Slashdot Mirror
New Study Links Video Games and Mental Problems
eldavojohn writes "A new study published today in Pediatrics Journal conducted in Singapore on three thousand children in grades third, fourth, seventh and eighth claims that one in ten are video game addicts and almost all of those suffer mental health problems. This comes conveniently after the suspect in the Tucson shooting has widely been reported as an online gamer. Among the accusations from the study are that playing video games leads to lower school performance and fewer social skills while exacerbating existing depression, anxiety and social phobias. Gamasutra reports that the Entertainment Software Alliance is already criticizing this study, saying, 'Its definition of "pathological gaming" is neither scientifically nor medically accepted and the type of measure used has been criticized by other scholars. Other outcomes are also measured using dubious instruments when well-validated tools are readily available. In addition, because the effect sizes of the outcomes are mainly trivial, it leaves open the possibility the author is simply interpreting things as negatively as possible.' It seems that the doctors are still disagreeing on whether or not gaming causes problems."
Or, you know, it could be that people with mental problems also have a predisposition to become video game addicts.
This is the kind of crap that keeps people from thinking straight. Video games do affect people. If you play them every now and then its normal. BUT if you play them to the point where you can't pry yourself away from them, then you have problems. For example if I eat like a pig and can't stop eating nobody would ever say, "oh no problem there." Or if you read, read, and read, and read to the point where you drone out reality everybody would say, "oh there is a problem." So why on this green earth can't people in slashdot admit that if you overdose on gaming then you have a problem!!!! Addiction, is an addiction and gaming is a vent for that addiction.
"You can't make a race horse of a pig"
"No," said Samuel, "but you can make very fast pig"
Actually, it seems to me you illustrate an even bigger problem.
The way I remember it, a correlation in statistics (as opposed to the usual "I have a couple of anecdotes and watch me leap to a conclusion") involve looking at the covariance of two variables vs their normal distribution for _both_ variables. Even in binary terms, you'd have to look at the set of people who, say, do bad in school, people who play games, and the intersection. Though a more useful correlation would look at something like SAT grades vs hours played, or some such.
And even then, you know, actual measured variables than someone's self-assessment. See for example Dunning Krueger for one problem with self-assessments.
Basically you don't have to look at just how many people skipped school for gaming, but basically at whether you're seeing more than the product of two unrelated probabilities. The relevant question is, basically, are people who play video games more likely to skip school than those who don't?
What I'm getting at is that asking "have you ever skipped school to play a game?" without also asking "have you ever skipped school?" is pretty worthless. A questionnaire like yours which asks, or _also_ asks, about the distribution of that variable without the conditional, would actually be a better exercise.
IOW, asking just "have you ever skipped school to play a game?" will produce a semblance of a correlation just because there is no way to say, "does it count if I skipped school to smoke behind the school instead?" It's like asking "have you ever masturbated in the bathroom?" and concluding that bathrooms cause masturbation. It's not a real covariance if they're together simply because the question is phrased to only allow a "yes" if they appear together.
A polar bear is a cartesian bear after a coordinate transform.