Home › Forums › General Help › adjusted pvalues for group comparisons
Tagged: fdr correction
 This topic has 4 replies, 3 voices, and was last updated 1 month ago by Mite Mijalkov.

AuthorPosts

johnsojpParticipant
Hello again Mite and company,
I’m a little confused about the fdr values that are shown when I look at group comparisons of nodal measures. I’m comparing two groups using weighted undirected graphs with anticorrelations set to zero and I have noticed that the fdr(1tailed) and fdr(2tailed) are the same across all of the regions.I’m guessing that these do NOT reflect fdrcorrected pvalues, but maybe represent the cutoff for significance after fdr correction–is that correct?
In other words, if the p(2tailed) for global efficiency is .002 and the fdr(2tailed) = .012, would it be correct to interpret that as a significant result (i.e., global efficiency for this region is significantly different between my groups)? On the other hand, if global efficiency = .02, the result would not be significant because the value is greater than the fdr of .012?
If all of that is correct, what does it mean when the fdr value is 0? Is that simply telling me there are no significant results for that measure after fdr correction?
Thank you very much for the clarification!
Best,
JeffMite MijalkovKeymasterHi Jeff,
Yes, your conclusions that you mentioned above are correct. In fact, we discussed these issues with the pvalues and fdr correction table in another post. There, I offered a detailed explanation about how to interpret the fdr values shown in the tables in Braph:
I hope that this information helps you. If you need to discuss any other possible problem, please do not hesitate to contact me.
Best,
Mite This reply was modified 1 year, 5 months ago by Mite Mijalkov.
JeffGuestThanks for your response Mite–very much appreciated!
DionGuestHi Guys,
So what does it mean if the p(1tailed) shows 0.1 and the value 1 is 0 ?
This is the same for p(2tailed) and value 2.Thanks,
Dion
Mite MijalkovKeymasterHi Dion,
I think that you are referring for the global measures table. If that is the case, then the relevant part of the table goes in the following way:
difference: denotes the difference between the average measure values in the two groups (this difference is tested whether is significantly different that zero)
p(1tailed): 1 tailed pvalue outputted from the permutation test
p(2tailed): 2 tailed pvalue outputted from the permutation test
value1: average value of the corresponding measure in group 1
value2: average value of the corresponding measure in group 2Hope this helps. If there is anything more I can help you with, please do not hesitate to ask me.
Best,
Mite 
AuthorPosts