> We are concerned that voxels with activation in one dataset may be
> dropped if they happen to be compared to zero value voxels in the
> second dataset.
Do you mean whether it is possible a voxel labeled "activated" for some subjects during individual analysis might end up showing no difference when compared to zero at group level? If so, yes it is possible. The reason is that the significance of a condition at individual level is evaluated by the condition effect relative to the noise level for that subject while the difference at group level is judged relative to the inter-subject variability.
AFNI Gang