Hi, I completed a study with sample size of 42 and has observed significant results. However one of the reviewers claims that my study is underpowered? I have carried out post hoc analysis that gives a 100% power, but I just learned that it is not good to use. How can I argue that my study is not underpowered??
I think that you set d of effect size from previous research and calculate Power (1 - β error prob) / sample size
If there are not previous research... you refer to similar research.
Yes, I agree. You should compute the necessary sample size for a given (expected) power before collecting the data and analysing it. The power depends on the type of analysis that will be conducted (and depends upon some parameters that will need to set, given previous studies results or expert thinking).
Thank you guys for your responses!
I will argue that we used sample size calculations before conducting the study.
Can I also argue that we have statistically significant results which is also a proof that our study isn't underpowered.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size