Author: Potter, W James; Smith, Stacy Source:
Original source: http://dorm.tunkeymicket.com/spcom111/controversy/tele_violence.html
This content analysis identifies the web of context that typically appears in the portrayal of television fictional violence. Highly graphic portrayals of violence are most likely in live action non-humorous programs with human perpetrators and targets. Graphicness was also found to vary across consequences to the victim, levels of reward, and use of weapons. The web of context that typically surrounds highly graphic portrayals is likely to lead to a fear effect more so than to desensitization or disinhibition.
The process of constructing explanations about how television violence influences viewers necessarily moves from content through viewer perceptions of the content to effects. This study focuses attention on the first of these components-content. We use the findings of the literatures of perceptions and effects to hypothesize patterns that should be prevalent in the violent content in order to lead to certain effects.
Key to this chain of reasoning is the context in which violence is portrayed on television. Contextual factors in portrayals cue viewers about how that action is to be perceived and hence interpreted. Reviews of the effects literature have concluded that exposure to television violence portrayed with particular contextual characteristics can lead to such negative effects as fear, desensitization, and disinhibition (for example, see Bryan & Schwartz, 1971; Comstock, Chaffee, Katzman, McCombs, & Roberts, 1978; Comstock & Strasburger, 1990; Hearold, 1986; NTVS, 1997; Paik & Comstock, 1994; Potter, 1999). The only contextual factor that researchers have been able to associate with all three of these major effects is graphicness. Exposure to highly graphic portrayals of violence are more likely to lead to fear as well as to disinhibition in the short term and to desensitization over the long term.
Because graphicness is consistently associated with a range of negative effects, and because graphicness is so strongly associated with violence itself, this study will treat graphicness as its focal construct. The purpose of this study is to answer two questions about graphic violence presented on television. First, what contextual factors are most likely to appear when violence is portrayed as graphic? Television narratives are complex. A contextual characteristic does not exist in isolation; instead, contextual characteristics appear in clusters-forming a web composed of meaning cues that supports each narrative.
Once we have found the answer to this first question, a second question becomes important: Given the web of context most prevalent with graphic portrayals of violence, which of the three major negative effects-fear, desensitization, and disinhibition-is most likely to occur when viewers are exposed to these portrayals?
Graphicness
Graphicness in a television portrayal serves to bring viewers closer to the action and thereby make that action more real or more shocking. With violent portrayals, graphicness puts viewers into the action and shockingly depicts physical harm to the victims. From a production point of view, graphicness is an issue of framing the action (National Television Violence Study, 1997). When the director shows action in close-up, it is more graphic than if the director chooses to show the action in long shots, off screen, or by implication. An extreme close-up of the perpetrator's finger squeezing the trigger, a close up of the perpetrator's evil eyes, and a close up of the bullet tearing through the flesh of the victim signal high graphicness. The same violent act would appear to viewers to be less graphic, if it were shown in a long shot. Graphicness can be further reduced by moving the action off screen, that is, viewers could hear the shot of a gun and the sound of a body hitting the ground, but the screen does not present this action. Thus the depiction of shootings can vary substantially in graphicness by changing the framing.
Another element of graphicness is degree of physical alteration to the victim. This is the concern of blood and gore. When a character is shot, for example, there are many ways in which that action can be shown to viewers. In a non-graphic manner, a victim can simply fall to the ground when shot: the victim looks the same as before being shot except she is lying down with her eyes shut. In contrast, a director wanting a highly graphic scene can show the victim lying on the ground in a growing pool of blood flowing from gaping holes in the victim's flesh.
From the viewers' point of view, graphicness is closely associated with violence, that is, the more a person perceives a portrayal as graphic, the more violent that person will rate the act. For example, Potter and Berry (1996) showed participants clips of different kinds of violence on television and asked them to rate the degree of violence along with 25 other contextual factors in those portrayals. Judgments of graphicness were found to be strongly and consistently related to viewer perceptions about how violent different types of portrayals were. Furthermore, these findings held across different types of participants (sex, experience with real life violence, amount of television viewing, and such psychological measures as religiosity and political orientation).
Why should graphicness be associated so strongly with violence? A schema perspective may provide an answer. Graphicness brings the viewer close to the action and thus elicits more attention. When violence is graphic, the schema for violence is quickly activated, and there is little chance for activation of a schema that does not have violence as its central element. Viewers experience a sharpened focus on the violence, because it occupies so much of the screen. Other contextual elements are secondary. Because the violence is so salient in the portrayal, the cognition of violence is strong. On the other hand, if the violence were presented in a non-graphic manner, viewers would be more likely to focus on the plot or characters and thereby activate a schema in which narrative or character were more central.
Graphicness has been identified as a key contributor to all three negative effects-fear, desensitization, and disinhibition. This is an intriguing finding, because these three effects are very different from one another. For example, the effect of fear is indicated by an increase in emotion, but the effect of desensitization is characterized by a decrease (almost to the point of absence) of sympathy or empathy, both of which have emotional components. The schema-based processes underlying these three effects explain how graphicness can contribute to each of them.
Fear. A fear effect is more likely when violence is portrayed graphically (Ogles & Hoffner, 1987). Graphic portrayals of violence engage emotions more strongly than non-graphic portrayals. The graphic portrayal is likely to put the viewer into the scene, thus triggering a flight-fight experience. A strong physiological change increases both emotions and cognitions. The shocking images capture the viewers' attention. Viewer awareness is high. Thus graphicness in violent portrayals strongly associates emotions with cognitions, as when either is activated, the other is immediately activated. For example, if a knife fight is portrayed as a dance from a long distance, graphicness is low, and the resulting fear is likely to be low. Viewers are likely to regard the violence as happening to someone else, and there is less likelihood that viewers will identify with the characters in the action. But, if the knife is shown in close up through subjective camera, viewers are more likely to identify with the characters. And if viewers see a knife slowly slicing through flesh while bright red blood spurts freely from the growing wound, graphicness is high, and the likelihood of viewers feeling fear is also high.
