Survival Parameter in Robust Design

[bass23]

After running several models using the robust design, my top models included survival probabilities that are very low and I didn't expect them to be that low. Does the survival parameter estimate only true survival or could it be low because several animals are permanently emigrating from my study area? This would make more sense if this was the case since we have evidence of this occurring in the population. Could anyone help me out here?

bass23

[Bill Kendall]

You are correct. Survival estimation under the robust design model is still apparent survival with respect to inherently permanent emigration (i.e., once the animal leaves it does not even 'flip a coin' to determine whether to come back. As you can imagine, if movement is temporary, but the probability of coming back is very small (i.e., almost permanent), then the results of the analysis can be questionable. To perform well, you need animals to be returning to be captured. To see a discussion of modeling where there are two types of emigration, one temporary and one permanent, see Lindberg et al. (2001, Biometrics).

[cooch]

You are correct. Survival estimation under the robust design model is still apparent survival with respect to inherently permanent emigration (i.e., once the animal leaves it does not even 'flip a coin' to determine whether to come back. As you can imagine, if movement is temporary, but the probability of coming back is very small (i.e., almost permanent), then the results of the analysis can be questionable. To perform well, you need animals to be returning to be captured. To see a discussion of modeling where there are two types of emigration, one temporary and one permanent, see Lindberg et al. (2001, Biometrics).

In fact, a careful reading of Bill's RD chapter in 'the book' would have given you the same information. The Lindberg model isn't discussed explicitly, but there are several different ways that you can 'get there from here' using MARK.