Testing for multicollinearity: different scales?

Hi all,

I have a question regarding multicollinearity. One of the assumptions of a logistic regression is the absence of multicollinearity. My problem is that I dont know the proper way to test this.

I have different sorts of variables, which I want to check for multicollinearity on each other.

The variables consists of dummy's (0 or 1), likert-scale (5-points), ratio (like 0,2 etc) and absolute values (1.....99). So I want to check for multicollinearity for e.g. the dumy with the likert-scales etc.

Could anyone tell me what test or tests I should perform?

Thanks in advance.


Less is more. Stay pure. Stay poor.
People typically test for VIF or tolerance, by running their model using linear regression with these test options solely just to get these statistics, then run final model back in logistic regression.
Thanks for the answer. I have another one.

When using the binary logistic regression there is the option 'Correlation of estimates'.
Using this option there is the output of a correlation matrix. Does anyone know what test is being used? And will it show significant correlation?


Less is more. Stay pure. Stay poor.
Don't use that program, but the p-value that usually comes with correlation is a test that the correlation is equal to "0", testing the null hypothesis. Would need to verify with that program's documentation.