Hi! I'm working with a data set in STATA, and I just discovered that my model has multicollinearity. I want to remediate this by removing a variable, but I want to make sure that it won't cause bias. Does anyone know how I can do this using STATA (I don't want to write it all out)?? Or what sequence of commands I have to go through to ensure that my new model is not biased?
THANKS!|||The whole problem with multicolinearity is that two variables are basically measuring the same thing. As long as you know which variables are colinear (and you can check this just by looking at their correlations) then you can safely remove one without causing bias, because the other similar variable is still measuring the same stuff. If removing a variable would introduce omitted variable bias, it wouldn't be causing multicolinearity problems to begin with.
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment