8.3 Effect Modification: Stratifying vs Modelling With Interaction Term

[@15798377]

The problem with stratifying is that you cannot directly compare the OR to see if they are different, at least not without using a complicated formula. The stratified models will have a smaller sample size, and could artificially inflate OR to make it seem like there are differences, no?

[@yuralee4953]

Thanks! your explanation is so clear. Let me ask two things. The OR for whites from the stratification analysis and the OR for whites from the modeling with interaction term should be different. Is this correct? If two ORs are different, is it also possible for the two ORs to have different significance?

[@artemisrahimi4703]

can u make a video specifically on how to check for effectmodification in a logistic regression?

[@marinstatlectures]

We have a few videos covering that. This playlist is the complete course videos: kzbin.info/aero/PLqzoL9-eJTNBDAG955KrzpduiPCj8-_3m the videos for 7.X (week 7 of the course) cover that…how to test it, what it looks like, calculating the OR with effect modification, etc

[@artemisrahimi4703]

@@marinstatlectures thank u so much!

[@samuelterkperahuno5294]

what is LWT?

[@marinstatlectures]

It’s weight at last menstruation...roughly weight at conception