that lead to increased or decreased exposure to the novel coronavirus. Although these data cannot show that the trust causes behavior to change, it is a strong marker of different behaviors, with real risk consequences. In particular, expressing higher trust in social distancing and lock-down policies is related to lower exposure, explained by a stronger belief that social distancing is effective. In contrast, expressing more skepticism about these policies is related to higher exposure, explained in part by a lower belief in effectiveness but even more so by a belief that concerns about the pandemic have been overblown. This divergent response speaks to the importance of a central, coherent, trusted policy and how it may help people to engage in difficult behaviors that will protect the community from infection. This pattern of findings may also indicate that political views are associated with goals and worldview—with those who are skeptical about social distancing and masks emphasizing economic stability and de-emphasizing infection risk, with those supporting the protective behaviors emphasizing strong measures to prevent the spread of infection.

It's worth noting that working for pay is associated with more risk, but that this variable is unaffected by the beliefs about effectiveness or trustworthiness, and is likely a more logistical rather than psychological variable accounting for greater risk.


The new coronavirus outbreak, named COVID-19 by scientists, poses a major challenge for our health system. As people become sick and need medical care, they will need resources like hospital beds and ventilators. However, if many people become sick in a short period of time, there will not be enough of these resources to care for them all. If we are to treat every sick person with the best possible medical care, we need to both prevent and delay new infections. We know from history and medical science that public behavior is the most important tool for this prevention. But for the public to help, they need to know what to do and how to do it, as well as to understand why these behaviors are so important. People take their cues from those around them in making sense of new, uncertain situations. This makes it very important to make sure that everybody is getting good information about the risks of COVID-19 and how to prevent it. Official messages need to reflect scientific knowledge, and myths that pop up in communities need to be addressed so that people can understand and debunk them.

Our research team has been studying how people are thinking about the risks of COVID-19, and what they are doing to protect themselves and their community. One key finding from that work is that when people are uncertain about the risk, they are more likely to rely on what other people are doing to determine what the right thing to do is. We also find that people's main concerns about social distancing are that they are worried about getting by without a paycheck and how they will get food and meet other urgent needs. These findings suggest that policies should make sure both workers and small business owners have enough money and food to get them through the near future, and that they can feel confident about this support continuing so that they can focus on staying safe and keeping other people safe.

We plan to do surveys and experiments to help us better understand these concerns and guide policy action. First, we need to know whether helping people understand how to prevent infection will actually lead them to protect themselves. We will conduct experiments to test and identify how best to help people understand and change their behavior, especially for those who are not fully engaging in social distancing. Then, over the next few months as the situation changes, we will keep our research up to date and develop messages to help people understand what is happening and how their behavior can help protect themselves and the people around them.