my research was intervention with 2 groups experimental and control group with dv 1 with five subscales one's value should decrease others value should increase has 3 levels pre post follow up, i have done 2x3 mixed factorial design as my sir guided me to go to general linear model than to repeated measure than i have given pre post follow up of 1st scale, define it and selected the options plots em means and run it likewise rest of 4 scales. and for result i had drawn 2 tables 1 with univariate discussing mean sd F significance and partial eta and the other table include 1st subscale and 1st subscale x group and values written were wilk's lamda F significance and partial eta. please guide me
Hello Jawaria Afzal. You said you were told to "go to general linear model than to repeated measure". Those sound like steps to take in the menu system for SPSS. Are you using SPSS?
Assuming you are, in the initial dialog that comes up for Analyze > GLM > Repeated Measures, use the Measures part of that dialog to indicate how many DVs you have and what you want to name them. I generated a simple example where I just named them dv1-dv3. Here is the basic code:
GLM A0 A1 A2 B0 B1 B2 C0 C1 C2 BY Group
/WSFACTOR=Time 3 Polynomial
/MEASURE=dv1 dv2 dv3
/METHOD=SSTYPE(3)
/CRITERIA=ALPHA(.05)
/WSDESIGN=Time
/DESIGN=Group.
In that code A0-A2, B0-B2, and C0-C3 would be the names of variables in the data file. The WSFACTOR line says there are 3 levels of time (baseline, post, and followup). The MEASURE line shows that there are 3 DVs: A0-A2 hold the baseline, post, and followup scores for dv1; B0-B2 hold the baseline, post, and followup scores for dv2; and C0-C2 hold the baseline, post, and followup scores for dv3.
So, that is how you could get the model that your supervisor has asked for. But whether that is the best approach is another question. Before you proceed, I suggest that you (and your supervisor) read the following articles, both of which suggest that very often, people who think they want or need to use MANOVA might be better off with some other approach.
Article Multivariate Analysis Versus Multiple Univariate Analyses
Article MANOVA: A Procedure Whose Time Has Passed?
See also this post by Thom Baguley:
- http://psychologicalstatistics.blogspot.com/2021/08/i-will-not-ever-never-run-manova.html
I hope this helps.
Thank u for your detailed answer Bruce Weaver can you please tell me if mixed factorial manova exsists or not and the procedure to it on spps is same as I have stated above
it will be a great help.
Regards
You can simplify this model by using the pre measure as a covariate and using centred post measures as the dv. Then you just run a MANOVA with your grouping variable. This approach is generally more powerful than treating pre and post as repeated measures
Personally I would rather either i) run separate univariate analyses with a post hoc correction for multiple testing, ii) a true multivariate regression (though I don't know if that's possible in spss).
I agree with both of Thom Baguley's suggestions. To use the baseline score as a covariate, and estimating a model for just one DV at a time, you would need something like this:
* Model for DV1.
GLM A1 A2 BY Group WITH A0
/WSFACTOR=Time 2 Polynomial
/METHOD=SSTYPE(3)
/CRITERIA=ALPHA(.05)
/WSDESIGN=Time
/DESIGN=Group A0.
* Model for DV2.
GLM B1 B2 BY Group WITH B0
/WSFACTOR=Time 2 Polynomial
/METHOD=SSTYPE(3)
/CRITERIA=ALPHA(.05)
/WSDESIGN=Time
/DESIGN=Group B0.
Etc.
I do know of any way to implement the adjustment for multiple testing automatically. You would have to do that after the fact, I think.
- Badges
- Science topic
- Similar topics
- Mathematics