How ‘Unequivocal’ is the Evidence Regarding Television Violence and Children’s Aggression?

In their review of the influence of media violence on youth, “The Influence of Media Violence on Youth,” published in Volume 4 of Psychological Science in the Public Interest, Anderson et al. concluded that there is “unequivocal evidence” that television and film violence “increases the likelihood of aggressive and violent behavior in both immediate and long-term contexts” (Anderson et al., 2003, p. 81). However, their review contained a number of problems that at least raise questions about their conclusion.

One problem involved the authors’ failure to provide accurate descriptions of measures and/or important methodological aspects of some key studies. For example, their description of the Columbia County study (Lefkowitz, Eron, Walder, & Huesmann, 1977) mischaracterized the dependent variable as “a boy’s exposure to media violence” (Anderson et al., 2003, p. 87, italics added), when in fact it really involved the child’s maternally-reported preference for a particular program (Lefkowitz et al., 1977). Similarly, the authors mischaracterized the two contrast communities in the Notel study (Williams, 1986) as being “without TV” (Anderson et al., 2003, p. 89), when in fact television had previously been available in these two communities for a number of years. Clearly, the authors misstated the facts regarding these two studies.

Anderson et al. (2003) also selectively reported results supporting their position while failing to report results from the same study that contradict their position. For example, in discussing the Columbia County study, the authors failed to note that females’ preference for violent TV at age 8 actually negatively predicted their aggression at age 19 (Lefkowitz et al., 1977). Also, in discussing the Belson (1978) study, the authors failed to note that only one of Belson’s 20 major hypotheses unequivocally supported the causal relationship between TV viewing and aggressive behavior. Finally, in discussing the Cross-national studies (Huesmann & Eron, 1986b), the authors failed to state that of the 52 possible correlations in the five countries, 13 (25 percent) were not reported, 21 (40 percent) were significant, and 18 (35 percent) were not significant (Moeller, 2001). They also failed to state that, considering all six multiple regression analyses that were conducted, the only instances in which actual earlier TV violence viewing significantly predicted later aggression were for girls in the United States and for both sexes in the Israeli city (but not in the Israeli kibbutz) (Bachrach, 1986; Huesmann & Eron, 1986a, 1986b; Moeller, 2001).

Anderson et al. (2003) also failed to include or describe important studies that contradict their position. One example involves a 10-year research program conducted by Gadow and Sprafkin (1993) with children identified as “emotionally disturbed.” In their field studies, these authors found that “nonaggressive content can exacerbate the symptoms of emotional disturbance to a greater degree than aggression-laden material” (Gadow & Sprafkin, 1993, p. 60). They concluded that, based on their studies, “evidence supporting a causal link between television violence and aggressive behavior is weak or inconclusive at best” (Gadow & Sprafkin, 1993, p. 61).

In view of these problems, readers should be skeptical of just how “unequivocal” and “consistent in overall findings” (Anderson et al., 2003, p. 81) this body of research really is.

References

• Anderson, C.A., Berkowitz, L., Donnerstein, E., Huesmann, L.R., Johnson, J.D., Linz, D., et al. (2003). The influence of media violence on youth. Psychological Science in the Public Interest, 4, 81-110.
• Bachrach, R. (1986). The differential effect of observation of violence on kibbutz and city children in Israel. In L.R. Huesmann and L.D. Eron (Eds.), Television and the aggressive child: A cross-national comparison (pp. 210-238). Hillsdale, NJ: Lawrence Erlbaum Associates.
• Belson, W.A. (1978). Television violence and the adolescent boy. Westmead, England: Saxon House, Teakfield Limited.
• Gadow, K.D., & Sprafkin, J. (1993). Television “violence” and children with emotional and behavioral disorders. Journal of Emotional and Behavioral Disorders, 1, 54-63.
• Huesmann, L.R., & Eron, L.D. (1986a). The development of aggression in American children as a consequence of television violence viewing. In L.R. Huesmann and L.D. Eron (Eds.), Television and the aggressive child: A cross-national comparison (pp. 45-80). Hillsdale, NJ: Lawrence Erlbaum Associates.
• Huesmann, L.R., & Eron, L.D. (Eds.). (1986b). Television and the aggressive child: A cross-national comparison. Hillsdale, NJ: Lawrence Erlbaum Associates.
• Lefkowitz, M.M., Eron, L.D., Walder, L.O., & Huesmann, L.R. (1977). Growing up to be violent. New York: Pergamon.
• Moeller, T.G. (2001). Youth aggression and violence: A psychological approach. Mahwah, NJ: Lawrence Erlbaum.
• Williams, T.M. (Ed.). (1986). The impact of television: A natural experiment involving three towns. New York: Academic Press.

— Thomas G. Moeller, University of Mary Washington

A Response from the Authors:

The Evidence that Media Violence Stimulates Aggression in Young Viewers Remains ‘Unequivocal’

Moeller [this issue] raised concerns about our review of media violence effects on youth (Anderson et al., 2003). These concerns deserve attention, but do not seriously challenge our main conclusions.

Moeller writes that we “failed to include or describe important studies that contradict their position,” and gave one example. But the purpose of the examples we included was to illustrate the general findings of relevant meta-analyses, which included every study published, even the negative example Moeller offered. The fact is that all major recent meta-analyses of media violence effects (reviewed in our 2003 article) show positive effect sizes for media violence on aggression. To focus on the occasional null or negative example would have misrepresented the general findings.

Moeller also claims that we used selective reporting to make our case look stronger. Citing Lefkowitz et al. (1977), Moeller says “the authors failed to note that in the Columbia County Study females’ preference for violent TV at age 8 actually negatively predicted their aggression at age 19.” Well, Moeller is simply wrong. As Table 4.5 in the Lefkowitz book shows, there is no significant correlation for females from age 8 to age 19.

Moeller also says that “the authors failed to note that only one of Belson’s (1978) 20 major hypotheses unequivocally supported the causal relationship between TV viewing and aggressive behavior.” Belson actually investigated 22 “principle hypotheses” involving effects of TV violence. Most of those hypotheses involved attitudes, opinions, sleep disturbance, or state of mind. Of the four principle hypotheses about TV violence effects on “the extent of violent behavior by boys” (p. v), Belson reported significant support for each, based on one-tailed tests. One of the four, involving the most serious types of violence, was significant with a two-tailed test.

Finally, Moeller criticized our presentation of the results from the first three years of Huesmann and Eron’s (1986) longitudinal study. He argues that in the five countries studied there were 52 possible correlations and only 21 were signifi- cant. It is unclear how he counted correlations, but as Table 8.1 showed, if one considers the correlations of average TV violence viewing with average aggression for boys and girls in the five countries (10 correlations), all 10 are +.13 or higher, 8 are significant with a one-tailed test, and five are significant with a two-tailed test.

Another concern is that “the authors fail to provide accurate descriptions of measures or other methodological aspects of some key studies.” Well, in a review one cannot provide the same detail as the original sources, particularly if the details do not pose any threats to the validity of the conclusions. Moeller offers two examples. First, in the first wave of the Columbia County Study exposure to violent TV was assessed by asking mothers what their child’s favorite programs were and scoring those for violence; Moeller suggests that it is misleading to call this variable exposure to violence. This argument has been addressed many times over the past 45 years in many different places. The bottom line is that although such mothers’ reports are not perfect measures of exposure, they do correlate substantially with other measures of exposure. Moeller’s second example concerns Williams’ (1986) study. Moeller says that we “mischaracterized the two contrast communities in the Notel Study as being ‘without TV’ when in fact television had previously been available in these two communities for a number of years.” However, Williams (1986, p. 319) wrote, “the longitudinal sample consisted of the 45 children observed when they were in grade 1 or 2, before Notel had television, [emphasis added] and again 2 years later when they were in grade 3 or 4.”

Overall, Moeller’s criticisms fit the pattern of prior criticisms faced by media violence scholars many times (e.g., Comstock & Scharrer, 2004; Huesmann & Taylor, 2004). The critic ignores the overall picture while selectively focusing on flaws or contrary results from a few unrepresentative studies. In fact, the major conclusion challenged by Moeller — that there is overwhelming evidence that exposure to media violence increases the risk of aggressive and violent behavior in youth — was based on multiple meta-analyses by multiple research groups, analyses that included relevant negative and null result studies. Thus, our original conclusions stand.

References

• Anderson, C.A., Berkowitz, L., Donnerstein, E., Huesmann, L.R., Johnson, J.D., Linz, D., et al. (2003). The influence of media violence on youth. Psychological Science in the Public Interest, 4, 81-110.
• Comstock, G., & Scharrer, E. (2004) Meta-analyzing the controversy over television violence and aggression (pp. 205-226). In D. Gentile (Ed.) Media Violence and Children. Westport: CT: Greenwood Press.
• Huesmann, L. R. & Taylor, L. D. (2003). The case against the case against media violence (pp. 107-130). In D. Gentile (Ed.) Media Violence and Children. Westport: CT: Greenwood Press.

— Craig Anderson, Iowa State University,  and L. Rowell Huesmann, University of Michigan

No ‘Spotless Mind’

STEVEN PINKER HAS GOTTEN a lot of mileage from his arguments about the blank slate, not the least of which is the financial and celebrity windfall from a very popular book. It also got him the Keynote Address at the 2005 APS Annual Convention. But the arguments he presents, that are summarized in the August Observer, “Empirical Science for the Spotless Mind,” are either not based on good science or are questionable interpretations of reasonable science.

