I have conducted a one-way ANOVA on 8 groups. I understand that the degrees of freedom for that test is 7, but I don't understand how to ascertain the degrees of freedom for the post hoc tests:

I conducted a series of t-tests as the multiple comparison post-hocs using the Holm-sidak method of familywise error rate reduction. How does one figure out what the degrees of freedom are for such tests? Is it simply n1 + n1 - 2, as it would be for regular t-tests?

If so, since my samples sizes are 893 and 169 respectively for the post-hoc t-test in question, are the sample sizes getting a little high, forcing my Type I error rate up?

  • 1
    $\begingroup$ Sidak-Holm is a step-down test. I believe df should be N-k where N is total sample size and k is number of groups. $\endgroup$
    – ttnphns
    Aug 10, 2011 at 1:35
  • $\begingroup$ Thanks for your comment! To clarify, in this case, N would be the sample size (n1 + n2), and k = 2, since only 2 groups are being compared in the post-hoc test? Or would k = 8, the original number of groups? $\endgroup$
    – Mog
    Aug 10, 2011 at 1:39
  • 1
    $\begingroup$ If you run ANOVA with 8 groups and and ordered to perform multiple comparisons on them it will be all 8 groups. $\endgroup$
    – ttnphns
    Aug 10, 2011 at 6:54


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.