Table 1. Vaccine increase per county.
Regression results. Sample size is 151,945 county-date observations. All regressions include fixed effects at the county and date levels. In columns 1 and 2, the ITT effect corresponds to the OLS-estimated coefficient on the interaction of a treatment assignment dummy (Treat) with a dummy for dates after October 14 (Post), the start of the campaign. Column 1 also includes the interaction of Post with county population. Column 2 replaces this with interactions of county population with (i) flexible dummies for each date within 2 weeks before to 2 weeks after the campaign (omitting the date before the campaign started), (ii) a dummy variable for 2 weeks or more before, and (iii) a dummy variable for 2 weeks or more after. Columns 3 and 4 report the IV-estimated coefficient on the interaction of the number of ads the county received (in 1000s) with Post, with this interaction instrumented for by Treat × Post. Column 3 mimics column 1 in controlling for differential trends by population, and column 4 mimics column 2. “***,” “**,” and “*” indicate significance (from a one-tailed test) at the 0.01, 0.05, and 0.10 levels. Standard errors, reported in parentheses below each estimate, are clustered at the county level. Randomization inference P values are from a one-tailed test based on 1000 permutations using the treatment effect as the randomization test statistic. Table S4 contains estimates of other coefficients from these regressions.
| ITT effect | ACR of 1000 ads | |||
|---|---|---|---|---|
| (1) | (2) | (3) | (4) | |
| Effect | 102.6* (78.74) |
101.4* (78.76) |
8.606* (6.608) |
8.500* (6.609) |
| Implied vaccines per dollar | 1.08* (0.828) | 1.07* (0.828) | 1.01* (0.773) | 0.99* (0.773) |
| County fixed effects | Yes | Yes | Yes | Yes |
| Date fixed effects | Yes | Yes | Yes | Yes |
| Population × Post dummy | Yes | Yes | ||
| Population × Date dummies | Yes | Yes | ||
| Randomization inference P value | 0.067 | 0.065 | ||