Hi all,

I am working on a pooled OLS to get fitted values of the regression. However, Stata gave me some negative predicted values which is not possible by definition of my dependent variable. Specifically, the dependent variable varies between 0 and 1(there are number of zeros, too).

Then I used GLM with long link and all of my fitted values are positive now. Is it acceptable to use GLM just because of this reason or am I violating some assumptions?

Thanks.