Expert Q&A: The U.S. Government’s Challenges for Ongoing Risk Communication About COVID-19
By Adam Siegel
Nathan Dieckmann, PhD is an Associate Professor at Oregon Health &
Science University and a Senior Research Scientist at Decision Research
Institute, a decision sciences research entity funded by U.S. government
agencies and private organizations. He conducts basic and applied research in
decision making, risk communication, and statistical methodology.
We’ve been working with Nathan on a U.S. Government research project, so we’re quite familiar with his work. I talked to him about how U.S. government officials are communicating and managing risk during the coronavirus crisis, what leads people to interpret risk differently, and how he thinks we can be presenting forecasted risk to minimize confusion and misinterpretation.
Here's a portion of that conversation:
Adam: Communities across the country are dealing with drastically different communication challenges as the COVID-19 outbreak wanes in some places, and ramps up in others. Communicating on the risks of work and lifestyle choices is not the same in New York City as in a small town of 25,000 people. At the Federal level, this presents a risk communication challenge so people from different places feel like they’re being represented in an appropriate way. How do you think the Federal Government should be communicating to both comfort people around the country, while also empowering state and local officials?
Nathan: We should be clear that there is no one-size-fits-all risk communication strategy. The communication needs are quite different in, for example, a disaster situation (e.g., tornado or earthquake) as compared to a healthcare provider communicating the risks of different medical procedures to a patient. In the disaster case, the main goal is to establish source credibility and provide clear directions for what to do next (e.g., shelter in place, evacuate, etc). In the case of the medical procedure decision, more specific and detailed risk information will need to be communicated to achieve the goal of helping a patient make an informed value-consistent choice.
These examples highlight two important principles of risk communication: First, any communication can only be considered good or bad in relation to the communication goals. It is important that communicators are very clear from the outset as to what they are trying to achieve. Second, any risk/uncertainty communication should only include as much detail as is necessary to achieve the goal and no more. If you present less information than is needed, then people cannot use the information to make the decision they need to make. If you present too much detailed information, you run the risk of overwhelming people and having them shut down entirely.
In our current situation, I think we all have questions as to what the different governmental entities are trying to achieve. I feel that this is potentially where many of these communications are going wrong. Different entities have competing goals or perhaps there are no clear goals for many communications. So to answer this question more directly, perhaps the goals at the Federal level should be to provide comfort and empathy to the public, and the goals at the local level should be to provide evidence-based recommendations regarding protective measures.
Adam: If you spend 30 minutes consuming any form of news, your interpretation likely falls in one of two extremes: (1) the situation is quite dire, or (2) there is a complete overreaction going on. How could the risk of COVID-19 be communicated while keeping in mind, even in a mildly hit locale, that social distancing and sheltering-in-place guidance must still be followed?
Nathan: One of the interesting things that social science research has shown is that an individual’s perceptions of risk is often more strongly related to individual difference and social factors than how a risk is communicated by experts. These factors include cultural worldviews, political ideology, trust in science, numeracy and literacy, social norms, to name just a few. I think there will always be extreme views and idiosyncratic perceptions of risk regardless of how these things are communicated. That being said, I think communicators can begin to craft consistent and helpful communications by following the principles discussed above.
Adam: An often cited example for the misinterpretation of risk in political forecasting is what led to the U.S. invasion of the Bay of Pigs in Cuba. When John F. Kennedy asked his advisers to evaluate the plan, they had quantified the likelihood of success as 30%. But, they conveyed to the President in writing that there was “a fair chance of success,” which Kennedy perceived as indicating favorable odds (greater than 50%), thus leading to the attack. It seems that when there is greater uncertainty, people are less willing to put numerical estimates to forecasts and instead are more comfortable relying on vague probability phrases, like “unlikely”, “even chance,” “may occur”. How should government agencies and public health leaders think about the role that both qualitative and quantitative forecasting play in conveying risk amid the pandemic to prevent similar misinterpretation from occurring?
Nathan: The use of verbal versus numerical expressions of risk/uncertainty has been relatively well studied. All things being equal, I think most experts would agree that verbal expressions of uncertainty are more vague and open to differing interpretations, although there is research suggesting that just presenting numbers does not solve all of these problems. The example you cite from the world of political forecasting is one where verbal expressions of uncertainty have been commonplace. This is for several reasons, including historical momentum and that forecasters might prefer vague expressions so they are not held accountable if things go wrong. In general, I would recommend presenting numerical point estimates of probability/risk (e.g., 35%) with some kind of additional expression that communicates the confidence or weight of the evidence underlying the forecast. However, in the context of COVID-19 risk communication, the situation is much more complicated. We are dealing with a rapidly changing, dynamic situation with a lot of incomplete knowledge about the nature of the disease and transmission and mortality rates. We are also actively changing the situation by enacting protective measures. The issues of verbal versus numerical expressions of risk/uncertainty are important but probably not one of the key issues in our current situation.
Adam: What are some examples of communicating risk that are cited as “master classes” in following best practices? What are some lessons we could learn from for today’s needs?
Nathan: This is hard to answer since what is good risk communication depends so much on the context and goals of the communication. Some additional rules of thumb that I think are relevant to our current situation are to let scientists and science communicators take the lead as opposed to politicians. Some countries and some states within the US have done a better job with this. The New Yorker recently had a story describing some of the differences in approach taken by Seattle and New York. The communication at the Federal level has obviously been weakened by the conflict between scientists and political actors. I would add to this team experts in decision making and decision analysis to help basic scientists and other stakeholders think through the impacts of different alternative courses of action. Deciding on optimal risk communications would be a lot easier and more consistent if all of the alternative courses of action and their likely impacts have been mapped out.
You may also be interested to read New Platform to Crowdsource Experts’ Forecasts on COVID-19 Pandemic.
Are you liking our blog? Follow us on Twitter @cultivatelabs!