0

There are 100 participants and I want to test if 50 metabolites change before and after intervention (antiobesity drug). First, I check the normality for each variable, and then I perform a T-test or Wilcoxon between the two groups. Should I use the FDR for all the final P-values, regardless of whether they derive from the T-test or Wilcoxon?

Thank you in advance for your help.

Geli
  • 1
  • 1
  • What are you testing and what are your goals? Testing 50 variables is begging for some sort of different approach. – Demetri Pananos Jan 13 '22 at 05:36
  • 1. Do you want to control the FDR across all the tests? Or just across each of two subsets of tests? $\quad$ $\,$ 2. also see [Is normality testing essentially useless?](https://stats.stackexchange.com/questions/2492/is-normality-testing-essentially-useless) – Glen_b Jan 13 '22 at 05:39
  • There are 100 participants and I want to test if 50 metabolites change before and after intervention (antiobesity drug). First I checked if the variables are normally distributed or not and then I performed t-test or Wilcoxon, respectively. My question is if I should gather all the P-values from these 50 comparisons and perform the FDR regardless of whether the P-value is from a T-test or Wilcoxon. Please let me know if any further clarification is needed. Thank you all for your help. – Geli Jan 13 '22 at 05:44
  • 1
    in relation to 2. above, also see https://stats.stackexchange.com/questions/100934/does-testing-for-assumptions-affect-type-i-error ... there are a number of other related questions on this issue. $\qquad$ My question 1. remains unanswered -- we can't tell you what you want to control the FDR over, that's your choice. (I would say that most people who worry about FDR would choose to control it for the whole set, but again, that's their choice, not necessarily yours) – Glen_b Jan 13 '22 at 05:46
  • Thank you for the link. Yes, I wanted to perform FDR for all the 50 comparisons. However, I was not sure if I should perform FDR separately for the P-values derived from the T-test and for P-values derived from Wilcoxon or for all the 50 comparisons irrespectively of the test. – Geli Jan 13 '22 at 06:03

0 Answers0