1
$\begingroup$

I understand Tukey's HSD is the most common post-hoc test for between-groups ANOVA. It appears some do not recommend or select it as a post-hoc test for within-groups ANOVA. Instead, they prefer Bonferroni test or Sidak test. But, both of those seem to just multiply whatever p-value you would get from a paired t-test by the number of hypotheses you are testing. This approach seems less than robust. Instead, if I could I would rather use Tukey's HSD test in this situation [within-groups ANOVA]. Could I do that? Or are there specific reasons not to.

$\endgroup$
5
  • $\begingroup$ What do you mean by “less than robust”? Bonferroni is always possible and guaranteed to provide strong control of the family-wise error rate. The problem is that it is often unnecessarily conservative/looses power. $\endgroup$
    – Gala
    Commented Jun 28, 2013 at 19:31
  • $\begingroup$ Not sure this addresses your question fully but you might find the following resource helpful: uvm.edu/~dhowell/StatPages/More_Stuff/RepMeasMultComp/… $\endgroup$
    – Gala
    Commented Jun 28, 2013 at 19:34
  • $\begingroup$ Gael, it is an interesting reference. But, it does not address the question. $\endgroup$
    – Sympa
    Commented Jun 28, 2013 at 20:44
  • $\begingroup$ AFAIK, only Bonferroni, Sidak and possibly Games-Howell post-hoc devices are safe to use with repeated-measures. Other traditional post-hoc tests won't do because they are ment for independent error terms. The test of spericity p-value should be 0.5 (not 0.05) or higher, for them, and this is seldom. $\endgroup$
    – ttnphns
    Commented Jun 28, 2013 at 21:18
  • $\begingroup$ ttnphns, thanks for the insightful comment. Actually, this seems like a good answer. Do you want to move it to the Answer section so we could vote on it? $\endgroup$
    – Sympa
    Commented Jun 28, 2013 at 23:42

1 Answer 1

1
$\begingroup$

Yes, you can do Tukey on a within subjects variable. You'll find the info in Keppel's book on ANOVA. SPSS doesn't include tukey for within subjects variables, but you can easily compute it by hand.

$\endgroup$
1
  • 1
    $\begingroup$ Welcome to Cross Validated. Thank you for your answer. Can you please edit you answer to expand it, in order to include the main points of the book you cite? It will be more helpful for people searching in this site. By the way, take the opportunity to take the Tour, if you haven't done it already. See also some tips on How to Answer, on formatting help and on writing down equations using LaTeX / MathJax. $\endgroup$ Commented Dec 31, 2019 at 3:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.