For a 2x2xk table, where k are strata from repeated cross-sectional studies over the course of 10 years, would it be suitable with a Mantel-Haenszel test of common odds ratio or would a logistic regression analysis be more apropriate? The 2x2 part are purely nominal data.


1 Answer 1


With large strata they should be identical. Only subtle differences arise between the two tests, more discussion here.

I would prefer the logistic regression model because it could be extended to use fixed or random effects to account for age / period / cohort effects of study follow-up.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.