observer effect on daily nest survival

[Hadu]

I am conducting a nest survival analysis to look in to an observer effect on nest survival and have a question on how an observer effect is estimated.

To get an estimate of an observer effect, I am creating an indicator variable for days on which nests were visited, as described in Rotella et. al 2004. However, I wonder how a difference in survival on visit vs. non-visit days can be estimated when the model makes no assumptions about when a nest fails during an interval. In other words, nest failure is not assigned to a specific day, but to an interval, so how can failure on a day a nest was visited be distinguished from any other day included in the interval during which the nest failed?

Thanks much for the help.

djr

Rotella et al. 2004. Modeling nest-survival data: a comparison of recently developed methods that can be implemented in MARK and SAS. Animal Biodiversity and Conservation. 27: 1-19.

pdf available at:

http://www.montana.edu/rotella/nestsurv/Rotella%20et%20al-NestSurv.pdf

[gwhite]

djr:
If you always visited nests with exactly the same interval between visits, then the observer effect would not be estimable. The nest survival model does estimate a separate survival rate for each day, and so a design matrix with the appropriate individual covariate that would be 1 for the days after a visit and 0 otherwise would only affect the days immediately after a visit. To estimate an observer effect, you definitely need to not have equal intervals. The information for estimating the observer effect is then coming from whether nests are more likely to be destroyed during short intervals between visits, i.e., whether there is a lower survival rate for the first day after a visit as compared to the other days during the interval.

Gary