Print

Print


Dear Noelia,

I wanted to second Sarika's suggestion.  But I am confused by your response:
 "in this region there is no response for unisensory conditions"; if there
is no response for either A>0 or B>0, then you don't have a conjunction of
(A>0)(B>0)(C>A)(C>B).  So isn't this negative result appropriate?

-Tom

On Tue, May 10, 2011 at 2:54 PM, Noelia Ventura <[log in to unmask]> wrote:

>  Hi Sarika,
>
> Thank you for your response.
>
> This does not function for our particular data set as we include a 3
> conditions: A = auditory, B = visual and C = audiovisual. We are now
> examining the multimodal region. In this region there is no response for the
> unisensory conditions and therefore this leads to a null result in the
> conjunction you suggested.
>
> Do you have a solution for this? Or anyone else?
>
> Thanks,
>
> Noelia
>
>
> El 09/05/2011 11:58, sarika cherodath escribió:
>
>  Hi,
>
>  I am not sure whether this could work, but you can make a subtraction of
> L1 and L2 for each condition and then do a conjunction to create the
> required contrast.
>
>
>  On Mon, May 9, 2011 at 2:47 PM, Noelia Ventura <[log in to unmask]>wrote:
>
>>  Hi all,
>>
>> We would like to compare directly the results of two conjunction analyses.
>> These conjunctions show overlapping clusters. One conjunction shows a
>> larger cluster then the other. We would like to examine whether the voxels
>> that show up for one conjunction but not for the other differ significantly
>> from each other.
>>
>> In more detail, we have two tasks in the native (L1) and non-native
>> language (L2).  Each task consists of 3 conditions, (A, B, C).  In the
>> random effects we use a full factorial within subjects design and create a
>> contrast : (  and we do this conjunction for L1 and L2.  We would like to
>> know in which voxels the conjunction of L1 > L2.
>>
>>
>>
>> However, in spm you cannot directly compare to conjunctions with each
>> other. Or can you?
>>
>> We would like to use imcalc to test where i1 = conj L1, i2 = conj L2 and
>> our expression is i1 -i2> 0. However, i think we obtain a binary image from
>> that but we need a statistical image. Can anyone provide a solution for
>> this?
>>
>> Thanks for your help,
>> Noelia
>>
>
>
>
> --
> Sarika Cherodath
> Graduate Student
> National Brain Research Centre
> Manesar, Gurgaon -122050
> India
>
>
>


-- 
____________________________________________
Thomas Nichols, PhD
Principal Research Fellow, Head of Neuroimaging Statistics
Department of Statistics & Warwick Manufacturing Group
University of Warwick
Coventry  CV4 7AL
United Kingdom

Email: [log in to unmask]
Phone, Stats: +44 24761 51086, WMG: +44 24761 50752
Fax:  +44 24 7652 4532