Hi Bike
On 24 October 2010 14:15, Michael Bynevelt <[log in to unmask]> wrote:
> Hi Eugene,
> Thanks for your comments.
> Just to clarify
> 2 different runs for each subject
> 2 different hand movements - A & B (Tasks)
>>> Run 1 rArArBrArBrBrArB
>>> Run 2 rBrArArBrArArBrB
>
> If I understand this correctly for run 1 for example, analyse each task independently, ..... following this performed a higher level analysis to obtain the mean effect.
>
> The number of EV's for each first order analysis therefore would be dependant on the variations within the run.
>
> Not sure how we accommodate for differences in the cycling.
>
> ??if for example the number of TR's (3sec) in each r, A and B is 4
>
> first order analysis
>
> EV1 A task Skip 0 Off 12 On 12 Phase 0 Stop after 48
>
> then
>
> EV2 A task Skip 48 Off 12 On 12 Phase 0 Stop after 72
>
> then
>
> EV3 A task Skip 72 Off 12 On 12 Phase 0 Stop after 96
>
> then
>
> EV4 A task Skip 96 Off 12 On 12 Phase 0 Stop after 144
>
> then
>
> EV5 A task Skip 144 Off 12 On 12 Phase 0 Stop after 168
>
> then
>
> EV6 A task Skip 168 Off 12 On 12 Phase 0 Stop after 192
> 4 contrasts (for each EV)
>
> 1 A 1 0 1 0 1 0
> 2 B 0 1 0 1 0 1
> 3 A-B 1 -1 1 -1 1 -1
> 4 B-A -1 1 -1 1 -1 1
>
> I'm probably completely on the wrong track here.
>
Yeah, a little bit! You need to use the 3 column custom regressor
approach to define a single regressor for each of A and B. You'll
need a separate text files for the two different orders.
Eugene
> Thanks,
>
> Mike
>
> On 24/10/2010, at 7:06 PM, Eugene Duff wrote:
>
>> Hi Mike
>>
>>
>> On 24 October 2010 05:51, Mike Bynevelt <[log in to unmask]> wrote:
>>> Need help ...suppose that's why I'm here.
>>> Involved with a study requiring ANOVA analysis
>>>
>>> 13 controls, 13 subjects
>>> 2 paradigms with 3 conditions
>>> 1 rest 2 differing hand movements
>>>
>>> Paradigm 1 r1r1r2r1r2r2r1r2
>>> Paradigm 2 r2r1r1r2r1r1r2r2
>>>
>>> Needing to compare r and 1, r and 2, 1 and 2 initially
>>
>> This shouldn't be too difficult to set up. By two paradigms, do you
>> simply mean two different orders? Do you have both for all subjects?
>>
>> If so, and you're just wanting to do the comparisons you specify,
>> model 1 and 2 with separate regressors at the first level, merge them
>> as shown in the Multi-Session & Multi-Subject (Repeated Measures -
>> Three Level Analysis) example on the web page, and then model an
>> average effect and between group differences at the third level.
>>
>> Cheers,
>>
>> Eugene
>>
>>
>>>
>>> Given the kind of randomness of the paradigm, unsure how to set up in modelling.
>>>
>>> Thanks in advance for any help.
>>>
>>> Cheers,
>>> Mike
>>>
>>>
>
>
|