estimating c-hat

[cassowary43]

Hello all,

I am running a CJS model with 5 capture occasions over 1 winter with four groups (HY-burn, AHY-burn, HY-unburned, AHY-unburned...refering to bird age and habitat type). If I'm not mistaken, there are two ways to estimate c-hat with bootstrap GOF: 1) observed c-hat/mean bootstrap c-hat and 2) observed deviance/mean bootstrap deviance. #1 gives me a c-hat of 0.633 (underdispersion) and #2 gives me 1.204 (overdispersion). The GOF df = 2.64 and the observed df=7. ??? What am I doing wrong or is this usual? Should I simply assume c-hat=1?

A sample encounter history:
11001 1 0 0 1; (am I even doing this right? - I can't find in the guide how to enter >2 groups)

Thanks in advance!
Erik Johnson
MS candidate, LSU

[cooch]

Hello all,

I am running a CJS model with 5 capture occasions over 1 winter with four groups (HY-burn, AHY-burn, HY-unburned, AHY-unburned...refering to bird age and habitat type). If I'm not mistaken, there are two ways to estimate c-hat with bootstrap GOF: 1) observed c-hat/mean bootstrap c-hat and 2) observed deviance/mean bootstrap deviance. #1 gives me a c-hat of 0.633 (underdispersion) and #2 gives me 1.204 (overdispersion). The GOF df = 2.64 and the observed df=7. ??? What am I doing wrong or is this usual? Should I simply assume c-hat=1?

Re-read *all* of Chapter 5 - this sort of thing is discussed in various places (for example, section 5.8) . Also, you should consider the median c-hat approach as well - it seems to have better performance than the bootstrap (which you'd know if you'd read all of chapter 5).

A sample encounter history:
11001 1 0 0 1; (am I even doing this right? - I can't find in the guide how to enter >2 groups)

Well, if you understood chapter 2 at all, you'd realize that if if have 4 groups, then you have 4 columns, as in your example. Not surprisingly, the guide doesn't give examples of every possible permutation - it assumes you'll grasp the basic idea, and apply to your basic situation.

In fact, some careful reading of section 2.1.1 would have given you the answer.