Because fear reactions are often unpleasant, viewers will employ coping strategies to reduce their fear, and this coping can lead to desensitization to graphic violence. Cantor (1994) explains that this can happen when viewers use cognitive strategies for coping with television violence. Cognitive strategies are those where viewers "think about the fear stimulus in a way that generally casts the threat in a different light" (p. 236) such as concentrating on the fantasy (non-real) nature of the portrayal or shifting one's cognition toward an expectation of an eventual happy ending. These strategies reduce the emotion by shifting the attention away from the violence itself and onto other elements in the portrayal. Thus from a schema perspective, the connection between cognitions and emotions is weakened, as viewers "think through" their fear by considering other factors in the context. Thinking it through tends to strengthen the inter-connections among cognitive elements (such as judgments about reality and projecting the next steps of the narrative).
Desensitization. Desensitization to the plight of victims can increase over time when portra44, no. 2 (Spring 2000): p. 301-323yals of violence are graphic, as in slasher films (Linz, Donnerstein, & Penrod, 1988). While this desensitization effect has been found to evaporate over time when exposure to graphic violence has ceased (Mullin & Linz, 1995), the desensitization effect is likely to continue or to increase when exposure to graphic violence is frequent.
Disinhibition. In addition to fear and desensitization, a third negative effect disinhibition-has been attributed to exposure to graphic violence (NTVS, 1996). There may be two processes at work in disinhibition-one that explains an immediate effect and another that explains a long term effect. As for the immediate effect, graphicness triggers physiological arousal. This arousal translates into a greater probability of behaving aggressively (Zillmann, 1971, 1991). Inhibitions are temporarily suppressed by the drive to act. In the long term, a steady diet of graphic portrayals will tend to socialize viewers that strong violence is a normal part of everyday life and that it is a successful tool in resolving conflict (Huesmann, 1986; Huesmann et al., 1984). Thus, inhibitions gradually erode a person's schema for violence.
Other Contextual Factors
Viewer schemas about violent narratives are composed of other elements as well. These other elements help signal viewers about how to interpret violent portrayals. They have also been found to be related to negative effects. A fear effect has been associated with factors of reward (Bryant, Carveth, & Brown, 1981 ), target attractiveness (Comisky & Bryant, 1982; Feshbach & Roe, 1968; Tannenbaum & Gaer, 1965; Zillmann, 1980, 1991; Zillmann & Cantor, 1977), justification of the violence (Bryant, Carveth, & Brown, 1981 ), and realism of the portrayal (teen, 1975; Geen & Rakovsky, 1973). The desensitization effect is also influenced by humor (Gunter, 1985; Sander, 1995).
The disinhibition effect is enhanced when the perpetrators are attractive as with heroes (Bandura, 1994), when the violent action is portrayed as being justified (Berkowitz & Geen, 1967; Hoyt, 1970), when weapons are portrayed (Berkowitz & LePage, 1967; Caprara, Renzi, Amolini, D'Imperio, & Travaglia, 1984; Frodi, 1975; Leyens, Cisneros, & Hossay, 1976; Page & O'Neal, 1977; Turner, Layton, & Simons, 1975) especially when guns are used (Berkowitz, 1990), when violence is realistic (teen, 1975; Thomas & Tell, 1974), when the violent act is not punished (Bandura, 1965), when it is portrayed with humor (Mueller & Donnerstein, 1977), and when violence is shown with no consequences (lack of pain or suffering) (Baron, 1971 a, 1971 b,1979; Berkowitz & Rawlings, 1963; Cantor & Hoffner, 1990; Cantor & Sparks, 1984; Comstock, et al., 1978; Dubanoski & Kong, 1977; Sparks, 1986).
Study Rationale
When a violent act is portrayed in a television narrative, it appears in a web of contextual factors that cue viewers to the meaning of the action. Which of these factors should we expect to be most closely related to graphicness in a portrayal? An answer was provided by Gunter (1983) who conducted a series of experiments to assess viewer judgments of the seriousness of various forms of violence. He asked participants to rate programs for seriousness of the violence they portrayed. He found that viewer ratings of the seriousness of the violence were not related to the number of violent incidents in the program as much as to the manner in which the violence was portrayed. Key among the contextual factors was the form of violence (shootings and stabbings), the degree of harm to the victim, and the degree of realism in the settings. These characteristics suggest a pattern for graphicness. While producers do not need to employ weapons in their violent narratives in order to make them graphic, the use of weapons provides greater opportunity for graphicness in portrayals. Close-ups of a bloody knife slashing across the screen or a subjective camera shot forcing viewers to look down the barrel of a gun as it fires into their faces are very graphic images. A scene that calls for a bomb exploding in a crowd offers the possibility to present body parts flying through the smoke and rubble.
Core Context. Given the finding that people strongly associate violence with graphicness (Potter & Berry, 1996) and given Gunter's findings about the characteristics viewers perceive to be most associated with serious violence, we can hypothesize a link between those perceptions about violence and a pattern in television content. Perhaps viewers have learned their general perceptions by watching a particular set of contextual characteristics repeated continually over time. If this is the case, we should expect to find in television content that the most serious acts of violence (most graphic) are likely portrayed with weapons (guns and knives rather than fists), where the victims are shown suffering great pain and harm, and where fantasy is avoided in favor of showing what happens in real life. This would appear to be a reasonable web of context for graphicness. The portrayal of consequences seems to reinforce a perception of high graphicness when producers show the look of pain on the victim's face, especially when the victim's expression is shown in close up. If the graphicness is high, there should be indications of blood, gore and other manifestations of great physical harm. Highly realistic depictions lend themselves to high graphicness. When depictions are fantasy (such as Bugs Bunny or the Roadrunner), graphicness is perceived to be very low, and the violence is perceived to be non-existent. Viewers-especially with fictional content-make assessments about whether something could happen as portrayed. That which could never happen is fantasy (Potter & Warren, 1998). So the judgment must move beyond the actualities of occurrence and consider the possibilities that different characters could be people encountered in real life and that different situations could occur. Violence performed by human type characters leads to more intense responses than violence performed by cartoon characters or puppets (Gunter & Furnham, 1984; Osborn & Endsley, 1971; Surbeck, 1975). For example, Osborn and Endsley (1971 ) found that participants responded more emotionally to the violence, when human characters were.involved. Human violence was also rated as most scary, and participants were able to recall more details from it. Cantor (1994) believes that children are especially sensitive to the similarity of depicted characters to real life people, so that when human characters perform violence, children are much more likely to experience a fear effect.
Therefore we hypothesize that the core of the contextual web supporting graphic violence on television would include the characteristics of weapons, harmful consequences to victims, and realism. If this contextual web is found to be prevalent in the presentation of graphic violence, then this finding will be a useful first step in building a chain of explanation as to how television viewers define violence and construct their perceptions of violence.
Secondary Context
There may be additional characteristics-beyond those Gunter found in the contextual web surrounding violence on television. Looking beyond the literature of perceptions to the research on effects of exposure to violence, we can identify several other potentially important contextual variables, such as reward/punishment, motives, justification, character attractiveness, and humor. Although none of these contextual characteristics seems to be related to graphicness as closely as presence of weapons, consequences, or realism, each has also been demonstrated to increase the likelihood of a negative effect from exposure to violence. Therefore these five characteristics also warrant examination as patterns of graphicness. Each of these has been found to increase the probability of a negative effect: Reward/Punishment (Bandura, Ross, & Ross, 1963; Rosekrans & Hartup, 1967; Walters, Parke, & Cane, 1965), motives of vengeance (Berkowitz & Alioto, 1973; Geen & Stonner, 1972, 1973, 1974; Hoyt, 1970); justification (Berkowitz & Powers, 1979; Berkowitz & Rawlings, 1963; Geen, 1979, 1981; Meyer, 1972); character attractiveness (Liss, Reinhardt, & Fredriksen, 1983), and humor (Baron, 1978; Bergen 1988; Zillmann, 1979).
To this "other set of variables" we add multiple acts of violence in a scene. This is a variable as yet untested in experimental effects literature, yet it seems intuitively to be related to graphicness. For example, a perpetrator could fire one bullet into a victim or the perpetrator could be given an automatic weapon and fire 20 bullets into a victim. Both examples would count as one instance of violence, but clearly there is a difference. The multiple acts should be associated with higher graphicness.
Characters
Finally we need to consider the role of different kinds of characters in the contextual web for television violence. We know from the effects literature that when violence is performed by an attractive character, the probability of aggression increases (Comstock, et al, 1978).
In summary, we examine three groups of variables that may contribute to the contextual web supporting graphic portrayals of violence. The variables in the core group (weapons, consequences, and realism) are those that have been associated with graphicness in past research. The variables in a secondary group (reward/ punishment, justification, humor, and multiple acts) are contextual factors that seem likely to be associated with graphicness but that have yet to be tested for that association. The variables in the character group are factors about the perpetrators and targets of television violence that have been associated with negative effects but which have not yet been tested in content studies for possible association with graphicness.
This study reports how these nine contextual factors are present in graphic portrayals of violence on television. The patterns among these contextual factors are contrasted between high graphic and no graphic portrayals of violence in an effort to map the pattern of context surrounding graphic portrayals of violence.
Method
Sample
The data for this analysis were from the second year of the three year National Television Violence Study (NTVS, 1997). The sample consisted of 3,235 programs representing a composite week of programming from 6 a.m. until 11 p.m. from 23 channels, including broadcast networks (ABC, CBS, Fox, and NBC), independent broadcast channels in the Los Angeles market (KCAL, KCOP, and KTLA), public broadcast, basic cable channels (A&E, AMC, BET, Cartoon Network, Disney, Family Channel, Lifetime, Nickelodeon, TNT, USA, VH-1, and MTV), and premium cable (Cinemax, HBO, and Showtime). Selection of programs was random from a sampling frame of 20 weeks of programming beginning on October 7, 1995 and ending June 7, 1996. Excluded from coding were the 478 programs that were religious, game, sports, infomercials, instructional, and breaking news shows. Atotal of 2,757 programs were coded.
Violence and Unitizing
The definition of violence used in this study is: any overt depiction of a credible threat of physical force or the actual use of such force intended to physically harm an animate being or group of beings. Violence also includes certain depictions of physically harmful consequences against an animate being or group that occur as a result of unseen violent means. Thus, there are three primary types of violent depictions: credible threats, behavioral acts, and harmful consequences of unseen violence.
For purposes of the analysis in this study, only violent interactions that were behavioral acts were included. Threats that did not escalate into a behavioral act were screened out. Also excluded were instances where the violence happened off screen such that the viewer could see only the aftermath of the violence. Violent interactions that were judged to have no possibility of harm were excluded. Thus the sub-sample used for the analyses in this study included only acts of violence clearly depicted on the screen and where there was an expectation of harm to the victims.
Contextual Variables
The variables used in this analysis come from a larger study where 26 variables were coded. This study focuses on graphicness and 17 other measures, grouped as follows: five core, four secondary, and eight character measures.
Graphicness was operationalized with three measures at the sequence level: blood/gore, framing of violence and framing of impact. Blood/gore had four values: extreme (heavy amounts of blood or gore or dismemberment), moderate (heavy bleeding or deep wound), mild (several drops of blood or cuts), and none. Framing of violence had three values: close-up (over 65% of screen consumed with an image of the perpetrator and/or weapon), long-shot (violent interaction shown in the background or without close-ups), and none (the violent interaction itself was not shown on screen). Framing of impact also had three values: close-up (over 65% of screen consumed with an image of the target while harm is being inflicted), long-shot (target shown receiving violent interaction in the background or without close-ups), and none (target not shown on screen). These three measures were used to construct a graphicness index by giving the blood/gore value a double weight then adding that to the sum of the values for the other two measures. Thus the graphicness index had a range of values from 0 to 10. A violent act that showed no blood or gore (0), and was shown off screen so there was no framing of the violence (0) and no framing of the impact, received a zero on the graphicness scale. At the other end of the graphicness scale was a 10, which required the blood and gore to be rated as extreme (thus the value of 3 doubled to make a 6), the framing of violence to be in close up (2), and the framing of impact also to be in close up (2). Only 129 (1.6%) of violent acts were portrayed off screen and did not meet any of the criteria for earning points on the graphicness index (see Table 1 ). Among the vast majority that were portrayed on screen, 75.9% earned two points -typically they were portrayed with the perpetrator and target not in close up shots. Most of the values are at the low end of the index with the median and mode both being 2.00. However, there were still 853 violent instances that earned a value of 6 or larger on the graphicness index.
The core set of five contextual factors included consequences (harm and pain), realism (genre and style), and weapons. Consequences was operationalized with two PAT level measures: harm and pain. Each of these measures had four values: none, mild (minor harm not needing medical attention), moderate (serious physical harm requiring medical attention), and extreme (life threatening physical harm). Realism had two operationalizations at the program level. First, realism was assessed with four values: actual reality (footage of real life events), recreation of reality (simulations), fiction, and fantasy. Then style of presentation was assessed with three values: live action, animated, and both. Presence of weapons was coded at the PAT level with six values: natural means (such as hitting, kicking, biting, etc.), unconventional weapon (such as baseball bat, chain, etc.), non-firearm conventional weapon (such as knives, police batons), hand held firearm, heavy weaponry (tanks, cannons, missiles), and bombs.
The secondary set of four contextual factors included justification, humor, reward/punishment, and multiple acts within a violent interaction (PAT). Justification was assessed at the PAT level with two values-yes and no. Humor was assessed at the sequence level with the values of yes and no. Rewards and punishments were assessed at the sequence level with seven measures: self praise, praise from others, material reward, self condemnation, condemnation from other, stopped by violent means, and stopped by non-violent means. Each of these was coded as either no (0) or yes (1 ). A reward/punishment index was constructed by summing the values from the three reward measures (+1 point) and the four punishment measures (-1 point). Thus the resulting values on this reward/punishment index ranged from +3 to -4. We also included a prevalence type measure of extent, which captured whether the violent interaction was composed of a single act of aggression or of repeated acts. It included four values: one act, some (2 and 9), many (10 to 20), and extreme (more than 20). For example, a single act of violence where the perpetrator stabs the target once was coded as one on the extent measure, but if there were 12 stabs in a single act of violence, it was coded as one act with an extent of "many."
Table 1
A set of eight character attractiveness measures was included. Four variables for the perpetrator included: hero status (yes, no), good/bad (primarily a good character, primarily a bad character, or a character with both qualities), type (human or fantasy), and an assessment of whether the perpetrator was a primary character (yes, no). Four similar variables were included for the targets of the violence.
Coding and Reliability
The coders were 58 undergraduate students at the University of California at Santa Barbara. Each coder was given about 40 hours of training over a 12 week period while their progress was continually tested through diagnostic reliability procedures. When they passed their training, they were certified to code.
There were 18 formal reliability tests-one conducted each week during the course of the coding. Intercoder agreement was quite good on the 41 variables monitored. Median level agreement ranged from a low of .61 to a high of 1.0 with over half of the medians being over .90. The median reliability coefficients for the variables used in this analysis areas follows: blood/gore, .90; framing of violence, .95; framing of impact, .90; realism, .96; style of presentation, 1.0; character type, .99; hero status, .96; good/bad, .79; primary character, .88; humor, .83; self praise, .83; praise from other, .95; material reward, .94; self condemnation, 1.0; condemnation from other, .90; stopped by violent means, .93; stopped by non-violent means, .95; harm depicted, .83; pain, .81; presence of weapons, .94; and justification, .87. For more detail on coding and reliability, see NTVS (1997).
Results
The first question guiding this analysis is: What contextual factors are most likely to appear when violence is portrayed as graphic? To answer this question, we use a two step analysis. In the first step, we determine if there is a difference in the degree of graphicness across the various values of the contextual variables in the three groups: core, secondary, and character. In the second step, we determine which of the contextual variables are most associated with graphicness by using a discriminant analysis to compare portrayals exhibiting low graphicness with portrayals exhibiting high graphicness.
The first analysis used the five core contextual factors as independent variables to compare the means of the scores on the graphicness index. All five of those core contextual variables are statistically significant (p < .01 ) in a simple factorial ANOVA testing for unique differences in a multivariate procedure (see Table 2). Higher degrees of graphicness are found with live action programs, especially in recreated reality programs. As should be expected, graphicness is higher when consequences of harm and pain are moderate and extreme. Also, graphicness is more likely when conventional weapons are used (such as guns and knives) more so than the use of heavy weaponry including bombs. It is interesting to note that, in general, when perpetrators use heavy weaponry, such as missiles, bombs, torpedoes, bazookas, and hand grenades, the portrayals are less graphic than when perpetrators fire a hand gun or stab a target.
Next, four secondary contextual variables were tested: three comparisons reached significance (see Table 3). Graphic portrayals are much more likely to be in non-humorous settings, with a higher number of acts of violence within an interaction, where the acts are motivated by protection of life, personal gain, or retaliation, and where the violent act is rewarded.
Of the eight character attractiveness measures, only one reached statistical significance (Table 4). When targets are human, there is a greater likelihood of graphicness.
Table 2
Table 3
When the six core measures were used in a stepwise procedure of discriminant analysis, four of those measures are accepted into the discriminant function (see Table 5). The consequence of depicted harm is the most important discriminator followed by likely harm, style, and pain. Realism and weapons were not accepted into the equation. The equation results in a fairly low Wilks' Lambda and a strong canonical correlation coefficient of .660.
When the four peripheral contextual variables were used in a stepwise procedure of discriminant analysis, all four of those variables are accepted into the discriminant function. This equation was not as strong a predictor as the previous one. The character set of eight variables resulted in a weaker still equation, which included four measures. Within those accepted variables, type of target was an especially strong predictor.
Table 4
Table 5
Finally, when all the measures were used in a stepwise procedure of discriminant analysis, six were accepted into the discriminant function. This is the strongest predictive equation. While it includes factors from all three sets of variables, the core measures of realism, depicted harm, and pain are the strongest predictors.
When we compare the contextual patterns in the most graphically portrayed violent interactions with those of no or very low degrees of graphicness, the core set of variables reveals the strongest prediction followed by the peripheral set, and finally the character set. This relative importance among variable sets is also reflected in the total equation where three of the seven stepwise selected variables are from the core variable set. The contextual profile of the most graphic violence is as follows: live action shows with human (not fantasy) perpetrators and targets; not engaging in multiple violent actions within an interaction; and where victims are shown with little harm or pain.
Discussion
We want to emphasize five findings from our analyses. First, there is a range of graphicness in the portrayals of violence on television, but most of the violence is presented with a low level of graphicness. Only about one violent act in 10 is shown with relatively high graphicness, which in our operationalization would mean an extreme close-up of the violent act and the impact on the target, a great deal of blood and gore, or some combination of these elements. While the percentage of violent acts that are violent is low, we should not conclude that the average viewer would rarely if ever see graphic violence. Because the base number of violent acts is so large, the number of acts in this relatively small percentage is still substantial. The average viewer who watches about three hours of television per day will see one of these acts of highly graphic violence every day. One per day sounds low, but 2,200 over the course of a child's elementary school education does not sound so small.
A second important finding of this study is that highly graphic portrayals of violence are presented in a different contextual web of factors than are low graphic portrayals. The contextual variable that is most strongly associated with high graphicness is high realism. Thus it appears that highly graphic portrayals of violence are much more likely to be in live action programs with human perpetrators and targets. Conversely, it is rare to find graphicness in fantasy programs with characters who are animals, superheros, or some other form of fantasy character.
Our third finding addresses the question: Given the web of context most prevalent with graphic portrayals of violence, which of the three major negative effects-fear, desensitization, and disinhibition-is most likely to occur when viewers are exposed to these portrayals? Using what the effects literature tells us about how individual contextual factors increase the risk of negative effects (fear, desensitization, and disinhibition), we searched for a pattern in television content to determine which effect typical viewers were most likely to experience.
We can rule out desensitization as the most likely effect of the three, because the prevalence of graphicness is very low and the prevalence of humor is less than half with violence in general (45.9%), and its prevalence drops dramatically (20.6%) when we consider only highly graphic portrayals (see Table 6). We do not make too much of this difference in profiles; because at this point little is known about the factors that lead to a desensitization effect. When more experimental research is conducted on this topic, the desensitization risk profile will be more elaborated, and a better comparison can be made.
The case for disinhibition is mixed. The patterns on punishment and weapons support the disinhibition effect. But the characteristics of graphicness, justification, attractive perpetrator, and consequences do not fit the pattern of effect. justification is relatively low (36.1%) and stays about the same with highly graphic portrayals. Perpetrators of violence (even highly graphic violence) are not likely to be heroes or good characters; however, they are likely to be human and primary characters. As for consequences, both pain and harm are just above the 50% mark with violence in general, but they increase dramatically (especially moderate and extreme) with graphic portrayals.
The case for a fear effect is stronger, but the pattern is still mixed. A fear effect has been associated with greater realism, a lack of justification, and a lack of punishment --all three of these characteristics are in the pattern of violent content. Live action is relatively high (58.9%) and fantasy is relatively low (44.2%) among violent acts in general and this pattern sharpens with high graphic violence. Also, punishment is low (24.7%) among all violent acts, and while it is higher among highly graphic violence, it is still relatively low (34.2%). In contrast, graphicness and target attractiveness do not fit the pattern of a fear effect. Few targets are either heros or good, and this does not change much with graphic violence. There is a high prevalence of human targets (especially with high graphicness) and primary characters being targets.
The case for a fear effect being most likely makes sense given our operationalization of high graphicness, which requires a high degree of blood and gore along with the perpetrator's actions as well as the harm to the victim to be shown in close up. When this is shown in a realistic setting (not fantasy) and where the perpetrator is rewarded for such actions, it is understandable how viewers would experience the emotion of fear. It is easy to understand how high graphicness would get viewers' attention and lead them to consider the portrayal as being more graphic (Potter & Berry, 1996) and more serious (Gunter, 1985). Judgments of high violence and high seriousness would seem to lead more to an effect of fear than to the effects of desensitization or disinhibition.
Also, the linkage of graphicness with reality would seem to increase the likelihood of a fear effect, because research has consistently shown that realistic settings and human type characters lead to more intense responses than do fantasy settings and characters (Gunter & Furnham, 1984; Osborn & Endsley, 1971; Surbeck, 1975), especially with children (Cantor, 1994).
Table 6
Would the absence of an attractive target and the presence of justification for the violence, which are two elements in the fear profile, prevent a fear effect from occurring? The answer would appear to be no, especially if the effect were regarded as being a long term effect as well as an immediate one. The immediate effect would seem to be more of an emotional experience, and as such, the more a viewer identifies with the target and the more undeserved (unjustified) the violence, the more fearful the viewer will become as the target is shown graphically experiencing the violence. But the long term effect seems to be more of a cognitive one where viewers come to believe that the world is a mean and fearsome place, as cultivation theory predicts (see Gerbner, Gross, Morgan, & Signorielli, 1980). With this long term perspective on a fear effect, it would seem that the contextual factors of attractiveness of the target and lack of justification for the violence would become much less important. The two contextual factors that would remain crucial to a long term fear effect would be that the violence is not fantasy but that it occurs in real life settings and that the perpetrators are not punished-so they are still out there rendering violence.
The fourth conclusion that we want to emphasize reflects more on what we did not find rather than what we did. This study has exposed many gaps in our understanding of effects and perceptions. We began with the literature on effects and perceptions in order to use the findings there to provide a foundation for our analyses of content. But our results illustrate that it may be time to reverse the direction of this reasoning. It appears that we know a good deal about the content of violence and its contextual web. Content findings such as those in this study can be used to form a basis and should be used to direct additional effects studies into specific areas. We need to know much more about the influence of various contextual factors (both individually and in combinations) on all three effects, as well as many other negative effects (see Potter, 1999).
In exploring some of the reasons why we did not find a stronger match between what we identified as the contextual web of violence and what the literature identifies as active factors in the effects process, it is clear that the effects literature is missing a great deal of what is necessary to make a strong assessment of the match between content and effects. Many of the contextual factors have not been tested as contributing to one of the three effects. The blank areas on the right side of Table 6 indicate areas where either no tests have been conducted or where very few tests have been run but we still do not know the relationship between those contextual factors and effects. For example, this is especially the case with the desensitization effect where we only have two factors tested thus far. Perhaps when more research is conducted, we will find that the resulting desensitization risk profile is a close match to the contextual pattern on graphicness. But as of now, we only have one point of comparison-humor-and on that point, there is not a match.
Another shortcoming of the effects literature is that rarely are these contextual factors tested in combination. We do not know much about how factors work together in interaction nor do we know which of these factors contributes more strongly to an effect. For example given our results, we were forced to conclude that the contextual web did not strongly support a disinhibition effect. Our conclusion was based on a simple comparison of the two factors present that would seem to increase the probability of an effect (high punishment and use of weapons) compared to the four factors absent flow graphicness, low justification, low attractiveness of perpetrators, and high pain/harm to the targets), But perhaps the two factors that were present were major contributors to a disinhibition effect compared to the relatively minor contribution of the other four factors that were not found to be especially prevalent in the contextual web of violence. If we knew the relative influence of the different factors, we might arrive at a very different conclusion. (For a more detailed discussion of this, see Potter, 1997, 1999).
Therefore the strongest contribution of this research about prevalent content patterns in graphic portrayals is to suggest a direction for the next step in effects research. We strongly recommend that effects researchers undertake more ambitious multivariate designs to assess the interactions among graphicness (high, low), realism (live action, animated), consequences (none, high), and characters (human, nonhuman). This would require 16 conditions in a 2x2x2x2 design. Of course more elaborate designs could be fashioned with the addition of other factors (such as multiples of acts, humor, justification, reward, etc.) and with using more than the minimum two values per variable. The dependent variables should be the three main negative effects (fear, disinhibition, and desensitization) and perhaps others. The fine of research beginning with these suggestions will extend our understanding of risk of negative effects of exposure to the most high profile graphic portrayals as they are currently portrayed in violent television narratives.
Finally our most intriguing finding may be that the contextual web surrounding highly graphic violence is more prosocial than the contextual web of television violence in general. Compared to all violent acts, the highly graphic violent acts are much more likely to be presented with the victims suffering more harm and pain as well as being much less likely to be presented in a fantasy or humorous context. Thus the contextual web of realistic, serious, painful action that surrounds graphic portrayals of violence serves both to outrage viewers to complain about these portrayals and at the same time protects them from negative effects, especially of disinhibition and desensitization. In contrast, it is the non-graphic violence that is surrounded by the much more antisocial web of context. While viewers are much less outraged by this "other" violence-which is much more prevalent on television -they are much more at risk of learning that violence is fun, successful, and non-harmful. This ironic situation is a clear case of what you don't know much about may be hurting you the most.
References
Bandura, A. (1965). Influence of models' reinforcement contingencies on the acquisition of imitative responses. Journal of Personality and Social Psychology, l, 589-95.
Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ: Prentice Hall.
Bandura, A. (1994). Social cognitive theory of mass communication. In J. Bryant & D. Zillmann (Eds.) Media effects: Advances in theory and research (pp. 61-90), Hillsdale, NJ: Lawrence Erlbaum Associates Publishers.
Bandura, A., Ross, D., & Ross, S. A. (1963). Vicarious reinforcement and imitative learning. Journal of Abnormal and Social Psychology, 67, 601-607.
Baron, R. A. (1971 a). Magnitude of victim's pain cues and level of prior anger arousal as determinants of adult aggressive behavior. Journal of Personality and Social Psychology, 17(3) 236-243.
Baron, R. A. (19716). Aggression as a function of magnitude of victim's pain cues, level of prior anger arousal, and aggressor-victim similarity. Journal for Personality and Social Psychology 18(1 ), 48-54.
Baron, R. A. (1978). Aggression-inhibiting influence of sexual humor. Journal of Personality and Social Psychology 36(2), 189-197.
Baron, R. A. (1979). Effects of victim's pain cues, victim's race, and level of prior instigation upon physical aggression. Journal ofApplied Social Psychology 9(2), 103-114.
Berkowitz, L., & Alioto, J. T. (1973). The meaning of an observed event as a determinant of its aggressive consequences. Journal of Personality and Social Psychology, 28(2), 206-217. Berkowitz, L., & Geen, R. G. (1966). Film violence and the cue properties of available
targets. Journal of Personality and Social Psychology 3, 525-530.
Berkowitz, L., & Geen, R. G. (1967). Stimulus qualities of the target of aggression: A further study. Abnormal and Social Psychology, 65, 197-202.
Berkowitz, L., & LaPage, A. (1967). Weapons as aggression-eliciting stimuli. Journal of Personality and Social Psychology, 7, 202-207.
Berkowitz, L., & Powers, P. C. (1979). Effects of timing and justification of witnessed aggression on the observers' punitiveness. Journal ofresearch in personality 13, 71-80. Berkowitz, L., & Rawlings, E. (1963). Effects of film violence on inhibitions against
subsequent aggression. Journal of Abnormal and Social Psychology 66(5), 405-412.
Bryan, J. H., & Schwartz, T. (1971). Effects of film material upon children's behavior. Psychological Bulletin, 75, 50-59.
Bryant, J., Carveth, R. A., & Brown, D. (1981). Television viewing and anxiety: An experimental examination. Journal of Communication, 31, 106-119.
Cantor, J. (1994). Fright reactions to mass media. In J. Bryant & D. Zillmann (Eds.) Media effects: Advances in theory and research (pp. 213-245), Hillsdale, NJ: Lawrence Erlbaum Associates Publishers.
Cantor, J., Hoffner, C. (1990). Children's fear reactions to a televised film as a function of perceived immediacy of depicted threat. Journal of Broadcasting & Electronic Media, 34, 421-442.
