A likelihood analysis of quantile-matching transformations

Abstract

Quantile matching is a strictly monotone transformation that sends the observed response values to the quantiles of a given target distribution. A profile likelihood-based criterion is developed for comparing one target distribution with another in a linear-model setting.