Fig. 4.

Moderating effect of pre-COVID-19 social contact with Hispanics or Asians. Each point is based on a different regression model and provides the estimate of the interaction between the COVID-19 salience treatment and prior social contact with a specific racial/ethnic group (and the 95% confidence interval). The estimates are based on linear regressions. Interactions that are significant at the 0.05 level are highlighted in red. The samples exclude respondents of the same race/ethnicity as that of the hypothetical room-seeker. All regression models control for age, gender, race/ethnicity, education, marital status, family income, logged county population size, and region of residency.