At present, likelihood ratios for two-level models are determined with the use of a normal kernel estimation procedure when the between-group distribution is thought to be non-normal. An extension is described herefor a two-level model in which the between-group distribution is very positively...