Hey everyone! Check out this recent work by @Shakil_Kashem_Kansas_State_University , Dwayne M. Baker, Silvia R. Gonzalez, and C. Aujean Leed titled, Exploring the nexus between social vulnerability, built environment, and the prevalence of COVID-19: A case study of Chicago.
They used SafeGraph to explore social vulnerability and COVID-19 prevalence in Chicago. They focused on 58 zip codes of the city for which COVID-19 cases are reported. One limitation of this type of study is that it’s based on where the people live and not where they were infected, but SafeGraph mitigates this by leveraging anonymized and aggregated data of people when they travel outside the city or even country. Their findings showed that there was no correlation to the prevalence of COVID cases and population density, but education, household size, and percentage of Latinx population showed a positive correlation with virus prevalence over time.
Great paper ! Thank you for sharing. Given that there likely will be long- and short-term impacts to how we structure cities following COVID this type of research is really valuable. I have a few questions about your analysis if you don’t mind:
I’m concerned about the potential of differential measurement error in the COVID case rates. Testing was not only in short supply earlier in the pandemic, but one might reasonably suspect that testing was more prevalent in higher SES communities for a variety of reasons. This might create heterogenous measurement errors in the short term (i.e., between high and low SES communities in May 2020) that are attenuated or eliminated in the long term. How big of an issue do you think this is and how might it change the interpretation of your short-term and long-term estimates?
While you control for multi-occupancy in your structural model, I don’t see anything that addresses multi-generational housing. Given that cases are more severe in the elderly and more likely asymptomatic in the young, I worry that the multi-occupancy measure might not capture this important distinction. For example, students living near U-Chicago multi cause a neighborhood to appear to have high rates of multi-occupancy housing that would be substantially different in nature from multi-generational housing in a more working-class neighborhood. Can you build a measure of multi-generational housing using data from the ACS?
It’s interesting that the proportion complying with stay-at-home guidance is insignificant in the short-term but significant and large in magnitude (relative to other factors) in the long-term. I wonder how much of this is due to changes in the composition of who stayed home. Have you done a supplemental analysis looking at this?