regression models for ordinal responses a review of methods

The regression coefficient corresponding to the

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: onstraints specified a priori: 1 = 0, 2 = 1, 3 = 2, and 4 = 7. These resulted in the following log-ratios: β, β + γ , β + 2 γ , and β + 7 γ , for the four logits, as described earlier. Our choice of constraints were based on examining the log odds ratios from the observed data, which were derived by constructing four 2 × 2 tables, with episiotomy (yes/no) as the two rows, and lacerations ‘any’ versus ‘none’ as the columns for the first table; 2° – 4° versus none plus 1° for the second table, and so on. A simultaneous test of H0: β = 0, γ = 0 (based on 2 d.f.) resulted in a χ2 = 85.6, implying good fit. Notice that the choice of different constraints will produce different parameter estimates and standard errors. The results of fitting the polytomous logistic and the adjacent-category logistic models are summarized in Table 6. The formulation of these models is more flexible when compared to the proportional odds and continuation-ratio models, in that the regression coe...
View Full Document

This document was uploaded on 02/25/2014.

Ask a homework question - tutors are online