On Oct 6, 2010, at 10:23 AM, 笑啸 wrote: > dear:teacher > i have a problem which about the polr()(package "MASS"), if the > response must have 3 or more levels? > and how to fit the polr() to 2 levels? Wouldn't you just use rms::lrm or glm( , family="binomial") in that case? -- David.