Extended Data Table 3 Randomization check of the second RCT
- Average and s.d. (in parentheses) of participant demographic variables in each condition of the second RCT, the P value of a two-sided F-test for joint significance across seven conditions for each variable, and the number of participants in each condition. The demographic variables reported here include participant age (‘Age (years)’), whether a participant was a man (versus a woman, other or unknown; ‘Male’), whether a participant was white (excluding Hispanic or Latino; ‘White’), whether Spanish was a participant’s preferred language (‘Preferring Spanish’), and whether a participant received the influenza vaccine in either of two recent influenza seasons (‘Received Flu Shot in Two Recent Seasons’). We predicted each demographic variable using an OLS regression as a function of indicators for the six conditions within the follow-through reminder arm (the holdout arm as the reference group) with robust standard errors to correct for heteroscedasticity. A two-sided F-test was then conducted for the β coefficients from each regression to compare the overall significance across conditions for a given demographic variable. If a given P value is greater than 0.05, it means the corresponding F-test does not allow us to reject the null hypothesis that all seven conditions in the second RCT have the same value for the corresponding demographic variable.
