Which is the best way to check for multicollinearity between two binary explanatory variables in logistic regression..? I use SPSS, if anyone could answer with special regards to that programme it would be extra helpful!
$\begingroup$
$\endgroup$
1
-
$\begingroup$ Note that the candidate for duplicate had a misleading title, as "dependent" really meant "independent". (Further ammunition for a personal campaign that these terms are past their sell-by date.) $\endgroup$– Nick CoxCommented May 10, 2017 at 17:20
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
Can you use R? You can quickly explore data by using CORRPLOT
Also running a PCA would be helpful to see how much redundancy/multicollinearity are in your features.
Finally, you can try to fit with a linear model and check if the data matrix is full rank / invertable.
answered Jun 22, 2016 at 0:25