Back

The Media Violence Myth
Page 3 of 5

So I looked up Rowell Huesmann at the University of Michigan and asked him about the mysterious loss of correlation at age 13. Rather than defend the failure of the study to find a measurable correlation, the professor of psychology blamed the anomaly on mistakes by his colleagues. "The little 8th-grade data they had collected was incomplete and clearly biased," Huesmann asserted in his response. "Once I joined the project in 1970 as Analysis Director, I argued successfully against analyzing or reporting at all on the 8th grade data."20 With a larger, "unbiased" sample, he added, the .31 correlation that turned up at age 18 might also have shown up in thirteen-year-olds. It's equally possible, of course, that it might not. The fact remains that the partial sample correlation at age 13 - published in 1972, with Huesmann's name on the paper - was effectively zero.

Despite these serious problems, Eron and Huesmann's investigation had hatched a result the NIMH could use to get media-muzzling Senators off its back, and the psychologists were encouraged to continue their followup studies with taxpayer support. "In 1980-82," Huesmann emailed me, "we .tracked down and reinterviewed as many of [the] boys [in the Columbia County study] as we could. We interviewed 198 males from the original 1960 sample of 436." By then the boys were 30 years old. A few of them had been convicted of violent crimes. Huesmann worked his statistical magic and came up with some impressive correlations.

In 1986, officially representing the American Psychological Association, he reported his team's new findings proudly to the Senate Judiciary Committee. "Because the National Institute of Mental Health was generous enough to give us funding," Huesmann told the senators, "we were able to go back 10 years later and 22 years later and track down these subjects, most recently in 1982 when these subjects were now 30 years old. We were able to look at the extent to which their early television viewing behavior related to their adult aggression and criminality..What we found was a strong relation between early television violence viewing and adult criminality. Television viewing in and of itself related to adult criminality, regardless of what the children were watching. But more specifically for boys, there was a strong relation between early violence viewing and later adult criminality."1 To make that twice-mentioned "strong relation" vivid, Huesmann presented the senators with a bar graph - "simply intended to be," he explained to me in his email, "a visual illustration of the correlation between age 8 TV violence viewing and adult criminality." The bar graph measured "Seriousness of Criminal Convictions by Age 30" on a scale of 1-10 against "Boys' Preference for Violent Television at Age 8." It showed three stark black bars stepping up from low preference (4.23 on the seriousness scale ) to medium preference (4.71 on the seriousness scale). The high preference group at 9.71 almost doubled in seriousness of criminal convictions, bumping the 10 limit.21 The clear implication was that an eight-year-old who watches mock violence on television is likely to grow up to be a rapist or a murderer.

Needless to say, Huesmann's bar graph was high drama and a call to arms. To the senators and the assembled press, it looked like clear evidence that how much violent television a boy watches in childhood will correspond closely to how heinous a violent criminal he will turn out to be two decades later. Since 1986, Huesmann has made that claim repeatedly. In 1996, defending his work in the Harvard Mental Health Letter under a headline calling media violence "a demonstrated public health threat to children," he claimed that his 1982 study found that "boys who spent the most time viewing violent television shows at age eight were most likely to have criminal convictions at age 30."22

But Huesmann has been curiously selective about where he reports his TV violence/criminal conviction finding. It went unmentioned in the final report on the 22-year aggression study that he and Eron published in the prestigious journal Developmental Psychology in 1984. Not one of the team's media violence findings appears there, not even the celebrated .31 correlation. Instead, the report affirms what psychologists have long known about aggressive behavior: that early aggressiveness predicts later violence and that violence runs in families. (Which doesn't make it hereditary. There's strong evidence that violence is learned behavior, and violence begets violence.) All the final report says about television, lamely, is that "examples of aggressive behavior are abundantly available in the media as well as at home, at school, and in the neighborhood."23 Watching violent television goes unmentioned. Evidently Eron's initial skepticism about the effects of television violence was justified.

Why should Huesmann's "strong relation" between violent television viewing and adult criminality have dropped out of his and Eron's final summary of twenty-two years of scientific investigation? The likeliest reason is that the independent scientists who reviewed the report when it was submitted to Developmental Psychology (in the evaluation process known as peer review) did not think the data justified the two psychologists's conclusions.

And what was that data? Huesmann has never published the crucial numbers that would make it possible to judge the significance of his age 8 violent television/age 30 violent criminal convictions correlation. The dramatic bar graph he showed to the Senate Judiciary Committee, with its low, medium and high TV violence bars plotted against seriousness of criminal convictions, doesn't give the number of boys for whom the two measures correlate. I found a clue to this puzzling omission in a paper Huesmann and a colleague published in a book Huesmann edited in 1994. The paper, portentously titled "Long-Term Effects of Repeated Exposure to Media Violence in Childhood," works all sorts of statistical sleight-of-hand to try to prove that watching TV turns boys into violent criminals. But buried in the text is a remarkable admission: "Unfortunately, the sample on which this conclusion was based was very small because of technical difficulties..While the results are significant, they mostly reflect the behavior of a few high violence viewers and must be treated very cautiously."24 Scientists are supposed to publish their data so that their claims to discovery can be checked, but even while grudgingly admitting that his data had problems, Huesmann chose in this 1994 book not to reveal the numbers.

I wondered what he was hiding. When I emailed him I bluntly asked him for the numbers. The answer was incredible. "The correlation between [age 8 TV violence viewing and adult violent crime]," Huesmann wrote me, "was entirely due to 3 boys who committed violent crimes and had scored high on age 8 TV violence viewing." Three boys! Huesmann's team had identified New York State records for 145 boys from the original age 8 study. Of these, 66 had committed crimes, but only 24 had committed violent crimes. The "technical difficulties" which Huesmann mentioned in his 1994 book, he now explained, were that "just 3 of the 24 boys arrested for violent crimes had contributed TV violence data [at age 8]." It happened that "all three had scored high on age 8 TV violence viewing." With serious violent crimes in adulthood and high TV violence scores in the third grade, the numbers on these three boys - the only boys with criminal convictions for whom age 8 TV data existed - poison the entire 145-boy sample. As Huesmann himself acknowledged, "if just these three boys had behaved differently, all the significant results could have vanished." David Sohn puts it differently. "For 142 of the individuals," he wrote me after reviewing my correspondence with Huesmann, "there is no relationship between TV violence at age 8 and arrests for violent crime. Huesmann knew from the very beginning that he did not have enough cases with data for the two key variables to permit a meaningful analysis. He does the analysis anyway and conceals the crucial facts about having only three cases. Of course, what he should have done is not to use such inadequate data."

But Huesmann went even farther. He made up a bogus bar graph that deliberately misrepresented his findings and used it to influence the Senate Judiciary Committee to pass a law intended to limit creative expression on television. With age 8 violent TV viewing data on only three boys with criminal convictions, he had no factual basis for presenting "Low" and "Medium" bars. All three boys scored "High" on TV violence viewing. The graph is a fraud.

Next Page