Dear all,
I expect there isn't a simple answer to this question but here goes anyway.
Say I'm building a regression model (in this case a logistic model with
a binary exposure). I have an important binary confounder that I need
to allow for.
My question is, when is it more appropriate to exclude all cases who are
positive for this confounder rather than adjusting for the variable in the
normal way?
To set the scene, the outcome is a measure of ADHD and the confounder
is developmental delay (IQ < 70).
many thanks
Jon
--------------------------------------------------
Jon Heron, PhD
Research Statistician
Avon Longitudinal Study of Parents and Children
[log in to unmask]
|