D. To check if the proportional odds assumption is
met:
i. Evaluate whether the crude odds ratios are
“close”.
ii. Evaluate whether the odds ratios from the
standard logistic models are similar:
a. Provides control of confounding but is
not a statistical test.
iii. Perform a Score test of the proportional
odds assumption.
E. If assumption is not met, can use a polytomous
model, consider use of a different ordinal
model, or use separate logistic regressions.
VII. Summary(page 481)484 13. Ordinal Logistic Regression