Sex and the Spectacle of Music Videos: An Examination of the Portrayal of Race and Sexuality in Music Videos

Abstract

This content analysis used two studies to examine sexual behaviors and sex role portrayals in music videos televised in the United States. The first study included 120 videos recorded from MTV, MTV2, VH-1, BET, and CMT and revealed African American videos were significantly more likely to portray sexual content and female characters in provocative clothing than White videos. The second study analyzed 20 videos from BET’s late-night program, Un:Cut. Results revealed Un:Cut videos depicted seven times as many sexual acts and featured significantly more discouraged sexual behaviors than videos from the five major music video channels. Findings from these studies clarify that race is an important factor when gender roles and sexual content are examined in music videos.

Introduction

This paper describes results from two quantitative content analyses of 140 music videos collected from five popular U.S. cable television stations to foster a better understanding of the ways race relates to differences in sexual content and sex role stereotyping in music videos. Social cognitive theory provides a theoretical foundation for the studies’ findings and directs attention to content most likely to be modeled by certain groups of young music video viewers.

The two studies described in this research add to the existing music video literature in a number of ways. First, Study I clarifies (for the 2000s) whether there are differences in the frequency of sexual content in videos performed by African Americans and Whites. Second, Study I examines whether differences exist between African American and White music video characters in terms of provocative clothing. There have been no known attempts to compare the frequency and level of undress of characters according to ethnic demarcations. Study I also compares the frequency and type of provocative clothing worn by females and males to better understand whether divergent sex roles are expressed in music videos through unequal portrayals of sexualized dress. Third, Study II represents a first look at a distinct form of hyper-sexualized music videos seen on Black Entertainment Television’s late-night program, BET Un:Cut. This examination will allow for a more complete understanding of the full range of sexual content and gender role portrayals that appear in Black music videos in the U.S.

The popularization of the music video is one of the most significant developments from the 1980s and 1990s cable television proliferation (Gow 1990) and it remains a prominent and important televised format today. Adolescent consumers continue to watch music videos regularly and remain the target audience for MTV, BET and the like (Kalof 1993; Lewis 1990; “MTV: Music Television Profile and Network Contact Info” 2005; Rideout 2003; Smith and Boyson 2002; Tiggemann and Slater 2003). Past research shows that music video content from the 1980s and 1990s was sexy (Sherman and Dominick 1986; Hansen and Hansen 2000; McKee and Pardun 1996) but that it also typically emphasized sexual innuendo and suggestiveness rather than overt/explicit depictions (Baxter et al. 1985; Gow 1990; Sommers-Flanagan et al. 1993). Past research also shows that music videos featured sex role stereotyping (Gow 1996; Hansen and Hansen 2000; Seidman 1992; Sommers-Flanagan et al. 1993), sexism (Vincent et al. 1987; Vincent 1989), and consistently fewer depictions of female characters than males (Gow 1996; Seidman 1992; Sommers-Flanagan et al. 1993). Finally, past examinations show that music videos by African American musical artists in the 1980s and 1990s tended to feature increased levels of sexual content when compared to videos by White artists (Brown and Campbell 1986; Jones 1997; Tapper et al. 1994).

Social cognitive theory (SCT) is the guiding theoretical perspective in this study. SCT’s (Bandura 1972, 1986, 1991, 2001, 2002) main thrust is that humans can learn vicariously through the observation of actions taken by others, called models in SCT. Individuals who attend to the behavior of models, then, can learn about the consequences of those behaviors, either reward or punishment, without the need to have those experiences firsthand (Bandura 1986, 2001). Through the inclusion of the vicarious learning concept, Bandura has conceived of a theory well equipped to discuss the implications of learning via symbolic modeling based on mediated formats like television (Bandura 1972, 1986, 2001).

Bandura (1986, 2001) argues that television and other visual media are particularly influential modeling agents because of their vast popularity and their disinhibitory powers. Bandura (2001) suggests that television and other media obscure and distort the relationship between actions taken and the effects they cause. The distortions of media messages tend to bring about better recall of possible benefits of a mediated behavior than memories of possible harmful effects, which are minimized during cognitive processing of the social information. Based on the strength of televised modeling, Bandura (1986) asserts, “modeling serves as a major conveyor of sex-role information” (p. 93). He notes, “some sex differences are biologically founded. But many of the stereotypic gender roles arise more from cultural design than from biological endowment” (Bandura 1986, p. 95). Television contributes to psychosocial sex role development by influencing adolescents’ understanding of their own gender identity through portrayals of their respective sexes acting in a larger society.

