2  Methods
Our data came from two nationally representative surveys of individuals living in China before versus during the coronavirus outbreak. Both surveys were administered by the Data Intelligence and National Development Lab of Peking University using the same nationwide participant panel.
The first survey (N = 11,131; from 32 provincial regions; 48% women; average age of 37.78; 66% married) was administered at the end of December 2019 (immediately before the coronavirus outbreak was publically reported in China). The second survey (N = 3,000; from 30 provincial regions; 50% women; average age of 34.7; 69% married) was conducted in mid-February 2020 (during the outbreak). Participants in both surveys completed an established measure of emotional well-being (Kahneman and Deaton, 2010). Specifically, they indicated whether they smiled or laughed a lot yesterday, and whether they experienced a lot of enjoyment/happiness/anger/sadness/stress/worry yesterday (1 = yes, 0 = no). Participants in both surveys also responded to demographic measures: age, sex, marital status, monthly household income, and residence region.
Participants in the second survey responded to additional measures pertaining to perceived knowledge about coronavirus infection. They indicated how much knowledge they had about how the coronavirus spreads from person to person, as well as how much knowledge they had about preventing coronavirus infection (1 = very little, 5 = very much). They also completed measures regarding their sense of control during the coronavirus outbreak. They indicated the extent to which they had control over the circumstances they were facing (1 = very much lack control, 5 = very much have control) and the extent to which they were confident that they could manage not to be infected by the coronavirus (1 = not confident at all, 5 = very confident). Finally, they completed five multiple-choice questions (adapted from the information officially released by China's CDC) that assessed their actual level of knowledge regarding preventing coronavirus infection. These questions covered effective handwashing, disinfectant usage, mask usage, avoidance of mouth/eye/nose touching, and prevention of droplet spread.
Following an established approach (Diener et al., 2010), we constructed an index of emotional well-being by subtracting the average of the negative affect measures (anger, sadness, stress, and worry; α = .769) from the average of the positive affect measures (smile/laugh, enjoyment, and happiness; α = .691). This overall index served as the dependent variable in our analyses. We coded whether participants were residing in Hubei Province, the epicenter of the coronavirus outbreak. Because the monthly household income measures used in the two surveys differed in the number of income brackets offered, we transformed them for use in our analyses. Specifically, due to the ordinal nature of the scale items (e.g., 5,000-7,999 RMB; 8,000-11,999 RMB; … 50,000 RMB or higher), we, following an established approach (Hout, 2004), recoded each response by taking the midpoint (e.g., 6,499 RMB) of the respective income interval when a fixed-range scale item was selected (e.g., 5,000-7,999 RMB). When an open-range item was chosen (e.g., 50,000 RMB or higher), we recoded the response using the lower bound (e.g., 50,000 RMB). We then linearly transformed the value by dividing it by 1,000 so that the monthly household income is measured in units of 1,000 RMB. Furthermore, we averaged the items for perceived knowledge (r = .595) and sense of control (r = .579), respectively, into a single measure. We also counted the number of objective knowledge questions each participant answered correctly, which served as a measure of actual knowledge.