I was reading this blog post ( Unadjusted Odds Ratios are Conditional – Statistical Thinking ) on conditional odds ratios and had a very basic question. In the post, Dr. Harrell states that: “The marginal OR depends on the distribution of the sex variable in the sample, and does not transport to populations with a different sex ratio than the trial enrollment achieved. It is conditional (adjusted) ORs that generalize to other populations.” But wouldn’t even the 'conditional OR depend on the marginal distribution of other variables in the trial?
Yes, we’re talking about relative quality of a measure. Both the conditional and the marginal ORs will be inaccurate when neither one uses a certain variable that is important in predicting outcomes.
3 Likes