Introduction

The aim of this paper is to shed some light on the different ways in which men and women are represented in violent episodes broadcasted on television. This analysis would contribute to an important debate in women studies: whether women’s media representation should be similar in terms of quantity as well as quality to men’s in the case of antisocial behaviors, like acts of aggression. The choice would be between fostering potentially negative roles for women and contributing to maintain the traditional feminine passivity. In this sense, it would be interesting to ascertain whether women are becoming more present, and/or whether the images of women as aggressors or victims are changing, in relation to up to date findings and common knowledge in the field. This research can also contribute to assess the extent to which gender differences are similar in different cultural contexts, in this case, Spain. Through the use of a content analysis methodology, several contextual variables will be used to try to further our knowledge of gender representations on television such as, the characterization of actors and victims, the characterization of violent actions, the consequences for the victim (harm done), as well as the consequences of the action for the perpetrator and the legitimacy or illegitimacy with which the violent action is presented.

Feminist scholars have always been worried about the socializing and normalizing effects of the differential media representation of men and women. As far as gender differentiation is concerned, media assign a different value to men and women, which could, in turn, have a socializing role (Gerbner and Gross 1976). In other words, it could influence people by framing (Goffman 1974) their experience and understanding of reality and result in greater difficulties when it comes to eradicating gender discrimination and to disseminating new models of gender identification. In different cultural contexts (USA, Europe, including Spain), women are underrepresented in cultural products such as books or children’s stories (Anderson and Hamilton 2005;Hamilton et al. 2006), in textbooks (Garreta and Careaga 1987;Oskamp et al. 1996) and also in audiovisual media (Goffman 1979; Frith and Mueller 2003; Ganahl et al. 2003; Reichert and Carpenter 2004), since the public sphere (the most widely portrayed in the media) is still perceived as something predominantly male. Moreover, the portrayals and images of men and women as put across by the media are clearly different. Different studies show that men’s characters tend to be more complex than women’s. In general, female characters are more likely to be placed in the private realm, while males are more likely to be shown in the public one. Women are shown as emotional, dependent, young, sexy, dominated by men. Men characters, in contrast, tend to be shown as knowledgeable, independent, powerful, successful, and tough (for reviews see Busby 1975; Cantor 1980; Coltrane and Messineo 2000; Chiricos et al. 1997; Davis 1990; Fejes 1992; López Díez 2001, 2005). As a result, women’s social activity is rendered invisible.

These representations of gender are generally consistent with traditional stereotypes, which nurture the oldest stories and cultural myths. In the case of violence, women tend to be represented as passive agents (seee below), in line with the alleged trait of passivity traditionally atributed to women (Fernández-Villanueva 2000). The concern with gender role models stems from the fact that the exclusion of women from certain models would reduce their opportunities to accede to certain social fields. In fact, the story of women, in Spain as in other parts of the world, has been one of a progressive but difficult access to different social spheres (see for Spain, Fernández-Villanueva et al. 2003). And this process has been facilitated by the emergence of new role models for women.

But what if the role models presented were not very positively evaluated by society, as is the case with aggressive behavior? In this case, equality in gender representation (women as present as men in any context of violence), would mean focusing on women’s violence and avoiding to reproduce women’s stereotypical passivity (Brush 2005). Nonetheless, it would also mean that the media would be portraying and fostering women’s violent behavior. In fact, much of the masculine involvement with media violence is composed of negative characters and models, like criminals, anti-heroes, monsters, etc., while women are spared of being depicted in this way, at the expense of not being represented as active participants in social conflict.

Media representation of men is, then, plagued with positive and negative violent models. The main archetypes of masculinity (the heroe, the patriarch and the monster; Gil Calvo 2006) can be related to violence: a) the heroe who, through the use of violence, overcomes the challenges placed on him by the community, thereby gaining social respect and power; b) the patriarch who uses violence to protect and dominate over the components of his group; c) the monster who exerts violence brutally or illegitimately.

Nonetheless, this traditional representation of masculinity is in crisis, as masculinity itself is in a difficult process of reconfiguration (see for Spain, Gil Calvo 1997). Patriarchal authority has been put into question and masculine figures of authority are denounced for the brutality of their violence. Even patriarchs become monsters when they use violence against the people they are supposed to protect without justification, as is the case in domestic violence (Gil Calvo 2006).

But media can contribute to reducing violence by broadcasting non-violent messages and models, accepting only those forms of violence that seem legitimate, regardless of the perpetrator’s gender, and showing that “violence does not pay”.

In Spain, as a result of both, recent national history and the questioning of traditional male dominance, news broadcasts and other reality programs (magazines, etc.) generally share a predominantly anti-violent theme and are mostly sympathetic to victims of violence. This can be traced back to the rejection of the Civil War -1936/39- (Díez-Nicolás 1999) and the horrors of war in general, that were brilliantly represented in Picasso’s “Gernika” (1937), or earlier, in Goya’s series “The Disasters of War” in nineteenth-century Spain.

These elements have given rise to the “pacifism of Spanish society” thesis. Nowadays, Spanish media are especially concerned about victims of terrorism (Basque separatism or Al Qaeda’s) and women victims of male domestic violence (each single murder of a woman is highlighted on TV news programs) and tend to reject major forms of violence. The presence of crime in Spain is not especially high, with an average crime rate of 50.7 crimes per 1000 inhabitants in 2006, well below the 69 per 1000 crime rate of the European Union (Ministerio del Interior 2006) and that of the United States (484,3; FBI 2007), and only 3.35 murders per 100,000 people (5.9 murders per 100,000 individuals in the USA in 2007; FBI 2007). On the other hand, gender equality is in the political agenda, and, therefore, so is in the media’s: laws are being passed and actions being taken to reduce gender inequality in the public and private realms.

However, in Spanish media, much like in other European countries (see for instance, Mustonen and Pulkkinen 1993), there is a very important presence of foreign fiction, mostly of US origin, especially in films but also in series and soap operas. This means that messages coming from a very different and certainly more violent society are included, such as views justifying war (Kendrick 1994; Iyengar and Simon 1993), for instance. This is not the case of Spanish fiction series, which can incorporate more contextualized scripts and messages.

For this study, the main variables will be those related to how men and women are represented as aggressors and victims of violence. An aggressive act is any intentional behavior causing harm, this being understood as any diminishing in the integrity of any of the agents. Studies of television violence based on content analysis have analyzed gender differences in television violence mainly from the perspective of the greater or lesser presence of men and women as aggressors and as victims. In the USA, Wilson and collaborators (Wilson et al. 1997; Wilson et al. 1998) found that around 75% of aggressive acts were carried out by men, while only around 10% of aggressors were women. Figures regarding victims were very similar, as these were mostly male (also around 75%). Still in the United States but from a different perspective, Eschholz and Bufkin (2001) found that in the most watched films in 1996, 65% of violent main characters were men. In the United Kingdom, Gunter and Harrison (1998) differentiate between aggressors and victims, both as individuals and in groups. They calculated that 62% were individual male aggressors, 14.5% were male aggressors in groups, 9% were individual female aggressors, an insignificant percentage were female aggressors in groups, and 3% were aggressors in mixed groups. Of the victims, 60% were individual males, 12% were males in groups, 9% were individual females, 6% were in mixed groups, and there were hardly any cases of females in groups. Although there is a lack of comparable data regarding TV violence in Spain, it can be hypothesized, based on general representation of women in Spanish media and cultural products (see above), that a similar uneven distribution will be found, based on data on general presence of men and women in news programming (López Díez 2001, 2005). Consequently, women’s active participation in violence is somehow minimized, which means that men keep their power and dominance over them, failing to acknowledge women’s potential to respond to aggression (Brush 2005; Fernández-Villanueva 2000). Also, groupness is an important issue when studying violence, so the first pair of variables will identify aggressors and victims in terms of gender and number (individual vs. group), in line with Gunter and Harrison’s work.

Nonetheless, it is important to use other variables to obtain a more complex view of gender differences in TV violence, and to get a better picture of the context in which aggressive acts take place. The importance of the use of contextual variables in content analysis has been expressed, among others, by Wilson et al. (1997). In their case, variables such as the justification of violence, the explicitness of images, the means used, rewards or punishments for the aggressor, the damage caused to the victim, the presence of humor, etc., are used, and these are relevant because they can modify the effect of violence on viewers.

However, from this perspective, the selected variables should consider some different aspects. First, the characterization of actors and victims should be considered. That is, the different roles that men and women perform in the context of violence, which is a fundamental issue in order to discover the different portrayals of men and women in the media. In their codification of TV violence, Gunter and Harrison (1998) used a pair of nominal variables that include categories as general categories (situation character, member of the public), different types of law enforcers and criminals, professions and entertainment, although they do not provide gender differences. These variables allow a rich analysis of the main roles of aggressors and victims and, therefore, have been used and adapted to our analysis. In this regard, it will be important to ascertain whether women as aggressors and/or as victims continue to be predominant in private (everyday and family) contexts, as has been traditionally the case (see for Spain Fernández-Villanueva et al. 2003), as against public (professional and law breaking and enforcement) contexts.

Second, the consequences for the victim (harm done) should also be considered. There are important discrepancies in the literature concerning this variable. Some authors prefer to consider only physical aggression, including threats and attempted aggression (Gerbner et al. 1980; Wilson et al. 1997), while others also take into account other types of aggression, such as verbal, psychological or against property (Donnerstein et al. 1994; Mustonen and Pulkkinen 1993; Williams et al. 1982). Focusing on physical aggression has the advantage of clarity and lack of ambiguity in such behaviors, but other important forms of aggression present in society and shown on TV are neglected, and they are even crucial to understanding physical aggression. As a result, a broader concept of aggression seems preferable and will be adopted for this analysis. This means that the types of aggression to be taken into account will be physical, as well as social, psychological, and against property (see below).

This question is important, since enormous differences have been found between men and women in this regard. In contexts of actual violence, male violence has been found to be mostly direct or physical, and female, mostly indirect or social (see, for instance, Olweus 1999; Björkqvist 1994; in Spain, Ortega Ruiz and Mora-Merchán 2000). Although the concepts of indirect violence and social violence are different (see Archer 2001), their similarities allow us to make a similar prediction in this case.

Third, there is a need to consider the consequences for the aggressor in a wider perspective, as this dimension is normally analyzed from a social learning approach, where rewards and punishments are thought to have important effects on spectators (see for instance Wilson et al. 1997). More than just a reward or punishment, the normative character of the violent episode is expressed through the consequences (positive, negative, both or none) of the aggressive act. If these consequences are positive, television is consistently showing the effectiveness of violence, thereby highlighting its logic, the reasons for it and its power as a resource to achieve goals. Something similar happens when there are no consequences. This category, far from being neutral, has a positive component, because, although the aggressor does not complete any objective, or obtain any particular benefit, neither does he suffer negative consequences of his act of violence, such as censure, criticism or reproach. Therefore, this ambiguity of consequences is an important matter. With ambiguity of consecuences we mean a double-sided effect, positive and negative at the same time, i. e., violence can achieve certain goals, but at a cost. This category contributes to reveal the complexity of television discourses. The literature in this precise subject is scant, but one would anticipate that, in line with the traditional representation of women as passive agents, women’s violence would produce worse consequences than men’s, as a sort of punishment for daring to transgress their expected role.

Fourth, it is important to consider the extent to which violent actions are seen as acceptable or legitimized. This aspect is especially relevant if we take into account the socializing role of television, since legitimized actions, those considered acceptable and justified, are the ones most likely to be learnt, imitated or internalized (Zelditch 2001). It is widely acknowledged that the reasons, intentions, motives, ends, or justifications offered for violence should have a crucial influence on viewers’ perceptions and behaviors (Paik and Comstock 1994). The problem is that there is no agreement regarding the most suitable way to study these phenomena. Wilson et al. (1997), and Wilson et al. (1998) talk of the reasons for violence (personal gain, protect one’s own or someone else’s life, anger, etc.). Potter et al. (1997) distinguish between maliciousness, and inconsiderate intent. Gunter (1985) distinguishes between positive (self-preservation, upholding the law, protect family, home or society) and negative (evil/destructive, ambition/power, desire for money) motives. Mustonen and Pulkkinen (1997) use the concept of justification, so they classify aggressive acts into: justified (defensive, externally motivated, and spontaneous), and unjustified (offensive, planned, and internally motivated).

One further problem is that most of these classifications are based on an attempt to penetrate the mind of aggressors in order to reveal their motives, intentions, reasons, or justifications, which is not always possible with the information supplied in the broadcast. Furthermore, it is also possible that there will be opposing justifications: what might be justified by the aggressor might not be justified by the victim or by the larger society, as Potter and Ware (1987; as cited in Potter 1999) have suggested. Consequently, it would be more useful to consider that the key to the significance that violence can have on viewers lies in the intentions of meaning of the specific program. In other words, it is not that the aggressor should consider his actions to be justified, or that the victim should consider that they are not, but rather to observe the audio-visual keys provided so that a particular interpretation of what is happening can occur.

The concept of legitimation is proposed for this approach. Legitimation can be understood as ‘a process by which cultural accounts from a larger social framework in which a social entity is nested are construed to explain and support the existence of that social entity, whether that entity be a group, a structure of inequality, a position of authority or a social practice’ (Berger et al. 1998, p. 380). As a result, legitimation is an evaluative process through which something is seen as correct and proper (Tyler 2006). Legitimation has to do with fair play, with doing the right thing (Tyler 2000, 2001, 2006; Tyler and Huo, 2002) and with morality. As Kelman (2001) asserts, “legitimization entails acceptance of a claim or a claimant into the domain of moral acceptability or moral obligation”. Legitimations may range from understanding and logical explanation, to moral justification. In this vein, Felson (1996) pointed out how television conveys the message that some forms of violence are necessary and legitimate while others are evil. Ball-Rokeach (1972) also noted the importance of connecting violence on television and its effects with the legitimacy or illegitimacy with which it is presented. It should be expected that mass media increase the likelihood of the audience accepting or rejecting (legitimating or delegitimating) the broadcasted conduct or social practice.

Unfortunately, this analytical approach of analyzing violence on television through the concept of legitimation has seldom been applied by the content analysis perspective of television violence. The only admirable exceptions found are a French report published by the Centre Superieur de l’Audiovisuel (CSA 1995), which uses the concept of legitimation without much conceptual development, as well as some studies that have performed qualitative analyses of the ideological messages emerging from certain programs, especially news programs (Kendrick 1994; Iyengar and Simon 1993; First 1998).

As a consequence, it would be interesting to identify cues in the broadcast which aim to lead the viewer to accept the action viewed as legitimate or reject it as illegitimate. Aggressive behavior in everyday life “always” requires justification, at least in Western society, where violence is expected to be monopolized by the State (Weber 1922). For this reason, when violent behavior is broadcast on audio-visual media, it is accompanied by cues that anticipate possible questioning and guide the viewer towards the same line of thought that caused the involved character to think of this idea or carry out this behavior. Thus, we will be analyzing the broadcaster’s “claims to legitimacy” (Kelman 2001), by which we broadly mean here any person or group responsible for agreeing on the way in which such behavior has been broadcast in a specific medium, in this case television.

Therefore, only an examination of all the elements used and appearing on the screen can give us an idea of what the message is meant to be. Images and words construct meaning, and together lend weight to an account that the audience is expected to accept. In this regard, specific legitimation mechanisms can be classified into mechanisms presenting actors (aggressors and victims), and those presenting violent actions (Fernández-Villanueva et al. 2004). It will be also interesting to ascertain whether these legitimation trends apply differently to the private and public contexts, and thus this issue will be explored through the cross-tabulation of the legitimation variable and the characteristics (roles) of aggressors and victims.

As this perspective is a new one, it is hard to predict the gender differences that will be found and the specificities of the Spanish media in this respect, but it is our wish that this analysis will help to enrich the picture. However, if men tend to be the sanctioned agents of violence, it could be expected that their violence would be legitimized to a greater extent. Furthermore, the above discussion on the questioning of masculine domination could influence these results in the direction of violence against women be less legitimized than that against men.

To sum up, the following predictions can be hypothesized: (1) TV violence will be predominantly male, so we expect a significantly higher percentage of male aggressors and victims (above two thirds of total acts of aggression), compared to female ones, as has been found in previous research. Similarly, female group violence should be rarer than male. At the same time, (2) women are expected to be significantly more present as victims than as aggressors, as an indicator of their traditionally female role of weak, passive actors. Moreover, the roles that male and female actors perform in the violent contexts will resemble the traditional gender division of roles, which means that (3) men will be significantly more present than women in the public sphere than in the private one, both as aggressors and as victims. In the case of women, a more numerous presence of women as victims is expected.

Regarding the normative context of violence, (4) a significantly higher percentage of male physical violence is expected, while women are predicted to make a significantly more frequent use of social violence. On the other hand, (5) better consequences for violent men are expected compared to women. (6) In the same vein, men’s violence is expected to be more legitimized than women’s. These predictions are based on the expectation that men are the sanctioned actors of violence, so women exerting violence will be more punished than rewarded, and their behavior will be less tolerated. Seemingly, Spanish media should show a higher concern with victims, especially female ones. This could be appreciated (7) in a significantly higher percentage of negative consequences when the victim is a woman than when it is a man, as well as (8) a significantly higher percentage of de-legitimating of violence against women.

Although these predictions may not seem too surprising, this research contributes to media violence studies with a thorough analysis of legitimating claims and of consequences to the aggressor that is unique in the literature. This allows further discussion on the consequences that different gender representations broadcasted on TV could have on the population.

Method

A method and a device for collecting information more appropriate to the peculiarities of television broadcasts were constructed. The methods used in previous studies seem too dependent on “programs” as the unit of analysis. A main feature of TV broadcasting is the fragmentation of programs by including publicity, promos of other programs on the same channel, brief news reports, etc. The fragmenting of TV broadcasts is such an important element that it cannot currently be ignored. Consequently, our collection of data, rather than choosing the kind of program shown, concentrates on a concrete period of time, perhaps comprising several program types. In short, the analysis unit is a 15–minute period, referred to as a “segment”. This method reflects more accurately current habits of watching TV, marked by “zapping” and consequently producing a fragmented reception of the transmitted messages.

On the other part, studying these issues in Spanish television can prove interesting, due to the characteristics of its programming. Until as recently as 1990, there were only two television stations in Spain, both of them State-owned. Since then, several private-owned stations have achieved a very important role in the broadcasting system, while several public regional stations also began broadcasting. These public stations combine commercial and non-commercial programming.

Data for this research project were taken at random from recordings of broadcasts by the most watched Spanish television channels in 2000 and 2005 (audience mean superior to 5%). Segments were randomly selected to make them representative without accumulating a vast amount of material making analysis impracticable. The day was divided in 4 time-periods of 6 h (08:00–14:00, 14:00–20:00, 20:00–02:00 and 02:00–08:00) based on audience rates, as measured by Taylor Nelson Sofres (TNS) digital-based audience measurement panel, and viewership habits in Spain. Audience rates are highest in the evening, with peaks of around 40% of the sample, and lowest in the night or early morning, around 1%. Thus, every week consisted of 28 periods, from which we randomly selected a segment of 15 min for each of the days of the 4 non-consecutive weeks we chose in a period of 4 months. A 15-minute segment was recorded in every weekday in each of the time periods. Time periods were consecutively assigned to each of the 28 days, starting from the first period up to the fourth, then back at one again, and so forth. In 2000 we started on Monday with the first period and in 2005 with the second. Next, each time zone was split in 24 sections of fifteen minutes and a number from a random number table was selected to choose the exact quarter of an hour to be recorded in each of the 28 days.So, 28 random numbers were needed, one for each day of the four weeks. The same quarter and time zone was used for each of the channels, so the recordings took place each day at the same time for the total number of channels. In 2000, a total of 28 fifteen-minute broadcasts by each of the four major then free-to-air channels, plus the local channel serving the autonomous region of Madrid, were randomly recorded between February and May at the rate of one per day during four non-consecutive weeks, making 140 fragments in all. In 2005, 28 segments were recorded in the same way, also between March and May, from each of the same four national channels as well as from the three biggest regional channels in terms of target population, namely those serving Madrid, Catalonia and Andalusia, making 196 fragments in total. All told, this provided a total of 84 h of recorded material (35 from 2000 and 49 from 2005) which were then analyzed. A decision was taken not to compare between years 2000 and 2005, in order to strengthen our data set and give reliability to our results.

From this material, we selected those episodes and aggressive acts that contained violence, according to the definition by Fernández-Villanueva et al. (2006). Each one of these episodes (minimal units of meaning) may contain several aggressive acts, which are intentional behaviors causing harm, as stated before. A different aggressive act was coded when there was some kind of discontinuity with the previous act, in terms either of aggressor, victim, type of harm or outcome. In contrast to other studies of television violence, this includes not only physical violence but all types of violence.

Next, coding was carried out. In order to test intercoders’ reliability in the identification of violent acts, we proceeded in the same way as other researchers who have analyzed TV violence (Mustonen and Pulkkinen 1997; Wilson et al. 1997, Wilson et al. 1998), namely, we made the five coders identify all the violent acts per segment according to our definition of violent act, and then classify them according to our coding sheet. To be precise, coder 1 selected all the violent acts in the whole sample; and each of the rest did the same for one station. Of all the violent acts 86% were selected by two coders. The remaining 14% of the cases were discussed to consensus. The reliability of our coding table was tested pairwise on the total sample of violent acts. Since the coding table is similar to the one used by the NTVS study, the percentage of intercoder agreement was applied as a reliability test. The percentage of intercoder agreement was computed by adding the number of times each pair of coders agreed in their coding, divided by the total number of variables coded (Wilson et al. 1998, p. 215).

The mean intercoder agreement for all the variables was 87%. The reliability coefficient for coder 1 vs. coders 2, 3, 4 and 5 for the 23 different variables in the coding table were as follows: all variables concerning general program information had a 100% reliability (year 100%, reference code 100%, amount of acts 100%, TV station 100%, time slot 100%, type of TV program 100%); variables with reference to the act had a mean of 89% (characteristics of aggressor 82%, genre, 89%); mean for variables referring to the victim was 85% (characteristics of the victim 78%, genre of the victim 83%); mean for variables relating to the consequences of violence 78% (degree of damage 86%, consequences to the aggressor, 80%), mean for variables referring to legitimacy of violence 75% (legitimized by type of aggressor 79%; legitimized by type of victim, 76%, legitimized by type of action, 71%), mean for variables concerning delegitimacy of violence 82% (delegitimized by type of aggressor, 83%, delegitimacy by type of victim, 86%, delegitimized by action, 75%). Agreement between coders was also calculated and it ranged from 92% (between coders 1 and 2) and 71% (between coders 1 and 4). This last relatively low figure could be due to a deficient training of that coder.

The characteristics of the final sample in terms of time recorded, number of aggressive acts and type of program (genre) can be seen in Table 1.

Table 1 Sample characteristics.
Full size table

In this study, we present the data relating to the agents of violence on television, aggressors and victims, as well as the normative context in which the violent acts takes place:

Aggressor and Victim

The aggressor is the person who intentionally inflicts harm on a victim. We include the concept of ‘intentionality’ in order to better analyze the “legitimacy claims” of the broadcaster, which would not be relevant in the case of accidents or natural disasters. We distinguish between human beings (a man, a woman, groups of men, groups of women and mixed groups), and other kind of actors, although here we are basically concerned with analyzing violence among human beings. Furthermore, different roles of aggressors and victims are identified (criminal, police officer, main character, member of the general public, in a professional or family role, etc.). The coding was made on the basis of identifying the main role the actors of violence were performing at the moment of the aggressive act to avoid the problem of multiple identities (and so multiple codification). As a consequence, each of the categories can be read as a private or public role and be analyzed accordingly.

Type of Harm

Different types of harm inflicted on victims were studied. The analysis of physical harm is fairly well defined in the scientific literature and we here distinguish among fatal harm, serious physical harm, and slight harm. Social harm has not always been taken into consideration, but from our point of view, it constitutes a form of violence and is fundamental when trying to present a complete picture of violence on television. We can define an act of social violence as one which causes harm to the social identity of the victim in the form of an insult (verbal abuse), disparagement or exclusion from a group. We also include detaining, any restriction of movement against the will of the victim if they prevent the subject’s normal interaction in relationships and the normal expression of their social identity. We also take into account damage to property, psychological harm (aimed at generating fear in the victim as a way of modifying their behavior) and symbolic aggression (destruction of symbols significant to the victim).

Consequences for the Aggressor

We considered the consequences that an act of aggression has for the aggressor. The categories applied are: positive consequences, negative consequences, ambiguous (positive and negative at the same time) and none (lack of reward or punishment). The consequences identified are mainly short term ones, which are those shown immediately after violent scenes. This coding strategy is similar to the one used in the NTVS (Wilson et al. 1997, 1998), who assessed the presence of immediate rewards or positive reinforcements for violence. Although, this is an under-researched question, the tendency towards fragmented viewing of broadcast products reinforces the value of these immediate consequences, as they are the ones finally perceived by the viewers.

Legitimation

Two sets of resources used in audiovisual media to fashion the broadcaster’s legitimation claims will be identified. Firstly, the characterization of the agents participating in the action: aggressor and victim. The cues that work towards the legitimation of the action of the aggressor are based on the attempt to establish an affective link between the viewer and the character, either because it is the main character in the broadcast and his/her behavior is a priori more likely to be accepted, given that he/she is shown with positive characteristics to encourage a closer identification with the viewer, or by making explicit a role that endorses violent behavior. The delegitimation of the aggressor would obviously be through a negative application of the same cues. In the case of victims, the resources would be similar, but with opposite effects. Establishing an attachment between the victim and the viewer would ease the rejection (deligitimization) of the act and the lack of it, on the other hand, its legitimation. Both the characterization of the character (positive/negative characteristics or those which foster closeness/rejection or identification/disidentification) and the visibility of the character serve this purpose. The identification of the character guides towards the rejection of the act of aggression, while anonymity eases acceptance.

Secondly, the way the action is presented guides towards different types of judgment. Hence, the order of the events leads the viewer to seeing the character who commits the first unprovoked act as guilty, with subsequent acts considered as responses to the original act. Alternatively, linking the action explicitly to a set of rules (the law), ethics (protection of the weak, obeying of generally accepted rules) or social contexts (rites, show business, sports) are again cues for accepting the act of aggression as legitimate, while the opposite induces rejection.

We have taken into account all the mechanisms of legitimation and delegitimation that appear in episodes of violence independently, in order to: a) describe the aggressor; b) describe the victim; and c) present the action carried out. In each case one or more cues can be identified for each act of aggression, and can even be contradictory with respect to making sense of the claim. The most frequently used legitimation or delegitimation mechanisms are those referring to the act (more than 40%) carried out by the aggressor, but arguments for legitimation or delegitimation referring simply to the perpetrator (more than 25%) or to the victims (around 30% approximately) can also be found.

Finally, we introduce a compound variable called legitimizing discourse, which is the one finally used for this analysis, that summarizes legitimation or delegitimation mechanisms used in each particular act. This variable has three categories: legitimate act (if it showed only examples of legitimation), illegitimate act (only delegitimation), or complex act (if it showed examples of legitimation and delegitimation at the same time).

These variables help us make a detailed analysis in an attempt to establish relationships between the aggressor and victim variables, and the rest of the variables in the normative context. The significance of the different cross-tabulation data is referred in the corresponding table (chi-square value and its probability –p-value). Chi-square was a valid test since in none of the analysis the threshold of a 20% of expected cell frequencies below 5 was reached.

Results

The Distribution of Men and Women as Aggressors and Victims

Data regarding involvement in violence are very conclusive. Confirming our Hypothesis 1 (at least two thirds of aggressors and victims would be men), violence on television is prototypically masculine. More than 75% (77.6%, see Table 2) of the aggressors are men, with women making up less than 15% (14.5%). The majority of victims are also male, basically acting individually (54.4%) or in groups (13.4%), and making up almost 70%. Female victims constitute 19.3%, mainly women on their own.

Table 2 Aggressors and victims of aggressive acts by gender and number.
Full size table

If we look at whether the origin of violence is individual or collective, we can see that group violence represents a relatively high percentage, around 30% of all cases. Groups of women aggressors are virtually non-existent, which renders their analysis virtually impossible in any aspect, although they do participate in some way in mixed groups, which represent 7.9% of human aggressors. As a consequence, it was decided to offer data on aggressive acts which have a group of women either as aggressor or victim, but they were not included in the calculation of statistical tests and its significance. So, chi-square, p-values and degrees of freedom are calculated without considering data on groups of women.

Relative to Hypothesis 2 (see Table 2), we can see that women (single and in group) appear more frequently as victims (19.3%) than as aggressors (14.5%), while men more often appear as aggressors (77.6%) than as victims (68.2%). This higher percentage of women victims of violence is significant, moderated by the high frequency of violence perpetrated by men (78%), and because of the relatively low number of cases of violence carried out by women (15%), as already mentioned. Nonetheless, it is a result that has not been highlighted in previous research, which usually finds a similar percentage of female aggressors and victims. These results show that a great proportion of violence is managed solely by men. This is why men are more frequently aggressors and more frequently also victims.

But how are men and women represented? The roles assigned to aggressors in terms of gender are very different (Table 3). It was predicted (Hypothesis 3) that men will be significantly more present than women in the public sphere than in the private one. As long as the public sphere is concerned, there are two well represented social fields: the maintenance or breakdown of social order and the professional world. Men generally appear to be more linked to the maintenance or breakdown of social order than women: 9.6% of men acting alone appear as police officers (26.2% of men in groups), as against 2.2% of women. Along the same lines, criminals make up 15.6% of men acting alone, against 3.1% (7 cases) of women acting alone. This difference is greater when we consider the roles of the army or terrorist groups, in which women only rarely appear. Therefore, in congruence with the percentage of male group violence shown before, a significant part of men’s violence is linked to collective violence (wars, terrorism) or State violence (for example, used by the police to maintain order). That is to say, we are dealing with instrumental violence, functional for some groups in pursuing certain objectives within the framework of social conflict. However, contrary to our expectations, there is hardly any difference in percentage terms as far as the portrayal of men and women as members of professions and jobs goes (between 8 and 9%). As these are also public roles, the percentage of male aggressors was expected to be higher.

Table 3 Characteristics and gender (aggressors and victims).
Full size table

Regarding the private sphere, women appear to a significant extent (55.8%) in more indeterminate roles as main characters or as famous people, as against 34.1% of individual men. People in this category appear more as individuals acting more in friendly or family-related contexts, and, therefore, more private than public. This confirms our predictions that women will be predominantly depicted in private realms. Also as expected, the presence of male aggressors in family roles is higher in total figures, although there is a percentage difference between them (between 4.9 and 6.6%) in favor of women.

In terms of the roles of victims, women again appear far more frequently in family roles (12.8% as against 9.8% of men) and in general private roles (main characters: 47.8% as against 38.3% of men; members of the public: 19% as against 9.8% of men). An important 13.5% of individual male victims are portrayed as criminals. Confirming our predictions, this implies a greater victimization of women in private spheres, as ordinary citizens, rather than as people playing positive (police officers, army) or negative roles (criminals, terrorist groups).

Normative Context of Male and Female Violence on Television

In the analysis of the normative context of violence, we include the consequences for the victim (harm), the consequences for the aggressor and the legitimation of violence. Given that these three elements are aimed at understanding the degree of acceptability granted to violence on television, the analysis of gender differences is highly relevant.

In relation to Hypothesis 4 (men aggressors use physical aggression much more frequently, while women use more frequently social violence), our data show that male violence has the most serious consequences (for the victim). Almost two thirds of violent acts carried out by men (alone or in groups) are physical (64.3%, see Table 4), while more than half of the female violence is of a social nature (53.9%).

Table 4 Type of harm and human aggressors and victims.
Full size table

A result that has not been highlighted is that men are mostly harmed physically (69.4%, see also Table 4), which is approximately in line with the percentage of harm inflicted by men as aggressors. Women, on the other hand, suffer more physical violence than the violence they inflict as aggressors.

Hypothesis 5 stated that consequences for male aggressors would more positive than those for female aggressors. In general terms, the use of violence is shown on television as being socially highly profitable, with almost 40% of the acts of aggression analyzed having positive consequences for the perpetrator. More than 18% have no consequences, almost 22% have ambiguous consequences and a little more than 22% have negative consequences (see Table 5).

Table 5 Consequences for aggressors and victims.
Full size table

Unexpectedly, the profitability or social functionality of violence seems to be even higher for women, since in the whole of our sample the consequences of the use of violence are more positive for female aggressors (44.9% in acts with positive consequences and 28.0% with no consequences), than for male aggressors (34.1% and 18.6% respectively) and more negative for male aggressors (24.7%) than female aggressors (17.3 %). But ambivalent consequences (positive and negative at the same time) are clearly more frequent in male violence, both individual and group (22.7% and 24.6%, respectively, as against 9.8% of the cases of women alone).

Along the same lines, and confirming Hypothesis 7, violence also has more positive consequences when the victim is a man than when it is a woman (38.8% as against 29.2%), while ambivalent consequences are much more similar.

In relation to Hypothesis 6, and although a higher legitimation for male violence was expected, female violence is more legitimized (51.8%, see Table 6) than male violence (35.4% when acting alone and 21.4% as part of a group). Gender difference diminishes a little if we take into account ambiguous legitimation, which is 6 points higher in males. We assume that qualifying an act of violence as ambiguous implies assigning some degree of rationality or understanding, some legitimizing action or reasoning. Furthermore, it means showing a more complex, less simplistic vision of the character.

Table 6 Legitimation/Delegitimation and human aggressors and victims.
Full size table

However, a different result is obtained if we cross-tabulate legitimation with the roles of male and female aggressors (see above). While women hardly appear as criminals, the opposite is true of men, with whom the deligitimization of the action or the aggressor and the negative consequences of the aggression are mostly associated. Nevertheless, the difference between men and women criminals is smaller than between men and women police officers, which is why this situation implies a greater deligitimization and negative consequences for male aggression, in itself a reflection of the high percentage of cases in which they appear as the “bad guys”. But a specific analysis of the category “main characters” (55.8% of women are in this category, as against 34% of men, and the number of cases is high enough to allow a separate analysis) is interesting, to the extent in which it is sufficiently general to allow us to see more clearly gender differences that are less conditioned by the specific role played in the program. In contrast with the global data, in this case the percentage of legitimized violence when the aggressor belongs to this category is greater among men than among women (50.6% as against 42.9%; see Table 6, bottom part), which coincides more clearly with the more typical gender differences in the literature.

Confirming our Hypothesis 8, violence against women is much more delegitimized than violence against men (60.9% as against 31.4%) and less legitimized (19.3% as against 40.0%). Therefore, it seems that women are normatively more protected against aggression, while they are also more legitimized to commit aggressive acts.

Discussion

The in-depth analysis of these data has yielded interesting results. First, the presence of women on television in violent roles is very low, in line with the literature on the subject. Gunter and Harrison’s data (1998), as shown before, are fairly similar to ours. Wilson and collaborators (Wilson et al. 1997; Wilson et al. 1998), Fejes (1992) and López Díez (2001, 2005) also refer equally to the prime involvement of males in television violence, as does Naylor (2001), who referred to male involvement in print media news.

The amount of violence used by men on television is more than three times higher in general (75% of all aggressive acts), six times higher in the case of physical violence (85%) and ten times higher in fatal violence (89%). Therefore, and in line with Fejes (1992), the television image of masculinity does not differ greatly from traditional stereotypes, which show men as being more involved in meting out violence and punishment. But although stereotypical, this gender distribution of violence is quite coincidental with that of real life. For instance, the US National Crime Victimization Study (NCVS 2005) finds that 90% of murderers are male. In Spain, more than 90% of convicted criminals are also male (INE 2007).

Second, the expected gender distribution of harm inflicted on victims has been found, as shown by the far greater percentage of cases of physical and fatal violence exercised by men, either acting alone or in groups. Women, in line with traditional images, use relational and social violence more frequently, which is less serious and has fewer consequences for the victim. Although in these contexts the number of cases of male violence is always higher, the above mentioned aspects are the most representative of female violence, its distinguishing features. Therefore, control over other people’s lives, the sign of power of physical force, is masculine. Again, this is in line with crime rates in real life (NCVS 2005), as men are more frequent aggressors and victims, due to the fact that social violence is hardly taken into account in this context. As mentioned before, this same pattern has also been found in other contexts of violence, such as school bullying (Olweus 1999).

Third, when they do use violence, however, women appear more often as aggressors (in percentage, not in total amount) in everyday contexts, portrayed with their own specific identity and as situation characters, in friendly or family contexts, confirming our expectations and that the traditional portrayal of women is dominant in TV violence. Consequently, female aggressors are more visible in program genres that remain closer to everyday contexts (series and magazines, especially). Part of this gender differentiation can be found in the different involvement of men and women in the enforcement and breaking of social rules. In the still very relevant mythic tale of the confrontation between good and evil, or law against crime, women are still outsiders; it is a job for men. This means equating heroic and violent masculinity (Katz 2003), but it also means that a most competitive scenery is depicted; thus, masculinity is construed as a competition between groups of equals: loyalty and solidarity within and violence without (Fernández-Villanueva et al. 1998).

However, there are reasons to believe that there is a certain process of making female violence invisible, in relation to the non-negligible presence of women in the army and police forces, in criminal groups, as well as in wars or terrorism, as highlighted by observers and war and terrorism researchers (Alcedo 1995; Coomaraswamy 1998; Fernández-Villanueva 2000, Höglund 2001), by institutions campaigning for peace (Panos Institute 1995) or by participants in armed conflict (Diana 1997). In these contexts, an emerging presence of women can be found in the violence exerted by mixed groups, some of which are portrayed as criminals or law enforcers.

Fourth, women appear more often as victims than as aggressors, so a certain victimization of women can also be found. Furthermore, the percentage of physical violence that men inflict as aggressors is somewhat smaller than what they receive as victims, while the same can be stated in relation to women and social violence, which means that women are victims of more serious harm. In terms of roles of victims, what accounts most for this result is the enormous difference between the number of female victims and aggressors in family roles. In fact, the portrayal of women in domestic violence highlights, perhaps excessively, the passive role played by women. Brush (2005) underlines the need to reassess the supposedly low involvement of women in domestic violence, and some data questions the “excessively passive” character of the victims of such violence. Figures from the U.S. Department of Justice (1998) show that among women who have suffered non-fatal male violence, 34% confronted the aggressor, although 23% did not put up any form of resistance.

This means that female violence is also being made invisible to a certain extent in this context. And this relative invisibilization not only reinforces the traditional unequal representation of women, but it is not even an adjusted representation of real violence, which is the main argument for this unequal presence of men and women in media violence. If it does not intend to be true to real violence, then one could ask media to foster new models of (active) women, as long as their violence makes sense, providing positive, active, albeit violent, models. This could a break in the equation of femininity with passivity (Katz 2003). In this sense, some authors argue that a new, more aggressive image of women is appearing, especially among the younger generation. This image treats female aggression of the “instrumental” type in a way similar to traditional male aggression, as in the case of women who are heroines by virtue of their fighting capabilities (“action women” such as Tank Girl or Lara Croft). However, it might be unlikely that this kind of new models will change the overall unequal gender representation in the media.

Fifth, and surprisingly, female aggressors get positive consequences more frequently than male aggressors. If the main message of TV violence is that women ought not to use violence and remain in their stereotype image of passivity (as victims to be protected by others), one would expect that women daring to transgress their role would be severely punished, or at least, more than their male counterparts. But this is not the case. Women are somehow encouraged, at least more than men, to become aggressors as a means to an end. As the analysis perspective followed in this study is new, it is impossible to contextualize these results, as this result is found in every kind of program. Consequently, further research on this issue is needed. Less surprising is that violence against women obtains negative consequences much more frequently, as a way to depict them as defenseless victims that must be protected by men. Another interesting result is that ambivalent consequences are much more prevalent in male aggressors. It could be said that men take the risk of a possible negative cost better than women, which would be coherent given their position of greater power and resources to face negative effects, or because of their greater capacity for questioning social order. It can also imply that men are portrayed as more complex, less simplistic and stereotypical than women.

Sixth and last, the results concerning the legitimation of violence are as interesting as they are difficult to contextualize. Female violence is more legitimized than male in general terms. But in this case, it is mainly men’s involvement in criminal violence that accounts for this difference. When comparing the legitimation of male and female violence when the aggressor is portrayed as a situation character (“main character”), the legitimation percentage of male violence is higher than that of female violence, unlike the results on consequences. This is especially surprising, as male violence is much more severe (physical). This point of view results in the expected gender differentiation, since women’s “normal” violence is less legitimized, instead of being less harmful. Therefore, in everyday contexts, the actions of women using violence tend to be more penalized than men, as women are not supposed to be aggressive, since it is men’s business. Consequently, those daring to go beyond their assigned role will be censured by showing the illegitimacy of their aggressive action.

From a different angle, this can be related to the dimension of irrationality, deviance, transgression or evil in female violence to which Naylor (2001) refers as dominant, as well as to the findings of Campbell and collaborators (Campbell et al. 1997; Campbell et al. 1999) which show female violence as an expression of mood, i.e. caused by a loss of control, by irrationality or senselessness, in contrast to male violence, which is shown as instrumental, a way to impose control on others, above all as a reaction to failing to reach goals.

However, an overwhelming result concerns the much higher delegitimation of aggressive actions when the victim is a woman, as compared to a man. This difference is much greater than the one found when comparing aggressors’ gender. As a consequence, it can be stated that women on television are also normatively protected from violence: they are especially not supposed to be the victims of violence whatever the context. Again, this result is in line with what could be expected from previous research: women, as an innocent character, must be protected from evil by men. But the previously stated characteristics of Spanish media in its concern to victims of violence may exacerbate this differential representation.

Conclusions

The panorama of violence on Spanish television, as far as gender is concerned, in general confirms what other studies had found. This very confirmation is significant in itself, and makes a noteworthy contribution to the literature since it comes from a particular cultural context (Spain), in a time when gender equality conditions have improved, and when television could be showing this change. In fact, in some other areas of social life, media amplify social tendencies that are still incipient. But this is not the case. The gender distribution of violence on television is mainly similar to statistics on actual violence, in terms of total percentages (violence is mainly male in origin), in terms of type of harm inflicted (male violence is physical vs. female being social), and in terms of victimization (women are, to a certain extent, more frequently victims than aggressors). Thus, television contributes to reinforce the world as it is, to the maintenance of an unbalanced representation of gender.

But, should media be responsible for changing society, especially commercial media? Or is it civil society and public institutions that should be accounted responsible for that? What at least one could expect is that media do not behave as a backward-looking institution. If media are to represent reality, they should not represent a reality that no longer exists.

There are two more relevant ways in which women are normatively protected from violence to a greater extent than men: violence against women is much more delegitimized and has worse consequences than violence against men. Since women are not expected to be aggressors, those who act against them are branded negatively.

However, there are reasons to believe that women’s violence is somehow underestimated in statistics on actual violence, as well as in media violence, with a failure to record the higher involvement of women in social life, in general, and violence, in particular. Female violence, therefore, remains invisible. Arguments in this regard have been provided before concerning two different contexts: law breaking and law enforcement, and the family. This does not mean that the involvement of men and women in violence in this context is equivalent or similar in nature, but that it is underreported. This could be seen by some as preferable from a gender equality perspective, but the result is that women are still depicted as powerless and passive, and not as powerful and active agents. As a consequence, the specific forms of women’s violence, surely less harmful than men’s, are hidden and go undebated. It is important to acknowledge the complexity of the issue, and further debate is needed in gender studies about the consequences of portraying women as powerful agents as against helpless victims. Brush (2005) indicates the “potential political costs” of raising these questions. However, what other ways are there to avoid depicting women as passive victims waiting for a hero man? Women should be represented as agents taking responsibility for their lives, which sometimes means using aggression.

In this sense, the normative context of violence gains relevance, as long as it adds important information to the picture, which has not been properly considered before. While in general women’s violence is less legitimized than men’s, it has more positive consequences. It is quite surprising that a behavior tends to be normatively rejected at the same time as it is more highly rewarded. The message from television, then, is that violence may not be the appropriate behavior for women, but those who actually use it benefit from its functionality and having used it. This may mean that television is portraying women as instrumental users of violence, although it tends to de-legitimize its use by women, reaffirming the traditional view of violence.

But it is difficult to answer whether this result is a consequence of a new portrayal of women’s aggressive behavior, or an aspect not yet considered in research. As a consequence, further research is needed to confirm or refute this result in similar or different contexts. Due to the lack of previous research on this issue, it is extremely difficult to ascertain if women’s violence had better consequences some time ago, when the move towards gender equality was less advanced than it is today.

Furthermore, the novelty of the randomizing methodology and of some of the variables used, recommend the replication of these findings in other contexts. The complexity of the categories identified also recommends the enlarging of the data set, as further cross-tabulating of variables would render most valuable information. On the other side, a growing part of TV broadcasting is out of this research, as it does not contemplate thematic channels or cable TV.