www.phidot.org

Hello- I am trying to fit multi-state robust design models in Rmark to a huge dataset. I want to fit different models for the PSI and R stratum, but can't figure out how. I know how to fit stratum specific models using 'stratum*predictor' and 'stratum:predictor' but it still means that the same predictors are included in each model. I want to fit something along the lines of:
psi:stratum1 = predictor.A+ predictor.B
psi:stratum2 = predictor.C+ predictor.D
etc.
I would greatly appreciate any help!
Cheers,
Jim Peterson

Create a 0/1 variable in the design data for the Psi parameter that is 1 for stratum i and 0 for other strata. Then use say ~d1*(preda+predb)+d2*(predc+predd)... where di is dummy (0/1) for stratum i. Now don't forget an intercept if pred variables are numeric. Also this would use a constant transition to each of the other strata (tostratum) so think carefully about what you are doing here.