RESULTS for 9337 runs indendent voxel


Results of a voxelwise analysis:
--------------------------------
  thresholds: 1.659782 2.363098 3.171250 
---> false positives voxelwise  at p=0.05 : 0.04999606 
---> false positives voxelwise  at p=0.01 : 0.009999029 
---> false positives voxelwise  at p=0.001: 0.001000540 

Thus the probalility of detecting a false positive 
voxel is well reproduced.

Results of Bonferroni-correction:
---------------------------------
  thresholds: 4.938354 5.319059 5.841054 
---> false positives bonferroni at p=0.05 : 0.05108707 
---> false positives bonferroni at p=0.01 : 0.00953197 
---> false positives bonferroni at p=0.001: 0.0007497055 

Good agreement with the theoretical values for false 
positive portion. This should be the case, since we have 
indeed independent observations in each voxel (up to the 
quality of the radnom number generator, which should be 
good in R)

Results of RFT and fmri-package:
--------------------------------
  thresholds: 5.39     5.7      6.115 
---> false positives rft at p=0.05        : 0.007604156 
---> false positives rft at p=0.01        : 0.002570419 
---> false positives rft at p=0.001       : 0.0002142016 
---> false positives fmri at p=0.05       : 0.007818357 
---> false positives fmri at p=0.01       : 0.002570419 
---> false positives fmri at p=0.001      : 0.0002142016 

It is not clear, why rft and fmri show sligthly results at
p=0.05, rouding effects?? The threshold is apparently more 
conservative than bonferroni (cf. thresholds). Thus the 
thresholds are not correct in this case. This may be because 
the assumptions of RFT are not met, because there is no 
smoothness in the image. Find details on RFT.




RESULTS for 1814 runs smoothed spm hmax=3.52 kern=Triangle

Results of a voxelwise analysis:
--------------------------------
  thresholds: 1.659782 2.363098 3.171250 
---> false positives voxelwise nonaws  at p=0.05 : 0.0517408 
---> false positives voxelwise nonaws  at p=0.01 : 0.01018692 
---> false positives voxelwise nonaws  at p=0.001: 0.000924612 
---> false positives voxelwise aws  at p=0.05 : 0.0620837 
---> false positives voxelwise aws  at p=0.01 : 0.01463301 
---> false positives voxelwise aws  at p=0.001: 0.001906407 

Results of Bonferroni-correction:
---------------------------------
  thresholds: 4.938354 5.319059 5.841054 
---> false positives bonferroni nonaws at p=0.05 : 0.01102536 
---> false positives bonferroni nonaws at p=0.01 : 0.001653804 
---> false positives bonferroni nonaws at p=0.001: 0 
---> false positives bonferroni aws at p=0.05 : 0.1405733 
---> false positives bonferroni aws at p=0.01 : 0.0429989 
---> false positives bonferroni aws at p=0.001: 0.005512679 

Results of RFT and fmri-package:
--------------------------------
  thresholds: 4.625 4.99 5.465 
---> false positives fmri nonaws at p=0.05       : 0.03969129 
---> false positives fmri nonaws at p=0.01       : 0.009371555 
---> false positives fmri nonaws at p=0.001      : 0.0005512679 
  thresholds: 4.655 5.02 5.495 
---> false positives fmri aws at p=0.05       : 0.2745314 
---> false positives fmri aws at p=0.01       : 0.115215 
---> false positives fmri aws at p=0.001      : 0.02315325 

