general question about the term "margin"

general question about the term "margin"

by Laura Koth -

I first heard the term "margins" in epi 203 in a homework question that involved an RCT.

I did not understand the use of the term then. Now, in lecture Part 2a_rec_Model Coeff_Interaction EffMod slide 4, the first line reads

"Marginal effect in whole population or conditional effect in a subgroup"?

I am wondering if there are many uses for the word "margins" in epidemiology and what you mean by the term in this sentence?


thank you

Re: general question about the term "margin"

by June Chan -

Hi,

The origin of referring to “marginal effect” is bc it has to do with examining the whole group, or the “total” group, which is often summed in the margins of tables.


“marginal”,  in that sentence, means the the effect in the broader overall group, vs. conditional, which means in a sub group... or conditioned on something.

I think this is consistent with general stats terminology, beyond epid. If it says “margin” instead of “marginal”, that may be just a typo. 

https://quizlet.com/34751806/chapter-1-definitions-ap-statistics-flash-cards/


Best,

JMC



Re: general question about the term "margin"

by Laura Koth -

thank you! that clarifies things for me

Re: general question about the term "margin"

by Laura Koth -

My biostats professors slides just stated

we interpreted equality of the adjusted (marginal) effect and the unadjusted effect to indicate lack of confounding

and on another slide stated

The marginal odds ratio (ORm) is defined as the odds ratio for the effect of exercise on hypertension obtained by averaging over the values of the model predicted outcome probabilities (e.g. using margins in Stata)

I will ask him about this in light of your prior definition which seems to conflict in my mind....