Just to add to Donalds answer:

As for the flexible factorial model, you can model the individual conditions (even if you have a factorial design) as one factor and the subjects as the other. Specify unequal variance but independence between subjects

In this case, you get as many columns in your design as you have conditions with the subject-factor accomodated in the model. Entering an F-contrast in teh form

 1 0 0

 0 1 0

 0 0 1

and the using plot -> contrast estimates -> select the entered F-contrast will show the contrast estimates, which correspond to the weighted (by the design matrix and hence the non-sphericity) mean of the images in each condition.

If you want to analyze effect-sizes, you will need to compare the first-level beta's multiplied by the regressor-hight with the individual session-mean. That is, you need to compare the signal change to the mean signal (for each subject) and then average across subjects. This can be done using several tools, e.g., MARSbar or the SPM Anatomy Toolbox.

Best

Simon