National identification and COVID-19 response: Calibrating replication claims made by Van Bavel et al.

In their impressive large-scale project, Van Bavel et al.1 reported global patterns of associations between national identification and public health support in two studies. Their findings indicate that higher levels of national identification are associated with greater COVID-19 protection measures both at the individual level (Study 1; N = 49,968, across 67 countries) and country level (Study 2, across 42 countries). The authors framed the country-level correlation reported in Study 2 as a conceptual replication of individual-level correlations from Study 1. As known from scholarly work2-6, associations at the individual level of analysis do not automatically imply the same associations will be observed at the aggregated level, and vice versa. Here I re-analyze their data to show that posing a country-level study as a conceptual replication of an individual-level study might not be warranted because it relies on untested equivalence assumptions that are not supported when empirically examined.