I am playing around with scripting conjunction analyses as I want to write out files I can use further. In this context, I am, for completeness sake, trying to implement the global null approach. In a message some time ago (09/19/2004), Tom suggested that the threshold that should then be applied to each image individually is the assumed significance, corrected for the number of effects ("If a0 is your desired false positive rate, use (a0)^(1/K) where K is the number of effects conjoined. For 0.05 and 3 effects this is 0.3684"). However, this was stated in the context of FDR-corrected thresholds, and I wondered if this is also true for uncorrected thresholds. Can anyone confirm that this approach is generalizable?
This is indeed the correct way to obtain an uncorrected threshold for K tests of the Global Null, using a "min T" (equivalently, "max P") conjunction test. Going further, if you want to create an image of uncorrected Global Null Conjunction P-values, then you simply take products of P-values: i.e. you would create uncorrected P-value images for each of the constituent tests and then take the product of those K images to get uncorrected Global Null P-values. If you wanted, you could then submit *those* P-values to voxel-wise FDR. (But they are completely valid, usual P-values otherwise).
Does this help?
-Tom
Thanks in advance,
Marko
--
____________________________________________________
PD Dr. med. Marko Wilke
Facharzt für Kinder- und Jugendmedizin
Leiter, Experimentelle Pädiatrische Neurobildgebung
Universitäts-Kinderklinik
Abt. III (Neuropädiatrie)
Marko Wilke, MD, PhD
Pediatrician
Head, Experimental Pediatric Neuroimaging
University Children's Hospital
Dept. III (Pediatric Neurology)
Hoppe-Seyler-Str. 1
D - 72076 Tübingen, Germany
Tel. +49 7071 29-83416
Fax +49 7071 29-5473
[log in to unmask]
http://www.medizin.uni-tuebingen.de/kinder/epn
____________________________________________________
--
____________________________________________
Thomas Nichols, PhD
Principal Research Fellow, Head of Neuroimaging Statistics
Department of Statistics & Warwick Manufacturing Group
University of Warwick
Coventry CV4 7AL
United Kingdom
Email:
[log in to unmask]Phone, Stats: +44 24761 51086, WMG: +44 24761 50752
Fax: +44 24 7652 4532
