 |
 |
Looks like you did the group comparison correctly, one thing that I might change is to have separate covariates for each group. Using a single covariate across groups assumes that the relationship between the DV and covariate is the same in both groups. The result is not surprising. I'd change the T to the FDR level of the first group -- this will give you a more direct comparison to the significance (without corrections). FDR is an iterative thresholding approach, so if the most significant voxel doesn't pass, then no voxels will. Furthermore, the absence of results in one group where results are present in another group DOES NOT imply group differences. For example, say visual cortex in group 1 has a T-value of 3.01 and group 2 has a T-value of 2.99. These are clearly not different, but using a threshold of 3 will show the result in one group, but not the other. Additionally, even if the T-statistics were significantly different between the groups, the groups could have the same mean value (T-stat difference is due to SD of the group), which is what is being tested with a 2-sample t-test. Hope this helps. On Wed, Jul 8, 2009 at 10:43 AM, Harsha Halahalli < [log in to unmask]> wrote: > Dear SPMers, > We have a query regrading group comparisons of PPIs. > We have performed a PPI analysis in a fMRI dataset acquired while subjects > were performing a cognitive paradigm which consisted of two tasks (say Task > A and and a baseline Task B). There were two groups of subjects (say Group > 1, n=23; and Group 2, n=23). We have extracted VOI-timeseries from a > chosen > seed region for each of the subject's 1st level SPM, derived the regressors > for the [Task A -Task B] contrast, set up the PPI GLM for individual > subjects and obtained con images for > positive and negative interactions. We then entered these 1st level con > images from subjects of the two groups into two separate second level > one-sample t-test models. This model included four confounding covariates. > One of the groups showed extensive areas of significant negative > interactions after FDR correction at p<0.05. The other group did not show > any such regions of negative interaction. However, when we entered these > same 1st level con images into a two-sample t-test model for a > between-group > comparison with the covariates of no-interest, we fail to see any > significant differences in the areas of negative interaction (See attached > file for images). > > Our quetions are: > > 1. Have we followed the correct steps for this between-group comparison of > PPIs? > > 2. Why does the two-sample t-test comparison fail to show any significant > group differences when the one-sample tests suggest such obvious > differences > in the pattern of negative interactions in the two groups? > > Any comments / suggestions would be greatly appreciated. > > Best Regards > Harsha > > ------------------------------------------------------------- > Harsha Halahalli > PhD Scholar > MBIAL, Department of Psychiatry > National Institute of Mental Health and Neurosciences (NIMHANS) > Bangalore - 560029, INDIA > -- Best Regards, Donald McLaren ===================== D.G. McLaren University of Wisconsin - Madison Neuroscience Training Program Office: (608) 265-9672 Lab: (608) 256-1901 ext 12914 ===================== This e-mail contains CONFIDENTIAL INFORMATION which may contain PROTECTED HEALTHCARE INFORMATION and may also be LEGALLY PRIVILEGED and which is intended only for the use of the individual or entity named above. If the reader of the e-mail is not the intended recipient or the employee or agent responsible for delivering it to the intended recipient, you are hereby notified that you are in possession of confidential and privileged information. Any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited and may be unlawful. If you have received this e-mail unintentionally, please immediately notify the sender via telephone at (608) 265-9672 or email.