SCT is used here to direct attention to content most likely to be modeled by certain groups of consumers. The focus of the current study is on “attentional processes” and “motivational processes”, two of the four subfunctions that govern observational learning (Bandura, 1986, 2001, 2002). Attentional processes suggest that for a mediated model to have a legitimate chance at gaining the attention of a potential attendee, that model must be salient, striking, conspicuous, and/or prominent (Bandura 2002). MTV and other music video channels are a popular form of entertainment for adolescents (Kalof 1993; Smith and Boyson 2002; Tiggemann and Slater 2003). It is estimated that 3 out of 4 young adults (16 to 24 years) watch MTV every day (Rideout 2003). This popularity should not be surprising as MTV is noted for targeting its content to a young audience, aged 12 to 34 (Lewis 1990; “MTV: Music Television Profile and Network Contact Info,” 2005; Sun and Lull 1986). Based on the popularity of music videos among young people, the musicians, performers, and characters portrayed in videos are by association prominent and conspicuous models for adolescents.

In addition to a model’s prominence, video producers gain young peoples’ attention by making the models (musicians and performers) attractive. Smith and her colleagues (1998) note, “viewers of all ages are more likely to emulate and learn from characters who are perceived as attractive” (p. 8), a belief resonated by a number of authors (see Bandura 1986, 2002; Kunkel et al. 1995; Mastro and Atkin 2002; Paik and Comstock 1994). Making attractive models might be as simple as including a “prestigeful model” (Bandura 2001, p. 283), such as an individual prominent in the public eye, like an actor, athlete, or a pop star. Durant et al. (1997) argue that prestigeful models in music videos make effectual role models, capable of exhibiting considerable influence on adolescents’ understanding of health risk behaviors. Another way to make a character appear more attractive and thus garner more attention might be to eroticize that character (Bandura 2001) through use of provocative clothing or depictions of that character engaging in sexual behavior.

Bandura’s coverage of the motivational processes of SCT suggest that a model’s similarity to the attendee, in terms of race, gender, and other observable parallels, increases the likelihood that the learned behavior will be incorporated into the attendees’ behavior-repertoire (Bandura 1986, 2001, 2002). Bandura’s theoretical claim has been supported in a number of research efforts. Viewers tend to feel similar to mediated models that are like them in terms of gender (Hoffner 1996; Hoffner and Buchanan 2005) and race (Hoffner and Buchanan 2005; Appiah 2001; Botta 2000). Perceived similarity and identification with media characters has been associated with wanting to become more like those characters in other ways, including emulation of appearance and behaviors (Hoffner and Buchanan 2005; Hoffner and Cantor 1991). In addition to taking place through observational learning, this emulation process is also theorized in SCT to come about via “social prompting effects” (Bandura 2001). Unlike observational learning, social prompting effects are not about the transference of new or unique behaviors but deal more with mediated models’ ability to support and reinforce beliefs and behaviors already common in a society, such as choices about clothing.

Do Sexualized Music Videos have Effects on Young Viewers?

Examining the effects of exposure to sexualized media messages on young people is an important line of research in communication studies. Since the popularization of music videos in the 1980s, scholars have extended this line of research to music videos. As content analyses, the two current studies cannot establish effects but the content does have implications for audience members. For that reason, a brief review of the music video effects research is covered below.

Beginning with survey research, Wingood et al. (2003) gathered 12 months of longitudinal survey data from 522 adolescent African American females. They found that adolescents who had greater exposure to rap videos were 2 times more likely to have multiple sexual partners, and over 1.5 times more likely to have acquired a new venereal disease. Johnson et al. (1995b) found that exposure to violent rap videos led young African American males to exhibit greater acceptance of violence against women when compared to exposure to nonviolent rap videos or no videos at all. In a similar study Johnson et al. (1995a) found that Black females who were exposed to “rap videos featuring women in sexually subordinate positions” (p. 603) reported greater acceptance of teen dating violence than those not exposed to the videos. Exposure to scantily clad female dancers in the videos affected the young women by increasing “the accessibility of constructs associated with female inferiority” (Johnson et al. 1995a, p. 603).

