Weighted restricted models: a new insight on sums of squars and hypotheses testing
Daniel Furtado Ferreira · Revista de matematica e estatistica · 2001
Gauss-Markov’s linear models have been used in many experimentation problems. To the unbalanced model, the sums of squares associated hypothesis are not so easily interpreted. The purpose of this paper was to present the weighted parametric restricted model and to obtain the sums of squares associated hypotheses. A 2-way crossed classification model was used. We conclude that researchers not so familiarized with statistical analysis of unbalanced data easily interpreted the presented hypotheses. W-restricted model lead to the sums of squares and associated hypotheses equivalent to those presented in type II SAS sums of squares and estimable functions.