I try to use the software[2] to calculate the needed sample size for a 2 way MANOVA.
The experiment is a mixed design, one Independent variable(IV) is between subject with 3 levels, and the other is within subject with 4 levels. And there is total 5 dependent variables to measure.
G*power provide several MANOVA, but I'm not sure about it's meaning. 1.MANOVA- Global effect 2.MANOVA- Repeated measure, between factors 3.MANOVA- Repeated measure, within factors 4.MANOVA- Repeated measure, within-between interaction
Question1：
What does it mean by repeated measure? If it denote to within subject design, then what is "repeated measure, within factor?" Does it means "Two way anova with within x within design" ? And the fourth one is a 3 way MANOVA?
If so, I think my experiment should use the 2nd one, which is "MANOVA- Repeated measure,
between factors".
But for the input parameter part I'm confused too. The following is the parameter I'm not sure how to give to the software：
1. Number of groups
2. Number of measurements
Question2 ：
For number of group, does it only denote to the level number for between subject? (In my case, 3) Then how about the within subject part? where to tell that?
or Can someone provide me with the reference for how to use G*Power?
[2]Buchner, A., Erdfelder, E., & Faul, F. (2001). G*Power. Retrieved April 25th, 2012, from http://www.psycho.uni-duesseldorf.de/aap/projects/gpower/
Hello Su-Fang,
With your design, there are three primary tests that may of interest to you:
1. Does the between-subjects factor make a difference in the profile of scores?
2. Do scores change across levels of the within-subjects factor?
3. Is there an interaction of the between-subjects and within-subjects factor?
Testing hypothesis #1 would call for MANOVA: Repeated measures, between factors. (Note that "repeated measures" refers to the within-subjects factor for your design). Number of groups = 3; Number of measurements = 4.
Testing hypothesis #2 would call for MANOVA: Repeated measures, within factors.
Testing hypothesis #3 would call for MANOVA: Repeated measures, within-between interaction.
Note that you could get different requisite N values across the three hypothesis types. If that were the case, then elect the largest of the values as your guide to suitable sample size.
Good luck with your work.
Dear Yeh,
You have asked a couple of questions on different things at a time. G*power is about the easiest software to use. But I don't understand your research questions at all.
If you have a mixed model (using multiple analysis on the same dataset or participants), it is safer to use the question with the highest sample size requirement, than using too small sample size.
Your design appears like you will use G*power MANOVA- Repeated measure, within-between interaction.
You asked how to input number of groups and measurements.
In a repeated measure ANOVA for instance, you are analysis wether there is a significant difference in IVs (groups A, B, C, D) when measured at times 1,2,3,4,5 or when treated with or subjected to interventions 1,2,3,4,5
Therefore the number of groups is 4 and number of measurements is 5.
Kindly seperate your research questions or hypothesis and see if MANOVA is the appropriate tool for each of them.
Good luck.
- Badges
- Science topic
- Similar topics
- Psychology