Hi Steve-
I have looked over my original email and you were right to be
confused as it wasn't clear.
Let me restate the situation:
2 groups (groupA, groupB) with 12 subjects each.
each subject did 8 runs (4 for condition C, 4 for condition P)
the lower level analysis was done for each run with 3EVs (ML, MH, SH)
and 4 contrasts:
1 0 0
0 1 0
0 0 1
0 1-1
then a higher level analysis was done to combine the 4 runs of each
condition for each subject.
this leaves us with 2 gfeat directories for each subject - one for
condition C and one for condition P. each has four cope.feat
directories corresponding to ML, MH, SH and MH-SH from the lower-
level analyses.
say what we want to compare:
groupA vs groupB, ML - MH for condition C
ML is cope1, and MH is cope2, from the gfeat directories.
therefore for the higher-level analysis we have 48 inputs:
first 12 are groupA condition C, cope1 (EV1)
next 12 are groupA condition C, cope2 (EV2)
then 12 groupB, condition C, cope1 (EV1)
then 12 groupB, condition C, cope2 (EV2)
would the contrast be [1 -1 -1 1]?
please let me know if this still isn't clear.
thanks again!
R
On Jun 23, 2007, at 2:13 AM, Steve Smith wrote:
> Hi, let's reconfirm what's in this higher-level analysis:
>
> In this particular analysis, all entries correspond to ML-MH
>
> The first 12 entries (modelled by EV1) are groupA condition C
> The next 12 (EV2) are groupA condition P
> Then group B condition C
> Then group B condition P
>
> Is this not correct? I wasn't 100% sure from your original email
> what was in this higher-level analysis.
>
> Cheers.
>
>
>
> On 21 Jun 2007, at 18:57, Remya Nair wrote:
>
>> Hi Steve-
>>
>> Thanks for your help.
>> I just want to confirm one thing though...
>>
>> I understand with the way you set up the EVs, but to compare
>> groupA vs groupB ML-MH for condition C, shouldn't the contrast be
>> [1 -1 -1 1]?
>> and not [1 0 -1 0] as you wrote.
>> My understanding is that would be a comparison of groupA cope1 vs
>> groupB cope1, right?
>>
>> Thanks again.
>> R
>>
>> On Jun 17, 2007, at 6:08 PM, Steve Smith wrote:
>>
>>> Hi,
>>>
>>> On 15 Jun 2007, at 17:14, Remya Nair wrote:
>>>
>>>> hi-
>>>>
>>>> i was hoping someone could help me set up an anova.
>>>>
>>>> let me describe the experiment:
>>>>
>>>> there are 2 groups with 12 subjects in each group.
>>>> each subject did 8 runs (4 run for condition C and 4 runs for
>>>> condition P, with 3 levels for each condition (ML/MH/SH)).
>>>>
>>>> i ran the first-level analysis with 3EVs (ML/MH/SH) and then
>>>> combined
>>>> the 4 runs of a single condition.
>>>>
>>>> now i have 2 .gfeat directories for each subject: 1 for condition C
>>>> and one for condition P (each has 4 cope.feat dirs - ML, MH, SH,
>>>> MH-SH).
>>>>
>>>> i want to look at the group differences across different conditions
>>>> e.g., groupA vs groupB, looking at ML-MH for condition C.
>>>>
>>>> so i would have 48 inputs in my higher level analysis:
>>>> 12 groupA cope1
>>>> 12 groupA cope2
>>>> 12 groupB cope1
>>>> 12 groupB cope2
>>>>
>>>> but i am not sure how to set up my EVs and contrasts.
>>>
>>> Something like:
>>>
>>> 1 0 0 0
>>> 1 0 0 0
>>> 1 0 0 0
>>> 1 0 0 0
>>> 1 0 0 0
>>> 1 0 0 0
>>> 1 0 0 0
>>> 1 0 0 0
>>> 1 0 0 0
>>> 1 0 0 0
>>> 1 0 0 0
>>> 1 0 0 0
>>> 0 1 0 0
>>> 0 1 0 0
>>> 0 1 0 0
>>> 0 1 0 0
>>> 0 1 0 0
>>> 0 1 0 0
>>> 0 1 0 0
>>> 0 1 0 0
>>> 0 1 0 0
>>> 0 1 0 0
>>> 0 1 0 0
>>> 0 1 0 0
>>> 0 0 1 0
>>> 0 0 1 0
>>> 0 0 1 0
>>> 0 0 1 0
>>> 0 0 1 0
>>> 0 0 1 0
>>> 0 0 1 0
>>> 0 0 1 0
>>> 0 0 1 0
>>> 0 0 1 0
>>> 0 0 1 0
>>> 0 0 1 0
>>> 0 0 0 1
>>> 0 0 0 1
>>> 0 0 0 1
>>> 0 0 0 1
>>> 0 0 0 1
>>> 0 0 0 1
>>> 0 0 0 1
>>> 0 0 0 1
>>> 0 0 0 1
>>> 0 0 0 1
>>> 0 0 0 1
>>> 0 0 0 1
>>>
>>> and for the question:
>>>> e.g., groupA vs groupB, looking at ML-MH for condition C.
>>> use the contrast [ 1 0 -1 0] etc.
>>>
>>>> also, what if i want to look at ML vs. (MH-SH)?
>>> Probably the best thing is to achieve this contrast at first-
>>> level making the higher-level analyses easy. At first level this
>>> would be:
>>> [1 -1 1] though I'm not sure whether you do exactly mean ML -
>>> (MH-SH)
>>>
>>>> likewise, is it possible to look at group differences across
>>>> conditions C+P?
>>>
>>> Sure - in a similar manner to the above.
>>>
>>> Cheers, Steve.
>>>
>>>
>>>
>>>>
>>>> thank you in advance!
>>>>
>>>> R
>>>
>>>
>>> --------------------------------------------------------------------
>>> -------
>>> Stephen M. Smith, Professor of Biomedical Engineering
>>> Associate Director, Oxford University FMRIB Centre
>>>
>>> FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
>>> +44 (0) 1865 222726 (fax 222717)
>>> [log in to unmask] http://www.fmrib.ox.ac.uk/~steve
>>> --------------------------------------------------------------------
>>> -------
>>>
>
>
> ----------------------------------------------------------------------
> -----
> Stephen M. Smith, Professor of Biomedical Engineering
> Associate Director, Oxford University FMRIB Centre
>
> FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
> +44 (0) 1865 222726 (fax 222717)
> [log in to unmask] http://www.fmrib.ox.ac.uk/~steve
> ----------------------------------------------------------------------
> -----
>
|