The conflict between measurement and selection invariance revisited: a response to Borsboom et al.

What is it about?

This note addresses issues related to fairness in testing. A previous article by Borsboom et al. has shown that two plausible ways of formalizing the idea of test fairness are inconsistent with each other. That is (under certain assumptions), if one notion of fairness holds, then the other fails, and vice versa. Here we note some minor errors in the mathematical proofs of Borsboom et al. and, more importantly, show that their findings hold even if the assumptions are significantly weakened. We also point out that in the intervening decade, similar results have been found in several other fields. We provide a brief comparison and highlight the potential for psychometrics research to contribute to other fields.

Featured Image

Photo by Dawid Małecki on Unsplash

Photo by Dawid Małecki on Unsplash