1

I am analyzing pre- and post- intervention data for which I have 6 dependent variables for two groups (treatment, no treatment). I want to know if, controlling for pre scores from the 6 measures, post scores differ between the two groups. (Some DVs show significant correlations with each other.)

I want to know whether a MANCOVA is appropriate, with all 6 DVs and the corresponding 6 pre tests entered as covariates, or whether running a separate ANCOVA for each DV and the corresponding pre test as covariate is more appropriate.

Other ideas for analysis are welcome, thank you in advance.

Glen_b
  • 257,508
  • 32
  • 553
  • 939
redracer
  • 67
  • 6

1 Answers1

4

Since you are using pre-treatment and post-treatment scores, I believe you should be considering that these scores are highly correlated due to within-subject variance. I would recommend doing 6 repeated measures ANOVA instead, with pre and post scores being the within subject variable. You could have the two groups as a factor in the repeated measures ANOVA.

This will allow you to see (1) how the 6 DVs change over time and (2) whether these changes over time differ between the two groups (i.e., if there is an interaction).

Behacad
  • 4,916
  • 8
  • 30
  • 48