I want to test if there is multi collinearity in my dataset. This is made up of multiple ordinal independent variables. Am I correct that this should be done as shown below

Model <- lm(BinaryVar~ OrdV1+OrdV2+OrdV3+OrdV4)


This yields GVIF values and GVIF^1/(2*Df) which I believe are equivalent to VIF.

Is this the correct way to treat ordinal variables and calculate multicollinearity?

I have read that you should turn ordinal variables into dummy variables to do this here Testing multi-collinearity of ordinal independent variables in R

I have found nothing else to confirm this. Is my method of calculating VIF correct?



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.