I have a repeated-measures experiment with 3 treatments. One of the within-groups variables I've measured differs significantly over these 3 treatments and I wish to add it into further analyses as a covariate. What analysis is the most appropriate? There doesn't seem to be a way to enter a covariate of this nature into a general linear model with repeated-measures. I am using SPSS.
Much thanks for your time.
Why is the "normal" procedure, using a rm-GLM, not appropriate? Maybe I did not understand your design exactly, would you therefore please add further informations.
You should use mixed models, or multilevel analysis (which is the same).
greetings, Pierre
I have noticed that repeated measure analysis allow both categorical and numerical covariates for adjusted analysis. Categorial variables are to be put into "Between-subject Factors" menu box and numerical variables are to be put into "covariate" menu box. I think your covariate variable should be adjusted since it differs across three groups (between subject factor). The result will show both significant changes across repeated measures and differences between three groups according to your request in the analysis option.
Thank you all, I am going through your suggestions, but to clarify:
The covariate is linked to the treatment, so there are 3 covariates (cov1, cov2, cov3). It's occurring to me that perhaps I have my data set up the wrong was as it's currently in long format?
Also, Kyaw Oo, what do you mean by adjusted analysis?
Sorry, I don't know spss any more. With another covariate, you may be asking the analysis of covariance to do more than it can do. Results may be difficult to interpret results. Ancovar is always tricky to interpret, even if you can get spss to give you something.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Applied Statistics