Cantor, )., & Sparks, G. C. (1984). Children's fear responses to mass media: Testing some Piagetian predictions. Journal of Communication, 34(2), 90-103.
Caprara, G. V., D'Imperio, Gentilomo, A., Mammucari, A., Renzi, P, G., & Travaglia, G. (1984). The intrusive commercial: Influence of aggressive TV commercials on aggression. European Journal of Social Psychology, 14, 23-31.
Caprara, G. V., Renzi, P, Amolini, P, D'Imperio, G., & Travaglia, G. (1984). The eliciting cue value of aggressive slides reconsidered in a personological perspective: The weapons effect and irritability. European/ournal of Social Psychology, 14, 131-322.
Cline, V. B., Croft, R. G., & Courrier, S. (1973). Desensitization of children to television violence. Journal of Personality and Social Psychology, 27, 360-365.
Comisky, P., & Bryant, J. (1982). Factors involved in generating suspense. Human Communication Research, 9, 49-58.
Comstock, G., Chaffee, S., Katzman, N., McCombs, M., & Roberts, D. (1978). Television and human behavior. New York: Columbia University Press.
Comstock, G., & Paik, H. (1991 ). Television and the American child. New York: Academic Press.
Comstock, G., & Strasburger, V C. (1990). Deceptive appearances: Television violence and aggressive behavior. journal of Adolescent Health Care 11 (1 ), 31-44.
Donnerstein, E., & Berkowitz, L. (1981 ). Victim reactions in aggressive erotic films as a factor in violence against Women. Journal of Personality and Social Psychology, 41(4), 710-724. Downs, A. C. (1990). Children's judgments of televised events: The real versus pretend
distinction. Perceptual and Motor Skills, 70, 779-782.
Dubanoski, R. A., & Kong, C. (1977). The effects of pain cues on the behavior of high and low aggressive boys. Social Behavior and Personality, 5, 273-279.
Eron, L. D. (1982). Parent-child interaction, television violence, and aggression of children. American Psychologist, 37, 197-211.
Feshbach, S. (1976). The role of fantasy in the response to television. Journal of Social Issues, 32, 71-89.
Feshbach, N. D., & Roe, K. (1968). Empathy in six- and seven-year-olds. Child Development 39, 133-145.
Frodi, A. (1975). The effect of exposure to weapons on aggressive behavior from a cross-cultural perspective. International Journal of Psychology, 10, 283-292.
Geen, R. G. (1975). The meaning of observed violence: Real vs. fictional violence and consequent effects on aggression and emotional arousal. Journal of Research in Personality, 9, 270-281.
Geen, R. G. (1981 ). Behavioral and physiological reactions to observed violence: Effects of prior exposure to aggressive stimuli. Journal of Personality and Social Psychology, 40, 868-875. Geen, R. G. (1994). Television and aggression: Recent developments in research and theory.
In D. Zillmann, J. Bryant, & A. C. Huston (Eds.), Media, children, and the family(pp. 151-162). Hillsdale, NJ: Lawrence Erlbaum.
Geen, R. G., & Berkowitz, L. (1967). Some conditions facilitating the occurrence of aggression after the observation of violence. Journal of Personality, 35, 666-676.
Geen, R. G., & Stonner, D. (1973). Context effects in observed violence. Journal of Personality and Social Psychology, 25, 145-150.
Geen, R. G., & Stonner, D. (1974). The meaning of observed violence: Effects on arousal and aggressive behavior. Journal of Research in Personality, 8, 55-63.
Geen, R. G., & Rokovsky, ). J. (1973). Interpretations of observed aggression and their effect on GSR. Journal of Experimental Research in Personality, 6, 289-292.
Gerbner, G., Gross, L., Morgan, M., & Signorielli, N. (1980). The "mainstreaming" of America: Violence profile no. 11. Journal of Communication, 30(3), 10-29.
Gunter, B. (1983). Personality and perceptions of harmful and harmless TV violence. Personality and Individual Differences, 4, 665-670.
Gunter, B. (1985). Dimensions of television violence. Aldershot, England: Gower Publishing Company Limited.
Gunter, B., & Furnham, A. (1984). Perceptions of television violence: Effects of programme genre and type of violence on viewers' judgements of violent portrayals. British Journal of Social Psychology, 23, 155-164.
Hawkins, R. P. (1977). The dimensional structure of children's perceptions of television reality. Communication Research, 4, 299-320.
Hearold, S. (1986). A synthesis of 1043 effects of television on social behavior. In G. Comstock (Ed.) Public communication and behavior (Vol. 1, pp. 65-133). San Diego, CA: Academic Press.
Hoyt, J. L. (1970). The effect of media 'justification' on aggression. Journal of Broadcasting, 6, 455-464.
Huesmann, L. R. (1986). Psychological processes promoting the relation between exposure to media violence and aggressive behavior by the viewer. Journal of Social Issues, 42, 125-139. Huesmann, L. R., & Eron, L. D. (1986). Television and the aggressive child: A cross-national comparison. Hillsdale, NJ: Erlbaum.
Huesmann, L. R., Eron, L. D., Klein, R., Brice, P, & Fischer, P (1983). Mitigating the imitation of aggressive behaviors by changing children's attitudes about media violence. Journal of Personality and Social Psychology, 44, 899-910.
Huesmann, L. R., Eron, L. D., Lefkowitz, M. M., & Walder, L. O. (1984). Stability of aggression overtime and generations. Developmental Psychology, 20, 1120-1134.
Huesmann, L. R., Lagerspetz, K., & Eron, L. D. (1984). Intervening variables in the TV-violence-aggression relation: Evidence from two countries. Developmental Psychology, 20, 746-775.
Jo, E., & Berkowitz, L. (1994). A priming effect analysis of media influences: An update. In J. Bryant & D. Zillmann (Eds.) Media effects: Advances in theory and research (pp. 43-60), Hillsdale, NJ: Lawrence Erlbaum Associates Publishers.
