Mapping culture with latent class analysis: A response to Eger and Hjerm

Abstract Eger and Hjerm's methodological critique of our 2016 study of Americans' sentiments towards the nation asserts that the latent class (LCA) models employed in our paper did not fit the data and that consequently, the paper fails to demonstrate the existence of multiple varieties of American nationalism. We challenge E&H's analyses and argue that their conclusions stem from erroneous assumptions, both about our models and about best practices for applied LCA‐based research. Based on a review of their results and additional analyses carried out with their preferred measures, we demonstrate that our model choices were justified and our 2016 findings are robust. In so doing, we offer a critique of unreflective adherence to inappropriate model fit criteria that ignores theory and concerns over the parsimony, interpretability, construct validity and external validity of model results.