Hi David & Rune

Sorry if I have not used the correct way to respond but I only get the digest so I'm not sure how to do it properly.

Anyway, to the point.

The only way I know how to fit cumulative link functions for mixed-model polytomous ordered logistic (or probit, complementary log-log) links using the GLM fitting algorithm is to use composite link functions as originally published for multinomial data by Thompson and Baker (1981) Applied Statistics 30, 125-131. The generalisation to mixed models using GLMMs is given ("buried in") in Candy 1997 Biometrics 53:146-160 (see pg 152) and uses a Poisson distn with constraints imposed on fitted values to sum to observed marginal frequencies. This composite link function approach is a bit messy to set up, so it might be less trouble to program it directly as simply an nlme problem. Also if the categories are not ordered this approach breaks down because its not possible to build the appropriate constraints into the GLMM fitting algorithm.

HTH, Steve

-------------------------------------------------------------
Steven G Candy
Senior Applied Statistician
Southern Ocean Ecosystems
Australian Antarctic Division
Kingston, Tasmania 7050
Australia
ph +61 (0)3 6232 3135




___________________________________________________________________________

    Australian Antarctic Division - Commonwealth of Australia
IMPORTANT: This transmission is intended for the addressee only. If you are not the
intended recipient, you are notified that use or dissemination of this communication is
strictly prohibited by Commonwealth law. If you have received this transmission in error,
please notify the sender immediately by e-mail or by telephoning +61 3 6232 3209 and
DELETE the message.
        Visit our web site at http://www.antarctica.gov.au/
___________________________________________________________________________

	[[alternative HTML version deleted]]