Hi Javier
> Hello,
>
> I have a 2X2 factorial design and therefore 4 possible experimental
> conditions. Each subject did two experimental runs such that in run 1 I only
> have conditions 1 and 2, and in run 2 I only have conditions 3 and 4.
>
well it's not really factorial isn't it?
> I was thinking about defining my factors at the first level analysis, but
> when attempting to estimate the model I get an error because there are only
> two conditions per run and SPM is looking for 4. I though that I need to concatenate the time series of both experimental
> runs in a single session within the model definition so that I can define
> the four conditions for this single session. I think that just adding the
> extra files to the scans is not a good idea, as concatenation of time-series
> is not so direct. Could you please tell me what would be the proper way to do this
> concatenation. Do I need to add an extra regressor? Is there any matlab
> function I need to run in my time series before?
> Thanks for your help,
> Javier.
>
>
It is generally not a good idea to concatenate time series because they
have different means, i.e. each session should be modelled separately to
estimate the cste term. You can still perform the analysis by 1.
computing a contrast by condition at the 1st level and 2. perform an
ANOVA at the second level ..
Best
Cyril
|