Introduction

Content Analysis of Gender Roles in Media: Where Are We Now and Where Should We Go?

The articles in these special issues of Sex Roles (Rudy et al. 2010a, 2011) and the information they provide represent a long stride forward in our understanding of the manner in which women are currently depicted in media (see Table 1 for a list of these papers). The studies cover a broad range of content and look at a variety of issues under the general theme of gender roles. In the context of this diversity, it is perhaps surprising that results converge as much as they do. When this body of work is viewed as a whole, some key messages emerge, as do some clear gaps in knowledge. While some of the contributions are theoretical or methodological, I will concentrate here on the substantive conclusion that most of them have in common: The studies make clear that women are under-represented in media, and that when women are present they are typically scantily dressed and relegated to stereotypical roles. While these patterns are clear, the articles also identify some variations in them that may be the starting ground for future avenues of research, including differences by race, sexual orientation, and nation. Below, I review and discuss the findings, noting some critical issues in interpreting and applying them, as well as pointing out where key gaps in our knowledge remain and providing suggestions for next steps in the field.

Table 1 Articles in the two special issues of Sex Roles and their coverage of select content
Full size table

Women are Under-represented

Under-representation Holds Across Multiple Media and Content Types

One message that emerges from these special issues, and does so with overwhelming clarity, is that women and girls are under-represented across a wide variety of media. Table 1 provides an overview of the articles included in the two special issues. In the table, articles are categorized to reflect coverage of each of several gender-role content areas: under-representation, sexualization, subordination, traditional roles, and body image. Of the 18 empirical articles appearing in the issues, half (nine) measure the frequency with which women versus men are portrayed. All of these find that women are under-represented in at least one content category. That women seldom appear on television is by no means a new finding (cf. Gerbner and Signorielli 1979). But it is noteworthy that the disparity in portrayals of males and females has persisted over decades, during which the roles of women in society have broadly expanded. Three decades ago, Gerbner and Signorielli (1979) found that television primetime males outnumbered females by a ratio of 2.5 to 1 in the years spanning 1969–1978. Producers and writers at that time might have argued that women did not serve in the societal roles television series wished to depict, constraining their ability to include them. Indeed, in 1950 there were 2.5 men for every 1 woman participating in the U.S. paid labor force (U.S. Department of Labor, Bureau of Labor Statistics 2000). Over the decade examined by Gerbner and Signorielli, rates were better—about 1.66 men for every woman (Fullerton 1999). But at last count (in 2008), the ratio was down to 1.2 to 1 (males to females) (U.S. Department of Labor, Bureau of Labor Statistics 2009). So, while women have progressed to nearly representative rates of participation in the working world, the “reel” and print worlds have continued to overlook them. This is particularly startling in the context of changes in the media environment over this period. The number of commercial television channels received in most homes has leaped from three to well into the three-digits, providing a much wider variety of programming than in the past. These shifts provide additional opportunity for women to be represented, but in the analyses presented in these issues of Sex Roles, there is little evidence this has taken place, and suggest that instead they are now relatively absent from a broader set of venues.

Hether and Murphy (2010) find women represent 38% of the major characters appearing in health storylines on popular primetime television programs. Male sources also appear almost three times more often than female sources in newspaper coverage of same-sex marriage (Schwartz 2010) and nearly twice as often in local television news coverage of a variety of topics (Desmond and Danilewicz 2010). In music videos drawn from five music-oriented television networks, male characters outnumber females by a ratio of 3 to 1 (Turner 2010). Smith et al. (2010) find that the ratio of males to females in top-grossing G-rated (general audience) films is 2.57 to 1. The place women were least likely to appear was video games. Downs and Smith (2010) find that only 70 of 489 (i.e., 1 in 7) characters appearing in the best-selling console games are females. These articles and analyses are far from documenting women’s under-representation in all media, or even in a representative sample of any one medium. But they focus on important content to which large numbers of persons are regularly exposed, and the relative absence of women in this diverse set of media is striking, suggesting a media world closer to the working-world reality of 1950 than to 2010 society.

