regression models for ordinal responses a review of methods

A situation under which this assumption does not hold

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: partial-proportional odds model6 was to relax the strong assumption of identical log-odds ratio for the Y by x1 association, in the proportional odds model. Violation of the assumption of identical log-odds could lead to the formulation of an incorrect or misspecified model. A situation under which this assumption does not hold is illustrated below. Analgesic trial data . For purposes of illustration, consider the analgesic trial data2 described in Table 1. ˆ The estimated log-odds ratios [β ], and their estimated ˆ )], for the logits are presented in standard errors [se(β ˆ Table 1, for comparisons between the drugs Z100 and EC4 versus C15 and C60. The results indicate that the ˆ log-odds ratio is largest (β = 2.6384) when the rating of the drug is dichotomized at Y = 4 ‘ less than very good’ ( 3) versus ‘very good’ (4); the dichtomization for the ˆ next largest (β = 1.5476) being at Y = 3, ‘poor or fair’ versus ‘good or very good’ (Y 3), and the log-odds ˆ ratio is smallest (β = 0.7013), when the dichtomization is made at Y = 2, ‘poor’ ver...
View Full Document

This document was uploaded on 02/25/2014.

Ask a homework question - tutors are online