regression models for ordinal responses a review of methods

Discussion choice of an ordinal model the choice

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: fficients corresponding to a covariate (such as midline episiotomy) is allowed to vary by every level of the ordinal response. The regression coefficient corresponding to the polytomous model for the first logit comparison (1 ° versus no laceration) is negative (–0.4661), resulting in a relative risk estimate of 0.6 (95% CI : 0.3–1.3). However, the coefficients for other logit comparisons are all positive, implying that midline episiotomy increases the likelihood of lacerations of 2°. Women with midline episiotomy are 5.4 (95% CI : 3.7–7.9) times at greater risk of a 4° laceration relative to women with no episiotomy. Notice the monotonic increase in the regression coefficients for midline episiotomy. The results of fitting an adjacentcategory logit model are somewhat similar to that of the polytomous model, although the underlying model assumption differ; the latter model involves comparing women with 1° to no laceration (first logit), 2° to 1° (second logit), and so on, while the logit comparisons for the polytomous model correspond to each category of laceration versus a ba...
View Full Document

This document was uploaded on 02/25/2014.

Ask a homework question - tutors are online