 |
 |
See in line responses below. Best Regards, Donald McLaren ================= D.G. McLaren, Ph.D. Postdoctoral Research Fellow, GRECC, Bedford VA Research Fellow, Department of Neurology, Massachusetts General Hospital and Harvard Medical School Office: (773) 406-2464 ===================== 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 (773) 406-2464 or email. On Tue, Dec 21, 2010 at 11:29 AM, Karl Liu <[log in to unmask]>wrote: > Hello, Spm users: > > > > I have some basic questions of how to run some data analysis. We have 4 > groups (group1, group2, group3 and group4). All the participants see a array > of neutral, fear and angry face pictures in an even-related design > (therefore, we have 5 contrast maps for each subject, including > netrual_activity, fear_activity, angry_activity, fear_minus_neutral, and > angry_minus_neutral) > > > > Here are my questions: > > > > 1) based on priori hypothesis, we want to draw ROI. In addition to > the typical anatomy-based ROI, some publication also used > functionally-defined ROIs (using the maximally activated voxels). My > question is whether one is better or more methodically valid than the other > in term of these two ROI methods. > Anatomical regions, whether drawn by hand or defined from an atlas, are better. The reason for this is that the p-values will not be biased in anyway. If you use yoru function data, then the p-values will be higher because you only selected certain voxels. Although, you could do it either way, it is important to clearly state the method and recognize that p-values from functionally defined regions are inflated. > 2) since we have 4 groups, ANOVA would be appropriate. How can we set > ANOVA in spm which include both within-subject variable (right and left > hemisphere) and between-subject variable (group1-4). > You'd have to extract the left and right hemisphere values to do this analyses. After extraction of the ROI, any statistical program would do the trick. > 3) How spm implements the correction of pair-wise group comparisons > after checking the main effects in ANOVA. If we choose ROI, we will use > small volume correction, but it is for voxel-wise correction, nothing to do > with the pair-wise comparison correction? > Divide the p-value by the number of tests. You can also read about this in a number of statistical text books that discuss post-hoc ANOVA tests. > 4) Since we have 5 contrast maps, it means we can have 5 possible > measures for group comparison. Any correction necessary for this > multivariate situation? > You wouldn't want to compare all 5 contrasts in a single model. You could do the first three in a single within-subjects model and then test for pairwise comparisons. Correction for pair-wise tests as above. > > > Thanks a lot! > > > > Karl >