See responses below.

On Tue, May 13, 2014 at 12:31 PM, Alla Yankouskaya <[log in to unmask]> wrote:

Dear SPMers,
I have data from two tasks assigned to different runs. I would like to compare, say, condition 1 of task 1 (runs 1,3,5,7) and condition 1 of task 2 (runs 2,4,6,8). I would appreciate your suggestions for the following questions:
1. Can i put the two tasks in one design matrix?

Yes. Just make a big design matrix of the 8 runs.

2. Can i make a comparison between conditions adding a run as a covariate?

You don't need a run covariate in your analysis. However, run will be a confound. There is no way to regress the confound out of the results. This is why it is preferable to have the tasks intermixed in each run, rather than split across runs.

3. What is general disadvantage of such comparison?

If you find a difference, you don't know if the neural activity of the tasks are different or if the neural activity during baseline is different.

Because the baseline activity might be different, you could end up not finding differences even when the brain activity is different.

thank you in advance
alla