0

I'm not sure about the following situation... In my test setting I need to adjust my p value because of multiple tests in the same population - this is clear to me... I have three hypotheses so my alpha level of 0.05 is reduced to 0.016.

So if any test is below 0.016 my result is significant. But what do I have to do with my results for my confidence interval of any test? Do I have to modify the upper und lower limit, which are in the output of my statistic program (f.e. of a simple t test in R) manually?

Thank you for any help!

MDStat
  • 11
  • 3
  • Adjustment for multiplicity of testing is not as straightforward as you appear to think. It is often not appropriate. See these previous questions: https://stats.stackexchange.com/questions/278414/multiple-comparisons-with-p-value-corrections-of-fiducial-limits/278426#278426 https://stats.stackexchange.com/questions/559132/do-i-have-a-multiple-comparisons-scenario-here/559134#559134 https://stats.stackexchange.com/questions/378021/post-hoc-corrections-for-multiple-t-tests/378181#378181 – Michael Lew Feb 12 '22 at 06:07
  • You already know how to decrease your significance level. What else do you need? Why bother with confidence intervals? – frank Feb 12 '22 at 06:29
  • Hi Frank, thank you for your answer! So I may use the CIs equally to a situation without multiple testing - the only thing I have to consider is IF I have to adjust my p value or not? – MDStat Feb 13 '22 at 20:00

0 Answers0