Leyens, J-P Cisneros, T., & Hossay, J-F. (1976). Decentration as a means for reducing aggression after exposure to violent stimuli. European Journal of Social Psychology, 6(4), 459-473.
Leyens, J-P., & Parke, R. D. (1975). Aggressive slides can induce a weapons effect. European Journal of Social Psychology, 5, 229-236.
Lieberman Research, Inc. (1975). Children's reactions to violent material on television: 5th year report. Unpublished manuscript, American Broadcasting Company.
Liebert, R. M., & Schwartzberg, N. S. (1977). Effects of mass media. Annual Review of Psychology, 28, 141-173.
Linz, D., Donnerstein, E., & Penrod, S. (1984). The effects of multiple exposures to filmed violence against women. Journal of Communication, 34(3), 130-147.
Liss, M. B., Reinhardt, L. C., & Fredriksen, S. (1983). TV heroes: The impact of rhetoric and deeds. Journal ofApplied Developmental Psychology, 4, 175-187.
Meyer, T. P (1972). Effects of viewing justified and unjustified real film violence on aggressive behavior. Journal of Personality and Social Psychology, 23, 21-29.
Mueller, C. & Donnerstein, E. (1977). The effects of humor-induced arousal upon aggressive behavior. Journal of Research in Personality, 11, 73- 82.
National Television Violence Study (1997). Thousand Oaks, CA: Sage.
Ogles, R. M., & Hoffner, C. (1987). Film violence and perceptions of crime: The cultivation effect. In M. L. McLaughlin (ed.) Communication Yearbook 10 (pp. 384-394) Thousand Oaks, CA: Sage.
Osborn, D. K., & Endsley, R. C. (1971). Emotional reactions of young children to TV violence. Child Development 42, 321-331.
Page, D., & O'Neal, E. (1977). 'Weapons effect' without demand characteristics. Psychological Reports, 41, 29-30.
Paik, H., & Comstock, G. (1994). The effects of television violence on antisocial behavior: A meta-analysis. Communication Research, 21, 516-546.
Potter, W. ). (1988). Perceived reality in television effects research. Journal of Broadcasting & Electronic Media, 32, 23-41.
Potter, W. ). (1997). The problem of indexing risk of viewing television aggression. Critical Studies in Mass Communication, 14, 228-248.
Potter, W. J. (1999). Media violence. Thousand Oaks, CA: Sage.
Potter, W. J., & Berry, M. (1996). Television viewers' perceptions of the seriousness of violence. Paper presented at International Communication Association.
Potter, W. J., & Ware, W. (1987). An analysis of the contexts of antisocial acts on prime-time television. Communication Research, 14, 664-686.
Potter, W. J., & Warren, R. (1998). Television aggression: Humorous and serious consequences. Journal of Communication, 48(1 ), 157-176.
Rosekrans, M. A., & Hartup, W. W. (1967). Imitative influence of constant and inconsistent response consequences to a model on aggressive behavior in children. Journal of Personality and Social Psychology, 5, 424- 434.
Sander, I. (1995). Now violent is TV violence? An empirical investigation of factors influencing viewers perceptions of TV violence. Paper presented at the International Communication Association Conference, Albuquerque, NM.
Speisman, J. C., Lazarus, R. S., Mordkoff, A., & Davison, L. (1964). Experimental reduction of stress based on ego-defense theory. Journal of Abnormal and Social Psychology, 68, 367-380. Tannenbaum, N, & Gaer, E. F (1965). Mood changes as a function of stress of protagonist and
degree of identification in film-viewing situation. journal of Personality and Social Psychology, 2, 612-616.
Thomas, M. H., & Drabman, R. S. (1975). Toleration of real-life aggression as a function of exposure to televised violence and age of subject. Merrill-Palmer Quarterly, 21, 227-232. Thomas, M. H., & Drabman, R. S. (1978). Effects of television violence on expectations of
other's aggression. Personality and Social Psychology Bulletin, 4, 73-76.
Thomas, M. H., & Tell, P. M. (1974). Effects of viewing real versus fantasy violence upon interpersonal aggression. Journal of Research in Personality, 8, 153-160.
Turner, C. W., Layton, J. F., & Simons, L. S. (1975). Naturalistic studies of aggressive behavior: Aggressive stimuli, victim visibility, and horn honking. Journal of Personality and Social Psychology, 31, 1098-1107.
Walters, R. H., Parke, R. D., & Cane, V. A. (1965). Timing of punishment and the observation of consequences to others as determinants of response inhibition. Journal of Experimental Child Psychology, 2, 10-30.
Wilson, B. J., Kunkel, D., Linz, D., Potter, J., Smith, S. L., Blumenthal, E., & Berry, M. (1997). Violence in television programming overall: University of California, Santa Barbara Study. In National Television Violence Study (pp. 3 - 204). Thousand Oaks, CA: Sage.
Zillmann, D. (1971). Excitation transfer on communication-mediated aggressive behavior. Journal of Experimental Social Psychology, 7, 419-434.
Zillmann, D. (1976). Excitation transfer in communication-mediated aggressive behavior. journal of Experimental Social Psychology, 7, 419-434.
Zillmann, D. (1979). Hostility and aggression. Hillsdale, NJ: Erlbaum.
Zillmann, D. (1982). Television viewing and arousal. In D. Pearl, L. Bouthilet, & J. Lazar (Eds.), Television and behavior: Ten years of scientific progress and implications for the eighties, volume fl, Technical reviews (pp. 53 - 67). Rockville, MD: U. S. Department of Health and Human Services.
Zillmann, D. (1991). Television viewing physiological arousal. In ). Bryant & D. Zillmann (Eds.), Responding to the screen (pp. 103-133). Hillsdale, NJ: Erlbaum.
Zillmann, D., & Cantor, J. R. (1977). Affective responses to the emotions of a protagonist. Journal of Experimental Social Psychology, 13, 155- 165.
W. James Potter (Ph.D., Indiana University 1979) is Professor of Communication at Florida State University.
Stacy Smith (Ph.D., University of California at Santa Barbara, 1999) is Assistant Professor in the Department of Communication, Michigan State University.