First, the so-called blank slate theory is not a scientific theory, but rather a philosophical position. Moreover, I would challenge Pinker or anyone else to identify one person alive who agrees with the description of the theory attributed to him, “that humans are shaped entirely by their experiences and not by any preexisting biological mechanisms” (italics added). Not even John Locke was a blank slate theorist based on this definition. Thus, this description is a total straw person, which should at least temper the debate if not quench it completely.

Let’s look at some of the “evidence” cited by Pinker. He refers to cognitive “mechanisms” such as a number sense and a spatial representation sense. These are not senses, but complex behavioral repertoires that arise only after complex interactions between behavior and environment. The so-called “ability to grasp the thoughts of others” (whatever that means) is based solely on our ability to learn verbal behavior. The language instinct for which Pinker is so well known is not a given, but is hotly debated, especially since there is overwhelming evidence that language is completely learned given, naturally, a brain capable of learning such behavior. The article states that “other human drives,” such as our desire for sugar and fat (and we can add salt and sex) “can only be understood in the context of evolution,” implying that the above mentioned cognitive mechanisms can also only be understood in the context of evolution. But there is no rational basis for comparing our desire for sugar to our abilities to talk and reason.

We are told that “the most devastating argument against the blank slate” comes from neuroscience and the work of Robert Plomin and his colleagues on separated identical twins. I’m sure this was just an editorial oversight, but Plomin’s research is not neuroscience. It is not even genetic science since it doesn’t deal with genes. In fact, as psychologist David Moore has so brilliantly shown in his book, The Dependent Gene, the concept of heritability is widely misunderstood even by behavior geneticists and, in fact, says nothing at all about the inheritability of traits. In fact, there are serious methodological and conceptual flaws with most behavior genetic research and the conclusions based on it as psychologist Jay Joseph has shown in his book, The Gene Illusion. Of course, there are many others (e.g., Stephen J. Gould, Leon Kamin, Richard Lewontin) who have pointed out the flaws in behavior genetic research. The bottom line is that this debate is by no means settled as Pinker would have us believe.

As I argued in, “The Almost Blank Slate: Making A Case For Human Nurture,” (Skeptic, Vol. 11, No. 2, 2004), “the scientific evidence for the influence of learning on human behavior is overwhelming.” This statement is supported by more than a century of basic experimental research on learning in humans and nonhumans.

Pinker is right about one thing, however: there are “innate mechanisms that do the learning.” How could it be otherwise? One of the defining differences between species, including between humans and all other species, is the capacity for adaptive behavior, otherwise known as learning.

So, are we born into the world as a blank slate? It depends on what one means by “blank slate.” If it means that neonates come into the world with not much more than basic reflexes, then the answer is yes. If it means that neonates are born with no evolutionarily-based, biological predispositions for learning the many complex behaviors that will ultimately define them as human, then, the answer is no.

Psychologists must realize that we can’t have science by persuasion or popularity but rather by simply doing good science.

— Henry D. Schlinger, California State University, Los Angeles

Steven Pinker Responds:

WHEN ONE PUTS ASIDE Schlinger’s ad hominem remarks and his objections to the word choices of the Observer’s writer, what remains in his letter is an exercise in nostalgia. The letter is reminiscent of 1950’s era Skinnerian behaviorism, with its phobia of the mentalistic explanations which are so indispensable both in everyday life and in our best psychological science. The ability to grasp the thoughts of others, for instance (which Schlinger professes not to understand) has been the subject of the most intensive research topic in cognitive development in the past two decades, the theory of mind. His dogmatic claim that it is “based solely on our ability to learn verbal behavior” is falsified by studies of mindreading in prelinguistic infants, which show that ability to be a prerequisite to, not a product, of the acquisition of language.

Schlinger also harks back to 1970s-style “radical science,” with its manifestos attempting to discredit politically incorrect research. Kamin’s claim that “there exist no data which should lead a prudent man to accept the hypothesis that IQ test scores are in any degree heritable” is just one example of this flat-earthism, amply documented in my book.

A third bit of nostalgia is setting up learning and innate abilities as a dichotomy. Schlinger announces that “the scientific evidence for the influence of learning on human behavior is overwhelming,” as if that had ever been doubted. What is at issue is not the existence of learning, but the number, specificity, and nature of our inborn leaning mechanisms and of the emotions and motives that drive them. The “almost-blank-slate” theory that Schlinger endorses, namely that “neonates come into the world with not much more than basic reflexes” (plus an unexplained “capacity for adaptive behavior”), shows that the targets of The Blank Slate are not made of straw.

— Steven Pinker, Harvard University