Vaccine hesitancy is a significant impediment to global efforts to vaccinate against the SARS-CoV-2 virus at levels that generate herd immunity. In this article, we show the utility of an inductive approach - latent class analysis (LCA) - that allows us to characterize the size and nature of different vaccine attitude groups; and to compare how these groups differ across countries as well as across demographic subgroups within countries. We perform this analysis using original survey data collected in the US, UK, and Canada. We also show that these classes are strongly associated with SARS-CoV-2 vaccination intent and perceptions of the efficacy and safety of the COVID-19 vaccines, suggesting that attitudes about vaccines to fight the novel coronavirus pandemic are well explained by latent vaccine attitudes that precede the pandemic. More specifically, we find four substantive classes of vaccine attitudes: strong supporters, supporters with concerns, vaccine hesitant, and "anti-vax" as well as a fifth measurement error class. The strong "anti-vax" sentiment class is small in all three countries, while the strong supporter class is the largest across all three countries. We observe different distributions of class assignments in different demographic groups - most notably education and political leaning (partisanship and ideology).
Keywords: COVID; anti-vax; latent class modeling; vaccine hesitancy; vaccines.