0
$\begingroup$

If not then where are we supposed to use that? If yes, then my vif values are all below 3, but statistically they are insignificant variables then what should I do next? How do I make myself understand this in easy terms? How do I remove the variables which creates the same effect?

$\endgroup$
  • $\begingroup$ 1) Exclude very strongly correlated variables. 2) Build your model with a model selection procedure. 3)Check the vif's and other diagnostics to check the quality of your model. $\endgroup$ – Knarpie Sep 11 '17 at 11:42
  • $\begingroup$ @Knarpie - Okay, will do that. If some variables are statistically insignificant but VIF factor is lower than 5, does that mean I have good model? $\endgroup$ – Osro_db40 Sep 11 '17 at 12:48
  • $\begingroup$ I'm not sure you completely understand the VIF. First eliminate the insignificant variables in your model building process, then check the VIF's of the final model, together with other diagnostics. $\endgroup$ – Knarpie Sep 11 '17 at 12:54
  • $\begingroup$ Okay, will learn more about that. $\endgroup$ – Osro_db40 Sep 11 '17 at 22:59
  • $\begingroup$ I fear there is some bad advice here. In general, you should not remove insignificant variables (see: Algorithms for automatic model selection) that's not what hypothesis tests are for. In addition, what to do w/ highly correlated variables is a substantive judgement, not something to be performed algorithmically (see: Logistic regression and inclusion of independent and/or correlated variables). $\endgroup$ – gung Sep 13 '17 at 16:01

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.