Table 4. Vaccine increase per county: Heterogeneous effects and causal responses.
Table reports the difference in the estimated effect in low versus high counties based on a given county characteristic. This characteristic is the 2016 Trump vote share in columns 1 and 2, the fraction of county residents with a college degree in columns 3 and 4, and the fraction of county residents who are white in columns 5 and 6. High refers to counties that are above the median level for that characteristic, and low refers to below the median, where the median is computed across counties in our sample. In odd columns, the effect is the ITT effect and in even columns it is the ACR. “***,” “**,” and “*” indicate significance (from a two-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 two-tailed test based on 1000 permutations using the effect in low-relative-to-high counties as the randomization test statistic. Table S5 contains estimates of other coefficients from these regressions.
| % Trump | % College | % White | ||||
|---|---|---|---|---|---|---|
| (1) | (2) | (3) | (4) | (5) | (6) | |
| ITT | ACR | ITT | ACR | ITT | ACR | |
| Effect in low relative to high county | 258.7* (154.5) |
17.28* (11.24) |
−56.76 (158.5) |
4.496 (10.84) |
216.3* (152.3) |
12.15 (12.18) |
| County fixed effects | Yes | Yes | Yes | Yes | Yes | Yes |
| Date fixed effects | Yes | Yes | Yes | Yes | Yes | Yes |
| Randomization inference P value | 0.096 | – | 1.00 | – | 0.149 | – |