Logistic Regression Notes

Odds ratio is relative to category 1 somewhat a

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: = = πjp /π1p πja /π1a πjp /πja = exp(β1j ) π1p /π1a or log (ORj ) = β1j . Odds ratio is relative to category 1. Somewhat a measure of factor effect. Wouldn’t it be easier to compare πjp with πja ? RRj = πjp πja UNM Ordinal Logistic regression Random variable Z difficult to measure (latent variable). Z could be ”income”, perhaps we know cutoff points −∞ < c1 < c2 < c3 . . . < cj −1 < ∞ Category j is established if cj −1 < Z < cj . Therefore πj = Pr (cj −1 < Z < cj ) is the probability associated to category j . Notice that larger Z values, the more ”preferable” the category. De...
View Full Document

{[ snackBarMessage ]}

Ask a homework question - tutors are online