Please provide complete references. Rory Wolfe's command -ocratio-
uses a "forward stopping" continuation ratio model, P[Y=j|Y>=j] (i.e.,
like "Armstrong and Sloan" below), but with reverse sign of the slope
compared with logistic regression on expanded data so that the sign of
the slope is the same as in the usual proportional odds model.
Anders Alexandersson
andersalex@gmail.com
Jon Heron (ALSPAC) <Jon.Heron@bristol.ac.uk> wrote:
I would like to fit a continuation ratio model to a 6-level outcome
using ocratio. I would then like to assess the constancy of model
parameters by fitting a series of logistic models
A quick lit search is showing variability in which comparisons are being
made with a CR model.
e.g. for 3-level outcome - 1/2/3
Armstrong and Sloan (1989)
1 versus 2 & 3
2 versus 3
Manor, Matthews and Power (2000)
1 versus 2
1 & 2 versus 3
Greenland (1994)
2 & 3 versus 1,2 & 3
3 versus 2 & 3
I imagine these are all permitted alternatives, but i would like to know
which particular comparisons are made using ocratio so I can fit the
appropriate logistic models.
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/