Sexual Violence in Three Pornographic Media: Toward a Sociological Explanation.

With almost three decades of social science research on sexual violence in pornography, many of its characteristics are now well understood. We know that pornography contains images of both violence and sexual violence (Palys, 1986; Scott & Cuvelier, 1993; Smith, 1976). We have some idea of the negative effect of sexually violent pornography, especially on men's attitudes towards women (Allen, D'alessio, & Brezgel, 1995; Brannigan & Goldenberg, 1987; Fisher & Grenier, 1994; Linz, 1989; Malamuth & Check, 1985; Mayerson & Taylor, 1987).

Unfortunately, however, the vast majority of previous research on sexual violence in pornography has approached pornography as an undifferentiated mass with regards to sexual violence. This study seeks to determine if this assumption is justified. Does pornography differ across media? In what ways, and to what degree? How can these differences be explained? These are the primary questions of concern to this paper. We begin our inquiry by reviewing past efforts to understand the content of pornography, and then present some tentative hypotheses concerning the levels of sexual violence across media. Through an empirical investigation of three contemporaneous pornographic media--magazines, videos, and Internet newsgroups (the Usenet)--we discuss differences in violent content, and speculate about some of the possible explanations for these differences.

THE CONTENT OF PORNOGRAPHY

Most social and behavioral science research on pornography has studied either the effects of pornographic images on viewers, or examined the content of the pornography itself. (As our empirical analysis focuses only on the changing content of pornography, we have confined our review of the literature to that branch of the research.) Although Smith (1976) found remarkable homogeneity in his sample of "adult-only" paperbacks, Malamuth and Spinner (1980) found increasing violence in their study of Playboy and Penthouse over a 5-year period, from approximately 1.5% of all pictorials for both magazines in 1973 to 6% in Penthouse and 4% in Playboy in 1977.

The debate surrounding these studies generated important additional research. Scott and Cuvelier (1987a, 1987b) disaggregated the two magazines and examined them over a longer time span, to eliminate the possibility that the original studies had conflated them and examined an anomalous period. They found a curvilinear relationship between year and sexual violence, from virtually no violence in its earliest years, with increases until the late 1970s (the end-point for the study by Malamuth & Spinner, 1980), when the violent content began to subside. It appears that Playboy responded to the proliferation of more violent magazines and the introduction of videotapes by returning to its traditional, more economically affluent and upscale consumerist readership.

Criticized for methodological synecdoche--letting Playboy stand for all pornographic magazines, especially at a time of proliferation--Scott and Cuvelier (1993) examined sexual violence in Hustler from its inception (July 1974) through July 1987. While they found a higher level of violent content (1.6% of all pictorials) in Hustler than in Playboy (.038%) overall, they found no changes in the magazine over time, and argued that their work "refutes the alleged increase in sexually violent portrayals" (Scott & Cuvelier, 1993, p. 367).

Other researchers used more synchronic approaches. Winick (1985), for example, examined 430 pornographic magazines found in an adult bookstore in New York City's Times Square. After classifying these magazines into 22 categories, Winick found that 4.9% of the magazines examined were dedicated to bondage and discipline, while violent content accounted for only 1.2% of his sample. However, lack of reliability of the sample (Winick only went into one store) and coding validity (all coding was done on the premises) make his findings less useful to other researchers.

The rapid development of video technology revolutionized the pornography industry. The rental of pornographic movies rose from 75 million in 1985 to 490 million in 1992 (Schlosser, 1997). Research followed suit. Comparing sexual aggression in "Triple-X" videos and in more mainstream "adult" videos, Palys (1986) found virtually no differences--6.6% of the scenes from adult videos and 6.4% of the scenes from Triple-X videos contained sexual aggression--but enormous differences between pornographic videos and pornographic magazines (where 1% was more typical). Virtually every study of pornographic videos has found similar levels of sexual violence, substantially higher than in magazines (see, for example, Brosius, Weaver, & Staab, 1993; Duncan, 1991; Yang & Linz, 1990).

Little research has been done on pornography and the Usenet (Internet newsgroups), although it has become a very controversial topic. One study (Rimm, 1995) was both methodologically and ethically suspect--supplemental data were collected from an adult bulletin board (which requires a credit card and age verification to use) and surreptitiously from students using the Usenet (DeLoughry, 1995; Elmer-Dewitt, 1995). Other, less typical forms of pornography--from pornographic cartoons (Matacin & Burger, 1987) to dial-a-porn recordings (Glascock & Larose, 1993)-- have also been examined. These studies tend to underscore the fact that all pornographic media contain violence, and that each displays violence differently.

Only one study has attempted to compare pornographic media. Dietz and Sears (1988) examined books, magazines, and films in adult bookstores in four cities. They found that 12% of the magazines, 20% of the books, and 7.7% of the films portrayed some violent theme. While this study proposed the reverse in the amount of violence in magazines and videos, Dietz and Sears only examined the covers and display boxes of the pornography being investigated, thereby making it impossible to understand the actual content of the material (see Linz & Donnerstein, 1988).

Studies of the content of pornography have thus been suggestive at best, but beset by both substantive and methodological problems that make reliability questionable and comparability impossible. Only the research on the effects of pornography on men's attitudes and behaviors have yielded any reliable, albeit often contradictory, results. We still know little about the differences in the actual content of pornography among various pornographic media. This study is an attempt to remedy this serious lacuna in our understanding of the content of pornography. We provide a careful methodological procedure to compare different pornographic media to ascertain the differing levels of violence.

The introduction of these different pornographic media at different historical moments (with magazines being introduced first, followed by videos, and then the Usenet) offers several intriguing possibilities that might affect the differing amounts of violence we might find within them. Violent content might remain constant across all three media. Levels of violence might increase in a linear way, moving from low (magazines) to intermediate (videos) to higher (Usenet). Or it might decrease in a linear way. Finally, the relationship might be curvilinear, with either a peak or trough in the middle. Given that previous research found a significant increase in level of violence between magazines and videos, we hypothesized that a constant level, decrease, or curvilinear (trough) model was unlikely. The plausible relationships in which we were interested were that the level of violence would continue to increase from videos to the Usenet, or that it would decline from videos to the Usenet. At the same time, we wanted to retest the relationship between magazines and videos.

METHODOLOGY

Our sample of magazines and videos was drawn from a suburban New York township. The population of this suburb is predominantly white and middle class. We collected 50 cases from each of the three pornographic media. The unique nature of each media, however, necessitated slightly different sampling schemes for each medium. Videos and magazines were chosen through a multilevel cluster sampling design. For both media, pornography was operationalized as being any sexually explicit material to which access was limited, either by signs or physical structure, to adults. At each of five stores, 10 videos were randomly selected to comprise the video sample. (See Appendix A for list of films.) Because the majority of stores selling pornographic magazines did not sell 10 different magazines, 5 magazines were selected from five different stores, and two stories were randomly taken from within each magazine to comprise our magazine sample. (See Appendix B for a list of magazines.)

Our Usenet sample was confined to stories taken from the newsgroup alt.sex.stories. It is important to remember that the Usenet portion of the Internet is only one portion of what is considered the Internet, and only one way to access pornography via the Internet. We decided to concentrate on this portion for several reasons. First, it provides a convenient data pool. While the World Wide Web has certainly caught the public's eye more than newsgroups, there is virtually no way to construct a list of all pornographic web sites from which to sample. Further, while some pornographic web sites contain stories, the majority primarily contain pictures, and thus do not provide the narrative elements important to this study. We used alt.sex.stories precisely for its narrative content. Finally, while many web sites with pornographic material have begun to charge for access, the Usenet remains free to all with access to the Internet.

The sampling population for this study was made up of all sexual stories of at least 250 words posted to the newsgroup during one month in 1997 (approximately 28% of the postings for this month fit this description). From this population, a random sample of 50 cases was drawn.(1)

Coding Scheme

The unit of analysis within each story (textual or visual) was the scene. We used Palys' (1986) definition as "a thematically uninterrupted sequence of activity in a given physical context" (p. 25). Each scene was examined for sexual and/or violent content. Coders were provided a list of specific violent acts (see Table 1) and recorded whether each scene contained each particular act and, if the scene did contain that act, the sex of both the person performing the violent act and the victim. Coders only measured the presence of the behavior. They were not instructed to measure the number of time it occurred within a scene or the intensity of that occurrence.

Table 1. Violence Categories

Violence category

Verbal aggression
Pushing, shoving
Being rough in an otherwise "normal" activity

Pulling hair/biting
Pinching
Open hand punch (e.g. slap, spank)
Threaten with weapon

Choking
Closed fist punch or kick
Confine, bondage

Use weapon on victim
Torture, mutilation
Attempted, actual murder

Table 1 is divided into 4 violence strata (indicated by the spacing between categories). These divisions were made during data analysis by placing the individual violent acts in a roughly hierarchical order and making logical divisions among them.

In addition to examining specific violent acts, coders were instructed to identify the sex of the participants; whether each was in a dominant, submissive, or ambiguous power position; and whether each appeared to be a consensual, coerced, or nonconsensual participant.

Two points are important to note here. First, no attempt was made to measure the level of intensity contained in any particular violent act. Thus specific instances of the same violent act may vary in intensity. While this is true for all forms of violence examined, it is especially noteworthy for those forms of violence that are more "culturally ritualized," such as verbal aggression, slapping/spanking, and confine/bondage, all of which oscillate between what might be considered light and heavy violence.

Five coders were employed to code the 150 cases. Four were undergraduate volunteers, and the fifth was one of the authors. Two of the coders were male and three were female. All were white and had some social science background, and all completed an extensive training seminar prior to participating.

Reliability

Ten percent of our sample was coded by two different coders in order ascertain intercoder reliability. Cohen's Kappa Coefficient was computed for each of these cases and their recode. This measure of association provides the percent agreement between coders rescaled to correct for chance. The average Kappa Coefficient for the 15 recoded cases was .8302.

RESULTS

The amount of sexual violence found in our sample is shown in Table 2. The first row aggregates the individual types of violence to give a summary of the percentage of scenes containing any violence, while each subsequent row summarizes the percentage of scenes containing particular forms of sexual violence.(2)

Table 2. Percentage of Scenes Containing Sexual Violence by Media

Type of violence                      Magazine(a)   Video(b)

Any violence                             24.8         26.9
Verbal aggression                         7.6         12.6
Rough in otherwise normal activity        2.9          4.1
Pushing, shoving                          7.6          3.8
Pinch                                     1.0          3.8
Pull hair/biting                          4.8          8.0
Threaten with weapon                      0.0          0.8
Open hand punch (i.e., slap, spank)       4.8          9.3
Choke 0.0 0.8 0.7 0.86 Punch/kick         1.9          0.8
Confine, bondage                          5.7          2.7
Weapon                                    1.0          1.6
Torture, mutilation                       1.0          0.3
Attempted/actual murder                   0.0          0.0

Type of violence                      Usenet(c)   [chi square]

Any violence                            42.1       12.83(**)
Verbal aggression                       15.0        3.11
Rough in otherwise normal activity      10.0        8.48(*)
Pushing, shoving                         3.6        3.06
Pinch                                    2.1        2.79
Pull hair/biting                         3.6        3.83
Threaten with weapon                     9.3       31.72(**)
Open hand punch (i.e., slap, spank)      8.6        2.23
Choke 0.0 0.8 0.7 0.86 Punch/kick        2.1        1.70
Confine, bondage                        25.7       69.59(**)
Weapon                                  15.7       48.17(**)
Torture, mutilation                      8.6       32.01(**)
Attempted/actual murder                  1.4        6.72(*)

(a) n = 105.

(b) n = 364.

(c) n = 140.

(*) p < .05.

(**) p < .001.

According to our findings, the Usenet is most likely to contain violence by a wide, and statistically significant ([chi square] = 10.34, df = 1, p [is less than] .0025), margin. Videos contain the second highest proportion of violent scenes, followed closely by magazines. However, while the videos in this sample contained a higher percentage of violence than magazines, these differences were not statistically significant at the .05 level. Although the difference between media in the percentage of cases containing any violence is statistically significant, more than half of the differences in individual violence items were not statistically significant. The majority of statistically significant differences occur in the most violent stratum, with one statistically significant form of violence occurring in each of the other three strata.

Table 3 illustrates the percentage of scenes containing violence in particular strata by medium. The row variables (low, medium, high, and extreme) are derived from dichotomous variables representing the presence (or absence) of violence from that stratum. Thus, any scene containing any verbal aggression, being rough in an otherwise normal activity, or pushing/shoving would be coded as containing violence in the low stratum.

Table 3. Sexual Violence in Scenes by Strata

          Magazine(a)   Video(b)   Usenet(c)   [chi square]

Low          14.3%       16.5%       23.6%        4.51
Medium        9.5        15.4        17.1         3.03
High          6.7         4.4        26.4        56.72(*)
Extreme       1.9         1.9        17.1        48.76(*)

(a) n = 105.

(b) n = 364.

(c) n = 140.

(*) p < .001.

Note that for both magazines and videos, violence decreases in a monotonic fashion. As violence becomes more intense, or extreme, fewer scenes contain it. For both media, approximately 15% of scenes contained low intensity violence. Both have fewer scenes with medium intensity violence (roughly 13%), fewer still with high intensity violence (roughly 5%), and very few scenes with extreme violence (2%). Differences between magazines and videos were not significant at [Alpha] = .05. The Usenet, however, does not follow such a linear pattern. Most of the violence in Usenet newsgroup pornography falls into the low and high categories, while the medium and extreme categories contain slightly less violence. For all strata, a higher percentage of scenes from the Usenet contain violence (the differences between the Usenet and the other two media combined are statistically significant at the [Alpha] = .05 level for all strata except medium).

When we examined the level of violence from more to less recent technology, we also found that the Usenet contains more coercive and less consensual sex in the scenes depicted. Over one fourth (26.4%) of scenes from the Usenet contain coercive or nonconsensual sex, followed by videos and magazines (4.9% and 4.8% respectively). While we can again note the Usenet-video-magazine pattern, the difference between magazine and video is not significant ([chi square] = .006, df = 1, p = .939).

When we examined the question of power in sexual relationships, we again find that the Usenet differs dramatically from the other media. While the majority of scenes in magazines and videos contained neither dominant nor submissive participants (i.e., portrayed egalitarian sexual relations, with 69.5% and 80.8% respectively), only 49.3% of Usenet scenes did. The difference between magazine and video here is significant (egalitarian sex, [chi square] = 6.055, df = 1, p = .014).

When these power positions are disaggregated by gender an interesting pattern emerges. Table 4 shows power positions broken down by gender. Percentages are for all scenes containing that type of power position; thus, of all magazine scenes containing a participant in a dominant power position, 9.7% of those scenes have a male in the dominant power position. (The percentage of males and females in a given power position in a given medium do not add to 100%, because some scenes contain two people of opposite sexes maintaining the same power position.)

Table 4. Percentage of Scenes Containing Nonegalitarian Sex by Sex of Participant and Media

              Magazine(a)       Video(b)       Usenet(c)

             Male   Female   Male   Female   Male   Female

Dominant      9.7    93.5    74.2    33.3    68.6    41.4
Submissive   74.2    35.5    26.5    77.9    16.2    85.3

(a) n = 31 (dominant), n = 31 (submissive).

(b) n = 70 (dominant), n = 68 (submissive).

(c) n = 66 (dominant), n = 68 (submissive).

Two things are worth noting here. First, while videos and the Usenet follow the expected gender role patterns (woman are more often submissive, men are more often dominant), in the magazines dominant power positions are virtually always held by women. At the same time, men are twice as likely to be in the submissive position as women. Second, across all media both sexes are shown relatively frequently in both dominant and submissive roles (with, perhaps, the exception of dominant men in magazines). This suggests a greater level of fluidity of dominance and submission than some theories, which hold that these images always accurately mirror gender stereotypes, might have predicted.

We then turned to an examination of the victims and perpetrators of violence. Where violence occurs it is disproportionately caused by men in Usenet scenes. Men make up a smaller percentage of aggressors in video scenes, and an even smaller percentage in magazines. In magazines, 38.5% of sexually violent scenes had male perpetrators and 65.4% had female perpetrators. However, in videos 60.2% of the sexually violent scenes had male perpetrators and 49% of the scenes had female perpetrators. In the Usenet sample 62.7% of the scenes had male perpetrators and only 42.4% of the scenes had female perpetrators. (Again, since a scene can contain perpetrators of both sexes, the percents do not sum to 100.) It is worth noting that magazines not only have a greater gender difference in the perpetrator of violence than the other two media, but they also display women inflicting violence more often then men.

The opposite gender relationship is expressed when we examine the victims of the violence. In magazines, 50% of the sexually violent scenes had male victims, and 61.5% had female victims. Percentages may not add up to 100% because the same scene may have had more than one victim of the violence. In the videos, 32.7% of the scenes depicted male victims, compared with 79.6% of the scenes depicting female victims. And on the Usenet, 23.7% of the scenes depicted male victims and 84.7% of the scenes had female victims. As we can see, the victims of violence in pornography are far more frequently female: Over 61 percentage points separate males and females at the extreme (Usenet), and 11.5 percentage points (magazines) is as close as the sexes come to parity. Magazines, which showed the greatest gender difference in terms of perpetrators of violence, are the most egalitarian in terms of the victims of violence.

Finally, we turn to the issue of consensual sex together with violence. To what extent were victims of violence consensual participants? What we found here was surprising. The vast majority of violence in magazines and videos occurs in the context of a consensual relationship. In magazines, 88.5% of the violence was depicted as consensual, compared with 3.8% as coercive and 7.7% as nonconsensual. In the videos, 91.8% of the violence was depicted as consensual, 5.1% as coercive, and 3.1% as nonconsensual. On the Usenet, however, 42.4% was depicted as consensual, while 10.2% was coercive and 47.5% was depicted as nonconsensual. Thus, less than half of the Usenet scenes that contained violence were consensual, statistically different from both magazines and videos ([chi square] = 51.38, df = 2, p [is less than] .001).

DISCUSSION

Our results clearly demonstrate differing levels of violence among the three pornographic media. What types of theoretical arguments would explain these differences? Although we cannot conclusively point to any single theoretical trend, we outline some of the possible interpretations of the different levels, and offer some evaluation of those explanations.

Several differences among the different media must be factored into the explanation. Magazines, videos, and the Usenet represent three different historical "moments" in the history of erotic representation, from the 1950s and 1960s (the rise of mass-market magazines), to the 1970s and 1980s (the dramatic explosion of videotapes), to the current era, in which the Usenet (and the Internet in general) has emerged as one of the most versatile and accessible venues for pornography. This historical progression also illustrates the increasing "democratization" of pornographic media. As new technologies (video, telephone, computer, etc.) have been adapted to pornography, the cost of production has dropped and control of production has diffused. Successful pornographic magazines require massive production and distribution requirements, making them the province of media elites. For example, while Hugh Hefner began Playboy with a initial investment of only $600 in 1953 (Leerhsen, 1986), a magazine start-up today can require as much as $20 million--the amount needed by John F. Kennedy Jr. to inaugurate George (Handy, 1995).

Adult videos, which became the preferred pornographic medium during the mid-1980s, are far less expensive to produce than magazines. A typical adult feature can cost as little as $3,000 (Schlosser, 1997). The relatively inexpensive production price of videos, compared to magazines, has allowed for an enormous upsurge in both the number of producers creating pornographic videos and the number of features made each year.

The Usenet offers the greatest access to the largest number of users at the least expense: One need only have a computer, a telephone line, and access to the Internet to enter its pornographic world. Usenet pornography production is also unique in that there is complete individual ownership of the means of production. Magazine production requires large printing presses as well as numerous specialized employees (typesetters, graphic artists, etc.). Video production requires cameras, lighting, sound systems, duplication equipment, and people to run them. Usenet pornography production requires nothing more than a computer, a telephone line, and an imagination.

In addition, pornographic magazines are dependent upon advertising, which may require the magazine to mute the violent content to express a more mainstream association with erotica and high-end consumerism. Videos and the Usenet are less dependent upon advertising, and may actually eschew it altogether. This independence, coupled with the lower costs of production and distribution, means that many of the participants in the Usenet world are as much creators of pornography as they are consumers of it. We believe that this blurring of the lines between consumer and creator contains some threads of explanation that may be worth exploring.

Democratization has led to an increase in violent scenes from magazines to videos to the Usenet, with the largest increase in violence in the move between videos and the Usenet. Perhaps this is the illustration of nothing more than simple psychological notions of addiction working themselves out on an aggregate scale. We might expect, for example, that just as individual consumers of pornography tend to tire of a certain level of explicitness and need more, so, too, would the market, acting as an individual. Thus, the more pornography is consumed at one level, the less arousing this material becomes, as the consumer becomes used to--satiated with--the material. This satiation leads the consumer to seek out newer, more explicit, and more violent forms of sexual material that will again arouse him/her (Russell, 1993; Zillmann & Bryant, 1984). Thus we might expect that as new pornographic technologies emerge, one will find them increasingly violent, because they must satisfy both the demand previously satisfied and an increased demand for even more.

Such an explanation, however, can take us only so far in terms of the historical sequence of pornographic form. Our second finding concerned the differing content of the victim and victimizer among the three media, and this difference could not be explained by a simple addiction model. We found that the Usenet shows men in dominant positions, as victimizer and not victim, in far greater proportion than do magazines and videos, which is also suggestive that democratization of pornography has increased both the violence and the amount of misogyny--women as victims--contained in the images. Such a finding confirms some of the arguments offered by radical feminist critics of pornography: that pornography illustrates a universal misogyny and that its function in male supremacist society is to eroticize male supremacy, to "... keep sexism sexy." (Stoltenberg, 1989, p. 129; see also MacKinnon, 1987).

Unfortunately, radical feminist critics often tend to treat pornography as an undifferentiated whole, a monolith, rather than as a set of diverse and often competing genres. Their critique is not

   ... presented against the background of any attempt to analyze pornography
   into different forms, some catering for more peculiar perversions and some
   not, some concerned with women, some wholly directed towards homosexual
   men. Pornography is treated as simply an individual phenomenon ...
   (Simpson, 1983, p.71)

How, then, can we explain not only the increasing violence in the progression of pornographic media but also the qualitative difference in the depiction of victims and victimizers between the Usenet, on the one hand, and magazines and videos on the other? We believe that several specific elements of the Usenet tailor it for a particular type of pornographic representation. As we noted above, it is more democratic, with greater mass access and far less dependence on commercial advertisers. It is as close as one can get to men's(3) direct expressions of their own fantasies, unconstrained by the demands of the marketplace or the high costs of producing and distributing those fantasies to others.

More than this, though, these Internet newsgroups are the closest things to the all-male locker room that exist in the pornographic world: A world, in a sense, entirely without women, a world in which men control absolutely all facets of the scene and in which women do not insert themselves as corporeal beings, even in the highly stylized forms offered by magazines or videos. Any adequate explanation of the increased violence and the shifting relationships of victims and victimizers, then, must take into account the distinctly, purely, and uncorruptedly homosocial element in the Internet newsgroup.

Surely, this homosocial element has been noted before. After all, most pornography is produced by men and for men; it is in this sense about masculinity (see, for example, Kimmel, 1990; Kimmel & Linders, 1997). If, as Kimmel argued, pornography is "gendered speech," it provides a communicative system among and between men.

This homosocial element--men communicating with other men using a particularly gendered speech--has been largely absent from most empirical discussions of pornography and its impact, in part because of the surface level "reading" that pornography is about men's relationships with women. Yet with the Usenet, the relationship between producer and consumer is so blurred as to become nonexistent. And perhaps what is different between the Usenet and the other pornographic media is that the apparent collusion between producer and consumer is broken. Magazine and video producers enter into at least a tacit alliance with consumers. This is necessitated by the pornography producer's desire for profits, and the pornography consumer's desire for arousal. No such alliance exists between the producers and consumers of Usenet pornography (at least not those who have continued to produce pornography free of charge). Perhaps what the Usenet offers is homosocial competition--a relationship among men in which the sexual victimization of women is a currency among men, used as a way to facilitate upward mobility in a masculine hierarchy.

Bird (1996) argues that homosociality is vital to the maintenance of hegemonic masculinity, acting to institutionally and interpersonally segregate men and women, and also acting to suppress nonhegemonic masculinities. Three characteristics of hegemonic masculinity are crucial to Bird's conceptualization, and may suggest some of the elements of an adequate explanation of our findings:

   (a) emotional detachment, a meaning constructed through relationships
   within families whereby young men detach themselves from mothers and
   develop gender identities in relation to that which they are not (Chodorow,
   1978); (b) competitiveness, a meaning constructed and maintained through
   relationships with other men whereby simple individuality becomes
   competitive individuality (Gilligan, 1982); and (c) sexual objectification
   of women, a meaning constructed and maintained through relationship with
   other men whereby male individuality is conceptualized not only as
   different from female but as better than female (Johnson, 1988). (Bird,
   1996, p. 121)

Perhaps the pornography found on Internet newsgroups is so much more violent than magazine and video pornography precisely because of the homosocial competition that exists between individual producers and between producers and consumers. Internet pornography, particularly the newsgroup stories used in this study, is written by men about their fantasies, and is intended for other men to read without marketplace considerations. The producers of this pornography are freed from formal market constraints and, therefore, participate in an informal masculinist marketplace, competing with other men in an effort to prove who can "do the most ...," "last the longest," "have the biggest ...," and so on.

Violence against women is thus a currency among men as they jockey for position in the eyes of other men. This is, of course, facilitated by the sexual objectification of women discussed by Bird (1996). The Usenet contains producers and consumers who are no less a social group because their interactions take place in virtual space. They have their own norms, values, symbols, and modes of interaction. We suggest here that this social group is particularly conducive to the hegemonic masculinity that promotes homosocial competition among pornography producers.

Neither magazine nor video pornography suggests such a relationship between producer and consumer. Indeed, constrained by the formal marketplace, magazines and videos promote homosocial collusion between producer and consumer. It is in the best interest of pornography producers from these media to provide their consumers with the most arousing material possible. The lines between consumers and producers is clear and well defined, thus facilitating their apparent collusion. By contrast, the boundaries between produce and consumer on the Usenet are blurred; consumers are producers and vice versa.

This collusion between producer and consumer helps explain the minor variations between videos and magazines. As we have noted, in most aspects videos and magazines are not significantly different. The homosocial competition model predicts this, since producers of these pornographic forms are trying to maximize their profits by providing the most arousing material to the largest proportion of consumers. This should be possible since hegemonic masculinity (and feminist theory) both suggest that men will tend to find the same material arousing. Marketplace adjustments in the amount of violent material contained in pornography would, therefore, eventually stabilize at relatively equal levels.

Of course, hegemonic masculinity is not absolute. Variations on the dominant masculine identity do exist. Hence, certain segments of pornography consumers would not be served by the homogenous pornography produced according to the demands of a hegemonic definition. Instead, microlevel variations would appear in some forms of pornography in an attempt to tap these consumers. However, these microvariations are just that--small scale. Variations from the hegemonic ideal still conform to its general ideology. Hence, women in magazines can be shown as causing violence, but they still are shown as suffering violence more often than men.

CONCLUSION

This article has provided the first systematic and methodologically coherent comparison among three contemporary pornographic media. We found not only increasing levels of violence as we moved from one medium to the next, but also a dramatic shift from magazines and videos on the one hand and the Usenet on the other when we examined both level of violence and the gender of victim and victimizer. While we offered no definitive explanation of these results, we suggested that explanations that stressed simple democratization, psychological addiction, or misogyny were inadequate to fully explain our results. We have therefore offered a sketch of a model that might explain the differences among these media by virtue of homosocial competition. This places the changing levels of violence within a context of men's relations with each other, and in their constant and ceaseless efforts to prove their masculinity in the eyes of other men. The changes in violent content among media, then, has more to do with the definition of hegemonic masculinity than it does with technological proliferation and democratization, or with the psychological propensity to require greater and greater thrills before satiation. Contrary to radical feminist theory, though, these changes in pornography may say more about men's relationships with other men than they do about men's relationships with women. One potentially fruitful avenue of further research would be an examination of commercial Internet pornography sites. Presumably, these sites operate under market constraints similar to those experienced by magazines and videos and, therefore, would experience the same collusion between producer and consumer leading to similar rates of violence. Only further research will be able to demonstrate the continued utility of this homosocial competition explanation of the changing violent content of pornography over different media and over time.

(1) While the pornographic videos and magazines we sampled may be nationally distributed, we sampled only those available in this one area. However, the Usenet, by design, allows users in many different countries to access the same material, with only language restriction. Thus the generalizability of these findings for the Usenet may be far greater than the findings for the other media. On the other hand, little research has addressed the geographic variations in pornographic content, and there is some evidence that suggests a great variation (Dietz & Sears, 1988). Care should therefore be taken in generalizing these results to other geographic locales.

(2) Thirty-six scenes (5.6% of all scenes) contained violence but not sex. Since the majority of the theoretical literature on the harm of violence in pornography has focused on sexual violence not violence per se, these scenes are not included in our subsequent analysis. Also included is the row's chi-square score. Chi-square tests the null hypothesis that there is no association between the presence of a particular form of violence and media. Thus, a significant chi-square indicates that there is an association between media and presence of violence.

(3) Unfortunately, it is virtually impossible to know what proportion of Usenet pornography is produced or consumed by women. We believe that the vast majority of producers and consumers are male. However we have no hard evidence to support this and, if incorrect, our explanation for differences in violent content would be weakened significantly.

REFERENCES

Allen, M., D'alessio, D., & Brezgel, K. (1995). A meta-analysis summarizing the effects of pornography II. Human Communication Research, 22, 258-283.

Bird, S. R. (1996). Welcome to the men's club: Homosociality and the maintenance of hegemonic masculinity. Gender and Society, 10, 120-132.

Brannigan, A., & Goldenberg, S. (1987). The study of aggressive pornography: The vicissitudes of relevance. Critical Studies in Mass Communication, 4, 262-283

Brosius, H.-B., Weaver, J. B., & Staab, J. F. (1993). Exploring the social and sexual "reality" of contemporary pornography. The Journal of Sex Research, 30, 161-170.

Chodorow, N. (1978). The reproduction of mothering. Berkeley, CA: University of California Press.

DeLoughry, T. J. (1995, July 21). Researcher who studied on-line pornography gets invitation from Congress, criticism from scholars. Chronicle of Higher Education, A19.

Dietz, P. E., & Sears, A. E. (1988). Pornography and obscenity sold in `adult bookstores': A survey of 5132 books, magazines, and films in four American cities. Journal of Law Reform, 21, 7-46.

Duncan, D. F. (1991). Violence and degradation as themes in "adult" videos. Psychological Reports, 69, 239-240.

Elmer-Dewitt, P. (1995, July 24). Fire storm on the computer nets. Time, 146, 57.

Fisher, W. A., & Grenier, G. (1994). Violent pornography, antiwoman thoughts, and antiwoman acts: In search of reliable effects. The Journal of Sex Research, 31, 23-38.

Gilligan, C. (1982). In a different voice: Psychological theory and women's development. Cambridge, MA: Harvard University Press.

Glascock, J., & LaRose, R. (1993). Dial-a-porn recordings: The role of the female participant in male sexual fantasies. Journal of Broadcasting & Electronic Media, 37, 313-324.

Handy, B. (1995, September 18). "Ich bin ein magazine editor." Time, 146, 127.

Johnson, M. (1988). Strong mothers, weak wives. Berkeley, CA: University of California Press.

Kimmel, M. (Ed.). (1990). Men confront pornography. New York: Crown.

Kimmel. M., & Linders, A. (1997). Does censorship make a difference?: An aggregate empirical analysis of pornography and rape. Journal of Psychology and Human Sexuality, 8, 1-20.

Leerhsen, C. (1986, August 4). Aging playboy. Newsweek, 108, 50-56.

Linz, D. (1989). Exposure to sexually explicit materials and attitudes toward rape: A comparison of study results. The Journal of Sex Research, 26, 50-84.

Linz, D., & Donnerstein, E. (1988). Methodological issues in the content analysis of pornography. Journal of Law Reform, 21, 47-53.

MacKinnon, C. A. (1987). Feminism unmodified: Discourses on life and law. Cambridge, MA: Harvard University Press.

Malamuth, N. M., & Check, J. V. P. (1985). The effects of aggressive pornography on beliefs in rape myths: Individual differences. Journal of Research in Personality, 19, 299-320.

