Derek Thompson: American elites still don’t understand how COVID-19 works
In late October, we conducted a randomized controlled trial on a sample of 3,021 American adults to assess how photos affected their interpretation of news articles. We chose real news stories about the pandemic from The New York Times, The Guardian, and the San Francisco Chronicle and randomly paired them with photographs of beaches and bars. (These pairings weren’t a stretch; some of the articles were originally published alongside beach photos.) After reading the article alongside the associated image, participants were prompted to report their overall fear of COVID-19 infection and to rank the following venues based on their perception of risk: beach, restaurant, public park, hair salon, and gym.
We found that when people saw coverage of COVID-19 accompanied by a picture of a beach, they placed beaches higher on their list of risky venues. Never mind that the text of the articles either didn’t mention indoor-versus-outdoor risk or explicitly noted that outdoor environments (such as beaches) have lower risk than indoor ones (such as restaurants). Furthermore, the proportion of readers who incorrectly believed that beaches are riskier than restaurants was about six percentage points higher when an article about the pandemic was accompanied by a beach picture.
In short, when accompanied by beach pictures, even factually correct articles made readers’ beliefs about pandemic risk significantly less accurate. These results held across age, race, gender, political affiliation, and media-consumption habits.
We also evaluated a separate set of control conditions in which an article about an unrelated topic (a form of architectural design) had no associated picture, a picture of a beach, or a picture of a bar. Just showing a picture of a beach alongside the control article about architecture did not significantly affect participants’ perception of COVID-19 risk at beaches. What mattered was the combination of beach photos and pandemic reporting.
Many news organizations have taken their responsibilities very seriously during the pandemic, whether by making their coverage of the pandemic free, maintaining open data sets to fill in government-information gaps, or providing real-time assessments of local COVID-19 risk. Commendably, many news organizations have also set an explicit goal of helping readers and viewers understand precautions (including masking, testing, and vaccination) that reduce the risk of coronavirus transmission and that mitigate the damage of the pandemic overall. Reporters and editors have consulted public-health experts to fact-check the guidance that news outlets are putting out.
But our experiment demonstrates that all of this may not be sufficient. The measure of good public-health messaging is not merely whether experts would deem an article factually correct but also whether a layperson takes away the correct message. In an epidemic, ripple effects can be large. A single misinformed person can touch off a series of infections and even deaths that might have been avoided.