On Tue, 26 Apr 2005 19:40:18 +1000, "CW" <CW@nospam.com> wrote:
> Under OLS, if an explanatory variable is not included in the model, the > resulting estimates would be biased. > > Under logistic regression, MLE is used. I wonder whether missing variable > would also cause misspecifications as well. > > thanx in advance
Cause bias? Certainly.
In data of the 1950s, lung cancer was correlated with sex; unless you thoroughly removed the effect of "How much do you smoke?"
That's a logical proposition which does not depend on the method of solution.