Luna

ALT Dear Staff, As the director of the Health Psychology doctoral program at Tulane University, I have substantial expertise in psychometrics as well as health justice. I strongly discourage switching the census disability questions to Test Version Definition 1. Using Definition 1 will artificially underestimate the prevalence of disabilities, reducing estimates from 13.9% to just 8.1%. Such a change would occur at an important historical time point where many people are developing new health conditions as a result of Long COVID. Thus, a change now that minimizes disability prevalence estimates would potentially conceal very serious societal shifts in health. Definition 2 suggests a prevalence rate of 31.7%, which is disconnected from prior estimates. I would suggest going back to the drawing board. The text anchors used in the Test Version appear flawed. If you include "Some difficulty" in the prevalence estimate, it's an overestimate. If you exclude, it's an underestimate....

ALT Comparison of the old survey (Control) versus the newly proposed survey (Test Version)

ALT The devil is in the details. The new version is more descriptive because it has more options, but the challenge is how to score it. When you count any level of difficulty as disability, the prevalence shoots up to 31.7%. If trying to only count the two most extreme levels of difficulty, it falls to 8.1%. The estimate is 13.9% under the old survey, so either scoring system is problematic. Likely, the response options need to be revised carefully.