For the DOS, use the three-level depression diagnosis and variable MS for marital status as defined in Section4.2.2. If MS is not ordered, we may not be able to talk about the uniform association. However, we may assess MS-effect association, the roweffect association when MS is treated as the row variable, as the depression diagnosis is ordered.

(a) Use goodness-of-fit tests to assess if the row-effect model is acceptable.

(b) Use the saturate model to assess if the row-effect model is acceptable.

(c) Use a generalized (or adjacent) logistic model to assess if the row-effect model is acceptable.