i am using proc glm with contrast to compare three means (i.e., one-way anova). see my SAS code:
proc glm data=ITT; class group; model BMD=group; means group;
contrast 'A>B>C: A vs B' group 1 -1 0;
contrast 'A>B>C: B vs C' group 0 1 -1;
contrast 'A>B>C: C vs A' group 1 0 -1;
contrast 'A=B>C: A+B vs C' group 1 1 -2; run; quit;
This produces output with:
1) a p-value for the three-way comparison,
2) means for each group and their corresponding SDs
3) a mean square and p-value for each contrast
I would also like to obtain:
1) means for each group and their corresponding 95% confidence intervals (instead SDs)
2) differences between the means and the corresponding 95% confidence intervals for the differences between the means. Is it possible to obtain this using contrasts?
You would be better off using LSMEANS rather than MEANS statement (under most circumstances). Then your request is also easy:
lsmeans group / cl pdiff;
will give you confidence limits and all possible pair-wise differences.