Nest survival-unknown fates

[jbruggin]

When using the nest survival data type, am I correct to assume that nests with unknown fates, or, for my purposes, radio-marked animals that disappear, be coded as 0 for successful (as in still active/alive on the last day checked) because exposure days are key, or is an “unknown” category needed?

Thanks.

John

[gwhite]

John:
As long as you KNOW the nest was alive on the last day checked, then it is okay to code the fate as successful. However, if the censoring is occuring BECAUSE of the radio's failure (e.g., a deer hit by a car that also destroys the radio), then you are going to get a survival estimate that is biased high. When data are censored, you either assume that the animal was still alive and the censoring was not associated with the animal's fate (and code the fate as alive), or else assume that the censoring occurred because the animal died, and code the fate as dead.
Either way, you have to make an assumption -- censoring of animals that disapper requires an assumption about either the independence of the fate and censoring, or the dependence of the fate and censoring.
Gary