I saw the research discussed in this Guardian piece earlier this week and it was actually on my list to blog about, but the trouble with the article has beaten it to it.

First the abstract of the actual study:

News reports and even some researchers suggest that there is no consensus on the basic question of whether violent media increase aggression in children. The purpose of this study was to test whether media researchers are in fact divided on this issue, and to compare the opinions of researchers with those of pediatricians and parents. Participants (n = 371 media psychologists and mass communication scientists, n = 92 pediatricians, n = 268 parents) completed an anonymous online survey about whether exposure to different types of violent media (i.e., comic books, Internet sites, literature, movies, music, music videos, sports, TV programs, video games) increase aggression in children, whether that effect is causal, and whether it is a major factor in real-world violence. All groups agreed that exposure to media violence can increase aggression in children (overall d = 0.49, a medium-sized effect). Ratings for violent video games and movies produced the highest level of agreement, whereas ratings for violent literature and comic books produced the lowest level of agreement. This pattern was highly consistent across all groups, indicating broad consensus on this issue. The only question on which groups differed in their opinions was whether media violence was a major factor in producing real-life violence: parents and pediatricians agreed that it was, media researchers did not agree. Although a few vocal researchers claim there is a “debate” on this issue, the overwhelming majority of researchers believe that violent media increase aggression in children, and that the relationship is causal. Pediatricians are even more convinced, and parents also have little doubt.

This made many journalists say that there is a broad consensus that violent video games have a bad influence (also because of the title of the press release). But… does the article supports this idea. Well, not really if you look at the data and it depends on which version you read of the article as Pete Etchells and Chris Chambers in their explain in their Guardian blog post:

“According to an anonymous source close to the process, the initial version of the paper went through peer review like any other paper. Four independent reviews were provided; in one of these, the reviewer pointed out a statistical flaw in the analysis of the data. Despite that, the editorial team at PPMC decided to accept the paper, and invited the authors of the negative review to write a brief reply that would be published alongside the main article. Remarkably, it appears that once Bushman and colleagues had seen this reply, they submitted a revised version of the manuscript to the editorial board. The new version, which was unsolicited, attempted to shift the goal posts by making changes to the manuscript to address the critical reply. This is a highly unusual practice – once a paper has been accepted it is generally considered “locked in”, especially if a reply to the paper has also been accepted.”