I would like to illustrate a way which omitted variables interfere in logistic regression inference (or coefficient estimation). These effects are different than what is seen in linear regression, and possibly different than some expectations or intuitions.
This is an interesting article and there’s a really good comment helping to explain this effect in epidemiology.
Comments closed