Profile likelihood and its modifications in the random effect model

S. Migliorati
In order to eliminate nuisance parameters several methods based on adjustments of the profile likelihood have recently been proposed. The present paper makes a compari-son among them analyzing their behavior in the random effect model from the analyti-cal and inferential points of view and through simulation. The conclusion is that the different adjusted versions do not behave so better as to justify the theoretical and computing difficulties they involve.