0
$\begingroup$

I want to undertake a Tukey HSD post hoc test on a Two-way Repeated Measures ANOVA in SPSS 25, but the option to select the test is unavailable.

  1. In SPSS I go to: Analyse > General Linear Model > Repeated Measures

  2. This brings up the "Repeated Measures Define Factors(s)" dialogue box where I define my factors.

  3. The "Repeated Measures" dialogue box then appears, where I transfer over my Within-Subjects Variables

  4. I click Plots which opens up the "Repeated Measures: Profile Plots" dialogue box where I set up my Horizontal Axis: and Separate Lines: and add them to the Plots box.

However, when I click on "Post Hoc", all of the options in the dialogue box are unavailable.

enter image description here

What am I missing?

Is it possible to do a Tukey test via: Analyse > General Linear Model > Repeated Measures?

Or do I have to use: Analyse > General Linear Model > Univariate?

$\endgroup$
  • $\begingroup$ Those well developed fancy post hoc tests are for comparison of independent groups (between subject factor) only. They are theoretically problematic, except Bonferroni or Sidak, to compare within subject levels. Go to button Options and request estimated means comparisons of the within s. levels, there the above two corrections are available. $\endgroup$ – ttnphns Jul 18 '19 at 16:48
  • $\begingroup$ Thanks @ttnphns. I followed your instructions but there was no `estimated means comparisons' under Options. The interface has changed slightly between SPSS 24 and 25. They have a new button in SPSS 25 called "EM Means" on the Repeated Measures dialogue box which now contains the Estimated Marginal Means that used to be under Options on SPSS 24. Is this what you mean? $\endgroup$ – Deepend Jul 18 '19 at 17:58
  • $\begingroup$ I think yes, sorry that I cannot check it myself now. $\endgroup$ – ttnphns Jul 18 '19 at 20:14
  • $\begingroup$ yes, select there Compare main effects $\endgroup$ – ttnphns Jul 18 '19 at 20:28
  • 1
    $\begingroup$ I've said in the first comment, that only Bonferroni and Sidak (besides LSD) are done for repeated measure factors. Tukey test is only for between-subject groups. $\endgroup$ – ttnphns Jul 20 '19 at 16:52
1
$\begingroup$

@ttnphns is right: the POSTHOC subcommand is only available for comparisons involving between-subjects factor levels, because many of the options there are only valid in that context. Also, if there are covariates in the model even with a between-subjects design, POSTHOC is not available. EMMEANS is the way to go in all of these cases.

| cite | improve this answer | |
$\endgroup$

Your Answer

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

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