This is a report on a plausible US-based study  on the effects of violent video games, with a fairly large number of subjects (all boys for the detailed study, though also investigating broader patterns across genders). It finds no causal impact of violent video games on violent behaviour towards people, though the researcher does note some short-lived increase in destructive behaviour towards property among some children after playing violent video games. They suggest that the fact that video games tend to be played at home means that players' temporarily heightened aggression from playing the game tends to be taken out on things around them rather than on others, and the fact that the effects are short-lived leads to no long-term harm nor any changes in existing patterns of behaviour in general. I do wonder what it means for games played in video arcades, or mobile games played in public, though! Maybe video game arcades would benefit from having a punchbag near the door. It might be interesting to compare such things with behaviours during and after watching sports (especially violent sports) or violent movies in public. I am also intrigued by the finding (noted but not actually discussed in the study, I guess because it has no material impact on the research questions) that releases of popular violent video games result in increased time spent gaming in boys (4-5 months after a game's release) but not in girls. Why would that be?

It is also interesting to think about why no significant long-term effects are seen for games, while there are plenty of studies showing widespread and profound harmful effects of using some other kinds of software, most notably social networking systems. I'd hypothesize that the difference may be due to the fact that identities in social media networks tend to be far more strongly associated with users' sense of self, personal identity, status, and feelings of belongingness in the real world even though, in fact, such identities are usually carefully curated and only represent chosen aspects of real identities, usually presenting a highly distorted and filtered view (even when entirely truthful). Games, on the other hand, are, well, just games, in which gamers play the roles needed for the games, very deliberately trying on different personas in fictional situations, that can easily be shrugged off at the end of the game. In that sense, it is much like how we identify with or recognize characters in movies and plays. They are not us, but they let us understand those around us more easily, to see the world through different eyes, and to imagine possible situations that we may never actually find ourselves in. This suggests that there may be a useful place for social networks where people can take on different personas and identities. Indeed, some social networks have done so. For a while, before being taken over by Tagged, Hi5 used to have a lot of fictional members (or members playing roles of historical characters), and some sites such as AncientWorlds (currently defunct but due for revival) have social networks consisting of nothing but fictional characters or characters from history.

The article I've linked to is just a summary of the study. You can read the original paper here (AU library login required). It's worth doing so for both for the much greater depth and detail, and to read the many caveats the researcher makes on the reliability of their results, which are scattered liberally throughout the paper, and that reveal a great deal of uncertainty and potential sources of inaccuracy around many of the measures used. It's a good study, but it is very far from definitive and it does have some obvious and less obvious limitations that the researcher discusses in some depth.