sdm-help-list > threshold (uncorrected p)
Showing 1-3 of 3 posts
Aug 25, 2017 02:08 PM | pingfan 2008
threshold (uncorrected p)
Dear experts on SDM,
1. I am trying to set a stringent threshold with probability=0.0.00001;peak height threshold=1;extend threshold=100. But I used the used the default threshold with probability=0.0.005;peak height threshold=1;extend threshold=10. So many brain areas survived but some areas is not reasonable.
But I am not sure whether it is reasonable to set such stringent threshold. As I see many published papers just used the default threshold with probability=0.0.005;peak height threshold=1;extend threshold=10. (Just like the paper mentioned "as this was found to optimally balance sensitivity and specificity and to be an approximate equivalent to corrected P-value = 0.05 (indeed 0.025) in ES-SDM."http://www.sciencedirect.com/science/article/pii/S0924933811000733#tbl0010)
2.If it is reasonable to use a stringent threshold with probability=0.0.00001;peak height threshold=1;extend threshold=100.What is the approximate equivalent to corrected P-value?
Many thanks in advance.
Best
Pingfan
1. I am trying to set a stringent threshold with probability=0.0.00001;peak height threshold=1;extend threshold=100. But I used the used the default threshold with probability=0.0.005;peak height threshold=1;extend threshold=10. So many brain areas survived but some areas is not reasonable.
But I am not sure whether it is reasonable to set such stringent threshold. As I see many published papers just used the default threshold with probability=0.0.005;peak height threshold=1;extend threshold=10. (Just like the paper mentioned "as this was found to optimally balance sensitivity and specificity and to be an approximate equivalent to corrected P-value = 0.05 (indeed 0.025) in ES-SDM."http://www.sciencedirect.com/science/article/pii/S0924933811000733#tbl0010)
2.If it is reasonable to use a stringent threshold with probability=0.0.00001;peak height threshold=1;extend threshold=100.What is the approximate equivalent to corrected P-value?
Many thanks in advance.
Best
Pingfan
Oct 11, 2017 02:10 PM | Joaquim Radua
RE: threshold (uncorrected p)
Dear Pingfan,
you can use more stringent thresholds and thus decrease the false positive rate. I don't think that there is a formal equivalent corrected P-value, you can find ways to correct for multiple comparisons, but as far as I know none is free from issues...
With the best wishes,
Joaquim
you can use more stringent thresholds and thus decrease the false positive rate. I don't think that there is a formal equivalent corrected P-value, you can find ways to correct for multiple comparisons, but as far as I know none is free from issues...
With the best wishes,
Joaquim
Oct 15, 2017 03:10 AM | pingfan 2008
RE: threshold (uncorrected p)
Thanks very much for your reply. I got the key points.