Presidential Statements Influence Health Behavior: Two Experiments in the US

• In Event: Relevance, Quality, and Expedience: Political Science Responds to COVID-19: COVID-19 and Political Behavior: Public Opinion and Political Participation

Abstract

There is a widespread belief that presidential statements can influence health behavior. This belief underlies the consternation with which many political opponents and medical experts have reacted to many of President Donald Trump’s public comments on the new coronavirus: By seemingly downplaying the COVID-19 threat, he may be leading individuals--especially his supporters--away from pro-health behaviors necessary to combat the pandemic. Unfortunately, we lack a solid research basis for evaluating whether and how presidential statements might influence health behavior generally, not to mention in a situation as important as a pandemic. And while the strength of partisanship in the US gives us good cause to worry that partisans will take their health-behavior cues from politicians over doctors, research has also found that differences in partisanship are associated with distinct poor health behaviors for other reasons (Kannan and Veazie 2018), generating risks of spurious causal claims. Experimental methods are well positioned to address such identification problems and thereby yield insight into how presidential statements might be influencing behaviors that could, in turn, impact the pandemic’s spread. This paper reports findings from a pair of survey-embedded experiments conducted in the United States during May 4-11, the early stages of the COVID-19 pandemic, using a nationally representative sample of 2000 participants in an online YouGov panel. Simple t-tests are used to evaluate average treatment effects, and Bayesian Additive Regression Trees (BART) is employed to calculate heterogeneous effects since it avoids many of the pitfalls associated with linear models and interaction terms (Green and Kern 2012). The first experiment asks a randomly selected half of the sample to reflect on what Trump has said about the virus (without reminding them what he has said) just before all respondents are asked about their propensity to engage in three pro-health behaviors. While this modest treatment had no effect on the propensity to wear gloves or to spread the view that the virus threat is overblown, it had a strong pro-health effect when it comes to personal contact: People having Trump’s coronavirus statements in mind were significantly less likely to report willingness to hug or shake hands with other people. The effect is not moderated by partisanship, is particularly strong among older people, and reflects (according to post-experiment questioning) a surprisingly widespread belief (as late as May 2020) that that Trump is sincerely trying to combat the pandemic’s spread. The second experiment focuses on the impact of two actual Trump statements (one appearing to minimize the virus threat and one that appears to take it gravely seriously) when combined and compared with statements by doctors. It is here that the dangers of presidential minimization of the threat emerges. The good news is that when the President calls on people to self-isolate in strong terms, he has a positive pro-health effect roughly in line with those of doctors, reducing the tendencies to visit public places and gather with friends and relatives. But when he appears to belittle the need to take personal self-isolation measures, he not only reduces pro-health behaviors but undercuts the effect of doctors’ strong advice. The effect is starkest when it comes to masks, where his statement that he personally chooses not to wear one significant dampens people’s propensity to wear them. It turns out, however, that these effects are also not moderated by partisanship, instead being most pronounced among people who are personally least at risk (the young) but who could spread the disease by failing to wear masks. One final finding emerges across both experiments: People in states with stay-at-home orders in effect are much more receptive to cautionary advice than those in other states, where rebellious responses are more common. Overall, while even weak presidential pro-health advice can have positive effects during the early stages of a pandemic, specific presidential rejection of pro-health behaviors can have damaging effects, undercutting the recommendations of medical professionals.

Author

• Henry E. Hale, George Washington University