When I run the VIF command after having run the Corr command STATA tells me "not appropriate after regress, nocons; use option uncentered to get uncentered VIFs". What does that mean? And also, the Hausman Test tells me to use the fixed effect model, but when I use it, STATA omits my dummy variables in the result. "Omitted because of collinearity" What can I do now?
You can do your on VIF. Just regress all explanatory variable one by one on the set of all other explanatory variables. Obtain the R squared. The VIF is then 1/(1-R squared). Or add a constant to your regression model. The implementation of the VIF command is a bit odd in Stata. Second problem: the fixed effects dummies seem to explain other dummies in the model. Stata drops these to avoid collinearity (the inverse of the data matrix does not exist as the matrix does not have full rank). Check the definitions of your dummies
The full course is available for USD 3.59 here www.yunikarn.com/p/data-science-using-stata-complete-beginners-course, including datasets and Stata do files. The slides are available as an E-book (129 pages) gerhard-kling-s-school.teachable.com/p/my-downloadable-14033 - combined offer: USD 5.40.
Straight forward and to the point. Thank you!
A pleasure. Thanks for watching!
Worth Watching
Thanks, Sourav. This is much appreciated!
When I run the VIF command after having run the Corr command STATA tells me "not appropriate after regress, nocons; use option uncentered to get uncentered VIFs". What does that mean? And also, the Hausman Test tells me to use the fixed effect model, but when I use it, STATA omits my dummy variables in the result. "Omitted because of collinearity" What can I do now?
You can do your on VIF. Just regress all explanatory variable one by one on the set of all other explanatory variables. Obtain the R squared. The VIF is then 1/(1-R squared). Or add a constant to your regression model. The implementation of the VIF command is a bit odd in Stata. Second problem: the fixed effects dummies seem to explain other dummies in the model. Stata drops these to avoid collinearity (the inverse of the data matrix does not exist as the matrix does not have full rank). Check the definitions of your dummies
The full course is available for USD 3.59 here www.yunikarn.com/p/data-science-using-stata-complete-beginners-course, including datasets and Stata do files. The slides are available as an E-book (129 pages) gerhard-kling-s-school.teachable.com/p/my-downloadable-14033 - combined offer: USD 5.40.