The Consequences of Under-Representation Are Unknown

One key goal of content analysis is to examine whether media have characteristics that might influence users’ attitudes and behaviors, either negatively or positively. As Rudy et al. (2010b) note, there are other reasons to conduct content analyses, but examining media for their potential effects is the one in which society has the greatest stake. Thus, the overwhelming pattern of under-representing women begs the question of how this affects consumers of media content. If young girls do not see themselves reflected in media, will this diminish their sense of importance and self-esteem? Will boys conclude that women and girls are unimportant, as well? Will girls lack role models? Will adult women feel disenfranchised? Does the under-representation of women constrict societal perspectives and information in important ways? Indignation over the lack of minority representation in media catalyzed a social movement (see Greenberg et al. 2002), and the lack of female representation recently led to the founding of See Jane (www.seejane.org). See Jane is a project of the Geena Davis Institute on Gender in Media that works with the entertainment industry to increase the number of girls and women in media aimed at children. These activities are founded on the belief that under-representation of females has meaningful and problematic effects. While this seems a reasonable conjecture, our theories provide little basis on which to rest hypotheses about the consequences of representation in media, or lack thereof.

Bandura’s social cognitive theory (2002) suggests that similarity to those portrayed in media is important to learning from their behaviors. Others argue that identification with a character (e.g., Dal Cin et al. 2007; Slater and Rouner 2002) and transportation by narrative (Dal Cin et al. 2004) are critical to media influence on beliefs and behavior. These processes may require similarity to operate, in other words, girls may require the presence of females in order to learn from their example. However, Greenberg and Atkin (1982) presented results indicating that while Black and White youth are equally likely to identify with White television characters, Blacks are far more likely than Whites to identify with Black characters. Other work indicates that girls are more likely to identify with male protagonists than boys are with female protagonists (Oatley 1996). Perhaps those who are represented less often in the media adapt by more easily identifying with dissimilar others, allowing them to be involved viewers even though they do not see themselves reflected in terms of gender or race. A greater understanding of the processes of identification and transportation and how they operate (if at all) among persons who seldom see themselves reflected in media is needed.

Other psychological processes might be triggered when viewers find themselves unrepresented. It has, for example, been suggested that self-esteem may suffer. Empirically, I know of only one study in which research has linked viewing content in which one’s demographic characteristics are not depicted and negative behavioral or attitudinal effects. McDermott and Greenberg (1984) linked the viewing of television programming with Black characters to higher self-esteem among Black fourth and fifth graders. But the study was small and preliminary in nature, and there are other data that appear inconsistent with its results—in spite of being heavier users of media than whites (Rideout et al. 2010), and being portrayed in those media far less often (Greenberg et al. 2002), Black Americans’ self-esteem tends to be higher than that of White people (Twenge and Crocker 2002). Of course, there may be other contributors to self-worth that override any negative effect of under-representation in media on Black viewers. But it is important that we develop theory on the question of under-representation, and that research begins testing for effects. This is as critical for the under-representation of women as it is for minorities, the study of which Greenberg et al. (2002) described as “acutely essential.”

Although it is not entirely clear what to predict about the general under-representation of women in media, it is easier to surmise effects of omitting them from specific areas of content. One study in the special issue suggests that a lack of female representation might affect the nature and quality of political debate surrounding gay rights. In a special case of the overrepresentation of men, Schwartz (2010) shows us that this pattern holds for men and women serving as sources for coverage of same-sex marriage appearing in four U.S. newspapers. The overall ratio of male to female sources was three to one. Males’ perspectives, which are more negative toward gay persons than females’ (Herek 2002), dominated the views expressed, and this may have helped to limit support for same-sex marriage.

Among a subset of sources, those who are gay, under-representation of women was present only in newspapers from cities where smaller percentages of gay men and women reside (Ohio and Oklahoma) (Schwartz 2010). It is my informal impression that lesbian service members have been largely absent from the discussion of repealing the U.S. military’s Don’t Ask Don’t Tell (DADT) policy, as well, but are present in the military in numbers as great as those of gay men (RAND National Defense Research Institute 2010). Lesbians are more accepted by the general public than gay men, and their service in the military and marriage to one another may also be more accepted. Including lesbians in news coverage might alter general reactions to these issues. More research on when and how lesbians are represented in discussions of same-sex orientation issues is urgently needed. Understanding how representation of women (or the lack of it) influences attitudes towards gay individuals is critical as the U.S. wrestles with issues of gay rights in these and other domains.

Another specific area where we might generate predictions regarding the impact of under-representation is in the realm of health-related content. Hether and Murphy (2010) find that women are less likely than men to appear in health storylines from popular fictional programming. Less than 40% of characters playing significant roles in health storylines were women, across the categories of sick person, caregiver, and bystander. The stories in which women and men appeared were of equal prominence, were equally serious in tone, equally educational, and equally likely to contain information about factors such as diagnosis, treatment and prevention—but women appeared less often. For viewers who learn about health problems through entertainment television (a not insignificant portion of the population) (Brodie et al. 1999), the less frequent presence of female characters undergoing serious medical problems may mean women know less about the health problems that may affect them and how to detect and address them. The authors note that stories about women often focus on maladies specific to women’s bodies, such as breast cancer or reproductive issues. This is interesting in that it parallels the focus of health research for many decades. Few studies of heart attack, for example, involved female samples, and as a result women’s risk for cardiovascular disease was largely overlooked for many years (Bird and Rieker 1999). In a media equivalent to this, fewer storylines about women may be equivalent to less information for women.

Here again, some testing of this hypothesis is a clear next step. Do women learn as much when they see storylines about men as when the same information is presented about women? Is it as likely to spur them to action—to get tested for a disease, take preventative measures, or look for more information about the issue? It seems likely, given what we know about the importance of identifying with a character in the process of observational learning (Bandura 2002; Dal Cin et al. 2007), but we need more evidence on this issue.

Women are Portrayed in Circumscribed Roles

Women are Sexualized

Although the articles in these special issues contain more consistent evidence on the under-representation of women than on any other point, when it comes to predicting effects on media consumers, how women and men are portrayed may be much more important than whether they are portrayed. Many of the Sex Roles studies indicate that when women do appear in media, they appear in sexualized or subordinated roles. As Table 1 indicates, five of the articles in the special issues code for and find sexualization of women, and four code for and document subordination of women. In the realm of “be careful what you wish for,” simply increasing the prevalence of women among characters in media might exacerbate any problematic effects of media use unless the manner in which women are portrayed is also addressed.

Much of what is presented on this topic in the special issues is about the ground where sex and gender overlap—in gendered sexual scripts and sexual gender scripts. For example, in their content analysis of best-selling console video games, Downs and Smith (2010) find that 41% of female characters wore revealing clothing and an equal number were partially or totally nude. And many of these women had unrealistically proportioned bodies. There was no parallel to these findings in the manner that male characters were portrayed. Notably, the researchers rarely found portrayals of either talk about sex or sexual behavior—only six instances of either were observed in 60 video games depicting 489 characters. The games were not about sex, but about sexy women.

Similar themes emerge in the article by Zhang et al. (2010). They find that the representation of thin women in rap music videos is equally common to White and Black female characters appearing in them. That is, about half of all female characters of either race (51%) were rated by coders as thin. The authors compare this to U.S. national data indicating that only 24% of women (across races) are thin. They might have further noted that Black women are about 50% more likely to be overweight than White women (U.S. Centers for Disease Control, MMWR 2009). So, rap video overrepresentation of thinness is probably greater for Black women than White, in spite of what their analysis indicates. However, it is probably more important that the authors find links between character weight and video theme. Thinness was associated with themes of sexuality and materialism, suggesting that it is a marker for the same sort of sexual objectification of women documented by Downs and Smith (2010).

Sex is also a theme in a second study comparing White and Black music videos. Turner (2010), finds that Black videos (characterized as such based on the genre of music as well as the apparent race of the characters) are more likely to include sexual content and to include female characters in provocative clothing. Overall, 59% of videos contained sexual content. Most characters did not wear provocative clothing, but those who did were more likely to be Black and more likely to be women.

These findings raise serious questions regarding the advisability of advocating for greater inclusion of women, and greater inclusion of Blacks, in media without simultaneously addressing other issues of content. Prior work indicates that women’s and girls’ self-esteem, body image, and emotional well-being may be adversely affected by exposure to media that are sexually objectifying (APA Taskforce 2010; Grabe et al. 2008). Social learning (Bandura 2002) and other theories of media use (Huesmann 2007; Steele 1996) suggest that these kinds of portrayals may also affect viewers’ sexual attitudes and behavior, particularly during adolescence when youth are actively coming to terms with their identities (gender and otherwise) and their sexuality. Empirical research bears this out; content that sexually objectifies or degrades women can alter adolescents’ sexual beliefs (Ward 2002) and behavior (Martino et al. 2006).

Sexual Content is Not the Same as Sexualization

Although the reports in these special issues are careful in their conclusions, these theories and findings have sometimes been stretched a bit thin, so it bears repeating that there are parameters surrounding these effects. Exposure to sexual content or partially clothed women is not, in itself, likely to foster problematic body image among girls, or reduce their self-esteem. Nor will sexual content or nudity necessarily lead boys to see girls and women as objects. But it is theoretically likely to have such effects if the body types portrayed are idealized (i.e., “hypersexualized”), if the lack of clothing is disproportionately characteristic of women, and if the images or other content subordinates women (APA Taskforce 2010; Calvert and Huston 1987; Grabe et al. 2008). Sexual content per se is not likely to foster negative attitudes toward anything except celibacy or virginity (Collins et al. 2004). Likewise, partial nudity may lead viewers to dress in scanty clothing, but whether it results in the objectification of women or of oneself depends on the way in which nudity is portrayed.

These distinctions sometimes get lost in discussions of results, even when researchers are meticulous in describing the processes involved and the bounds on their empirical findings. Reporters, advocacy groups, and the public sometimes translate the research into something more like the morality-based concerns they are accustomed to hearing, or use it to advocate for censorship of sexual content. For example, Fox News (2007) titled its report of findings from the APA Taskforce on the Sexualization of Girls (APA 2010) “Report: Sexual Images Psychologically Damaging Young Girls.” The report did not argue that sexual images damage youth, but rather, that images in which young girls are portrayed as sexual objects might damage young girls who are exposed to them. To be fair, the accompanying text correctly described the findings, but those who paid attention only to the headline would have been seriously misled. More problematic was the use of the report to advocate for a boycott of Dove beauty products (LifesiteNews.com 2007). Dove’s parent company, Unilever, had launched a television advertising campaign that featured nude older women and the slogan “beauty has no age limit.” American Family Association noted the study’s call for reducing sexualized images of women and girls and the possible damaging effects of exposure to these images as if these conclusions applied to the Dove campaign. But the definition of sexualization set out clearly by the report was not met by the campaign. In fact, the Dove campaign might be considered a positive media image, in that it countered an idea that was cited as problematic in the report—that only young women and those with particular body types are attractive (APA 2010).

In short, emphasizing the subtleties involved in sexualization is critical. This is true first, because overdrawing or oversimplifying findings, or allowing others to do so, ultimately undermines the credibility of researchers. And second, because, should policy or practice be based on the research, it is critical that it target the right set of portrayals. Content analysis has the ability to capture some of these subtleties and make them clear. Many of the articles in these special issues represent fine examples of this, and future work should follow and build upon them.

Broader Gender Stereotypes Are Also Present

Many of the studies in these special issues focus on stereotypical portrayals of women apart from the role of “sex object.” Nine of the studies appearing in the special issues code for some sort of traditionally female (i.e., stereotypical) role and find evidence that it appears commonly in media. Das (2010) finds that the percentage of women versus men shown in Indian television advertising is not dramatically discrepant (women were far less likely to appear in voice-overs for the ads, however). Of the 627 characters identified and coded in their analysis, 43% were female and 57% male. And women were not more typically portrayed as housewives, but they were more likely to be depicted in relationship roles, and to represent products for the body or clothing than were men. Men also represented stereotypically male products, such as automobiles and electronics. This again suggests that simple solutions such as increasing the rates at which women are represented will not produce “better,” “fair,” or realistic portrayals. Indeed, in their cross-national analysis of television ads, Paek et al. (2010) show that in the single country where women were not under-represented, South Korea, the roles in which women appeared remained those traditionally associated with females. Women were overwhelming more likely to play the role of housekeeper and men were overwhelmingly the choice to portray professionals and even office workers.

Research Frontiers: What We Still Need to Know about Sex Roles in Media

Theory and Research on Consequences of Under-Representation

As noted above, we are short on theory and empirical evidence regarding the consequences of consuming media in which one’s gender is chronically under-represented. Are there likely to be effects, on whom, and through what process(es)?

New Media

In their introduction to these special issues, Rudy et al. (2010b) point out that published content analyses of gender roles have seldom examined gender depictions in so-called “new media,” and the same is true of the studies in this special issue. The papers presented focus on traditional media such as television, film, music, and newspapers. But the media landscape is evolving at a startling pace, with a greater diversity in content, new types of media, and new platforms for delivering media constantly emerging. The variety of content available on the Internet is practically limitless, and includes what were previously “other media” such as music, television, games, and films. Moreover, content can now be viewed or used on computers, mp3 players, handheld video players, and cell phones, as well as on television sets. Adults use a variety of media and increasingly engage with these media on diverse platforms (Simmons New Media Study 2009). But adolescents, in particular, are immersing themselves in such content.

In 2009, Kaiser Family Foundation (Rideout et al. 2010) surveyed more than 2,000 young people ages 8–18 years from across the U.S. regarding their media use. Youth spent a total of 10 h and 45 min each day using various media. Television was still the medium of choice and viewing it takes up the largest chunk of adolescents’ media use, accounting for nearly 4.5 h of media time. Music/audio was second, accounting for about 2.5 h of use. Nonetheless, newer media—computer use and video game use, accounted for 1.5 and 1.25 h of youths’ media time, respectively. And a large minority (20%) of media consumption (about 2 h) took place using mobile platforms such as cell phones, laptops and handheld game players. As with traditional media, Black and Hispanic youth spent somewhat more time with most of these activities, relative to White youth. We know almost nothing about content of newer media, or about how it varies depending on the platform used to access it. Thus, it is critical that researchers begin to study them systematically.

New media use permits exposure to content, as with traditional media. But it also affords a chance to create and distribute media. Sixty-four percent of online teens ages 12–17 have created and posted content on the web, ranging from having a personal webpage to blogging and posting artwork, photos and videos. New media also allow an opportunity to publicly comment on portrayals. Many online activities are highly social relative to traditional media use, including social networking, email, chat room participation, and instant messaging, as well as video games that can be played by multiple online players and using voice over protocols that permit conversation between these players.

These unique features of new media raise a few additional questions worthy of study. In particular, just as the content of professionally produced media is likely to be influenced by the gender of those who create it, content created by users in interactive environments, whether videos, blogs, or comments on such, is likely to be influenced by gender. Thus it is important to track this content and the gender of participants in these media platforms and content types. Presently, girls spend more time on social networking sites than boys, and less time game playing and watching or posting videos. Does this influence gender portrayals in these media?

Positive Portrayals

There was little discussion in any of the papers regarding content that might depict women particularly positively. Are there positive images of women in media, or images that run counter to traditional gender roles or stereotypes? If so, where do they reside? Given what we know from the articles in these issues, there are likely to be few such portrayals, making it difficult detective work to track them down, and obtaining enough of them for a quantitative analysis even more taxing. But discovering where they reside, and why they are there, may be as important as identifying problematic portrayals.

Linking Content to Effects

That my own research (e.g., Collins et al. 2004; Martino et al. 2006) has focused on testing the effects of media is perhaps obvious from the emphasis I have placed on outcomes of exposure in this overview. That work has relied heavily on collaboration with colleagues who know far more about the science of content analysis than do I, and on methods that meld content analysis with survey research. Those who study media effects increasingly come from different disciplines than those who study content, and the two camps often have different perspectives and theories and are trained in different methodologies. Many media-effects researchers are ill-equipped to conduct rigorous, scientifically sound analyses of gender content. As a consequence, they tend to focus on testing associations between use of a medium (e.g., hours of television viewing) and outcomes—an approach that is inadequate in a diverse media universe, and that under the best of conditions can provide only indirect evidence of content effects. Conversely, many who conduct content analyses are at least as interested in the effects of the content they are describing as the content itself, but have little experience in the kinds of large scale surveys used to test associations between media exposure and outcomes. Interdisciplinary work that pairs these methodologies has pushed forward our understanding of media effects in realms like sexual content (e.g., Brown et al. 2006; Collins et al. 2004) and portrayals of substance use (e.g., Heatherton and Sargent 2009). Rudy et al. (2010b) note in their overview that such approaches are rare, but I believe we will see much more of them as interdisciplinary collaboration increases in the sciences. Forging relationships between survey and experimental research and content analysis provides the possibility of realizing the greatest potential from both fields and having the most substantial and informed impact on media practice.

Conclusion

On their own, each of the articles in these special issues adds to the body of science addressing gender roles in media. Some of the articles further theoretical understanding, others add methodologically, and some enhance the general body of evidence with applications to new arenas or caveats regarding the differences in portrayals by nation or race. When combined, their contribution is a substantial shift in the state of knowledge and stage of research in this area. The least surprising of their findings, that women are under-represented and women are sexualized, are so clearly documented across such a variety of media and settings that it is clearly time for the next stage of research—testing the impact of this on media audiences. Although there is existing research on this topic, it is in a relative stage of infancy.