Tukey Test Error

Moderators: statman, Analyst Techy, andris, Fierce, GerineL, Smash

sloopylew
Posts: 2
Joined: Sun Mar 08, 2015 5:12 pm

Tukey Test Error

Postby sloopylew » Sun Mar 08, 2015 6:26 pm

Hello all - having an issue running a Turkey test. I have five IVs. I put them in to the ANOVA with my DV and mark the post hoc Tukey. However, when I run the test I get the error "post hoc tests on (IV) because there are more than 50 groups". I haven't separated out the variables or selected cases. The data is pretty forward - answers on a survey that combined 3 assessments. The questions gave me 5 variables (one assessment with 3 sub scales and one with 2). Any ideas what I'm doing wrong here?
iopsych
Posts: 25
Joined: Tue Feb 24, 2015 11:58 pm

Re: Tukey Test Error

Postby iopsych » Sun Mar 08, 2015 11:53 pm

Hello,

Using five IVs at the same time in an ANOVA makes it a very complicated design. Depending on the number of levels each IV has, your number of groups may become very large, resulting in the error message you mention. Specifically, assuming that each IV is a between subject factor with two levels (the best case scenario), you would have 2x2x2x2x2 = 32 groups. Apperantly you have some IVs with more than two levels, making the number of your groups over 50.

In most cases, it is not necessary to run an ANOVA with some many factors. Are you sure you are using the appropriate analysis?

Yalcin
sloopylew
Posts: 2
Joined: Sun Mar 08, 2015 5:12 pm

Re: Tukey Test Error

Postby sloopylew » Mon Mar 09, 2015 1:36 pm

Thanks for the response. I may not be going about it the right way. The 5 IVs come from continuous scales (items ranked 1-7 with 11-14 items in each scale). I'm trying to find out how these IVs impact the DV which is also on a continuous scale. I'm running a regression so I'll have parts and partials to look at but I was interested in comparing the means before running the regression - which is where the ANOVA came in to play. But I'm open to suggestions.
iopsych
Posts: 25
Joined: Tue Feb 24, 2015 11:58 pm

Re: Tukey Test Error

Postby iopsych » Mon Mar 09, 2015 6:14 pm

Hey,

If you just want to compare means to each other, I think you can run t-tests for each pair using the means and the SDs.

You can find calculators for that online.

Yalcin

Who is online

Users browsing this forum: No registered users and 2 guests

cron