On Tue, Aug 7, 2012 at 12:53 PM, Ming-Tsung Tseng <[log in to unmask]> wrote:
> Thanks Mclaren. Big help!
>
> If my understanding is correct, that means if I want to know the average activation of condition 3 as well as activations attributed to its early and late phase, I have to do 2 SEPARATE models: one modelling condition 1, 2 (both are EVs of no interest), and 3 (condition 3 as a whole) and then using the contrast (0 0 1), and the other modelling condition 1, 2, early phase of condition 3, and late phase of condition 3 then using, for example, contrast (0 0 1 0) to know activation related to early phase of condition. Right?
>>>> Yes. That would be the easiest way to model things.
>
>
>> My question is:
>>
>> (1) To know activation related to early phase of condition 3, the contrast should be (0 0 1 0)
>> (2) To know activation related to late phase of condition 3, the contrast should be (0 0 0 1)
>> (3) To know activation related to the entire condition 3, the contrast could be (0 0 1 1) (I know 1 1 means the average activation of early and late phases of condition 3)
>
> 0 0 1/2 1/2 would be the average. 0 0 1 1 is twice the average. Both
> will give the same statistics and group maps though.
>
>>
>> Are (1) and (2) correct?
>> Does the result of (3) equal to Result A?
>
> No. Result A will only equal (3; 0 0 1/2 1/2) when the activity is
> constant over the whole event, which is unlikely since you are telling
> me that there is a early phase and late phase of the response.
>
>
|