repeated measures MANOVA syntax

[Michel Passager]

Hello everyone,
A request for help from a researcher in psychology who loves STATA but only partially masters it. Here is the problem...
In my study I have 8 dependent variables, 100 participants divided in two groups (n=50, n=50). Each group is given a specific learning program (0-1). Participants are tested at pre-test and posttest. Thus, we have:
8 VD (VAR1 to VAR8)
2 GR (0-1)
Pre-test scores - post-test scores.
The purpose of the analysis would be to test the scores of the two groups are different between the pretest and the posttest and if there is a interaction GRX pre-test-posttest
I wrote the following lines to do a repeated measures ANOVA for each variable (VAR1 is the dependent variable here). I repeat this 8 times (VAR1, VAR2, etc.).

anova VAR1 GR / id|GR PREPOST GR#PREPOST, repeated(PREPOST)
estat esize
anova VAR1 GR##PREPOST
contrast PREPOST@GR, effect
margins GR#PREPOST
marginsplot, x(PREPOST)
graph export marginsplotEL.pdf, replace

But it would be more elegant to do a repeated measures MANOVA. This is where the trouble starts for me. I have not been able to find the orders to do this. Could you help me write the necessary orders? I use Stata 17
Many thanks.

[Joseph Coveney]

But it would be more elegant to do a repeated measures MANOVA. This is where the trouble starts for me. I have not been able to find the orders to do this. Could you help me write the necessary orders? I use Stata 17

The syntax for MANOVA is directly analogous to that for ANOVA.

manova VAR? = GR / id|GR PREPOST GR#PREPOST

The repeated(PREPOST) option is not available for manova, but in your case, with only a single posttreatment observation, you don't need it. It won't do anything with anova, either. (And I'm not sure what you're trying to accomplish with anova VAR1 GR##PREPOST afterward, but it doesn't seem legitimate in this context.)

With repeated measurements, you can also set up a so-called doubly multivariate MANOVA, but postestimation contrasts in your case will be involved.