How can I test collinrarity between categorical, not continues, variables in SPSS program before running logistic regression to fulfill requirements of the logistic regression ?
Someone is liable to point out that measures like variance inflation factor (VIF) have to do with the relationships among explanatory variables, and have nothing to do with the dependent variable. So the usual response to a question like yours is to estimate your model via the REGRESSION procedure, even though the outcome variable is dichotomous, and look at VIF and the other multicollinearity diagnostics that the REGRESSION procedure provides. (Note that the suggestion is to use this linear regression model ONLY to get the multicollinearity diagnostics.)
That advice is fine when all of the explanatory variables are quantitative. But you are asking about categorical explanatory variables. In that case, I believe you would want to look at the so-called Generalized VIF. AFAIK, SPSS currently has no built in method for computing GVIF. But see the old discussion forum post linked below for some info about an R package that can be used (and called from within SPSS). If you have further questions about that, you might consider posting to the SPSSX-L mailing list (see Nabble link below), where users with expertise on calling R from within SPSS may see your post.
HTH.
https://groups.google.com/forum/#!topic/comp.soft-sys.stat.spss/h_5nlSj9P7o
http://spssx-discussion.1045642.n5.nabble.com/
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS