Experimental Design
Respondents are first asked about their level of concern about the health risks of COVID-19 for themselves and close family/friends; then similarly they are asked about concern about economic risks. We also elicit their estimates of current and future (year-end) number of fatalities attributed to COVID-19; the US unemployment rate; and their assessment of stay-home/lockdown policies. They are then asked if they would prefer to read an article about the health risks of the COVID-19 or the economic risk. Within each of the domains (health/economic risks), they are asked to rank two different headlines in terms of which article they would prefer to read. The headlines (and the associated articles) differ in terms of their tone, with one clearly more optimistic than the other. They are randomly shown one of the articles, and asked to assess it in terms of informativeness and trustworthiness. Then, they are asked again for forecasts of number of fatalities and the unemployment rate, and about their policy preferences. They are also given the option to allocate money (provided by experimenter) to (1) the promotion of the news source from which they read the article on social media, and (2) donate to one of two organizations that either supported or opposed lockdowns.
There are two treatment arms (run sequentially): in the first arm, the sources of the news articles are not provided. In the second arm (which will be run after the first arm), the sources of the news articles will be provided to a subset of respondents, and the tone of the article descriptions (which here are generic rather than in the form of headlines, and only about health topics) will be randomized within news source. The samples for the two arms will be non-overlapping.