# Testing and comparing contingency coefficients

Hello everyone,

I have administered an attachment test to the two partners of married couples and of couples in LTR; this test yields four attachment styles (nominal scale). I wish to test:

1. If there is an association between the attachment styles of the two partners. I guess I will do this by calculating a contingency coefficient (or Cramer's) for the 4x4 table (rows: attachment styles for partner A, columns: attachment styles for partner . But how can I test these coefficients for statistical significance?

2. If the association is the same for married partners and for partners in LTR. I guess I will do as in (1), but include the relationship status as a layer. But how can I compare the contingency coefficients with each other (the way a Fisher z-test would do for Pearson r coefficients)?

As for the rest, congratulations for this amazing program! A fresh way of (studying again and) doing statistics!

## Comments

Dear gvleioras,

Sorry for the tardy response. Yes, you can do exactly as you suggest. For the second problem, you can also use loglinear modeling. If you analyze this using JASP the relevant tests should become available. Have you tried?

E.J.