 |
 |
The only way to interpret the contrast is that the sum of A and C is greater than B. In my opinion, this is a bad contrast. If A=B=C, then you will still find an effect with the contrast. The reason for this is that A+C>B will always be true is this case as 2*A>A, by substitution. Also if A=C=0 and B is negative, then you would also get an effect, but I wouldn't call it activation as A and C are 0. The better contrast for investigating A and C > B, would be A/2+C/2>B. This means that the average of A and C is greater than B. Hope this helps. Best Regards, Donald McLaren ================= D.G. McLaren, Ph.D. Research Fellow, Department of Neurology, Massachusetts General Hospital and Harvard Medical School Postdoctoral Research Fellow, GRECC, Bedford VA Website: http://www.martinos.org/~mclaren 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 Mon, Dec 1, 2014 at 9:42 AM, Daniela Rabellino <[log in to unmask] > wrote: > Dear Helmut, > > Thanks for your clarification and possible solutions. > > Just to be sure that I understood correctly, in case I also keep > investigating the T contrast A-B+C, is it correct to state that we are > answering the question: > "Which brain areas show activation > 0 in response to A OR in response to > C (-B), as an additive effect?" > > In this case, we cannot know, whether it is activation > 0, it is due to > A, C, or both (if A=0 and C>0 I would still obtain a result), correct? > > Thank you for your help. > Daniela > > > > On 2014-11-30 12:48, Helmut Nebl wrote: > >> Dear Danila, >> >> No. With [1 -1 1 0 0 0 0 0 0] you test whether (A - B + C) > 0 in case >> of a T contrast or whether (A - B + C) is sig. different from zero in >> case of an undirected F contrast. You can also think of "Is the sum of >> A and B larger than B". There are instances in which such a comparison >> makes sense, but usually it's not what one wants to look at >> >> It seems you're interested in >> 1) voxels/clusters associated with higher activations for A relative >> to B, this would be [1 -1 0 0 0 0 0 0 0] >> 2) voxels/clusters showing a positive linear relationship with >> regressor C, which would be [0 0 1 0 0 0 0 0 0] >> 3) the conjunction of 1) and 2) = voxels/clusters with an effect A > B >> AND a positive correlation with C. For setting up a conjunction, you >> would first specifiy the contrasts for 1) and 2). Then press the >> Results button once more, select the two contrasts (while holding >> [control]) and choose "conjunction". However, it has been argued on >> the mailing list that these conjunctions are statistically invalid for >> within-subject designs. As a solution, you can save the SPMs for the >> two contrasts at a certain threshold and generate an image that shows >> the overlap. This is no proper statistical test of course. >> >> Best, >> >> Helmut >> >