Mixtures of regressions for size estimation of heterogeneous populations

We propose a capture-recapture model which exploits finite mixtures of logistic regressions to account for latent heterogeneity between groups of individuals, in order to better understand their different propensities to the capture as well as different behavioral patterns. The additional behavioural variation in capture probabilities among individuals within a group is expressed by a suitable time-dependent covariate, which summarises the past individual experience. A real data example and a simulation study illustrate how the proposed model performs.