RAPID: COVID-19 Information Campaigns for Vulnerable Populations

Project Outcomes Statement

The grant funded interdisciplinary work during the COVID-19 pandemic to study how to reach underserved populations with public health messaging. In the US, racial and ethnic minorities experienced disparate impacts of COVID-19 from the beginning, and during the early phases of the pandemic, there was a paucity of public health messaging that directly addressed communities of color. To investigate how to reach underserved communities more effectively, a team of economists partnered with physicians and nurses and conducted three related experimental studies. In the first study, "Comparison of Knowledge and Information-Seeking Behavior After General COVID-19 Public Health Messages and Messages Tailored for Black and Latinx Communities," published in Annals of Internal Medicine, in 2020, we tested the impacts of messages about COVID symptoms and precautions recorded on smartphones by a diverse group of physicians on knowledge and information-seeking behavior among Black and Latinx individuals. We experimentally varied the racial or ethnic identity of the messengers along with elements of the message content, for example, whether the physicians acknowledged inequalities in either the healthcare system or economic opportunity. We find that all videos, regardless of content or the identity of the messenger caused an increase in knowledge. However, information-seeking only increased with race-concordant messages for Black respondents. None of the treatments impacted information-seeking for Latinx respondents, and tailoring the content did not change the impact on information-seeking for either group.

In "Effect of Physician-Delivered COVID-19 Public Health Messages and Messages Acknowledging Racial Inequity on Black and White Adults' Knowledge, Beliefs, and Practices Related to COVID-19," published in JAMA Open in 2021, we show that similar DIY video messages recorded by physicians and shown to a random sample of Black and white adults in an online survey experiment again caused an increase in COVID-19 knowledge. Moreover, we show that the videos also improved the practice preventive behaviors measured several days later. Notably, despite large baseline differences in both knowledge and behaviors by both race and political affiliation, the positive treatment impacts on health-preserving behaviors cut across demographic groups.

Given the promising individual-level impacts of physician videos across groups, in a third experiment, we asked whether scaling messaging campaigns by healthcare professionals could lead to changes in behaviors and health at the population level. In "Effects of a Large-Scale Social Media Advertising Campaign on Holiday Travel and COVID-19 Infections: a Cluster Randomized Controlled Trial," published in Nature Medicine in 2021, we report the results of a randomized controlled trial in which messages recorded by doctors and nurses urging people to stay home for the Thanksgiving and Christmas of 2020 were broadcast using a Facebook ad campaign. Zip codes in 820 counties and 13 states were randomized either into the ad campaign or a control group in a two-stage research design. Over both holidays, more than 30 million users received messages on their Facebook feeds, and those we reached received 3 ads on average. Facebook users in counties with more randomized exposure to the campaign reduced the distance traveled in the three days prior to each holiday by approximately 1 percentage point. We also find that COVID-19 infections in the two-week periods starting five days after each holiday declined in treatment zip codes by 3.5 percent. Again, despite average differences in behaviors by political leaning, we find no differences in the treatment effect of the messages in red versus blue counties.

In the three projects, we show that messages from trusted and credible sources can play an important role in the early public health response to a pandemic and can cut across demographic and political lines. In follow-up work, we show that video messaging does not have any detectable impacts at encouraging vaccination a full year later in the pandemic, when views and beliefs had further calcified.

Intellectual Merit. The interdisciplinary team of investigators executed and published a series of randomized controlled trials to probe how to best improve knowledge and health-preserving behavior among diverse populations during the COVID-19 pandemic. The research adapted to the changing goals of health-care professionals across several different stages of the pandemic.

Broader Impacts. 1) Scalability of results: The results of the study have clear implications for conducting public health messaging. The research itself included a test of messaging at scale -- reaching tens of millions of Americans. 2) Interdisciplinary investigators: The study team includes medical doctors, nurses, public health economics, and experts in social networks. The research builds on concepts from social learning, epidemiology, econometrics, experimental economics, and medicine. (3) Training: PhD students worked on all aspects of the project and coauthored the final publications. (4) Data: the design, instruments, code, and de-identified data from all of the published studies have been made public.