I came into the comments to query this because, yeah, my limited stats knowledge (I did stats 108, makes me a pro, I know) was at odds with OP's article. 109 kids seemed like a reasonable sample size. It'd be nice to get a link to the actual study though, you're absolutely right.Requia said:To all: THERE IS NOTHING FUCKING WRONG WITH THE SAMPLE SIZE. Jesus Christ you can do good statistics at n=10 if the effect size is strong enough, n=45 is fine. Small sample size issues don't even work that way, the problems you can get from small sample sizes are incorrect effect size estimates (not reported here, so not an issue) and false negatives (so for example it's possible nonviolent games have the same effect, but because the effect is weaker it was not detected).
It's difficult to tell with the correlation =/= causation stick. It is overused on the internet to handwave stats, but I'm not sure if it's entirely invalid in this case, although you didn't necessarily suggest that it was. It could simply be that children who happen to be morally underdeveloped (whatever that means) are more likely to enjoy and play violent video games.
Edit:
http://www.tandfonline.com/doi/full/10.1080/09523987.2013.836367