Hi FSL experts,
I am trying to use randomise for a Group x Age interaction, with additional covariates of non-interest. I have already run this analysis in the standard higher level FEAT and there are clusters surviving the threshold of 3.1. However, when running randomise for the same analysis, the output file 1-p of 0.3 or lower in randomise. This is true with randomise -C option at 3.1, and for the randomise -T option. In some cases, it is yielding files with a range of 0-0.006.
I am using the following command to run randomise:
randomise -i cope1.feat/filtered_func_data -o randomise_results_T -d design.mat -t design.con -m mask -T -n 1000
OR
randomise -i cope1.feat/filtered_func_data -o randomise_results_C -d design.mat -t design.con -m mask -C 3.1 -n 1000
Note that filtered func, design.mat and design.con here came from the group level Feat analysis. I do not expect the results to be exactly the same as Reat, but I do not understand how the output can be so discrepant when exactly the same model and data are used for randomise. Please advise if there are any trouble-shooting tips with the way I am running the randomise command or with the files that are input. My design files are attached.
Thank you,
Michelle
|