4
$\begingroup$

I'm currently working on a statistics paper for practice and cannot figure out how to control for multicollinearity if almost all my variables are ordinal. VIF doesn't seem to work for instance. It cannot be that difficult, can it?

I found a helpful post from 2012 that suggests using "perturb" or "jamming your data into an ill-fitting linear model and request the collinearity diagnostics".

How to test for and remedy multicollinearity in optimal scaling/ordinal regression with categorical IVs

Now I don't quite know how to do either of this with my dataset:

Independent variables: V9 - ordinal, V19 - ordinal

Control variables: V242 (age),V240 (gender)

Dependent variables: V211 - ordinal, V214 - ordinal

My regressions: 

  polr(V211 ~ V9 + V242 + V19 + V240, data = finalvariables, Hess=TRUE)

  polr(V214 ~ V9 + V242 + V19 + V240, data = finalvariables, Hess=TRUE)

If anyone could tell me how to check for multicollinearity, I'd be very grateful.

P.S.: I had a look at the kendall tau for V9 and V19, which suggests, that I will probably end up with some kind of multicollinearity.

P.P.S.: Almost forgot: I'm using R.

Improve this question
$\endgroup$
1
  • $\begingroup$ I had a second look at it today, and the "ill-fitting" linear model has worked best so far. If anyone is familiar with perturb, feel free to add a comment or a more elaborate answer for future reference. $\endgroup$
    – Amb
    Jul 2, 2017 at 11:54

0

Reset to default

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.