Comparison between the Parametric Mixing Distribution with Mover-stayer Model and the Nonparametric Mixing Distribution for the Analysis of Tower of London data

摘要

The Tower of London data have one serious problem. The problem is the high proportion of stayers, because parametric estimation of random effects tends to underestimate the number who are stayers. In this paper, we will use alternative estimation procedures like the parametric mixing distribution with mover-stayer model and the non-parametric mixing distribution methods. In this paper we try to answer how well these alternative procedures compare with each other?