I'd follow the following steps.
(1) Identify your null hypothesis.
(2) Write the null hypothesis (e.g. H0: A=B)
(3) Make the equation equal to 0 (e.g. A-B=0)
(4) Take the weights and insert them into the contrast.
In this case, [1 -1].
In your case, since you are comparing A and B, the contrast will also
be [1 -1].
The number of trials does not change the amplitude estimate of the
BOLD response -- assuming that all trials should have the same BOLD
response.
On Thu, Feb 28, 2013 at 8:00 AM, Shengchuang Feng
<[log in to unmask]> wrote:
> Hi SPMers,
> I want to compare two conditions with different trials: one condition has 20
> trials, and the other one has 60 trials.
> Can I simply set the contrast as A: [1 -1/3]
>
> or B: [3 -1]?
> Which one is right or both are wrong?
> Thanks in advance for your response!
>
>
>
>
|