Parameter-expanded data augmentation for Bayesian analysis of capture–recapture models

摘要: Data augmentation (DA) is a flexible tool for analyzing closed and open population models of capture–recapture data, especially which include sources hetereogeneity among individuals. The essential concept underlying DA, as we use the term, based on adding “observations” to create dataset composed known number This new (augmented) dataset, includes unknown individuals N in population, then analyzed using model that reformulation parameter conventional observed (unaugmented) data. In context models, add set “all zero” encounter histories are not, practice, observable. augmented zero-inflated version either binomial or multinomial base model. Thus, our DA provides general approach both all types. doing so, this unified framework analysis huge range treated unrelated “black boxes” named procedures classical literature. As practical matter, by MCMC greatly simplified compared other methods require specialized algorithms. For example, complex an can be fitted with popular software packages (WinBUGS JAGS) providing concise statement model’s assumptions usually involves only few lines pseudocode. paper, review basic technical concepts data augmentation, provide examples analyses closed-population (M0, Mh, distance sampling, spatial models) open-population (Jolly–Seber) individual effects.