Table 3.
The effect of the first local COVID-19 Diagnosis on racial animus Twitter Post Index.
| (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) | |
|---|---|---|---|---|---|---|---|---|---|
| VARIABLES | Ch-word | Ch-word | Ch-word | Exclude | Dynamic | Severity | Asian | Exclude | Exclude |
| index | level | per capita | counter-hate | DID | control | control | states | bots | |
| 6w | 0.075 | 0.022 | 0.061 | 0.037 | 0.070 | 0.053 | 0.127 | 0.098 | |
| (0.159) | (0.276) | (0.307) | (0.097) | (0.159) | (0.255) | (0.165) | (0.151) | ||
| 5w | 0.030 | 0.069 | 0.036 | 0.801 | 0.085 | 0.027 | 0.091 | 0.056 | 0.039 |
| (0.143) | (0.158) | (0.267) | (1.252) | (0.091) | (0.143) | (0.242) | (0.142) | (0.153) | |
| 4w | 0.098 | 0.128 | 0.117 | 0.328 | 0.025 | 0.095 | 0.248 | 0.113 | 0.075 |
| (0.140) | (0.165) | (0.240) | (1.181) | (0.107) | (0.140) | (0.239) | (0.141) | (0.144) | |
| 3w | 0.004 | 0.024 | 0.100 | 0.450 | 0.082 | 0.006 | 0.095 | 0.018 | 0.014 |
| (0.121) | (0.091) | (0.195) | (1.152) | (0.081) | (0.121) | (0.213) | (0.129) | (0.138) | |
| 2w | 0.150 | 0.065 | 0.412 | 0.361 | 0.120 | 0.149 | 0.331 | 0.136 | 0.242 |
| (0.137) | (0.050) | (0.308) | (0.967) | (0.094) | (0.137) | (0.212) | (0.146) | (0.180) | |
| +0w | 0.158 | 0.012 | 0.390** | 5.154*** | 0.120*** | 0.163 | 0.168 | 0.169 | 0.203 |
| (0.112) | (0.069) | (0.170) | (1.005) | (0.094) | (0.159) | (0.171) | (0.122) | (0.142) | |
| +1w | 0.707*** | 0.227** | 1.037*** | 5.075*** | 0.689*** | 0.718*** | 1.077*** | 0.572*** | 0.952*** |
| (0.169) | (0.105) | (0.197) | (1.046) | (0.159) | (0.166) | (0.238) | (0.162) | (0.228) | |
| +2w | 0.460*** | 0.348*** | 1.140*** | 2.855*** | 0.428*** | 0.478*** | 0.763*** | 0.389** | 0.538*** |
| (0.142) | (0.109) | (0.252) | (1.039) | (0.111) | (0.145) | (0.199) | (0.151) | (0.173) | |
| +3w | 0.297** | 0.631*** | 1.331*** | 2.688*** | 0.181* | 0.315** | 0.526** | 0.300* | 0.255* |
| (0.141) | (0.193) | (0.396) | (0.842) | (0.095) | (0.152) | (0.204) | (0.154) | (0.137) | |
| +4w | 0.286* | 0.789** | 1.947** | 1.521 | 0.122 | 0.307* | 0.361 | 0.273 | 0.132 |
| (0.173) | (0.310) | (0.771) | (1.257) | (0.103) | (0.184) | (0.269) | (0.187) | (0.157) | |
| +5w | 0.394* | 0.683*** | 1.650*** | 1.158 | 0.240 | 0.421* | 0.535* | 0.385 | 0.144 |
| (0.221) | (0.201) | (0.466) | (1.396) | (0.154) | (0.248) | (0.323) | (0.240) | (0.178) | |
| +6w | 0.459** | 0.696*** | 1.664*** | 2.264 | 0.340** | 0.489* | 0.533* | 0.479** | 0.373* |
| (0.222) | (0.223) | (0.469) | (1.566) | (0.150) | (0.252) | (0.315) | (0.243) | (0.198) | |
| Observations | 7930 | 7976 | 7976 | 3141 | 103,694 | 7930 | 5578 | 7188 | 11,811 |
| R-squared | 0.121 | 0.809 | 0.270 | 0.611 | 0.112 | 0.121 | 0.142 | 0.123 | 0.060 |
| Outcome mean | 0.591 | 0.681 | 1.075 | 6.779 | 0.591 | 0.591 | 0.591 | 0.591 | 0.569 |
Notes: The table presents the effect of the first local COVID-19 diagnosis on the prevalence of ch-word tweets in an area. All columns report the estimates of coefficients on the event dummies in Eq. (3), except for column (5). Column (1) corresponds to Fig. A4, panel B. The outcome variable in column (2) is the number of ch-word tweets, and the regression controls for the number of “the” tweets. The outcome variable in column (3) is the number of ch-word tweets per one million county population. Column (4) uses an alternative Twitter post index, which removes counter-hate tweets (see Section 4.2.1). Column (5) presents the estimates from a dynamic DID event study (Sun and Abraham, 2020). Column (6) controls for the number of COVID-related new cases and deaths, and whether the state has any stay-at-home orders in place. Column (7) controls for the Twitter post index for “Asian(s).” Column (8) excludes Washington, New York, and California. Column (9) excludes tweets from users who are likely Twitter bots. All regressions control for county fixed effects and year-month fixed effects. Standard errors are clustered by county. *** 0.01, ** 0.05, * 0.1.