I am planning an experiment 2x2 within-subjects factors and additionally I would like to consider a covariate, so that a 2x2 repeated measures ANCOVA is the intended analysis. Does anyone know how to calculate the sample size for that? G*Power does not offer this option and the R function pwr.f2.test doesn't help either (as far as I understood).
Or do you think that another kind of analysis is more appropriate? Thanks!
Hello Claudia Menzel. If there is no canned power analysis routine for your model, you can always use simulation. Here are some slides that explain the general idea. If you say what software you use as your main stats package, readers may be able to give more specific advice.
- http://meeting.spsp.org/2016/sites/default/files/Lane%2C%20Hennes%2C%20West%20SPSP%20Power%20Workshop%202016.pdf
HTH.
Hello Claudia,
Bruce Weaver 's suggestion of going the simulation route is a good one. Here is another link that is helpful for getting into a software package that was developed to evaluate a multitude of anova type designs: https://cran.r-project.org/web/packages/Superpower/vignettes/intro_to_superpower.html
As well, you could use G*Power, if you were willing to:
1. Go the generic F test route; or
2. Use F-test / ANOVA options, with the understanding that you'd be evaluating results for residualized (e.g., covariate-adjusted) scores.
Good luck with your work.
Thanks, Bruce Weaver and David Morse for your fast replies! I'll have a look into the suggested slides and links. Usually I use G*Power and R, but I am also familiar with the basics in SPSS.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size