Hi I did Friedman's RM ANOVA and found data is significant.Then Wilcoxon posthoc analysis is showing significant differences, p=0.001.But Bonferroni adjusted significance is bringing p value to 0.063. Theoritically I should consider this to be nonsignificant. Because it is a thin line will I be missing important signals because of technicalities??
I would try the Benjamini-Hochberg approach under these conditions. It is described in the attached. Best, David Booth
PS BTW is 0.06 much different than 0.05? or in confidence interval language is 94% sure much different than 95 % sure. you choose alpha but DON"T p-hack. I would go with 0.06 myself. Best, David Booth
First, why do you call this technicalities? Second, of course you risk missing valuable effects and reporting effects that you should not. Your question should be why you are weighting controlling for Type 1 error as more important than type 2 error. FInally, as David Eugene Booth says, there are alternatives to the conservative Bonferroni's.
- Badges
- Science topic
- Similar topics
- Correction