Theresa,

(1) You can't select only the parametric regressor as it wouldn't make any sense. In your task, you have the canonical HRF + parametric modulator that has a HRF with a variable amplitude for each trial. This generates the mean response, plus the slope of the parametric regressor. Likewise, in gPPI, you need the mean connectivity of each task and the connectivity change related to your parametric modulator. Thus, you want your task and the parametric modulator. To select both the task and the PM, simply list the task and it should select the PM as well as the task (see lines when it gets the task data, ~540). Hope this answers the question and is a suitable solution.

(2) Thanks for catching the typo, I'll fix it in the next release of the manual.

(3) Your understanding is correct. In your case, was the VOI falling outside of the first-level mask? If the VOI is inside the mask, you won't notice any effect. In general, I recommend using the group mask to threshold the VOI file that is generated. If you could send me the VOI that is outside the mask and the the first-level mask, I can look into this further.