0
$\begingroup$

How do we eliminate variables after calculating VIF ? I have read about step wise removal of variables with high VIF till we reach VIF values below 5.

On the other hand , my professor told me that it is good idea to look at ratio of standard errors and Beta coefficients after calculating VIF to remove variables. Can any one explain how should I go forward with this ratio ?

$\endgroup$
  • 1
    $\begingroup$ There is no agreement on the procedure. You could remove the hard-to-get variables and keep the easier-to-get ones for practical purposes. $\endgroup$ – user2974951 Dec 6 '18 at 9:16
  • $\begingroup$ The ratio is generally meaningless. (That might explain why I have never seen an account of a regression analysis that employed it.) For procedures that have real meaning and theoretical support, see Chapter 3 of Belsley, Kuh, and Welsch, Regression Diagnostics. $\endgroup$ – whuber Dec 21 '18 at 16:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.