I would like to combine data from a series ( 7 seasons) of experiments that have a split plot design with nitrogen fertilization as main plot and genotypes as subplots. Therefore I have to check homogeneity of variances for error A and error B. I have preformed the individual anovas model 1 fixed effects, for each experiment and used Bartlett´s test. But I would like to use Levene´s test being this test less sensible to departures from normality.
How can I do a Levene´s test for k-way (more than 1)Anova.??? what statements in Proc glm would I have to write????
Thanks in advance
Eugenia