Hi,
I'm dealing with data where I'm testing several 2 and 3-way interactions. However, I also want to be able to determine the statistical power of those that are statistically significant (sample sizes are known). Could someone possibly offer some advice? known. I only know how to did this with t-tests in G-power where I compare two means.
Thanks in advance,
D
Simulation is always an option. The details of how to go about it will depend on what software you have. But you can get some ideas by Googling . HTH.
Do you want to find statistical power when interaction is significant in a 2 or 3-way ANOVA?
I just noticed now that (somehow, inexplicably) I missed the words post-hoc in the question. Post-hoc power analysis (as it is typically done) is not a good idea. See Russell Lenth's commentary on it in this report:
- https://homepage.divms.uiowa.edu/~rlenth/Power/2badHabits.pdf
See also this nice article by Len Thomas.
Article Retrospective Power Analysis
HTH.
https://academic.oup.com/beheco/article/14/3/446/257432
This is a good article by Colegrave and Ruxton (2003), discussing the relative merits of post-hoc power analyses vs confidence intervals
Thanks everyone. I managed to code into SAS and get it to spit the power output in the fixed effects table. I understand that this is not ideal for analysis post-hoc but I recently had a paper come back and a reviewer wanted to know the power for each effect tested.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA