Multiple Pairwise T-test Comparisons against a reference group

pkfamily — Fri, 14 Apr 2017 18:50:48 GMT

Is there a way to do multiple pairwise T-test comparisons against a reference group? For example, let's say our group is race with white, black, and hispanic. We then choose our reference group to be white. And our continuous variable to be blood pressure. What is the best way to get the "Diff (1-2) Pooled" output from proc ttest for white vs black and white vs hispanic?

In the end, I want to get the difference means, CL, and P-value for multiple variables (race, education, income, etc..) and ouput into a table. (But I can figure this part out myself)

Re: Multiple Pairwise T-test Comparisons against a reference group

pkfamily — Fri, 14 Apr 2017 21:09:43 GMT

Looks like the answer is Proc GLM with the lsmeans statement and options cl, pdiff=Control, and adjust=t.

Proc GLM data=mydata;

class race;

model bp=race;

lsmeans race/cl pdiff=control adjust=t;

run;

topic Re: Multiple Pairwise T-test Comparisons against a reference group in Statistical Procedures

Multiple Pairwise T-test Comparisons against a reference group

Re: Multiple Pairwise T-test Comparisons against a reference group