*3.4. Methods*

Our panel data were derived from a national probability sample weighted on gender; age; whether the participant was born in the US; education; race/ethnicity; census region; whether the participant resides in an urban, rural, or mixed zip code; employment status; number of members in the household; and household income. We weighted observations for each participant by the final post-stratification survey weights relative to the survey mean as described in the dataset.

As our dependent variables were transformed into interval measures, we conducted cross sectional analysis through ordinary least squares (OLS) regression to determine the relationship between identification as Hispanic/Latinx and immigration status on perceptions of health risks from COVID-19, with the survey data weighted as described above. Our model can be expressed by the following:

Perceived Risk = β<sup>0</sup> + β<sup>1</sup> Hispanic Latinx + β<sup>2</sup> Immigration Status + . (1)

An alternative approach would be to use ordered logit regression to analyze the data given we have ordinal dependent variables. While we present results from ordinary least squares regression due to its more intuitive results, we also report results from ordered logit regression in Tables S1–S4 in the Supplementary Materials. In short, our results are robust to these alternative specifications, with the main findings consistent across these models.

#### **4. Results**

Collectively, the results from our analysis support our theoretical expectations that identifying as Hispanic/Latinx and as a first- and second-generation immigrant is associated with increased perceived health risks from COVID-19. However, our results also demonstrate the complexity of race, ethnicity, nativity, and risk perceptions related to health during the pandemic. Digging deeper into heterogeneity in the results provides us with insights into both the interaction between race/ethnicity and immigration on health risks, as well as important differences in the risk perceptions among different major subgroups of the Hispanic/Latinx community in the US. Just as political scientists are increasingly calling for greater attention to heterogeneity between different Hispanic/Latinx communities in politics [48–51], our results suggest this attention to heterogeneity should also guide our understanding of health risks during the pandemic.

In this section, we first present results of the perceived risk of infection before turning to the perceived risk of dying from COVID-19. We also present results from subgroup analyses, demonstrating the variation within the Hispanic/Latinx community and the need for scholars to pay close attention to different risk perceptions among these groups.
