Treatment of Partial Homogeneity in Population Variability Analysis of Data

摘要

This paper considers partial homogeneity of a population within the context of Bayesian population variability analyses. Partial homogeneity is defined as a condition in which within an otherwise non-homogeneous population, homogeneous sub-populations can be recognized. Partial homogeneity can be modelled using a revised likelihood model. Doing so results in an increase of the inferential strength of the available evidence and reduction of uncertainty bounds, even though a failure to recognize partial homogeneity does not invalidate the population variability analysis.