bad GOF but not too bad models

[jdanadon]

hello,

i have data set with terrestrial tortoises (low detectability).
models in MARK seems to be not too bad, however i can't do the TEST2
of GOF due to lack of data ¿how much credible are the models?

second question ¿how can i see the %explained deviance of the models?
(not just the absolute values of deviance of the models)

yours sincerely,


jose daniel anadon (spain)

[cooch]

i have data set with terrestrial tortoises (low detectability).
models in MARK seems to be not too bad, however i can't do the TEST2
of GOF due to lack of data ¿how much credible are the models?

Read chapter 5 of the MARK book - use either the bootsrap or the median c-hat approach to assess fit to whatever general model you can reasonably fit to your data. The general model in the candidate model set need not be the fully time-specific model - it merely should be a model which adequately fits the data, and has the relevant factors of interest in it.

second question ¿how can i see the %explained deviance of the models?
(not just the absolute values of deviance of the models)

If I understand your question correctly, you want to read on ANODEV in the MARK helpfile.