Back

The Media Violence Myth
Page 2 of 5

Two legal scholars at the University of California at Berkeley, Franklin E. Zimring and Gordon Hawkins, refuted Centerwall's findings in a 1997 book, Crime Is Not the Problem: Lethal Violence in America. Zimring and Hawkins point out first that there are awkward problems with Centerwall's basic assumptions. How can television set ownership tell you anything about murder rates? Isn't television program content supposed to be the issue? And comparing white murder rates in the U.S. and Canada with white murder rates in South Africa, where whites represent fewer than five percent of the murder victims, is probably comparing apples and oranges.

Zimring and Hawkins tested Centerwall's theory more fundamentally by looking at homicide rates in four other industrial democracies - France, Germany, Italy and Japan. They found that the incidence of murder in those countries either remained more or less level (Italy) or actually declined (France, Germany and Japan) with increased television exposure. These counterexamples, they write, "disconfirm the causal linkage between television set ownership and lethal violence for the period 1945-1975."6

I sent Zimring and Hawkins' analysis to Centerwall for comment. He hadn't seen it before, but he told me he'd heard similar arguments. He was quick to offer reasons why he was right and the legal scholars were wrong. He said he interpreted the French and Italian graphs as confirming his theory - he thought they showed a longterm upward trend. Germany he acknowledged was different, "but since many other European countries that I didn't include in my paper had increased homicide rates, it doesn't bother me all that much." Japan isn't a Western country, Centerwall reminded me, arguing that "culture overrides television if it has a mechanism for dealing with physical aggression."

To explain the recent declines in homicide in the U.S. and England despite continuing and even increasing exposure to media, Centerwall redrew the theory of his study, claiming that it really should have been a two-factor model, factoring in not only television exposure but also economic conditions. Economic conditions affect the murder rate, he said: It goes up in bad times and when times are good it goes down. He said the television effect eventually saturates, after which its influence on the murder rate is steady-state. Thus, he claimed, rising postwar prosperity probably retarded somewhat the influence of television on the murder rate. Then, when that influence saturated, further prosperity kicked in to bring the rate down. He pointed to a particularly dramatic drop in English homicide rates between 1978 and 1981 as evidence of the success of Margaret Thatcher's economic policies, which he said had increased per capita income in England by 80 percent.

I passed along Centerwall's explanations to Franklin Zimring at Berkeley. In an emailed response Zimring barely restrained his scorn. Since Centerwall's theory is generated by U.S. and Canadian data patterns, he wrote, "it should be tested elsewhere." One way to do that is to look at U.S. and Canadian data after 1975. After 1975, it turns out, despite the continuing and increasing exposure to television, the homicide rates leveled off and declines. Centerwall claims the television effect saturates. "Why and how this might be," Zimring responds, "is anybody's guess" - that is, Centerwall offers no evidence for his saturation theory; it looks like something he made up to explain why the data don't fit his model. Zimring added that he'd never seen any evidence that economic conditions immediately impact homicide rates, but in any case, "the big drop in English homicide rates was between 1978's high and 1981's low. Mrs. Thatcher took office in 1979." Causes are supposed to precede effects, but homicides were already declining before the British economy improved.

As for culture overriding television in Japan, Zimring wrote, "says who, and when?" The French murder rate trends upward between 1980 and 1985 and then trends downward, "but all of this leaves French homicide in 1990 at 35 percent lower than it was in 1960" when Centerwall's theory would predict it to double. Italy, similarly, "goes up in the 1970s, drops back from 1981-1986, and then goes up again. How this pattern fits the Centerwall thesis is his secret." Centerwall told me he based his claim that other European countries also experienced doubled murder rates (a claim Grossman also makes) on Interpol data. Wrong data, Zimring advised: "Most Continental countries report homicide and attempted murder together, which led our current drug czar to assert recently that Holland had a higher homicide rate than the United States. But even General McCafferty would not use Interpol data, which is unaudited and notorious." In conclusion, Zimring wrote, "the off-hand and ad hoc quality of the responses that you report reinforce my disinclination to buy a used car from Dr. Centerwall." Yet Centerwall's theory has been a mainstay of American Medical Association and Congressional claims that television violence is destroying American youth.

Psychiatrists have been prominent players in the media violence controversy; though they have no special training in assessing broad social trends, people take them seriously because they're medical doctors. An illustrious predecessor of Brandon Centerwall's, the psychiatrist Frederic Wertham, indicted comic books in the 1940's and 1950s as fervently as Centerwall has condemned electronic media. (Every popular art form - the novel, the circus, Punch 'n Judy shows, comic strips, movies, rock 'n roll, video games, now the Internet - starts out condemned as trash. One generation's trash is the next generation's art form.) Wertham had worked with juvenile delinquents in New York City in the immediate post-World War II years when juvenile delinquency was on the rise and Congress was looking for answers much as it looked for answers in the 1970s and 1980s when the homicide rate was going up. "If it were my task, Mr. Chairman, to teach children delinquency," he testified before a Congressional committee in 1954, "to tell them how to rape and seduce girls, how to hurt people, how to break into stores, how to cheat, how to forge, how to do any known crime, if it were my task to teach that, I would have to enlist the crime comic book industry. Formerly to impair the morals of a minor was a punishable offense. It has now become a mass industry. I will say that every crime of delinquency is described in detail and that if you teach somebody the technique of something you, of course, seduce him into it. Nobody would believe that you teach a boy homosexuality without introducing him to it. The same thing with crime."7

In those days being gay was believed to be a serious mental illness, and Wertham was convinced that Batman and Robin were a blatantly homosexual couple created to entice new recruits. (Robin, he wrote, "is buoyant with energy and devoted to nothing on earth or in interplanetary space as much as to Bruce Wayne. He often stands with his legs spread, the genital region discreetly evident.")8 The psychiatrist thought Superman was a fascist and worried that the muscular Krypton native gave children "a completely wrong idea of.basic physical laws" by leaping tall buildings at a single bound.9 He called comic books "the marijuana of the nursery." Like Grossman and Centerwall, Wertham demonstrated that literal-minded humorlessness is a requirement for media bashing, but Congress and the public took all this unsupported slander seriously. The comic book industry, which published 130 million copies a month, including at least 30 million devoted to crime and horror, capitulated after the 1954 Congressional hearings and thereafter published only G-rated stories. Fortunately for popular culture, the writers and artists laid off at EC Comics, the hardest hit when the industry crashed, went on to found Mad magazine.

Medical authorities, medical organizations and state and federal legislators have awarded statistical studies of media violence broader endorsement than Dave Grossman's exaggerations or Brandon Centerwall's purblind graphs. Statistics are said to correlate when they change together. When Centerwall showed the U.S. murder rate and U.S. television-set ownership increasing during the same period of years, he was graphing a positive correlation between those two variables. If one variable had gone down when the other went up (as Centerwall claims murder rates and income do), that would be a negative correlation. That two variables correlate doesn't necessarily mean they're influencing each other; they may both be changing because of some third factor, or the change may be simply coincidental. Raincoats and umbrellas appear on the streets in increasing numbers on certain days of the year (a positive correlation), but raincoats aren't influencing umbrellas: Both appear because their owners believe it might rain. Correlations by definition can't reveal the cause of anything. They're simply interesting information which can sometimes offer clues about where to look for a cause.

The most celebrated correlations in the annals of media violence studies emerged from longterm investigations of aggression in school children conducted across twenty-two years (from 1960 to 1982) by psychologists Leonard D. Eron and L. Rowell Huesmann, both now professors at the University of Michigan (Huesmann joined the investigations in 1970). (Click here to read Huesmann's and Eron's response to this article.) According to David Pearl, who administered media research at the National Institute of Mental Health (NIMH), when the U.S. Surgeon General appointed a committee to review research on television violence at the beginning of the 1970s, Eron and Huesmann's investigation "was a key study leading to the Surgeon General's Committee conclusions."10 Two decades later, when Congress passed the Telecommunications Act of 1996 which requires all new television sets to be equipped with a V-chip enabling parents to block out programs they don't want their children to see, the text of the Act implicitly invoked Eron and Huesmann's findings to justify its intrusion: "Studies have shown that children exposed to violent video programming at a young age have a higher tendency for violent and aggressive behavior later in life than children not so exposed.."11

