Richard Rhodes Responds

Rowell Huesmann and Leonard Eron's rebuttal to my report "The Media Violence Myth" is a characteristic example of their style of argument, complaining of ad hominem attack while attacking ad hominem. Much of their rebuttal my report already answers. A few issues merit further exploration.

Science is not a branch of religion. H&E can appeal to authority until the cows come home, but the proof is in the data, and their data don't prove much of anything except the old truism that aggressive children tend to grow up to be aggressive adults -- whether they watch mock violence on television or pick daisies.

H&E like to cite the Comstock and Paik meta-analysis because it tends to support their theory. They don't mention the several other meta-analyses that offer no such support. University of Toronto professor of psychology Jonathan Freedman has gone even further recently (in press) and done a detailed critical analysis of every identifiable published scientific investigation in English of the effect of media violence on aggression. (He found about 200 such studies, not the thousands usually claimed.) Freedman, formerly a professor at Columbia University, told the U.S. House of Representatives Bipartisan Task Force on Youth Violence on 13 October 1999: "The available studies provide no convincing evidence that viewing violence on television or in the movies causes aggression or crime and quite a bit of evidence that it does not. The research demonstrates either that media violence has no effect on aggression or that if there is an effect, it is vanishingly small."

H&E defend crediting only one of three measures of aggression they used in their Columbia County study: peer ratings, which produced their famous .31 correlation, but not self-reports or personality tests, which failed to show statistically significant results. Self-reports and personality tests aren't as reliable as peer ratings, they say. Well, I didn't choose which measures to use; they did. If they didn't think all three measures were reliable, why did they decide to use them in the first place? Discounting results that turn out not to fit your theory isn't science; it's special pleading. (Notice that H&E dodge explaining how a meager .21 correlation between age 8 "violent" TV viewing (we're talking about watching "77 Sunset Strip" and Three Stooges cartoons, by the way) and "aggression" at age 8 can disappear at age 13 and then grow to .31 at age 18. They don't explain these weird correlations because they aren't explainable in terms of human behavior.)

The 3 boys problem: H&E put on their dancing shoes to dance away from this one. They introduce the red herring "self-reported aggressiveness at age 30." That's not what I was talking about and that's not what Dr. Huesmann was talking about when he testified before the Senate Judiciary Committee. I suggest H&E reread my report. Dr. Huesmann's statement in the Senate was, "There was a strong relation between early violence viewing and later adult criminality." Not "self-reported aggressiveness." H&E had records for 145 boys. By age 30, 66 of these had committed crimes (for 42 of them, drug possession, drunken driving, perhaps theft and so on -- not crimes which H&E categorized as violent). Only 24 had committed violent crimes, which is what the study is supposed to be assessing.
But H&E found to their dismay that they only had age 8 TV viewing data on 3 of the 24. H&E's own professional colleagues consider such a paucity of data "inadequate." And if, as they say, they "never hid the exact results," they certainly hid the fact that the results were based on only 3 boys out of 145. I read through every one of their papers looking for the magic number, and in the end had to ask Dr. Huesmann to supply it. My professional consultants were amazed that he did, since it gave his game away.

There is indeed causal evidence that violence is learned in personal violent encounters. I report that evidence -- the work of American criminologist Dr. Lonnie Athens, associate professor of criminal justice at Seton Hall University -- and explore its meaning in my book Why They Kill. Dr. Athens interviewed several hundred incarcerated violent felons at length and identified a pattern of violent socialization common to them all. The pattern was not present or not complete in two other groups of people who had experience of violence but were not personally seriously violent whom he interviewed as controls -- battered women and common criminals. Athens's methodology was retrospective, of course, and is therefore always provisional. But to argue, as H&E do, that causation can only
be determined prospectively is to argue that we can never know the cause of a disease because it's unethical to infect a random group with the presumptive disease agent. In fact, of course, disease agents are often discovered retrospectively, by examining people who have the disease and identifying what's different about them from people who don't have the disease. This old and reliable method of establishing (provisional) causal connections was named (by the 18th-century Scottish philosopher David Hume) the "method of universals." Today it's called analytic induction. It worked for malaria and many other diseases.

It worked for Dr. Athens, who found a common pattern in every violent felon he interviewed that began -- began -- with being brutalized (usually but not necessarily within the family; sometimes on the streets, in gangs or in the military) but included three other, later necessary stages of violent experience, each leading to the next by a toxic combination of personal choice and social feedback, culminating in violent criminality. I certainly experienced violence as a child at the hands of a violent stepmother. I never made the choices, or experienced the feedback, that Dr. Athens found common to violent criminals. Neither I nor Dr. Athens claims that child abuse inevitably leads to violence later in life. That would be as absurd as claiming that watching television makes kids violent.

H&E imagine that I chose to expose the inadequacy (to put it politely) of their media violence research because my hobbyhorse is community mental health centers (CMHC). They imagine so because I pointed out that the money granted them for their research was taken from the NIMH budget allocation for CMHC. CMHC, at least in Vermont, actually do have documented success reducing child abuse, teenage pregnancy and other ills. My personal hobbyhorse is reducing the brutalization of children, followed closely by blowing the whistle on sloppy, marginal, biased social science research.

Why They Kill has not yet earned back its publisher's advance, nor is it likely to. Buy my book or read it at the library, it's all the same to me; you'll reduce my publisher's loss if you buy a copy, but I've already been paid.

I do confess to taking testosterone, at physiological replacement level, as prescribed by my endocrinologist for hormone replacement therapy -- the male counterpart to women taking estrogen after menopause. If H&E kept up with the literature, they would know that the supposed correlation between testosterone levels and aggression (speaking of unsupported correlations) has been discredited. That's not surprising, since violence is learned through violent socialization, not excreted hormonally or radiated electronically.

The false analogy between smoking/cancer and media violence/violence I dealt with in the Q&A that accompanies my report. A brief summary: an effective scare tactic, but it's apples and oranges. And why are supposedly respectable and ethical professors of psychology at the University of Michigan trying to scare us with false analogies?

Two last points. First, I'm hardly alone in arguing that the evidence does not support the theory that media violence is a risk factor for violent criminality. Anyone who would like to explore the question further should look up Jib Fowles, The Case For Television Violence (Sage: 1999), or, a British review edited by Martin Barker and Julian Petley, Ill Effects: The Media/Violence Debate (Routledge: 1997). These careful reviews by professional social scientists are, if anything, far more skeptical and scathing than I chose to be.

Second and finally, U.S. media -- movies, television, video games, popular music -- are seen and heard not only by American children but also by children in Canada, Mexico, Europe and indeed throughout the world. Yet violence rates vary greatly from one country to another: much lower in the U.K. and Europe than in the U.S., higher in Mexico, lower in Canada. U.S. homicide rates have declined dramatically in the past decade, from a high of nearly 10/100,000 in the late 1980s to a low of 5.8/100,000 in 1999 -- that is, by more than a third -- despite increased media outlets and thus exposure. These variations alone show that the media violence theory is false and the data that purports to support it poor science at best, self-serving and, yes, fraudulent science at worst. If Rowell Huesmann and Leonard Eron believed their own findings, their .31 correlation, their "later adult criminality," they would have included them in the final report on their Columbia County longitudinal study published in Developmental Psychology in 1984. They left them out, or their peers disagreed that the data justified including them. Q.E.D.