Malamuth, N. M., & Spinner, B. (1980). A longitudinal content analysis of sexual violence in the best-selling erotic magazines. The Journal of Sex Research, 16, 226-237.

Matacin, M. L., & Burger, J. M. (1987). A content analysis of sexual themes in Playboy cartoons. Sex Roles, 17, 179-207.

Mayerson, S. E., & Taylor, D. A. (1987). The effects of rape myth pornography on women's attitudes and the mediating role of sex role stereotyping. Sex Roles, 1987, 321-338.

Palys, T. S. (1986). Testing the common wisdom: The social content of video pornography. Canadian Psychology, 27, 22-35.

Rimm, M. (1995). Marketing pornography on the information superhighway: A survey of 917,410 images, descriptions, short stories, and animations downloaded 8.5 million times by consumers in over 2000 cities in forty countries, provinces, and territories. Georgetown Law Review 83, 1849-1934.

Russell, D. (1993). Against pornography: The evidence of harm. Berkeley, CA: Russell Publications.

Schlosser, E. (1997, February 10). The business of pornography. U.S. News & World Report, 122, 43-52.

Scott, J. E., & Cuvelier, S. J. (1987a). Violence in Playboy magazine: A longitudinal analysis. Archives of Sexual Behavior, 16, 279-288.

Scott, J. E., & Cuvelier, S. J. (1987b). Sexual violence in Playboy magazine: A longitudinal analysis. The Journal of Sex Research, 23, 534-539.

Scott, J. E., & Cuvelier, S. J. (1993). Violence and sexual violence in pornography: Is it really increasing? Archives of Sexual Behavior, 22, 357-370.

Simpson, A. W. B. (1983). Pornography and politics: A look back to the Williams Committee. London: Waterloo.

Smith, D. D. (1976). The social content of pornography. Journal of Communication, 26, 16-24.

Stoltenberg, J. (1989). Pornography and freedom. Portland, OR: Breitenbush Books.

Winick, C. (1985). A content analysis of sexually explicit magazines sold in an adult bookstore. The Journal of Sex Research, 21, 206-210.

Yang, N., & Linz, D. (1990). Movie ratings and the content of adult videos: The sex-violence ratio. Journal of Communication, 40, 28-42.

Zillmann, D., & Bryant, J. (1984). Effects of massive exposure to pornography. In N. Malamuth (Ed.), Pornography and sexual aggression, (pp. 15-138). New York: Academic Press.

Manuscript accepted January 14, 2000

APPENDIX A

VIDEOS IN THE SAMPLE

The following videos comprise the sample reviewed by this study:

All Little Women
Amateur A Cuppers
Anal Mystique
Anal Persuasion
Anal Playground
Analtown USA #3
Asses Galore
Bad Attitude
Bad Girls #2
Bad Girls #5
Because I can
Bisexual Anal
Buttfucker #3
Butt Banged Bicycle Babes
Casting Call #18
Cum in my Holes
Cumming Clean
Damp Spot
Danish Hard Core 104
Dirty Dozen #67
Ebony Experience
Eros Extreme
Filthy Fuckers #60
Girls who Love it Ass Backwards
Hidden Obsession
Impact
Jizz and Tonic
Junkyard Dogs
Kink-o-Rama
Pleasure Zone #22
Plum and Dumber
Private Profits
Sex Freaks
Sex Spa
Snakedance
Spunk Suckers
Strap on Sally #9
Super Diamond Double XX #6
Super Tramp
Superstar Sex Challenge #1
The Big One
Tits a Wonderful Life
Totally NakedUFO Tracker
Up and Cummers #32
Upbeat Love
Wall to Wall #33
White Chicks #13
Wicked Women
Zane's World

APPENDIX B

MAGAZINES IN THE SAMPLE

The following magazines were comprise the sample reviewed by this study:

40 +
Adult Cinema Review
Asian Beauties
Big Busts
Celebrity Skin(a)
Club Confidential
College Girls(a)
Confidential Letters
Dirty
D-Cup
Dolly(b)
Family Fun
Gallery
Girls(a)
Girls of Penthouse
Graham Cracker(a)
Hot Buns
Hot Stuff
Just Eighteen
Kinky Babes(a)
Leg Action
Leg Sex
Live Nude Girls
Lovers in Heat(a)
Nugget
Original Porn Stars(a)
Panther(a)
Penthouse
Penthouse Letters
Playboy(a)
Playboy Playmates(a)
Playgirl
Portfolio(a)
Rage(a)
Ravers
Score
Turn-ons

(a) contained no stories

(b) text not in English

Martin Barron and Michael Kimmel State University of New York at Stony Brook

Address correspondence to Michael Kimmel, Department of Sociology, SUNY at Stony Brook, Stony Brook, NY 11794-4356; e-mail:[e-mail].