A. Analysis of the Coreopsis experiment.
PROC GLM; /* use GLM for unbalance designs */
CLASS TREAT; /* TREAT = treatments */
MODEL NUM MASS = TREAT/SS3; /* use type III sums of squares for unbalanced designs */
CONTRAST 'H VS M' TREAT -1 1 0; /* contrast statements for obtaining the tests of pairwise differences */
CONTRAST 'H VS L' TREAT -1 0 1;
CONTRAST 'M VS L' TREAT 0 -1 1;
MANOVA H=SITE/PRINTH PRINTE; /* MANOVA statement for obtaining the multivariate and the coefficients analysis of the greatest characteristic roots. PRINTH and PRINTE request the among-group and within-group SS&CP matrices */
MANOVA H=SITE/CANONICAL; /* alternative MANOVA statement for obtaining the multivariate analysis and the canonical coefficients */
B. Analysis of the protozoan-bacteria experiment.
CLASS COLPODA CYCLID BODO;
MODEL A B C D = COLPODA|CYCLID|BODO; /* the "|" specifies a model with all possible interactions */
MANOVA H=_ALL_/CANONICAL; /* the "_ALL_" requests multivariate tests for all model effects */