I am currently trying to determine the required sample size for an experiment. As I want to use the statistical power concept (for a t-test) to determine the required sample size, I want to specify a power level, significance level and effect size. Using these inputs, e.g. the G*Power tool will allow me to determine the sample size.

A complicating factor is that the results of the analysis will be analyzed using multiple testing. More specifically: 12 tests will be conducted. This caused some doubt on how to specify the inputs for the power analysis:

  1. I am aware of the fact that the significance level needs to be corrected for multiple testing. To determine the sample size, I plan to use a Bonferroni correction. Is my statement correct that the corrected significance level needs to be used as an input for the power analysis?

  2. Is a similar correction required for the power level, i.e. do I have to set a higher power level for the power analysis to obtain a particular power over all 12 tests?

Thank you for the input.

  • $\begingroup$ Just use the Bonferroni-adjusted alpha for the significance level, and you'll be fine. No beed to adjust the target power as well. $\endgroup$
    – Russ Lenth
    May 20, 2016 at 20:59


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy