Legal Niches

Researcher Disputes Study Equating Violent Games w/Aggression & Prosocial Games w/Helpfulness

By gamepolitics

Yesterday GamePolitics reported on a study detailed in the current issue of the Personality and Social Psychology Bulletin which found that violent game players displayed aggressive behavior while those who player more prosocial games exhibited helpful behavior. The study actually encompasses three seperate research projects which took place in Japan, Singapore and the United States.

But a researcher from Texas A&M disputes those findings. Prof. Chris Ferguson, who has frequently studied video game issues, commented on yesterday's report which was authored by, among others, University of Michigan's Brad Bushman and Douglas Gentile of Iowa State.

Of the Bushman-Gentile study Ferguson told GamePolitics:

You know trouble is brewing right in the beginning as they start with the false premise that there is an established relationship between video games and aggression. The authors engage in what's called citation bias, which means they only cover research they like and ignore anything they don't like. This is just not good science. Since this literature review is so slanted, that worries me about how they collected and analyzed their data.  

In [one study] they note that there is a high correlation between prosocial exposure and violent game exposure. This suggests that these may be some of the same games that have both kinds of content! They then suggest that there wasn't a problem with multicollinearity (basically means if you include 2 predictors that are too similar it can screw up your results), yet they only say they had no VIF less than 10...yet even something as low as 4 or 5 is pretty high. So multicollinearity may have been a bigger problem than the authors try to suggest.  Therefore, there may be some serious problems with their analyses here.  

[Also] the authors say that prosocial exposure and violence exposure were very highly correlated and then claim they have completely opposite effects. That is just highly unlikely.

In [another study] the standardized coefficient between playing prosocial games and prosocial behavior... suggests that playing prosocial games had almost no overlap with prosocial behavior one year later. Here we have yet another example of a "significant" finding being touted even though it's so small you'd never notice it in the real world. They also assert causality from correlational data which they can't do no matter how they analyze it.

The final study is probably the best of the three, but it's also the most artificial. Indeed, a fair number of their participants express suspicion about what went on. These kinds of studies have a high risk of "demand characteristics" In other words, students will give you the results they think you want and they won't admit to it afterward. Also the resultant effect sizes are all pretty small.

So, at best, a mountain is being made out of a molehill here, and at worst there are some pretty serious flaws in all analyses. I do worry about the "tone" from this research group. They do not comprehensively cover the literature honestly, and appear to have a hypothesis that they favor from the get-go. That tone would lead me to question their objectivity and, as such, the quality of their analyses.

Bottom line  - I doubt you'd see prosocial games solve the world's ills anymore than violent games have caused any outbreak of youth violence. 

Full post as published by GamePolitics.com on June 18, 2009 (boomark / email).