Watching violent and sexist music videos, then, not only can affect young people’s view of overt sexual and violent attitudes and behaviors but can also affect viewers’ more subtle attitudes towards their own gender’s sex roles as well as those of their partners in intimate relationships (Bandura 1986; Gow 1996). Specifically, past research has shown that even brief exposure to particularly sexualized and gender-stereotyped images in music videos can lead to liberal sexual attitudes (Calfin et al. 1993). Perhaps just as important, however, is the notion that exposure to this type of imagery can also lead to adversarial beliefs about gender roles and unhealthy and inaccurate views of the opposite sex that go beyond issues of sexuality (Kalof 1999). To clarify, continued exposure to sexual and sexist images in music videos (and elsewhere) might lead young women to overvalue their physical appearance and lose their sense of identity beyond male-defined corporeal desires (Stephens and Few 2007). Additionally, these types of images “could exacerbate gender opposition and antagonism in relationships among young people” (Kalof 1999, p. 384). Furthermore, underrepresentation and a lack of diversity in female characters’ roles in music videos might have an effect on young women’s view of gender role diversity in society generally. For example, Gow (1996) found that women in music videos were presented in lead roles much less frequently than men and were more likely to be cast as “posers and dancers” (p. 159), thus needing to affect an attitude and demonstrate physical talents in the background rather than call upon potentially more admirable talents like that of the lead actor or musician.

Watching violent and sexist music videos, then, can affect one’s gendered view of self in relation to the larger society and that includes one’s view of people of other races. Gan et al. (1997) found that exposure to “sexually enticing” rap music videos led White females and males to report unfavorable evaluations of African American women when compared to exposure to videos featuring “devoted love” or no videos at all. Sexualized music videos also appear to influence consumption behaviors and purchasing intentions. Cummins (2007) argues, “sex in music videos, [is used] to increase their enjoyment and effectiveness as a marketing tool” (p. 97). Through a series of experiments, Hansen and Hansen (1990) found that increased sexual imagery in music videos led subjects to report increased liking of both the music and visuals. These findings support Bandura’s (2001) assertion that erotic content makes a product more attractive to a potential buyer.

Examinations of the effects of sexualized television programming generally echo the concerns covered in music video research. A number of studies have linked adolescents’ exposure to media depictions featuring objectification of women and sexually explicit materials to adolescents’ beliefs about women as sex objects (Peter and Valkenburg 2007; Ward 2002; Ward and Friedman 2006). Collins et al. (2004) found that after controlling for more than a dozen other predictors of sexual behavior, exposure to sexual content on TV was a strong predictor of adolescent sexual behavior. Using a three-year longitudinal design Brown and Newcomer (1991) found “evidence of a significant relationship between the proportion of sexy programming an adolescent watches on television and the adolescent’s sexual activity status” (pp. 87–88). In addition to the ways exposure to sexually explicit content can impact young viewers understanding of sexual norms and predict future sexual activity, Brown and L’Engle (2009) found that early exposure can also lead to less progressive gender role attitudes in early adolescent males and females as well as increased likelihood of sexual harassment in males.

Together, the above findings show that sexist, sexual, and violent music videos, like other television content, can affect young audience members in myriad deleterious ways. To fully understand the possible effects of sexualized television content, scholars need to continue to analyze the content itself. There are recent indications that sex is growing increasingly prominent, frequent, and explicit on television (Arnett 2002; Kunkel et al. 2003; Lampman et al. 2002), in music videos (Industry Ears 2008; Turner 2006) and in rap music (Kitwana 2004; Neal 2005; Stephens and Phillips 2003).

A great deal has changed in and around music videos since the 1980s and 1990s when most content analyses of videos took place. In the 1980s and 1990s, MTV, along with its competitors and siblings, looked and felt more like radio stations; audience members were more likely to treat music videos as background entertainment while engaging in other activities (Williams 2003). Today, videos offer more complex visual arrangements akin to short films and are accompanied by increasingly sophisticated programming, promotions, and logos that command more attention from the viewer than in the past (Williams 2003). It is also important to re-examine sexual content in music videos, especially African American artists’ videos, after Viacom’s purchase of Black Entertainment Television in 2000. This change is significant because until 2000 BET was owned by its founder Robert L. Johnson, an African American businessman (Smith-Shomade 2007). As a result of the multi-billion dollar sale, BET joined Viacom, a White-owned major media conglomerate that also owns MTV, a channel that has been widely criticized in the past for its exclusion of music videos performed by Black artists (Brown and Campbell 1986; Garofalo 2002a; Peterson-Lewis and Chennault 1986; Williams 2003). As discussed in more detail below, this economic arrangement, then, has implications for the way African American performers are portrayed on BET. Finally, in the 2000s (unlike in the 1980s and early 1990s) television has to compete with streaming video websites (where there is a possibility for more sexually explicit content) for the attention of music video consumers. The two current studies offer an opportunity to further examine the frequency and nature of the sexual content portrayed in the evolving terrain of music videos against the backdrop of both racial and gender role considerations.

Study I

Hypotheses and Research Questions – Study I

Past music video content analyses have shown that typically African American genres tend to feature increased levels of sex when compared to typically White genres (DuRant, et al. 1997; Jones 1997; Tapper et al. 1994). In their analysis of 161 music videos, Tapper and his colleagues (1994) found that Soul videos displayed the greatest level of sexual appeal, followed by Rap videos. These genres are both classified by Tapper and his colleagues as exhibiting a high presence of Black performers, while so-called White genres of music (Alternative Rock, Classic Rock, Country, and Heavy Metal) showed little more than slight sexual appeal. DuRant and his colleagues’ (1997) analysis of 518 music videos showed that R&B and Rap videos both displayed higher degrees of sexuality and eroticism than Rock and Country videos. Jones’ (1997) content analysis of 203 music videos showed that compared to other genres, Hip-Hop and R&B were greatest in all sexual variables. Jones argues that these results support the notion that musical genre can be used to predict differences in frequency of sexual behavior in music videos. Based on the previous literature the following hypothesis was tested in the current study:

• H1 African American music videos will feature significantly more sexual content than White music videos.

As discussed above, published content analyses of music videos have typically classified race in music videos based on genre of music. This decision makes sense in that popular music genres in the U.S. have traditionally been drawn along racial (specifically Black and White) lines by record companies, radio stations, and record stores (Garofalo 2002b; Negus 2004; Shuker 2002). The decision to use the race of the performer as an organizing principle in musical genres is based on both racism in the music industry and the music listening audience as well as hard-line marketing realities (Garofalo 2002b; Negus 2004). As a result of historical antecedents as well as business and institutional practices in the recording industry, “black music has been relegated to a separate and unequal marketing structure” (Garofalo 2002b, p. 112). These racial consistencies in musical genres are not based merely on business decisions. Like Blues before it, Rap music is the musical manifestation of a much larger African American cultural movement, namely Hip-Hop (Boyd 2004; Stephens and Phillips 2003). Like Rap and Blues, Country and R&B have similarly homogeneous racial origins both in terms performers and audiences (Garofalo 2001, 2002a; Smethurst 2001).

In their analysis of White college students’ musical preferences, Christenson and Peterson (1988) argue, “the racial origin of music is a powerful organizing principle” (p. 297). Tapper and his colleagues’ (1994) results confirm Christenson and Peterson’s argument. The scholars report the following about their sample of 154 music videos: Rap and Soul videos featured 100% Black performers, Country and Classic Rock videos featured 100% White performers, and Heavy Metal videos featured 92% White performers. Only Pop videos showed any real racial diversity, featuring 60% White performers. See Table 1 below for a breakdown of genres by race of the musical artists in the current sample.

Table 1 Number of videos in sample, sorted by genre and artist race (weighted).

Previous content analyses show that traditionally Black genres of music videos feature increased levels of sexuality when compared to traditionally White genres. Based on the historical indicators and previous music video research the first hypothesis was tested by comparing the frequency of sexual content of African American genres of music videos (Rap, R&B, Mixed Rap/R&B) with the sexual content of White genres (Rock, Country).

