I am trying to analyze a dataset that has two dichotomous variables (placstat and hiv) predicting a continuous variable (lplacwbc). The data are unbalanced. I am interested in seeing a matrix for pairwise comparisons, with p values, for all possible comparisons. The following code yields the output provided (not showing the model output, only lsmeans output). I may be missing something simple here, but I cannot figure it out!   proc glm; class placstat hiv; model lplacwbc=placstat hiv placstat*hiv/ ss1 ss2 ss3 ss4; lsmeans placstat hiv/pdiff=all adjust=tukey; run;       ... View more