Multi-state model with transition prob = 0

[rasrage]

Hi all,
I would like to analyze robust design CMR data collected over multiple years at multiple study sites, which were subject to different kinds of treatment. I am interested in obtaining survival and abundance estimates for each site (i.e., understand the effect of treatment on these parameters), but I would also like to be able to share some parameters across sites (for example, climate might affect survival at all sites similarly). The multi-state robust design model (with the study sites representing the states) seems like a logical choice for this task. However, there is no movement between sites in this data set. In practice, I know I can fix state transition probabilities to 0 (and fix the probabilities of remaining in one's state to 1). Does that seem like a sensible approach, or is there a simpler/better option to perform an integrated robust design analysis on multi-site data (I know I could just analyze sites separately, but that would not allow me to share parameters across sites; and if I included "site" as an individual covariate in a single-state RD model, I would still not be able to get site-specific abundance estimates)?
Thank you
Rahel

[jlaake]

Why not use just a Robust design and use sites as groups?

[rasrage]

That sounds like the simpler solution, Jeff. Thanks. I wasn't sure whether a group covariate would translate to group-specific abundance estimates.