Thanks Stephen. I did notice that the smaller clusters (size = 8) are the ones that had the higher p-value. Although they are still significant, I was wondering if this could be do to an increase in variability due to the fact that I have a smaller ROI. Would that also contribute? 

From: Stephen Smith <[log in to unmask]>
To: [log in to unmask]
Sent: Fri, June 24, 2011 7:25:38 AM
Subject: Re: [FSL] Randomise Results P-value

Hi - I guess the two differences are 1) the difference between this partialled correlation and multiple regression, and 2) the difference between doing voxelwise regression and the ROI-average-data feeding into the regression.   Both things should be minor but either could be causing the change you're seeing.

Cheers


On 23 Jun 2011, at 11:33, R. Duke wrote:

Hello Experts,

I have successfully completed a multiple regression analysis with 3 covariates of no interest and 1 covariate of interest. Upon completing the analysis I identified several clusters from my tbss_vox_p_tstat1.nii.gz file (uncorrected at 0.001) and determined the mean FA for each region across all of my participants. However, when I use SPSS to explore the partial correlation between mean FA in each region and my covariate of interest (controlling for my 3 covariates of no interest) a couple of my p-values do not reach a significance level of 0.001. Does this mean that I have calculated something incorrectly, or is this just the result of two different statistical techniques (multiple regression vs. correlation)? Also, should I even control for my 3 covariates of interest when running the correlation since these regions were selected from the multiple regression analysis.

Cheers



---------------------------------------------------------------------------
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director,  Oxford University FMRIB Centre

FMRIB, JR Hospital, Headington, Oxford  OX3 9DU, UK
+44 (0) 1865 222726  (fax 222717)
[log in to unmask]    http://www.fmrib.ox.ac.uk/~steve
---------------------------------------------------------------------------