I have pre/post data where all subjects underwent the intervention. The outcome is a score. I am also looking to stratify my analysis by gender. I understand I can stratify my paired t-test in SAS via a simple class statement, but how do I determine if the difference between the genders are significant? I would normally use an interaction for this but given the data is paired and I am using a t-test, i am not sure how to test the significance of the difference.

  • $\begingroup$ Why not just conduct a t.test between the original data of the two groups based upon gender (that is, not just a t.test between male pre and post treatment and female pre and post treatment, but pre-treatment male and female and post-treatment male and female)? $\endgroup$ – Tavrock Mar 8 '17 at 5:33
  • $\begingroup$ well if i would to approach it like that, i would probably need to do a t test between genders for the difference, but then do i have to adjust for baseline scores? $\endgroup$ – Lucy Mar 9 '17 at 21:45
  • $\begingroup$ The t.test does not rely on previous estimates of the distribution. You could also do an F.test through an ANOVA, which would be able to look at the interactions as well. $\endgroup$ – Tavrock Mar 9 '17 at 22:20
  • $\begingroup$ If i decided to use an F.test then would my outcome be the difference, and would I need to adjust for the (pre) scores? $\endgroup$ – Lucy Mar 9 '17 at 23:37
  • 1
    $\begingroup$ I do not believe that you would need to adjust for the pre scores for either test. In either case, the test is between the subgroups and detecting if the is a statistically significant difference between the samples. $\endgroup$ – Tavrock Mar 10 '17 at 0:01

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.