Study Finds Violent Games Reduce Violence - Hmmmm

Polygon reports news of a study mentioned in the New York Times that says it demonstrates the rise in sales of violent videogames does not cause a spike in the rates of violent youth crime. In fact, they say, it may even lower it. Hurrah! you might cry. But let's stop and do some science.

When it comes to this debate, there's no greater cry we should be emitting than, "correlation is not causation!" At everything. No matter the question. Because as inappropriate it might be when asked if you'd prefer chips or a side salad, it might just permeate the discussion to the point where we can stop both sides from making spurious claims.

This latest study, from the Social Science Research Network (downloadable), is being reported by both the New York Times and Polygon as demonstrating a correlation between a drop in violent crime by youths, and the rise in popularity of violent videogames. And at first glance, that seems like good news. If violent games were causing violence, we'd see an increase in reported crimes wouldn't we? Without an increase - with a drop, in fact - then surely we can throw the case out of court and assume games innocent?

But no, of course not. For just the same reasons that we mock the idiotic attempts to draw wild conclusions from correlating data to prove that games are destroying our young, we need to take the same attitude to claimed data that might suit us more.

"We found that higher rates of violent video game sales related to a decrease in crimes, and especially violent crimes," claims one of the paper's authors, Dr. Michael Ward of the University of Texas, after studying communities and monitoring sales of violent games and incidents of violent crimes. The paper concludes that the sale of violent games sees violent crime reduce by 0.3%.

But this is entirely based around correlation. In fairness, in this study, it's designed to. The researchers set out to do what they call "tests of external validity", investigating whether the findings of laboratory studies are reflected in the general population. The laboratory studies cited, as we've reported before, have shown short-term minor raises in aggression levels in participants who spend prolonged time playing violent games. Minor raises that would not cause a person to actually become violent (ie. they are such that a person not prone to violence would never behave violently as a result of gaming). And this seems to be the biggest issue with this study. In attempting to investigate if these findings do not apply to society, they seem to have set their sights on the wrong target.

The study measures for increases in violent crimes, alongside increased sales of violent games, on a weekly basis, over a number of years. And they found none. In fact, they found a pattern that seemed to suggest that when a violent game is released, through what they propose as catharsis, violent crimes go down. And thus, they conclude, the two are "associated". But based on what evidence?

"We calculate that video game unit sales increased by an average of 9.6% per year. Assuming this applies to both violent and non-violent games, our estimated violent video game-to-violent crime elasticity of approximately -0.03 would predict almost 0.3% fewer violent crimes per year due to violent video game sales."

What's the difference between this claim, and the so-called Center For Successful Parenting's daft declaration that television causes murders? It'd sure seem good if the matching lines on a graph meant it were true, but we can't just assume it is. Because what other factors are at play?

Let me pluck the first that came into my head: Perhaps it's about money. Maybe when a violent person spends $60/£40 on a violent videogame, they can't afford to go out at night for that week, so don't commit crimes on their way home from the bar? That's just as good a conclusion, isn't it? The paper's authors would likely point out that the weeks during high sales of non-violent games didn't have an effect on violent crime rates. But my argument is now that perhaps violent people only buy violent games. The reason crime doesn't change during high sales of a non-violent game is because they were all bought by toddlers and grannies, who weren't going to commit crimes anyway.

This detachment from the laboratory experiments reveals an enormous flaw. Without knowing some actual direct science, without a demonstration that there is a catharsis for a potentially violent person when playing a violent game, we can't just make this assumption because it seems like it might be nice. (It's worth noting that the paper does mention the possibilities of misdiagnosing causation, although with the rather flippant example of weather being the influencing cause, but then seems to dismiss it once again.)

And this is why they have the wrong target. When the credible laboratory experiments are demonstrating minor, short-term raises in aggression levels that stop far short of causing someone to become violent, what does this investigation demonstrate by looking for increases or decreases in violent crimes in relation to gaming? None of the papers they cite were suggesting there would be! The effects mentioned wouldn't cause there to be fewer or more violent crimes at all! To look for "external validity", they'd have to perform a far more specific and complicated study, following specific individuals and looking for incredible minor changes in behaviour - something that would be almost impossible to do, without relying on the individuals to report their own behaviours.

Further, we should be incredibly hesitant before concluding that their findings are a positive result. Assuming that their unevidenced implied causation were true, what are we to learn from it? That we need to keep a constant supply of violent games in the hands of the violent, in order to lower their chances of committing violent crimes? If anything, trying to diagnose a positive effect of gaming only serves to distract attention away from something that might meaningfully prevent violence in society. Certainly it's not the responsibility of a research group looking for specific results to prevent others from making a moral interpretation. But it doesn't help when authors give quotes to the press like the one above.

If you look at the very long-term figures, it could seem reasonable to conclude that violent games haven't caused an epidemic of violence. Violent crime in the US more than halved between 1994 and 2010, says the New York Times, while of course videogame sales massively increased over the same time. But we have to be careful here too. Yes, draw those two lines on a graph and you have a very satisfying-looking result. But doing so is to pluck two statistics out of the ether, and ignore absolutely every other factor in existence. What we do not know from such comparisons is whether those violent crime figures might have dropped even further were it not for the increase in the popularity of violent games. Any number of other factors could be the cause of the drop in crime numbers, with gaming perhaps inhibiting their fall. Or not. That's the point - these studies just don't demonstrate a meaningful result from which we can draw useful conclusions.

These are the same methods used by those who wish to draw spurious connections between a thing they don't like, and a thing a portion of society doesn't like. "The rise in the sale of iPhones directly correlates with the increase in immigration..." And they're not methods that should be embraced when the results lean in what be thought to be our favour, either.

And this is all ignoring that they took their sales data from VGChartz.