Like Carlo says, when complicated models fail I suggest starting with simple models and gradually building them up. You may then be able to identify what is causing you grief.
In addition to what others have said, collinearity problems can occut if you use variables computed from each other. For example, if
faminc = hinc + winc
You can't include all three of faminc, hinc, and winc in your model I see things like this occasionally when the original collectors of data have added computed variables to the data and the users of the data do not understand how variables were computed.
In addition to what others have said, collinearity problems can occut if you use variables computed from each other. For example, if
faminc = hinc + winc
You can't include all three of faminc, hinc, and winc in your model I see things like this occasionally when the original collectors of data have added computed variables to the data and the users of the data do not understand how variables were computed.
Comment