Inference in mixed linear models with four variance components - Sub-D and Sub-DI

This work approaches the new estimators for variance components in mixed linear models Sub-D and its improved version Sub-DI, developed and tested by Silva (2017). Both estimators were deduced and tested in mixed linear models with two and three variance components; the authors gave the corresponding formulations in models with an arbitrary number of variance components but no one had ever tested their performances in models with more than three variance components. Particularly, here we aim to give the explicit formulations for both Sub-D and Sub-DI in models with four variance components, as well as a numerical example testing their performances. Tables containing the results of the numerical example will be given.