I would like to estimate the sample size for a 3b x 2b x 2w x 3w x 6w ANOVA [b=between subject factor; w=within subject factor].
I do not know much about the estimated effect size or correlations, so i would always assume a standard value. For instance, r=.5 and peta²= .1. I would like to set power=.8.
This ANOVA seems to be far to complicated for GPower, so i am looking for alternative calculators.
You have lots of dfs in your model, and therefore there will be lots of individual effects. You will need to decide both the minimum effect sizes to detect for each of these, and whether you want to design the study with power for detecting all of them, some specified proportion of them, or say one of them. This is the hard part, before you get to the stats bit.
If G*Power and other power packages (e.g., https://cran.r-project.org/web/packages/pwr/index.html ) do not allow this many factors, you can use simulation.
This should help with simulation
http://www.bristol.ac.uk/cmm/software/mlpowsim/
Hello Stephan,
I do agree with Daniel Wright and Kelvyn Jones that simulation can be quite informative, and offer one more option for your consideration: https://cran.r-project.org/web/packages/Superpower/vignettes/intro_to_superpower.html
The design you propose is not too complex to be evaluated in G*Power, at least as regards: (a) main effects tests; or (b) interaction tests, at least at the two-way level. The tricky part is, always, picking defensible values for ES (effect size) and for the correlation among measures (for each of the repeated measures factors).
How to get them?
1. Guess. I don't mean to suggest that's what you've done in your query, but this is the least satisfactory of all methods.
2. Run a quick and dirty pilot study to get some insight as to what plausible values (especially for correlations) might be.
3. Ask domain experts (especially as regards a plausible target ES).
4. Refer to the literature; related studies may provide insight as to both aspects.
Good luck with your work!
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA