by L. Rowell Huesmann and Laramie D. Taylor
2003
NOTICE: THIS WORK MAY BE PROTECTED BY COPYRIGHT
YOU ARE REQUIRED TO READ THE COPYRIGHT NOTICE AT THIS LINK BEFORE YOU READ THE FOLLOWING WORK, THAT IS AVAILABLE SOLELY FOR PRIVATE STUDY, SCHOLARSHIP OR RESEARCH PURSUANT TO 17 U.S.C. SECTION 107 AND 108. IN THE EVENT THAT THE LIBRARY DETERMINES THAT UNLAWFUL COPYING OF THIS WORK HAS OCCURRED, THE LIBRARY HAS THE RIGHT TO BLOCK THE I.P. ADDRESS AT WHICH THE UNLAWFUL COPYING APPEARED TO HAVE OCCURRED. THANK YOU FOR RESPECTING THE RIGHTS OF COPYRIGHT OWNERS.
PART 1 OF 2
Even before the introduction of television into everyday life over 50 years ago, the question of whether exposure to violence in the media made the viewer more violent was being debated. But it was the introduction of television into the average American home in the early 1950s that really stimulated an explosion of scientific research on the topic. In this chapter we are not going to review that large body of research in detail; other chapters in this volume do that and show that the accumulated research indicts media violence as a cause of viewers' aggressive behavior. Rather, in this chapter we are going to deal with the writings of those who argue that media violence has no effect on aggression. Specifically, we will (1) summarize some of the most common flaws in their arguments and criticisms, (2) respond in detail to the criticisms of several of the most vociferous "naysayers," and (3) try to explain the psychology of why these naysayers find it so difficult to accept conclusions regarding media violence that are supported by large amounts of evidence while they find it easy to accept conclusions about other threats to public health supported by less compelling evidence. However, to accomplish these goals we need to begin by briefly summarizing what the empirical evidence shows, how the integration of different methodologies leads to a particularly strong indictment of media violence, and what psychological processes explain the effects of media violence on aggression. Let us start with an explication of the psychological processes through which exposure to media violence has an effect on viewers' violent and aggressive behavior. Understanding these processes is the key to understanding what the body of empirical data really shows.
PROCESSES ACCOUNTING FOR EFFECTS OF MEDIA VIOLENCE
To begin with one must realize that different processes explain short-term effects and long-term effects. Short-term effects are due to (1) priming processes, (2) excitation processes, and (3) the immediate imitation of specific behaviors (Bushman & Huesmann, 2000; Huesmann, 1988, 1998).
Briefly, priming is the process through which spreading activation in the brain's neural network from the locus representing an external observed stimulus excites another brain node representing aggressive cognitions or behaviors (Berkowitz, 1993). These excited nodes then are more likely to influence behavior. The external stimulus can be inherently aggressive, for example, the sight of a gun (Berkowitz & LePage, 1967), or something neutral like a radio that has simply been nearby when a violent act was observed (Josephson, 1987). A provocation that follows a priming stimulus is more likely to stimulate aggression as a result of the priming. While this effect is short-lived, the primed script, schema, or belief may have been acquired long ago and may have been acquired in a completely different context.
To the extent that observed violence (real world or media) arouses the observer, aggressive behavior may also become more likely in the short run for two other possible reasons -- excitation transfer (Zillmann, 1979, 1983a, 1983b) and general arousal (Berkowitz, 1993; Geen & O'Neal, 1969). First, a subsequent provocation may be perceived as more severe than it is because the emotional response stimulated by the observed violence is misattributed as being due to the provocation (Zillmann, 1979, 1983a). Such excitation transfer could account for a more intense aggressive response in the short run. Alternatively, the increased general arousal stimulated by the observed violence may simply reach such a peak that the ability of inhibiting mechanisms such as normative beliefs to restrain aggression is reduced (Berkowitz, 1993).
The third short-term process, imitation of specific aggressive behaviors, can be viewed as a special case of the more general long-term process of observational learning (Bandura, 1986; Huesmann, 1998). In recent years the evidence has accumulated that human and primate young have an innate tendency to imitate whomever they observe (Butterworth, 1999; Meltzoff & Moore, 2000; Rizzolati, Fadiga, Gallese, & Fogassi, 1996; Wyrwicka, 1996). These theories propose that very young children are likely to imitate almost any specific behaviors they see. Observation of specific aggressive behaviors around them increases the likelihood of children behaving exactly that way (Bandura, 1977; Bandura, Ross, & Ross, 1963). Granted, not all aggression is learned; in children two to four years old, proactive-instrumental behaviors that might be called aggressive (e.g., pushing another child without any provocation to get a desired object) appear spontaneously (Tremblay, 2000), as may hostile temper-tantrums. However, the observation of specific aggressive behaviors at that age leads to the acquisition of more coordinated aggressive scripts for social problem-solving and counteracts environmental forces aimed at conditioning the child out of aggression. As the child grows older, the social scripts acquired though observation of family, peers, community, and mass media become more complex, abstracted, and automatic in their invocation (Huesmann, 1988, 1998). Additionally, children's social-cognitive schema about the world around them begin to be elaborated. In particular, extensive observation of violence around them biases children's world schemas toward attributing hostility to others' actions (Dodge, 1985; Gerbner, Gross, Morgan, & Signorielli, 1994). Such attributions in turn increase the likelihood of children behaving aggressively (Dodge, 1980; Dodge, Pettit, Bates, & Valente, 1995). As children mature further, normative beliefs about what social behaviors are appropriate become crystallized, and begin to act as filters to limit inappropriate social behaviors (Huesmann & Guerra, 1997). Children's own behaviors influence the normative beliefs that develop, but so do the children's observations of the behaviors of those around them including those observed in the mass media (Guerra, Huesmann, Tolan, Van Acker, & Eron, 1995; Huesmann, Guerra, Zelli, & Miller, 1992; Huesmann, 2003). In summary, social-cognitive observational-learning theory postulates long-term effects of exposure to violence through the influence of exposure on the development of aggressive problem-solving scripts, hostile attributional biases, and normative beliefs approving of aggression.
Long-term effects are due to (1) observational learning of social scripts for behavior, of schemas about the world (e.g., is it hostile or benign), and of normative beliefs about the appropriateness of aggressive behavior; (2) emotional desensitization to violence; and (3) justification processes based on social comparisons (Bushman & Huesmann, 2000; Huesmann, 1988, 1998). There is some overlap with short-term effects; long-term effects are also quite likely increased by the habituation process called "desensitization." Most humans seem to have an innate negative emotional response to observing blood, gore, and violence. Increased heart rates, perspiration, and self-reports of discomfort often accompany such exposure (Cline, Croft, & Courrier, 1973; Moise-Titus, 1999). However, with repeated exposure to violence, this negative emotional response habituates, and the observer becomes "desensitized." One can then think about and plan proactive aggressive acts without experiencing negative affect. Consequently, proactive aggression becomes more likely.
One other long-term process is probably important. Social comparison theory suggests that humans evaluate themselves by comparing themselves to others. The aggressive child is generally (with some exceptions) not accepted because others do not like to be around aggressive peers (Anderson & Huesmann, in press). Huesmann (1988, 1995, 1998) has suggested that, to counter this threat to self-worth, aggressive children seek out aggressive media. Observing others behaving aggressively makes the aggressive children feel happier and more justified. Viewing media violence makes them feel happier because it convinces them that they are not alone in being aggressive. Of course, the ultimate consequence of such a turn toward more exposure to violent media, is more observational learning of aggressive scripts, schemas, and beliefs, and more desensitization to violence.
INTEGRATION OF EMPIRICAL RESEARCH RELATING MEDIA VIOLENCE TO AGGRESSION
Once these processes are understood, the wealth of empirical evidence implicating exposure to media violence as a cause of aggressive behavior does not seem so surprising. However, to understand how compelling the evidence really is, one needs to integrate the evidence from all the different empirical approaches that have been employed.
The methodologies used in studying the relation between media violence and aggression fall into three major classes: (1) experiments in which the researcher manipulates exposure to media violence, (2) correlational studies, or one-shot observational studies in which exposure to violence and concurrent aggressive behavior are measured with surveys or observations, and (3) longitudinal observational studies in which exposure and behavior are measured on the same sample repeatedly over long periods of time. It is critical to integrate the findings of all three bodies of research in reaching any conclusion.
Generally, experiments have demonstrated consistently that exposing children to violent behavior on film and TV increases the likelihood that they will behave aggressively immediately afterward (see reviews by Bushman & Huesmann, 2000; Geen and Thomas, 1986; Paik & Comstock, 1994; Strasburger & Wilson, chapter 4, this volume). The typical paradigm is that randomly selected children who are shown either a violent or nonviolent short film are then observed as they play with each other or with objects. The consistent finding is that children who see the violent film behave more aggressively immediately afterward. They behave more aggressively toward persons (Bjorkqvist, 1985; Josephson, 1987) and toward inanimate objects (Bandura, 1977). The effects occur for all children -- from preschool to adolescence, for boys and girls, for black and white, and for normally aggressive or normally nonaggressive. The average size of the immediate effect produced is about equivalent to a 0.4 correlation (Paik & Comstock, 1994). In these well-controlled laboratory studies there can be no doubt that it is the children's observation of the violence that is causing the changes in behavior. As described above, the psychological mechanisms operating are priming, excitation transfer, and simple imitation.
The question then becomes whether these causal effects observed in the laboratory generalize to the real world. Do they have real significance in the world? Do they extend over time? Does real media violence cause real aggression in the real world, not just in the short run but in the long run as well?
Empirical correlational studies of children and youth behaving and watching media in their natural environments have demonstrated that the answer to both of these questions is "yes." The great majority of competently done one-shot survey studies have shown that children who watch more media violence day in and day out behave more aggressively day in and day out (Paik & Comstock, 1994). The correlations obtained usually are between 0.15 and 0.30. Such correlations are not large by the standards of variance explained, but they are moderate by the standards of children's personality measurement, and they can have real social significance (Comstock & Scharrer, chapter 11, this volume; Rosenthal, 1986). In fact, as Rosenthal (1986) has pointed out, a correlation of 0.3 with aggression translates into a change in the odds of aggression from 50150 to 65135 -- not a trivial change when one is dealing with life-threatening behavior. Moreover, the relation is highly replicable even across researchers who disagree about the reasons (e.g., Huesmann, Lagerspetz, & Eron, 1984; Milavsky, Kessler, Stipp, & Rubens, 1982) and across countries (Huesmann & Eron, 1986).
Complementing these one-time survey studies are the longitudinal real-world studies that have shown correlations over time from childhood viewing of media violence to later adult aggressive behavior (Eron, Huesmann, Lefkowitz, & Walder, 1972; Milavsky, Kessler, Stipp, & Rubens, 1982; Huesmann, Moise, Podolski, & Eron, 2003; for reviews see Huesmann & Miller, 1994; Huesmann, Moise, & Podolski, 1997). Analysis of longitudinal data has also shown that early habitual exposure to media violence predicts increased aggressiveness beyond what would be predicted from early aggressiveness.
In conjunction with the theories described above, the results from these three kinds of research-experiments showing unambiguous causation, one-shot surveys showing real aggression correlates with concurrent habitual exposure to violent media, and longitudinal studies showing that childhood exposure predicts increased adult aggression independent of childhood aggression -- should lead objective scientists to conclude that exposure to media violence increases a child's risk for behaving aggressively in both the short run and long run. So why is there still a body of public intellectuals who refuse to accept this conclusion?
THE PSYCHOLOGY OF DENIAL
To begin with, one must note that there is a clear consensus of opinion among scholars who actually do research on the topic that exposure to media violence causes aggression. Most major health professional groups have issued statements citing exposure to media violence as one cause of youth violence. Two Surgeon Generals of the United States (in 1972 and 2001) have warned the public that media violence is a risk factor for aggression. For example, in March 1972, then Surgeon General Jesse Steinfeld told Congress, it is clear to me that the causal relationship between [exposure to] televised violence and antisocial behavior is sufficient to warrant appropriate and immediate remedial action.... there comes a time when the data are sufficient to justify action. That time has come. (Steinfeld, 1972)
Surveys have consistently shown that over 80 percent of those doing research on the topic have concluded from the evidence that media violence is causing aggression (Murray, 1984).
So who are the vocal minority denying that there can be any effects? The best-known social scientists who deny there are any effects (e.g., Cumberbatch, Fowles, Freedman, Jenks) generally have never done any empirical research on the topic. However, they are glib and compelling writers, and their opinions cannot simply be dismissed. Furthermore, there is a large body of other intellectuals who deny that there are any effects. They range from the president of the Motion Pictures Producers Association (Jack Valenti) to the president of the Entertainment Software Association (Doug Lowenstein); from movie directors (e.g., Rob Reiner) to comic book producers (e.g., Gerard Jones); from science writers (e.g., Richard Rhodes) to booksellers (e.g., Chris Finan, president, American Booksellers Foundation).
