Hello, thanks for the video. I have a question , what if all ordered p-values are below the cut off of bonferroni , so none is above holm's cut off, in this case bonferroni will reject all nulls but holm's will reject none of them, so we're not really making more discoveries with holm's method here. I hope i'll get a clarification . Thank you
If all p-values are below Bonferroni and therefore all nulls get rejected, they will also be below Holm and get rejected by that method too. The Holm cutoff only gets bigger. So if all p-values are below the first Holm cutoff (the smallest one, which is equivalent to Bonferroni), they will also be below all later, larger ones.
Thanks for the video. What are the steps to deal with a tied p-value with the Holm's method?
Hello, thanks for the video. I have a question , what if all ordered p-values are below the cut off of bonferroni , so none is above holm's cut off, in this case bonferroni will reject all nulls but holm's will reject none of them, so we're not really making more discoveries with holm's method here. I hope i'll get a clarification . Thank you
If all p-values are below Bonferroni and therefore all nulls get rejected, they will also be below Holm and get rejected by that method too. The Holm cutoff only gets bigger. So if all p-values are below the first Holm cutoff (the smallest one, which is equivalent to Bonferroni), they will also be below all later, larger ones.