Nov 25 2008
Correlation, Causation and Media Violence
Scientists and science reporters should know that because two variables show correlation — change in one associated with change in another — does not mean that the one caused the other. The educated person who forgets this ought to write “correlation is not causation” 100 times on the nearest blackboard.
That goes for the researchers and writers of this article over at ScienceDaily: Media Violence Cited As ‘Critical Risk Factor’ For Aggression.
Okay, let me state outright that I believe media violence can contribute to aggressive behavior. But its a complicated issue and we’re still trying to sort it all out. I was very glad to read this nod to the fact with the statement that violence is a “multiply determined behavior.”
Here’s the article’s lead graph:
You are what you watch, when it comes to violence in the media and its influence on violent behavior in young people, and a new paper, lead-authored by Rutgers University, Newark, researcher Paul Boxer, provides new evidence that violent media does indeed impact adolescent behavior.
Hmm. Violent media does impact behavior? How do they know?
The research shows that even when other factors are considered, such as academic skills, encounters with community violence, or emotional problems, “childhood and adolescent violent media preferences contributed significantly to the prediction of violence and general aggression” in the study subjects.
Did you catch that? Media preferences contribute the the prediction of violence. Yes, and having a left arm is a good predictor that an animal will also have a right arm. Does having a left arm cause a person to grow a right arm?
The word preferences is also very telling. A preference for violent media may be the result of a third variable that influences both the watching of violence and engaging in violence.
How was the research conducted? I’m glad you asked, for it is a very important question.
It reports the results of the research team’s extensive interviews of 820 adolescents from the state of Michigan – 430 high school students from rural, suburban and urban communities, and 390 juvenile delinquents held in county and state facilities. The adolescents were about evenly split between male and female, minority and non-minority. Parents or guardians of 720 of the youths also were interviewed, as were teachers/staff of 717 of them. Each subject was asked about favorites TV shows, movies and video/computer games, both as a child and as a teen, and questioned to determine if they had engaged in specific antisocial behaviors, such as throwing rocks or using a weapon.
Extensive interviews!!! Poor quality data here, folks. Not much a person could reasonably conclude from it.
“You are what you watch.” Or maybe “You watch what you are.” Of course, both may be true. May be. We don’t know yet.
Interesting take on the press release. Of course I agree wholeheartedly with you that correlation is not causation. If you read the original article you will see that we consider that and related caveats when attempting to interpret our data, and we are quite up front about the fact that the study was correlational in design. The experimental and longitudinal studies that are out there are far more conclusive regarding the causal influence of violent media. So your statement “we don’t know yet” is true only insofar as actual violent behavior is concerned. We certainly do know with respect to experimental analogs of aggression, and more general, lower-intensity forms of aggression, that you do tend to be what you watch.
What we show in the new study is a link between actual violent behavior and violent media preferences, controlling statistically for other risk factors implicated in the development of antisocial behavior. These two facets of the investigation are novel contributions and need to be replicated and extended. Happy to discuss this further with you.
Paul,
Thank you for commenting. I am pleased you did. I have two points and two questions.
Points:
- Unfortunately, and somewhat disturbingly, to read many journal articles a person must either be a member or subscriber. I hope this changes in the future. Just as precision is essential to the scientific endeavor (readers of this blog know that one of my pet peeves is that precision should be extended to the use of language) so too is transparency. It doesn’t make sense to me, beyond financial considerations, that scientific journals make it difficult to “look inside” a study.
- Upon re-reading my post, I see I may have undersold the current state of evidence for the influence of violent media on behavior. It is my tendency to err on the side of tentativeness. This, I think, both reflects the reality of science and best models a scientific attitude. I strive to make my tentativeness inversely proportional to the weight (and quality) of evidence.
Questions:
- On that note, could you point me to the best, recent studies conducted on the topic? I may have missed something.
- Who writes the news releases for your work? Is it generally the job of the institution’s “public relations” department and/or science journalists?
It seems to me a problem with the communication and consumption of science is that risks for populations are frequently interpreted as risks for individuals.
A case in point would be the statement “You are what you watch.” Whether or not an individual is or becomes what they watch will likely be determined by whether they choose what they watch, their personality/temperament, whether they identify with the protagonist or antagonist, etc.
Again, thank you for commenting.
Andrew,
By far, the most comprehensive (and, actually, quite readable) reference to which I could refer you is the monograph lead authored by Craig Anderson and published in the periodical, “Psychological Science in the Public Interest.” This is the policy-oriented outlet of the Association for Psychological Science (formerly American Psychological Society). I am almost certain that Craig has a copy of this posted on his website through the Psychology Department at Iowa State University. Excellent reviews and meta-analytic works also have been authored by Brad Bushman and Rowell Huesmann, both from my group at the University of Michigan. Brad posts his articles on-line too.
These works will lend credence for you to the idea of “you are what you watch.” It is true that choices drive consumption, but over time that effect washes out in comparison to the effect of the consumption per se. For example, in Rowell Huesmann’s 2003 paper which appeared in Developmental Psychology, early TV violence viewing predicted later aggression but early aggression did not predict later TV violence viewing.
As to the press release, mine was written by our PR office and I know this is the case at least for the University of Michigan as well. The headlines and leads are designed to be attention grabbers, I’m sure, which is partly why they include links and citations for the full studies.
[...] number of weeks ago I made this post: Correlation, Causation and Media Violence. I discussed a problem with a study finding and was pleased to have one of the authors comment on [...]
[...] and researchers know this. Reporters do not, and it’s a great source of misinformation (see here and here for [...]