SPSS - ANCOVA Tukey Test?

[Prodigal Moon]

I was hoping someone on here might know how one goes about running a Tukey (really a Tukey-Kramer) test on an ANCOVA in SPSS. Tukey is one of the ANOVA post-hoc options, but when you add in a covariate it blocks you from the Post Hoc tab, so you can only use the 3 contrast options from a different tab (LSD, Sidak, and Bonferroni, iirc). Is there a statistical reason why you wouldn't want to use a Tukey test with an ANCOVA, or is this just a quirk of SPSS? If it's the latter, does anyone know a workaround, or have access to a script?

Thanks in advance for your help!

[Dragan]

Is there a statistical reason why you wouldn't want to use a Tukey test with an ANCOVA...?

Thanks in advance for your help!

Yes, there is a reason. It is because the standard error associated with the studentized range statistic (e.g. Tukey) is not correct.

[Prodigal Moon]

Ah, okay. I think I get it. The Tukey test would be based on adjusted scores, so the standard error would be thrown off and the test statistic wouldn't mean anything?

[Dragan]

Ah, okay. I think I get it. The Tukey test would be based on adjusted scores, so the standard error would be thrown off and the test statistic wouldn't mean anything?

In short, the standard error has to include a term that includes the differences between means of the covariate (X_i , X_j) associated with the group means (Y_i, Y_j) that are being tested in the numerator for each pairwise comparison.

[Prodigal Moon]

Okay, thanks so much for your help!