In addition to testing the first hypothesis through an examination of genres, a second test was also administered to analyze whether videos performed by African American artists featured significantly more sexual content than videos performed by White artists. Social cognitive theory suggests that a model’s similarity to an attendee increases the likelihood that the learned behavior will be incorporated into the attendee’s behavior repertoire (Bandura 2002). For this reason, identifying whether performers from one ethnic group are more likely to include sexual content in their videos than others has implications for particular audience members.

Testing the first hypothesis by looking at the race of the musical artist not only holds theoretical relevance but can also work to overcome somewhat misguided assumptions regarding popular musical genres and their relationship to race. Through the inclusion of the second test, a more accurate and one-to-one relationship between race and sexuality is engendered. Genres, which are merely labels, cannot completely and effectively account for the performers’ and fans’ fluidity across musical formats (Christenson and Peterson 1988; Shuker 2002); popular music is not so black and white. Looking directly at the race of a performer in the video will work to overcome this potentially incongruent arrangement. Although no known research has measured sexual content in terms of the race of the main performer, previous research does suggest sexuality will materialize on these terms. In addition to the notion that African American music video genres are more sexualized than their White counterparts, Hansen and Hansen (2000) and DuRant et al. (1997) both found that BET, which features Black artists, offered a more sexualized product than the racially mixed MTV and VH-1, and the homogeneously White CMT. Based on past music video research and the increased theoretical and methodological significance of the inquiry, the first hypothesis was tested according to the race of the main musical artist(s) performing in the videos as well as along race-based genre lines.

Despite relatively little research about levels of undress analyzed along racial lines in music videos, a trend does emerge. Jones (1997) found that Hip-Hop and R&B displayed the highest levels of “scant dress and sexual dancing” (p. 353). Hansen and Hansen (2000) found that sexually provocative dress was significantly more prominent on BET than MTV and CMT. Additionally, Emerson (2002) found that music videos performed by Black women emphasized those women’s bodies and constructed a one-dimensional sexualized Black womanhood. Given the results in past music video research and the previously discussed theoretical implications from social cognitive theory, the following hypotheses emerged:

• H2a African American regular characters will be significantly more likely to dress in provocative clothing than White regular characters in music videos.

• H2b African American background characters will be significantly more likely to dress in provocative clothing than White background characters in music videos.

Method

Sample

A sample of music videos was recorded on videotape during a five-week period (weekdays only) in December and January 2004–05. The sample included content from MTV, MTV2, VH-1, BET, and CMT. The decision to analyze videos shown on these five channels was based, in part, on a pretest in the form of a nonprobability sample of college students (n = 389) at a mid-size, state university on the east coast. The pretest revealed that MTV, MTV2, VH-1, and BET were the music video networks most frequently watched by the students. Other channels on the pretest questionnaire, which included CMT, Fuse, MHz, MTV HITS, VH-1 SOUL, and VH-1 CLASSIC, were not as highly ranked by the students. Even though CMT was not highly ranked, Newcomb (2004) reports that “CMT is the number-one choice for cable programming among women aged 18 to 49” (p. 611). Furthermore, Wilson (2000) convincingly contends that country music videos are an important source of information about gender roles and sexuality and to that end should also inform discussions about race in relation to these considerations. For those reasons, and because the pretest did not include college students from rural areas in the American south, where CMT’s audience reach is strongest (Miller 2005), this study’s sample included music videos drawn from CMT.

Sampling took place in the mornings when music video programming (rather than reality television and other nonvideo content) was most likely to air on each of the five channels. Two hours of music videos were recorded each day between the hours of 6 a.m. and 8 a.m. for MTV, MTV2, VH-1, and CMT and between the hours of 9 a.m. and 11 a.m. for BET. Just after the 6 a.m. to 8 a.m. time block was implemented, BET stopped airing videos at that time and began their daily video rotation with equivalent programming at 9 a.m. This change led to the above discrepancy. Please see Table 2 for the complete sampling schedule.

Table 2 Sampling schedule.

Each channel was recorded for one two-hour time slot per week, with a different channel recorded each day. The channels were randomly assigned to weekdays the first week, and then the same channel order was rotated by one day in each of the following weeks. Recording two hours from each station on each day of the week over a five-week period rendered 10 h of recorded programming from each of the five stations. A total of 50 h and 480 music videos were recorded; of the 480 videos, 172 were unique. In order for this sample to parallel the average number of videos studied in previous music video content analyses, 120 of the original 172 unique videos were randomly selected to form the final sample.

Duplicate videos were accounted for statistically during data analysis. Each video was weighted according to the number of times it appeared in the original 50-hour sampling frame. All results reported below are derived from the weighted sample. The videos were weighted in both of the current samples (Study I and Study II) for two related reasons. First, by statistically analyzing a video in relation to its frequency of appearance, the likelihood that a viewer would come into contact with that video in an actual viewing situation is taken into account. Music videos that were most likely to be seen during the selected time blocks carry the most statistical weight. Second, retention processes in SCT (the second subfunction of observational learning) suggest that the repetition of a model’s behaviors can impact whether the modeled information will be incorporated into the viewer’s memory codes and cognitive rehearsals (Bandura 2001). In the case of music videos, the behaviors of models that appear more frequently (via repetitions of a video) are more likely to be attended to and retained by a viewer when compared to those that appear less frequently.

Content Categories

There were two units of analysis in this study: (a) the video, and (b) the characters.

The Video

Genres

The following genres of music were coded for each video: (a) Rap, (b) Pop, (c) Rock, (d) Rhythm and Blues (R&B), (e) mixed Rap/R&B, and (f) Country. This nominal variable used genre codes based on Jones’ (1997) coding scheme, which included: Rap, Hip-Hop, Rock, R&B, and Country & Western. Unlike Jones, this study did not differentiate between Hip-Hop and Rap because Rap music can be thought of as a subset of the broader Hip-Hop culture (Pough 2004; Stephens and Phillips 2003). Two categories were also added to Jones’ original list (Pop and mix Rap/R&B) to reflect changes in music since the 1990s. Since the time Jones conducted his study, a new genre that blends Rap and R&B has emerged (Pough 2004). In this study it is called Mix Rap/R&B but has been more elegantly termed, “hip-hop soul” (Pough 2004, p. 172). The Pop category, which is somewhat common on MTV and MTV2 but not included in Jones’ coding scheme, was added to account for videos that do not fit in any other categories and videos that are noteworthy for their inclusion of bubbly, light, and innocuous music and lyrics. Table 3 gives a breakdown of the videos in this sample by genre and channel.

Table 3 Number of videos in sample, sorted by genre and channel (weighted).

Sexual Content

Silverman’s (1979) sexuality scale was used to code sexual content in each video. Prior research in music videos has not measured sexual content in a consistent way. However, the Silverman sexuality scale has been used in multiple content analyses of music videos (Gow 1990; Jones 1997; Sherman and Dominick 1986) and displayed strong intercoder reliability therein (Gow 1990; Sherman and Dominick 1986). The scale consists of 18 continuous variables that allow coders to count the number of instances of those sexual behaviors in each video. Specifically, the scale has two main parts. The first part focuses on 7 possible behaviors including: (a) kiss, (b) hug, (c) interpersonal touching, (d) suggestiveness and sexual innuendo, (e) heterosexual intercourse, (f) nonsexual aggressive bodily contact, and (g) physical contact with children. The second part of the Silverman sexuality scale deals with socially discouraged sexual practices. Eleven behaviors are included: (a) homosexuality, (b) incest, (c) pedophilia, (d) prostitution, (e) aggressive sexual contact, (f) exhibitionism, (g) fetishism, (h) masturbation, (i) transvestism, (j) voyeurism, and (k) other unnatural sexual behavior (including group sex).

From the Silverman sexuality scale a Sexual Behavior subscale was created. The subscale collapses a number of variables into one, allowing for a more graspable examination of the frequency of a variety of sexual behaviors. This Sexual Behavior subscale was used in data analysis to assess the frequency of instances of sexual behavior by combining all the Silverman behaviors from both the first part and second part of the scale, except those that were decidedly nonsexual (nonsexual contact with children, nonsexual aggressive contact). For example, if a video contained two instances of suggestive innuendo and one instance of group sex, it would garner a score of ‘3’ on the continuous Sexual Behavior subscale. Unless otherwise noted, all reported means for sexual acts reflect a video’s score on the Sexual Behavior subscale. One concern was the scale’s dated view on homosexuality as a socially discouraged sexual practice. Homosexuality does not have a place among the other 11 categories in the second part of this scale and for this reason it will not be connected with the other discouraged sexual contacts in any data-driven or theoretical way.

The Characters

Regular Characters

The first variable used to code regular characters was dichotomous and used to determine whether characters in the foreground of the video played a major or supporting role. Major and supporting characters were collapsed into one type, called “regular characters” in data analysis, due to a lack of salient differences between the two.

Demographics of Regular Characters

Basic demographic information was gathered for the regular characters, including gender and ethnicity. The ethnicity variable was nominal and demarcations were as follows: (a) African American, (b) Asian, (c) Hispanic/Latino, (d) White, and (e) Other. This variable was coded based on the performers’ appearance and coders’ previous knowledge about the racial identity of the performers.

Provocative/nonprovocative Clothing of Regular Characters

Level of undress of the regular characters was coded using an ordinal variable that measured provocative clothing on a 4-point scale: (a) nonprovocative clothing, (b) somewhat provocative clothing, (c) very provocative clothing, and (d) nudity. The 4-point scale was collapsed to a dichotomous variable (provocative or nonprovocative) in the first analysis of H2a to test the frequency of appearance of all types of provocative clothing at once. In the adjusted dichotomous clothing variable, the provocative clothing level included three of the four levels in the original 4-point scale: (b) somewhat provocative clothing, (c) very provocative clothing, and (d) nudity.

Background Characters

The first variable used to code background characters was dichotomous and used to determine whether or not a video contained this type of character. Background characters were a distinct type in this study and were kept separate from other characters during analysis. Background characters were defined as the group of people who make up the moving and living atmospheric backdrop of a music video. Large groups of fans, friends, dancers, and partygoers are common in all genres of music videos. Background characters were not present in all videos in Study I. The examination yielded a total of 367 background characters in the weighted sample.

Demographics of Background Characters

The same gender and ethnic demarcations were applied to background characters as were applied to regular characters with one additional level – Mixed. If the background of a video was made up of mostly Whites (an estimation of greater than 50% of the background characters), it was coded as a White background. If the background was made up of mostly African Americans (an estimation of greater than 50% of the background characters), it was coded as an African American background. However, if the background was made up of a mixture of Whites, African Americans, and/or other ethnicities (where an obvious majority could not be determined), that video’s background characters were coded as Mixed. Background characters were coded as Female, Male, or Mixed using the same methodology.

Provocative/nonprovocative Clothing of Background Characters

Level of undress of the background characters was coded using the same ordinal variable that measured provocative clothing of the regular characters. The 4-point scale has been collapsed to a dichotomous variable (provocative or nonprovocative) in the first analysis of H2b to test the frequency of appearance of all types of provocative clothing at once.

Coding Procedures

The main researcher in this study was also the first coder. Consistent with content analysis methods (e.g., Riffe et al. 1998; Wimmer and Dominick 2006), 20% of the 120 sampled music videos (24 videos) were randomly selected and independently coded by a graduate student enrolled in a communication program to provide information for intercoder reliability analysis. Before coding began, the second coder received extensive instruction on how to use all elements of the recording instrument, including more than 10 h of intensive training with a subset of videos not included in either Study I or Study II. After training and coding were completed, intercoder reliability was estimated using Krippendorff’s (1980) alpha. Percent agreement was calculated for those items that had very little variance or were artificially low (i.e., when an item was coded in limited numbers by one or both coders) (see Lombard et al. 2002). Because the variables were generally coded reliably, all data analyses are based on the coding of the first coder. The range of reliability coefficients of the coded variables was .52 to 1.00, the average was .90, and the median was .93. For the coding of a variable to be considered reliable it was required that Krippendorff’s alpha be .70 or higher and percent agreement be .90 or higher for items with minimal variance (see Lombard et al. 2002). One variable (implicit intercourse) fell just below this .70/.90 dual cut point, with a percent agreement score of .89, but was left in during analysis because the relatively low incidence of intercourse in the current sample and past music video research (Baxter et al. 1985; Gow 1990; Sommers-Flanagan et al. 1993) makes the reliability score appear artificially low. It is important to note that these reliability coefficients include combined intercoder data from both Study I and Study II. Please see Table 4 below for reliability coefficients and percent agreement (where necessary) for all the variables used in Study I and Study II.

