I'm having trouble running post hoc comparisons on my data for the final project. I am running an ANCOVA and I have three groups in my IV. Before I covaried out my covariate, the differences between the means in the three groups were not significant, but the adjusted means used in the ANCOVA are significantly different from each other. The problem I'm having is finding a way to run post hocs using the adjusted means and not the raw means. SPSS won't let me run post hocs as a part of the Univariate analysis so I have to run independent sample t-tests, but they use the raw means to compute their results...
posted by Grant Goodman @ 7:00 PM
2 Comments:
At 6:43 PM,
Grant Goodman said…
I think I figured out a way to compare the adjusted means in the three groups in my IV.In the options section of Univariate analysis, there is a little box that says "compare main effects" under the larger box that is labeled "Display Means for:"
After checking this smaller box, I am then asked to choose my confidence interval adjustment and I chose Bonferroni. The analysis now gives me what appear to be t-test comparisons of the three groups using the adjusted means.
My question now is: with these three comparisons am I supossed to adjust my alpha level? I assume so, but I was just wondering if I could get some feedback about whether I am on the right track or not. We've never done post hocs with ANCOVA before.
Sorry this is so long winded. Thanks for the help!
At 11:00 AM,
David said…
Good find on the "compare main effects."
The Bonferroni already adjusts alpha based on the number of comparisons made. To see that this is true, you can run the analysis twice, once with the Bonferroni adjustment and another with LSD (which makes no adjustments). The p values you get for those will differ by a factor equal to the number of comparisons made.
Post a Comment
<< Home