Getting back to Ferguson’s results, he concludes that players of Medal of Honor: Allied Assault were not significantly more aggressive in the CRT than players of Myst III, but there were other factors that did seem to matter. Interestingly, this experiment included a group that was given the open to choose they game they would play, offering insight into how a possible preference for violent games might be exacerbated by then playing them into subsequent aggressive responses in the CRT. Turns out that the only variable that predicted anything was gender. Men were more aggressive than women. No other variable – not perceived excitement of the game, not perceived frustration, not ethnicity, not violence levels, not preference for Medal of Honor – had any effect on aggressive behavior. Even aggressive personality didn’t seem to correlate, a result noted as “theoretically puzzling,” but weirdly consistent with previous similar research on such topics.
The same study also included a large survey section broken into several parts, and I’ll just skip to the results. The variables that seem to statistically predict aggressive personality were exposure to verbal abuse, exposure to physical abuse, absence of parental affection, and most importantly, gender. Experience with violent video games did not correlate, and interestingly, neither did witnessing domestic violence. The other part involved a survey of violent crime history, which was predicted by aggressive personality and exposure to physical abuse. Video games were insignificant, except – and this is important – in the case of subjects with the most aggressive personality. That might have been a checkmate for Anderson and Bushman if it weren’t demonstrated in the previous section that video games have no role in formation of aggressive personality.
In fact, this final result supports Ferguson’s “catalyst theory,” which a broader, competing theory to the General Aggression Model. Ferguson suggests that a violence-prone personality is the result of genetic predisposition toward violence then shaped by direct environmental factors like family violence. A violent media environment is not direct enough to have casual links. However, these individuals often tend to gravitate toward violent entertainment, a hypothesis backed up by Ferguson’s results with the most violent personalities in this experiment.
Ferguson and company have done a lot of thorough work on this subject, especially with very nuanced statistical analysis of surveys. Here’s one that focused on kids, and here’s one that focuses on 8th graders. It should be noted that while the experimental results into this topic are not consistent (Anderson finds links, Ferguson doesn’t), the survey analyses are consistent – and in Ferguson’s favor.
Now, if it seems like I did my own hole-poking in Anderson and Bushman’s work without applying the same rigor to Ferguson, that’s because Ferguson’s work doesn’t list the same sort of limitations. The burden of proof is on the positive; scientists have to be more cautious about concluding causality than they do about concluding there’s none. Again, perhaps an unfair advantage to Ferguson, but there it is.