Michael,
Is there something else I'm missing. Because when I look at my activationmaps:A-B zstat contrast map - I see activation in areas I expectB-A zstat contrast map - I see nothing at allA-B zfstat contrast map - I see nothing at allB-A zfstat contrast map - I see nothing at allI just think there is something I'm not doing right.
It is not clear what is worrying you about these results.
Firstly, the t-tests are directional, e.g. 1 -1 looks for A>B and -1 1 looks for B>A. In contradistinction, the f-tests test the power explained by the contrasts, and so lose directional information, e.g. an f-test containing 1 -1 or an f-test containing -1 1 are identical and both look for where either A>B or B>A.
Because t-tests in Feat are directional, they are one-tailed tests (we only look for significance in the positive tail of the null distribution). However, as f-tests are non-directional, an f-test containing a single t-contrast is only equivalent to a _two-tailed_ t-test on that contrast (where we look for significance in both tails of the null distribution).
Hence, you would need to double your p-value for the f-test in order to get the same probability under the tails, i.e. for positive and negative t-contrasts thresholded at p=0.05 you need to use a p=0.1 for the f-contrast to capture all of the former.
In short, your A-B zfstat map and B-A zfstat map will be identical (as f-tests are non-directional). And both should be the same as the combination of the A-B and B-A maps if you doubled the p-values used for the t-tests when you do the f-contrast. Without that adjustment, voxels could be passing one of the t-tests but not the f-tests - as appears to be happening in your data.
Cheers, Mark.
Hi, it looks like you're missing the subject-specific mean modelling- your EVs are all empty from EV2 onwards - see the example in themanual.Cheers.Oh no, I just didn't add my subject-specific mean modelling in message thatI posted.Here is my example: 10 subjects 2 conditions, paired-test.Group EV1 EV2 EV3 EV4 EV5 EV6 EV7 EV8 EV9 EV10 EV111 1 1 0 0 0 0 0 0 0 0 01 1 0 1 0 0 0 0 0 0 0 01 1 0 0 1 0 0 0 0 0 0 01 1 0 0 0 1 0 0 0 0 0 01 1 0 0 0 0 1 0 0 0 0 01 1 0 0 0 0 0 1 0 0 0 01 1 0 0 0 0 0 0 1 0 0 01 1 0 0 0 0 0 0 0 1 0 01 1 0 0 0 0 0 0 0 0 1 01 1 0 0 0 0 0 0 0 0 0 11 -1 1 0 0 0 0 0 0 0 0 01 -1 0 1 0 0 0 0 0 0 0 01 -1 0 0 1 0 0 0 0 0 0 01 -1 0 0 0 1 0 0 0 0 0 01 -1 0 0 0 0 1 0 0 0 0 01 -1 0 0 0 0 0 1 0 0 0 01 -1 0 0 0 0 0 0 1 0 0 01 -1 0 0 0 0 0 0 0 1 0 01 -1 0 0 0 0 0 0 0 0 1 01 -1 0 0 0 0 0 0 0 0 0 1And then my contrasts are:title EV1 EV2 EV3 EV4 EV5 EV6 EV7 EV8 EV9 EV10 EV11A-B 1 0 0 0 0 0 0 0 0 0 0B-A -1 0 0 0 0 0 0 0 0 0 0Shouldn't this be correct, no?Also I want to add F-tests to my contrasts because I want to detectactivation for both postive and negative responses.So my inital setup wiht contrasts was:title EV1 EV2 EV3 EV4 EV5 EV6 EV7 EV8 EV9 EV10 EV11 F1 F2A-B 1 0 0 0 0 0 0 0 0 0 0 * 0B-A -1 0 0 0 0 0 0 0 0 0 0 0 *But something strange is happening.Is there something else I'm missing. Because when I look at my activationmaps:A-B zstat contrast map - I see activation in areas I expectB-A zstat contrast map - I see nothing at allA-B zfstat contrast map - I see nothing at allB-A zfstat contrast map - I see nothing at allI just think there is something I'm not doing right.Your help would greatly be appreciated.Thanks