Q: The data for your study were collected in April 2020 — the early days of worldwide COVID-reduction measures. Can you describe the circumstances under which this study came together?
Clara Bicalho, Melina Platas, and Leah Rosenzweig: We had been starting to plan an in-person survey in Uganda over the summer of 2020, looking at the role of social norms in explaining political behavior. When it became clear that the pandemic would disrupt not only our research plans but also life as all of us knew it, we pivoted to turn this into an online survey to look at social norms and health behavior. We also wanted to look into how the political context might shape behavior–including the presence of lockdown policies and trust in government. We decided to broaden the survey to cover Uganda, Kenya, and Nigeria because of the variation in lockdown policies experienced in these countries as well as variation in historical experience with infectious diseases such as Ebola and HIV/AIDS.
Q: The article details a large gap between the relatively high levels of knowledge about how COVID-19 spreads and relatively low levels of actual behavior change. What do your findings tell you about the cause of that disconnect?
CB/MP/LR: Behavioral change is hard. There are many examples in our everyday lives in which we have good information to guide our behavior — for example, we know that exercising regularly and eating a balanced diet are important for our health — but we nonetheless behave in a manner inconsistent with that information. So it is not too surprising that we find despite the fact that respondents know how to prevent COVID-19 — regularly washing hands, maintaining physical distancing — it is still challenging to constantly adhere to these best practices. We have all probably experienced this disconnect in our own lives at some point over the course of the pandemic. In other words, having accurate information is often not sufficient for behavior change. In fact, the study focused on the idea of social norms shaping health behavior precisely because we anticipated that information alone would not motivate people to practice the health behaviors being promoted by governments.
Q: In a discussion of social norms, the influence of perceived peer attitudes and behaviors are important. Your study looked at the difference between respondents’ attitudes towards distancing measures and their best guess of other respondents’ attitudes. What conclusions did you draw from these gaps?
CB/MP/LR: At the time of the survey in April 2020, respondents in all three countries greatly underestimated their peers’ support for lockdown policies. Hence, there is a gap between actual support for these policies and expectations of others’ support. We observe the largest gap in Uganda, where respondents guess that 56% of others taking the survey are supportive of lockdown policies, but in actuality 77% of respondents say they are. This kind of gap can be problematic when trying to encourage widespread compliance with public health policies, such as physical distancing, because you might see little point in practicing it if you don’t think others are going to do so. While our study was not designed to examine whether misperceptions reduce compliance with health behavior, such an effect would be consistent with existing research and worth pursuing in future work. Though there is potentially a role for social desirability in explaining some of the gap–respondents might think they are “supposed” to say they favor lockdown policies–we don’t think this is driving our results. We see the gap in beliefs in places with and without lockdown policies, the survey was online and anonymous, and very few people refused to answer this question.
Q: What recommendations do you have for policy makers and practitioners based on the findings of this study?
CB/MP/LR: We found that that most respondents were willing to write a message encouraging the public to maintain physical distance in their respective countries. Combined with the fact that respondents consistently underestimated others’ support of policies that were designed to facilitate physical distancing, we suggest (and would like to test in future work) that using citizens’ own words may be one way to create common knowledge about the extent of support for these practices, which may in turn increase compliance. Communication from citizens to peers may be particularly useful when citizens do not trust information coming from the government. Moving forward, it would be interesting to see whether this kind of peer-to-peer communication could be helpful in overcoming vaccine rumors and hesitancy.
Q: Can you share a few best practices for anyone who might be conducting similar types of research during the COVID-19 pandemic?
CB/MP/LR: We initially tried to run this online survey without offering a monetary incentive for participation, given a bit of a logistical hurdle to do so in all three countries. It failed miserably and we got about 40 participants total. When we began offering phone credit as an incentive (even of less than US$1 equivalent), response rates increased. Collecting data at this time, when some respondents were experiencing lockdowns and spending more time at home, proved relatively quick. Besides also being low-cost, this participant recruitment approach allowed us to easily target respondents by gender and country of residence, track engagement with our recruiting posts, and quickly adapt our participation incentives.
There were also tradeoffs that we considered in terms of our sampling strategy. Advertising the survey through social media (Facebook and Twitter) limited our ability to make claims beyond our study population. Nevertheless, we think that the sample–which tended to be relatively more urban, educated, and able to work from home than the general population — is a particularly interesting one for the question we were interested in. Our respondents were living in some of the most densely populated areas in their respective countries. It was in these urban centers that physical distancing was especially important, and our respondents, being relatively well-off, were in theory in a better position to practice it. Online surveys in countries with relatively low internet penetration have limitations in terms of their representativeness, but there are still important questions that can be answered in surveying this population.