A new study (which is actually based on an old study) by Dr. Craig Anderson of Iowa State University and two other researchers is making news today because it suggests a link between violent video games and real-world aggression. I have written extensively about such studies here in the past, and have included a list of relevant links down below. But let me just use the opportunity to restate the fundamental problem with the way the press reports these things.
In the literature review the authors suggest that research on video game violence is consistent when this is hardly the case. The authors here simply ignore a wide body of research which conflicts with their views. A bibliography of research studies finding either null results for video game violence or results that suggest that violent game play reduces aggression is appended to this review.
The authors fail to control for relevant "third" variables that could easily explain the weak correlations that they find. Family violence exposure for instance, peer group influences, certainly genetic influences on aggressive behavior are just a few relevant variables that ought either be controlled or at minimum acknowledged as alternate causal agents for (very small) link between video games and aggression.
Overall results are very weak with effect sizes ranging from (.07 to .15). Video game exposure overlapped in this study approximately half a percent to 2% with the variance in aggression, which is as close to zero as one can get without being zero. If anything it is remarkable how little effect that violent games had on trait aggression, considering that other relevant variables were not controlled. Likely if other variables had been better controlled, such small effects may have vanished.
Lastly the authors link their results to youth violence in ways that are misleading and irresponsible. The authors do not measure youth violence in their study. The Buss Aggression measure is not a violence measure, nor does it even measure pathological aggression. Rather this measure asks for hypothetical responses to potential aggressive situations, not actual aggressive behaviors. [...] the authors appear to generalize their results to youth violence, but offer no compelling reason why this should be, particularly in light of the weak results they achieve. The authors also fail to note that during the period in which violent video games became increasingly popular, youth violence has plummeted approximately 66% to levels not seen since the 1960s (childstats.gov, 2008; FBI, 1951- 2007). Although I suspect the authors would simply try to argue that this does not matter, such arguments are disingenuous, particularly as they raise the issue of youth violence themselves.
In short, given the weak effect sizes, the lack of control of relevant variables, and the failure of the authors to acknowledge data and research which contradicts their hypothesis, I am left with little confidence that the results of the current study provides much meaningful information on the impact of violent games.
It's that fourth point he mentions above about defining "youth violence" or "aggression" that I think is most important. When you read through many of these studies that claim to find a link between violent games and real-world aggression, the definitions of "aggression" or "harm" are often a scandalous stretch of the imagination. (One study I read two years ago referred gossip among peers and siblings as a form of aggression!) More importantly, as Ferguson rightly notes, at some point the research needs to be compared against real-world data if we are going to take this "monkey see, monkey do" theory of media effects seriously. Of course, when you do look at real-world data, you find the exact opposite story: Juvenile crime rates have plummeted even as exposure to video games has exploded. I have documented this in my big paper on "Fact and Fiction in the Debate Over Video Game Regulation" as well as some of the essays I list down below.
Bottom line: There is another side to the story and the press needs to dig a little deeper to find it. At a minimum, they need to stop blindly accepting every assertion and assumption made by the authors of these studies and instead start asking some tough questions about the details. Most simply, they should (1) start by asking whether how the researchers account for other variables that influence human behavior and then (2) make them account for why their theories do no match up to real-world data.
Anyway, you can find Dr. Ferguson's own work on the matter here and in his letter to Pediatrics he references some other literature on the subject you should be reading for the other side of the story. Also, here are my relevant blog entries on the subject from past years: