And I was just about to comment on the lack of reliability of such a study that includes ONLY 30 participants. Any break-down within this group (by aggressiveness level of the games, by gender, by SES, etc) leaves you with two or more groups, each of less than 30 participants. When social scientists say N>=30, they refer to the N of each subgroup in a statistical analysis.
Looks like a qualitative study might have been more appropriate, considering the sample size and the various reservations you mentioned.

I AM referring to random sampling. No parametric statistical test is valid when your subgroups contain less than 20 cases each.