Eron himself has candidly called the television violence component of his longterm aggression studies "the tail that wags the dog." He said he and his colleagues "got a lot of financial support through [investigating television violence]" - hundreds of thousands of taxpayer dollars, in fact - but that doing so had not been part of his original research agenda, because he didn't think it was important.12 "More than 35 years ago," he reminisced in 1995, "when I started to do research on how children learn to be aggressive, I.was skeptical about the effects of television violence."13 In 1960, Eron and his colleagues began studying 875 third graders - boys and girls eight or nine years old - in rural Columbia County in upstate New York. They wanted to identify what childhood experiences correlated with mental health problems later in life, and they decided to use aggression as an marker, since it was something they believed could be measured objectively. They asked the children who started fights, who got into trouble, who said mean things. They questioned parents and teachers. They measured popularity, anxiety, IQ and family values. One measure they recorded almost as an afterthought was how much violent television each third grader watched.

In 1963 Eron reported finding a correlation between aggressive behavior at school (as estimated by classmate peers) and violent television watching at home. A correlation only emerged for boys; there was no such connection for girls. To further confuse the issue, kids who watched the most television overall turned out to be the least aggressive.14 Eron calls the finding for boys "unsuspected." He adds: "We didn't have too much confidence in the finding by itself" - nor should they have, given the zero finding for girls and the negative correlation overall. "You couldn't tell by these data alone," Eron explains, "whether aggressive boys liked violent television programs or whether the violent programs made boys aggressive - or whether aggression and watching violent television were both due to some other third factor."15 Nor had the federal government yet become interested in the problem. Eron's requests for grant support were turned down twice in the 1960s by the NIMH and once more by another government agency.16 But in 1970, when the Surgeon General's committee noticed the 1963 positive correlation for boys, it realized that the Columbia County third graders would now be graduating from high school, raising the possibility that a correlation between childhood exposure to violent television and adult aggression could now be measured. So the NIMH awarded Eron's team, now including Rowell Huesmann, a grant of $42,000, the first of several lucrative grants, and the psychologists were able to locate and reinterview 436 of the original 875 subjects. (The money the Surgeon General granted for such speculative media studies - $1.5 million in all - was gouged from the NIMH budget by eliminating or postponing the construction of community mental health centers, at a time when mental institutions were being closed all across America and tens of thousands mental patients were being turned out onto the streets.17)

In their followup, Eron and Huesmann found a correlation of .31 between boys' preference for violent television at age 8 (based on their mothers' estimates) and their peer-rated aggressiveness at age 18. In other words, the psychologists found that a preference for viewing mock violence on television in the third grade might account for 10 percent (the square of the correlation) of the childhood influences that led the boys to become aggressive adults.

Assessing this famous correlation, Jib Fowles points out that Eron and Huesmann had looked at two other measures of adult aggression besides peer reports: self-reports and the results of psychological tests which they administered. These two other measures did not correlate significantly with age 8 television preference. Nor did any of the three measures correlate for girls. Another research team, Fowles says, might conclude from such a poor showing - only one of six possible correlations turning out to be significant, and that one only weakly - that their data failed to support their theory. Eron and Huesmann chose instead to highlight the one correlation that might. "It is difficult to believe," Fowles concludes, "that a study with such a weak single finding has been taken so seriously by so many thoughtful people."18

A bold, savvy psychologist at the University of North Carolina at Charlotte, David Sohn, points to even more damning problems with Eron and Huesmann's famous correlation. If watching television is influencing an eight-year-old boy to be aggressive, Sohn argues, you would expect such influence to be more intense at the time than ten years later. But the correlation Eron and Huesmann found between age 8 TV exposure and aggressive behavior at the same age was only .21 - 4 percent. Ten years later, despite years of intervening experiences, the correlation of age 8 exposure with age 18 aggression had grown to .31. How could that be? Influences weaken as time passes and other experiences intervene - they don't strengthen. Even more weirdly, Sohn points out, the correlation disappears in between: a partial sample of 64 boys in the study, reinterviewed in the mid-1960s, revealed no correlation between age 8 exposure and aggressiveness at age 13.19 Which would mean that an eight-year-old's TV exposure influences his aggressiveness immediately, has no measurable influence five years later, then mysteriously reemerges five years after that to influence an 18-year-old's behavior even more than it did when he was eight - an obvious absurdity.

Next Page