To establish causation, researchers must rule out these "other factors" by performing a lab experiment. For example, Anderson ran an experiment in 2000 that had college students play a violent game (Wolfenstein 3D) or a nonviolent game (Myst). Then each subject played a game in which they could punish a student in another room with a blast of noise - though the game was rigged and the other student did not exist. The subjects who had played Wolfenstein chose longer blasts of noise than those who played Myst.

Anderson concluded that the violent game caused this sample of college students to behave more aggressively, because the game was the only factor that changed in the experiment. Everything else - the lighting in the lab, the length of play and the way the noise-blast game was run - was kept the same. Unfortunately, the extremely specific setting of the experiment makes it hard to apply the results to the real world, and the experiment proves only that the game made people more aggressive immediately after playing it.

Anderson's findings are a lot less meaningful than you might think - in fact, most can be explained by a commonly observed, widespread and powerful phenomenon called "priming." When someone thinks about something, the brain "warms up" the parts that relate to that topic. For example, if a fire alarm goes off in your school, you suddenly become much more sensitive to things like heat and smoke. If a test subject reads a passage about fruit, that person will probably pick words like "apple" or "tree" from a list of random nouns. Priming predicts behavior, too. Showing people things that relate to violence, like pictures of guns, causes them to behave more aggressively on tests immediately afterward.

On the crucial question of long-term effects, however, studies have not proved that playing violent videogames makes aggression more likely. Such a study would take a huge amount of time, money and effort - researchers would have to find a group of people who were similar in every way but the videogames they play. Then they would have to tightly regulate these subjects' game play and measure their tendency for aggression over months or years. Even then, there are so many ways for the data to be tainted that the experiment would have to be repeated many times before it could show a convincing connection.

Researchers like Anderson have not overcome this major hurdle. They have not shown a causative link between videogames and long-term increases in aggression. In light of this shortcoming, Anderson dismisses the importance of causation in his article for the APA, and points out that correlation can prove "inherently causal" theories. For example, astronomers can use correlational data to prove the existence of planets. This works because astronomers' assumptions of causation come from deeply reliable laws of physics that have been proven beyond a reasonable doubt.

Unfortunately, our understanding of the laws governing the human mind is murkier than that of the laws of physics. Like an astronomer with Einstein's theory of relativity, Anderson uses his belief in the gaming-violence connection as a cornerstone of his research. In other words, he accepts as fact the very hypothesis that he is attempting to prove. For his research to effectively test the link between games and long-term aggression, however, Anderson must be prepared to accept results that disprove this connection, an impossible feat when he implies the connection from the outset.

Comments on