Table 2.
Predicted margins of desire to call the police and perceived threat (N = 2038)
| Model | (1) | (2) | (3) | (4) |
|---|---|---|---|---|
| Outcome | Desire to Call the Police | Perceived Threat | ||
| Regression type | Simple | Multiple | Simple | Multiple |
| White Low-Severity Vignette | 41.5 | 42.5 | 38.3 | 38.7 |
| (1.37) | (1.29) | (1.20) | (1.11) | |
| Black Low-Severity Vignette | 42.0 | 42.0 | 38.5 | 38.2 |
| (1.42) | (1.33) | (1.25) | (1.14) | |
| White High-Severity Vignette | 81.9 | 81.5 | 72.5 | 72.5 |
| (0.97) | (0.96) | (1.02) | (1.01) | |
| Black High-Severity Vignette | 81.5 | 80.8 | 74.1 | 73.7 |
| (1.04) | (1.05) | (0.98) | (0.98) | |
| Covariates | ||||
| Demographics | No | Yes | No | Yes |
| Community views | No | Yes | No | Yes |
| Police views | No | Yes | No | Yes |
| State fixed effects | No | Yes | No | Yes |
Robust standard errors. Bolded confidence intervals indicate that the marginal contrast between the adjoining group is significant (p < 0.05). Predicted margins can be interpreted as the average conditional mean of desire to call the police or perceived threat, on a [0,100] interval, after fractional logit regression on covariates (if any)