Hi Chirag,

I was actually thinking on the images (i.e., brain images). In any case, let's recap: you are saying that, using exactly the same input data, reversed t-contrasts are giving the same result. This is really impossible unless there's an error, or an F-test.

Perhaps the contrasts were flipped at the 1st level, and then again at the 2nd level? Then each flipping cancels the other.

Cheers,

Anderson



On 9 September 2015 at 23:21, Chirag Limbachia <[log in to unmask]> wrote:
Hi Anderson,

Here are pictures of how I setup the contrast.

Thanks
Chirag

On Fri, Sep 4, 2015 at 3:20 AM, Anderson M. Winkler <[log in to unmask]> wrote:
Hi Chirag,
Sure, feel free to send if the file is small.
All the best,
Anderson



On 3 September 2015 at 22:37, Chirag Limbachia <[log in to unmask]> wrote:
Thank you Anderson! 

I retried doing the analysis, this time unpaired t-test. Still, I am noticing similar trend. I did not flip the data, I think so, and I am not using the f-test. I can send you the pic of my first and second level setup if you want to look. Your help is much appreciated. 

Thanks 
Chirag Limbachia 
Sent from my iPhone

On Sep 3, 2015, at 5:03 PM, Anderson M. Winkler <[log in to unmask]> wrote:

Hi Chirag,

Flipping the direction of a contrast in the paired t-test should reverse the direction of the effects and not give identical maps.

Identical maps can appear if either:
- By mistake the data was flipped twice, e.g., the input data somehow, as well as the contrast.
- Instead of a t-test, it's the result of an F-test, which is two-tailed and has no direction.

All the best,

Anderson



--
Anderson M. Winkler
FMRIB / Analysis Group
[ Blog | Twitter ]


On 3 September 2015 at 21:00, Chirag Limbachia <[log in to unmask]> wrote:

Hello all,

I had a question regarding group analysis (paired t-test),

There is a group of 5 subjects, scanned under two different conditions. Let say, an Easy and a Hard.

I performed a  group feat paired t-test to compare the two conditions. So, there were 5 input feat directories (5 subjects) for each condition, total of 10 inputs. The contrast defied earlier in the first level feat analysis for individual runs and later for the group run are as follows,


Individual                     Group
a                    
b                                Easy- Hrad
a-b                             Hard- Easy
b-a


After running the group feat, in the results, I found the spatial activation maps for,
a-b > Easy-Hard  to be exactly same as the b-a > Hard-Easy.

My question is why is there this redundancy and are these results reliable?


Thank you!