Hi,
The third stage DR does first perform a standard t-test based on your design and contrast matrices before using randomisation to build up the associated distributions and testing for significance. You can simply use the uncorrected maps and subject those to your special type of inference procedure... just make sure you a very clear about what you have done when writing this up ;)
hth
Christian
On 23 Jun 2011, at 11:41, Alisha Janssen wrote:
> I am currently working with a resting-state data set comparing two groups. I have run dual_regression using the randomise inference method to output stage3 results based on my glm design. However, this is a very stringent inference method and I am lacking results for many of my contrasts. I am currently trying to take the outputs from dual_regression stage2 and run them through a different inference method. I would like to use the cluster command to threshold my end result, but I need to apply my design matrix first that will output my contrasts (basically a two group unpaired design with 4 different contrasts. group1>group2, group2>group1, group1 mean, group2 mean).
>
> I have attempted to use feat, but the inputs from stage2 are not suitable. I have also tried fsl_glm, but have not been successful there either. Does anyone have any ideas?
|