Hi - please copy in old emails to save people having to look for them….I've done that below now.
Yes, I would go with option 1 (fixed-effects second-level analyses per subject).
Cheers.
On 27 Nov 2012, at 00:38, Jennifer Townsend wrote:
> Hi Steve,
>
> I think it would be meaningful in some sense. We interested in looking at response inhibition using the exact same stimuli and frequency of the "inhib" stimuli using both a block design and an ER related design. Both design-types are widely used in the lit and largely seem to activate similar (not not exact) networks. We'd like to formally test this in the same subjects to see if there is a significant effect of paradigm type (ER vs Block) and thus look at to what degree task design may influence the similarities/discrepancies in the current lit. Just from eye-balling it, the results look somewhat similar but I'd like a way to statistically test this.
>
> Do you have a recommendation regarding which is the correct setup (1 or 2, from my prior message)?
>
> Many thanks in advance for your insight into this issue,
> Jen
>
>>
>>
>> On 25 Nov 2012, at 00:01, Jennifer Townsend wrote:
>>
>>> I am trying to compare activation from an event-related version of a paradigm to a block version. Our subjects have 1 run of the block and 3 runs of the ER. I have 1st level analyses for each of these 4 runs per subject and am now wondering which is the correct way is to setup the higher-level Block vs ER analysis:
>>>
>>> 1.) Do I first create a 2nd level ER analysis for each subject (with the 3 runs as the input) and then set up a 3rd level analysis with the copes from the block 1st level and the ER 2nd level (so 2 copes per subject) and run a paired t-test?
>>>
>>> or
>>>
>>> 2) Do I create a 2nd level analysis with 4 inputs per subject (1 block FEAT and 3 ER FEATs) and then set up the paired t-test with contrast weights 1 (block) and .33 (for each of the 3 ERs)?
>>>
>>> Many thanks,
>>> Jen
>>>
>>
|