ABSTRACT
This chapter applies the developed tools to calculate the prediction variance for a log-linear model to model the capture-recapture situation in a multiple-list or multiple-source setting. It also applies the imputed bootstrap to the deer mice example. The chapter finds using a bootstrap replication of 5000 resamples, a prediction standard error of 1.78 which compares well with the analytical approach taken in 26.3, being slightly on the conservative side. The bootstrap approach also offers a better way to compute prediction intervals, for example a 95% prediction interval based upon the percentile method delivers 0–6.88 which deviates from the normal approximation 95% prediction interval of 0–4.9. A classical Lincoln- Petersen experiment requires sampling the target population on two occasions at which each sampled unit of the target population is marked as identified at occasion 1 or 2 or both, respectively.
