*
chapter_9_table_7

The following data represent a three group pre-post design, where 30 depressives
were randomly assigned to one of three conditions. The primary question of 
interest is: "do individuals in some groups change more on their measures of 
depression than do individuals in other groups?"

As before, we will make use of the GLM procedure. The difference in the case of an
ANCOVA as compared to an ANOVA is that the effect of the covariate will be included 
on the right hand side of the equal sign, as was the case previously with the grouping 
factor. The difference for covariates is that they are (generally) not classification 
variables but quantitative variables and are thus not included on the class line as 
grouping variables generally are. 

Note that the /SOLUTION option is given on the model line. This is necessary in order 
for SAS to print the parameter estimates. Also notice that we have included an
LSMEANS option. The LSMEANS option outputs the adjusted mean for the class variable(s)
that are specified. The options following the the "/" on the LSMEANS line are for 
pairwise adjusted mean comparisons (obtained by including PDIFF) and that the 
confidnece limits that are adjusted using Tukey's WSD (or HSD) method (obtained 
by includeding ADJUST=TUKEY).

The results given by the SAS syntax below yield the results given in Tables 9.8 and 9.9
and 9.10.
;

LIBNAME md 'd:\data files by type\sas data files\tables';
     Data c9t7;
       SET md.chapter_9_table_7;
PROC GLM DATA=c9t7;
CLASS cond;
MODEL post = cond pre / SOLUTION;
LSMEANS cond / PDIFF CL ADJUST=TUKEY;
RUN;