regression models for ordinal responses a review of methods

Model logitpr y yj j episiotomy table 5 maximum

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: us 1° – 2°, and 4° to none plus 1° – 3° are β + γ2, β + γ3, and β + γ4, respectively. The corresponding log-odds ratios with the standard errors are presented in Table 5. A 1329 REGRESSION MODELS FOR ORDINAL RESPONSES TABLE 4 Maximum likelihood estimates: Results of fit of proportional odds (PO model) and continuation-ratio (CR model) modelsa Variable PO model ˆ β ± se (β ) ˆˆ Intercept1 (α1) Intercept2 (α2) Intercept3 (α3) Intercept4 (α4) Midline episiotomy P-value –2.8452 ± 0.0336 –3.0363 ± 0.0363 –3.1680 ± 0.0383 –5.0027 ± 0.0883 0.7423 ± 0.0946 52.9 Model fit Likelihood Ratio test, χ2 1 CR model ˆ β ± se (β ) ˆˆ P-value 0.0001 –3.9387 ± 0.0484 –0.8260 ± 0.0213 –0.8260 ± 0.0213 –0.8260 ± 0.0213 0.2968 ± 0.0345 0.0001 0.0001 74.7 0.0001 a Response variable is degree of laceration: no laceration, 1°, 2°, 3°, and 4°. Model: logit[Pr (Y yj)] = αj – β (Episiotomy). TABLE 5 Maximum likelihood estimates: Results of fit of partial-proportional odds modelsa Variable Parameter a. Unconstrained model Intercept1 Intercept2 Intercept3 Intercept4 Midline episiotomy Laceration 2° Laceration 3° Lacer...
View Full Document

Ask a homework question - tutors are online