Later in the paper we will deal with the individual criticisms of the most visible critics, but we first want to offer a psychological perspective on why well-intentioned, generally intelligent, and well-informed people can hold and promote attitudes on this topic that are so discrepant from what the majority of scientists, health care providers, and parents believe. We see three explanations for the discrepancy which are all grounded in psychological theory: (1) the need for cognitive consistency, (2) reactance against control, and (3) susceptibility to the "third person effect" of human behavior. However, all of these psychological processes depend on two underlying facts -- one grounded in economics and one grounded in political principle, American history, and constitutional law. The economic fact is that violence in entertainment attracts audiences and makes large amounts of money for its purveyors (see Hamilton, 1998). The political principle is the sacredness of free expression in American society and law.
We propose that individuals involved in the production or marketing of violence will find it difficult to believe that viewing violence could be damaging to audiences because that belief would be cognitively inconsistent with their behaviors. Cognitive consistency is a remarkably powerful force that affects behaviors and beliefs (Abelson et al., 1968); so it would not be surprising that the behaviors shaped by subtle economic forces would shape beliefs. Only if the economic forces are blatant, or other beliefs relieve the inconsistency, can effects be admitted. For example, the director Oliver Stone says that "of course his [violent] movies [e.g., Natural Born Killers] are dangerous." His movies are intended to affect people, and that is one of the costs of free expression (BBC Panorama, 1997). Similarly, the cognitive consistency process can lead to a denial of effects for those who believe strongly in free expression in the mass media. Many individuals with strong liberal beliefs about free expression in the mass media also have strong beliefs about society having a duty to protect children. If they accepted the fact that media violence harms children, they might have to rethink their beliefs about balancing freedom of expression and consider protecting children. It is easier for them to avoid this cognitive dissonance by denying that media violence has effects than it would be for them to resolve the dissonance. Furthermore, if the purveyors of violence accepted that violence has serious effects on children, they would have to categorize themselves with other purveyors of products that threaten health, for example, tobacco, which would produce even more dissonance.
The second psychological process we see as relevant applies only to the producers of violent media. Most humans at a young age develop an aversion to being controlled and respond to such attempts with reactance, or attempts to regain or increase their own control (Brehm & Brehm, 1981). We suggest that artists, writers, and producers are particularly susceptible to displaying such reactance when attempts are made to control their creative products, in which their egos are heavily involved. Artists often view as threats of control statements that their programs or films harm viewers. Suppose a researcher tells an artist that a program of theirs, which is a financial and critical success, is bad because it stimulates violence in the children watching it. The artist, rightly or wrongly, consciously or unconsciously, may interpret this statement as a threat of control. Therefore, a plausible response of the artist according to reactance theory would be to attack the researcher's thesis that the program has bad effects on the viewer.
The third psychological process we offer is intended more to explain a frequent opinion one hears from violence viewers who believe viewing violence cannot be bad. The opinion is that "media violence may affect some 'susceptible' people, but it will not affect 'me' or 'my children' because we are impervious to such influences." It is common in opinion surveys to find people reporting that a media message or personal communication might affect some people, but not the respondent. This phenomenon has been labeled the "third person effect" (Davison, 1983). Of course, as the research reported elsewhere shows, media violence can affect any child. The third person effect may be related to reactance theory. If viewers admit that they are being influenced by messages in the media, then they would have to admit they are being controlled to some extent by the media. Reactance would demand some action then. But if one denies that one is being controlled by the media, one does not need to act according to reactance theory.
We offer these three psychological processes only as suggestions that may help explain how many informed and intelligent people can read the reports of the studies done and still sincerely deny that media violence has serious effects on viewers. We certainly cannot offer any empirical evidence that these processes do operate among the noted critics. However, they are well-established psychological processes that are likely to operate among any of us. Of course, there are other plausible contributors to the dissents of specific critics. Some critics have invested a great deal in denying effects for a long time and have been paid by the violence purveyors to write dissenting books (e.g., Freedman). Obviously, it would be difficult for them to change. Others have alternative theories of violence that they are invested in and see media violence as a competitor (e.g., Rhodes). Probably, it is a combination of multiple processes that leads to the most vociferous dissents, which are obviously sincerely held.
Four Common Flaws in Critiques of Media Violence Research
Let us now turn to a discussion of a few of the most frequently repeated errors of reasoning that have been made by critics who challenge the conclusion that exposing children to media violence puts them more at risk to behave aggressively in the short run and in the long run.
1. Assuming that the question is whether TV violence is the "only" cause of aggression, and arguing that TV violence can't matter because people who see the same TV shows differ in aggressiveness. For example, critics say that "Detroit, Michigan and Windsor, Ontario see essentially the same TV shows. But the murder rate is much higher in Detroit. Consequently TV violence cannot be increasing serious aggression." This argument has been repeated over many years by intelligent people ranging from social scientists criticizing the research to politicians to network vice-presidents. Of course, this argument would only make sense if nothing except TV violence influenced murder -- not guns, not poverty, not social support, not peer attitudes, not child rearing, not biological predispositions. The murder rate in Detroit has been higher for a lot of these reasons. It's puzzling how anyone could seriously offer this as an argument that TV violence has no effect. No reputable researcher of media violence has ever suggested that media violence is the only cause or even the most important cause of aggression. Serious aggressive behavior only occurs when there is a convergence of multiple predisposing and precipitating factors (Huesmann, 1998).
2. Ignoring laboratory experiments. It has been common among the naysayers (e.g., Freedman, 1984) to ignore completely the well-done laboratory experiments that have shown that exposure to violence stimulates aggressive behavior in the short run. Ignoring any of the different types of research (e.g., experiments, cross-sectional surveys, longitudinal studies) on media violence would be risky, but ignoring laboratory experiments is particularly inappropriate because it is the kind of study that most clearly tests causation. One typical rationale for ignoring laboratory experiments is the supposed artificiality of the aggression measures used in the laboratory, such as giving shocks to another person. The critics most often mention only the measures with the least face validity and never offer any empirical evidence that the lab measures are not valid indices of real-world aggression. The truth is that there is substantial empirical evidence that the measures of aggression used in laboratory studies are quite valid indices of how aggressively the person would behave outside the laboratory (Anderson & Bushman, 1997; Berkowitz & Donnerstein, 1982).
The other common complaint used against experiments without any justification offered is that "laboratory work suffers from strong experimenter demands" that bias the results in the direction of showing effects of media violence (Freedman, 1984). In fact, this criticism runs counter to the empirical evidence that suggests that participants in aggression experiments are likely to inhibit aggressive impulses because they fear being negatively evaluated by the experimenter (Turner & Simons, 1974). Finally, excluding laboratory experiments in favor of focusing only on field research reflects critics' misplaced confidence in such studies. While field studies may often (but not always -- see Berkowitz & Donnerstein, 1982) have greater external validity, it is much harder to confirm the internal validity of the conclusions of field studies. And causation can never be tested as conclusively with field research as with a well-controlled laboratory experiment. The critics who ignore experiments conveniently overlook this fact.
3. Selective reporting of negative results and changing criteria for accepting results. Another common flaw in the critics' analyses of the research is their tendency to change the criteria for reporting a study or evaluating it depending on how the results came out. One study with positive results is discounted because of supposed demand characteristics on participants to behave aggressively, while a study with negative results is praised despite the fact that there were clear demands placed on the participants not to aggress. Another study with positive results is discounted because the stimulus films may have differed in attractiveness to the viewers, while a similar study with negative results is praised even though differential attractiveness of the films could have accounted for the negative effects. While meta-analyses that systematically combine all studies on a topic uniformly show positive and significant effect sizes for media violence on aggression (see Comstock & Scharrer, chapter 11, this volume), the reviews of the naysayers often convey the impression that most studies do not have positive effects simply because, for one flimsy reason or another, they exclude many studies with positive results and include every study with no results.
4. Analyzing studies in a theoretical vacuum. Perhaps the most egregious common error made by the naysayers is to evaluate the research on media violence as if it is completely disconnected from our existing knowledge about learning, social cognition, and aggression. We have outlined earlier in this chapter the established theory that explains how media violence influences aggression. The psychological processes that account for the effect were not invented to account for the effect; they had been established independently. Given what we know about priming of social cognitions (Bargh, 1982), it would be incredibly surprising if media violence did not prime aggressive cognitions. Given what we know about arousal processes and excitation transfer (Zillmann, 1983a, 1983b), it would be startling if media violence did not produce such effects. Given what we know about the innate propensity of primates to imitate (Meltzoff & Moore, 1977, 2000) and the developmental course of observational learning in the real world (Bandura, 1977, 1986), it would be a shock if children did not acquire social scripts, world schemas, and normative beliefs from the mass media. Given the research on how hostile attributional bias (Dodge, 1980, 1985) and normative beliefs promoting aggression (Huesmann & Guerra, 1997) influence children's behavior, it would be surprising if such cognitions acquired from the mass media did not influence the children's behavior. And finally, given the established continuity of aggression from childhood to adulthood (Huesmann et al., 1984), it would be very surprising if the effects of media violence on children were not detectable when they were adults years later. Yet, the naysayers seem totally unaware of such psychological facts.
Given this background and overview of the most common errors in the critics' reasoning, let us now turn to a discussion of the specific views of several of the most prominent critics who challenge the conclusion that exposure to media violence increases the aggressive tendencies of the viewer.
The Freedman Chronicles
Over the past 20 years no critic has played a more prominent role in denying that media violence has any effect on behavior than psychologist Jonathan Freedman at the University of Toronto. Since 1984 (Freedman, 1984), he has written and published numerous articles disputing the fact that media violence has any significant effect on aggression and has culminated that work with a recent book (Freedman, 2002). Like most critics, he has never done any empirical research on the effects of media violence, and his critiques contain numerous examples of the four general flaws of thinking that we described above. Additionally, Freedman's objectivity, at least in his recent summative book, must be questioned as he was paid to write the book by the Motion Picture Association of America. Freedman accuses some of the scholars who have done empirical research on media violence of "basing their whole careers on showing that television violence is harmful" and therefore of being biased. In fact, a large amount of the important empirical work on media violence and aggression has been done by psychologists whose careers were devoted to understanding social behavior or aggressive behavior and who branched out to study media violence (e.g., Anderson, Bandura, Berkowitz, Bushman, Eron, Geen, Huesmann, Huston, Lagerspetz, Malamuili, and Parke, to name a few). Freedman, on the other hand, has based an entire 20-year career on trying to show that media violence has no effect on aggression. Nevertheless, Freedman deserves some additional response because he is a social scientist and has attracted considerable attention. Because previous responses to his critiques (Huesmann, Eron, Berkowitz, & Chaffee, 1992; Huesmann & Moise, 1996) have focused on his early essays, we focus on his recent book here (2002).
The book is engagingly written, accessible to a nonscholarly audience, and presents glib arguments for not believing that media violence affects aggression based on some truths, many selective distortions and exaggerations, and a few outright untruths or misunderstandings. As always, Freedman points out some real flaws in research on the topic. Freedman also avoids some of the errors of his earlier reviews -- he does now devote attention to laboratory experiments, for example. However, as before, his approach to the issue is a-theoretical, and he employs selective reporting and shifting criteria for evaluation of studies. Because he is an engaging and talented writer, probably sincere in his beliefs that there are no effects, and probably unaware of how selective his reporting is, he builds a bond with the reader. This allows him to shift from the third person to the first person and engage in "conversations" with the reader. When faced with a result inconsistent with his thesis that he cannot explain away, he relies on this bond to allow him to simply say things like, "It is a complicated study with very complicated results. I am confident that, overall, these results do not show that exposure to media violence increases aggression" (p. 29). Or in response to a published assertion he does not like, he may simply exclaim: "This is incredible," or "Scandalous," or "It is junk science." These are just some of the phrases Freedman uses to convey his feelings. His feelings are clear, but the justification for them is not made clear. What is worse, while he carefully qualifies every statement he makes concerning any kind of positive effect that might possibly be found between media violence viewing and aggression, he forgets about qualifying negative statements; as a result, he has a number of clearly false statements. To offer a few of his statements that are patently false unless qualified: "Virtually, no research shows that media violence desensitizes people to violence." "There is virtually no research on changes in attitudes due to viewing violence." "A majority of studies show no ill effects." "None of these reviews looked at hundreds of studies." (Paik and Comstock [1994] looked at 217, to give one counterexample.)
While his style contributes significantly to the dismissive message about media violence that he conveys, it is the more substantive review techniques he employs that really distort the truth. To begin with, Freedman completely dismisses meta-analysis as a viable review technique; for some time meta-analysis has been the accepted scientific mechanism for combining the results of many social science studies on a particular topic (see Comstock & Scharrer, chapter 11, this volume). In fact, in the early 1990s Paik and Comstock published an outstanding meta-analysis on this topic covering research through 1990. Essentially, Freedman ignores this analysis with the comment that "they do not provide any details for how they classified results; so it is not possible to comment" (p. 31). This statement in itself implies a lack of understanding of the nature of meta-analysis. In meta-analyses quantitative effect sizes are combined according to mathematical principles, and the outcome does not depend on an individual classifying any particular study as supportive or not supportive. The aggregate effect size speaks for itself. Freedman's rationale for dismissing existing meta-analyses are not only weak, but also self-serving, as Paik and Comstock's conclusions are quite at odds with Freedman's thesis. For example, Paik and Comstock report an overall effect size (as a correlation) of 0.37 for experiments and 0.19 for surveys (see Comstock & Scharrer, chapter 11, this volume, for details and an explanation of how to interpret how large these effect sizes are). Furthermore, while dismissing this fine meta-analysis of 217 studies without comment, Freedman spends some pages taking apart a less sophisticated meta-analysis of 23 studies by criticizing the author's conclusions about each study on the basis of nothing but his word that he agrees or disagrees. By dismissing meta-analyses and refusing to do one himself, Freedman greatly diminishes the value of his review. However, he could still have had an impact if he had presented consistent scholarly arguments in analyzing each study independently, but he did not.
Freedman's approach to analyzing a study is to mix together some quite appropriate methodological issues of concern with other unique value judgments of his own into a glib mix that might fool the lay reader but will not carry much weight with the informed reader. It is not that he is misleading the reader. He is quite straightforward. But the assertions he makes without supporting facts just do not hold up. Also, despite his statement that he attempted to overcome his own biases and review studies fairly, his biases remain readily apparent. He shifts his criteria for criticism depending on whether the study shows positive or negative effects. He emphasizes minor methodological errors and statistical errors in studies finding a positive effect but minimizes them in studies showing no effect. We can't go through every study he reviews, but let's consider two.
In his discussion of a study by Leyens, Camino, Parke, and Berkowitz (1975), Freedman dismisses the positive results, which fit perfectly with what theory would have predicted, by saying that "there are a number of serious problems ... that make its result almost impossible to interpret" (p. 101). In this study Leyens reported that boys who watched a week of violent films in their cottages were observed behaving more physically aggressively afterward compared both to their behavior during a previous week and to boys who saw nonviolent films. What are the serious problems, according to Freedman? First, Freedman says, "it was a mistake" that the observers were instructed to code observed behaviors (hitting, etc.) without reference to what they imagined might be the intent of the act (p. 101). Freedman does not mention that this is the typical instruction used everywhere for behavior observations to reduce observer bias. No -- instead he says that it is wrong because the boys may be hitting each other without real intent to hurt the other person, and that should not count as aggression. But, of course, no one knows the boys' intentions except the boys, and, regardless of intent, if boys hit each other more after watching the films, one has demonstrated exactly the effect that observational learning theory predicts. Second, Freedman says that "the boys were more aroused by the violent movies than by the non-violent" (p. 102). This seems reasonable, though the authors collected data showing that this is not true (Leyens, Camino, Parke, & Berkowitz, 1975). If it is true, it simply suggests that the effect may be partially due to "excitation transfer" rather than observational learning. That is still an important result. Freedman would not see it as such, however, because he has taken a completely a-theoretical approach to the topic. He does not seem to realize that many researchers arguing publicly for concern about media violence believe that "excitation transfer" is an important component to the short-term effects of violence. Finally, Freedman says simply: "The statistical tests ... are all inappropriate" (p. 101). Why? The children in each condition were divided between two cottages and cottages were the units assigned to condition. The analysis should have treated the subjects as nested within cottages but did not. Indeed this is a statistical error, but how serious? The key interaction is statistically significant at the 0.001 level in this study. It seems likely that, given that level of significance, the results would have held with the correct analysis. However, the point about all these criticisms is that Freedman ignores or minimizes similar flaws in a study when it suits his purpose.