"Correlation" between multiple dichotomous/categorical variables

[Melk]

What is the best way to measure correlation between 2 dichotmous/categorical variables? I have been using tetrachoric/polychoric, but I wonder if phi might be better. This is being used for collinearity between independent variables and association between multiple outcome variables.

[Ksharp]

If they are all category variables then correlation is called association.

Check contingency table analysis

PROC FREQ + chisq + cmh

PROC CATMOD