Table 4 Reliability results for variables used in study I and study II.

Results

Using the Sexual Behavior subscale, which collapsed all the instances of sexual acts outlined in the Silverman (1979) sexuality scale, revealed sexual content of some kind occurred in 58.5% of all the music videos sampled in this study. Of the six musical genres, videos that mixed Rap and R&B displayed sexual content the most frequently (82.9%), followed by R&B by itself (78.9%), and Rap by itself (78.0%). Country videos depicted sexual content least often (35.6%), followed by Rock (36.8%) and Pop (52.9%). Table 5 gives further detail about the frequency of sexual content broken down by genre.

Table 5 Frequency of sexual acts in music videos by genre (weighted).

Of the 1,256 regular characters in the weighted sample, Whites held a slight majority in music videos (51.2%), African Americans made up more than a third of the total sample (42.1%), and other ethnicities were coded in limited numbers (6.7%). The majority of the regular characters, 73.6%, were male and only 26.4% were female.

H1 predicted African American music videos would feature significantly more sexual content than White music videos. This hypothesis was tested in two ways. The first test compared the number of the sexual acts included on the Sexual Behavior subscale in music videos classified in African American genres (Rap, R&B, Mixed Rap/R&B) with videos classified in White genres (Rock, Country). An independent-samples t-test comparing the mean number of sexual acts per video found a significant difference between the two groups: t (371) = 8.50, p < .001. The mean number of sexual acts in African American videos was significantly higher (M = 9.45, SD = 7.77) than the mean in White videos (M = 3.04, SD = 6.67).

It is important to note that the Pop genre was excluded from this first analysis of the first hypothesis because it included a mix of African American, Latina/o, and White artists; this racially mixed status in Pop videos is consistent with past music video research (Tapper et al. 1994). As the Pop category was the least frequently used genre in the current sample, accounting for only 8.4% of the weighted sample (n = 34/407), a relatively small number of videos are excluded by excluding the Pop genre.

A second analysis was also conducted to test the first hypothesis. Specifically, the frequency of sexual acts appearing in music videos performed by Black musical artists was compared to the frequency of sexual acts in videos performed by White artists. An independent-samples t-test comparing the mean number of sexual acts in videos found a significant difference between the two groups: t (380) = 8.85, p < .001. The mean number of sexual acts in videos that feature Black musical artists was significantly higher (M = 9.80, SD = 7.99) than the mean number of sexual acts in videos that feature White musical artists (M = 3.11, SD = 6.78).

H2a predicted African American regular characters would dress in significantly more provocative clothing than White regular characters. Two separate analyses were conducted to test this hypothesis. The two analyses allowed for this hypothesis to be tested both in terms of frequency of appearance of provocative clothing and degree of provocative clothing on a four-point scale. For the first analysis, a two-way chi-square test was calculated to compare the frequency of African American regular characters and White regular characters appearing in provocative clothing. A significant difference was found between African American regular characters and White regular characters in frequency of provocative dress: χ ² (1, N = 1172) = 26.11, p < .001. At the regular character level, which often included the main artist and supporting musicians, African Americans were significantly more likely to appear in provocative clothing of some kind than Whites. Table 6 gives further detail about the frequency of provocative clothing among regular characters.

Table 6 Crosstabulation of race of regular characters by type of clothing.

For the second analysis of H2a, a two-way ANOVA was run to test the interaction of gender and race in terms of the degree of provocative clothing worn by regular characters. A main effect for race emerged, as expected, F(1, 1168) = 55.34, p < .001, partial η ² = .05. Black regular characters were shown in clothing that was significantly more provocative (M = 1.60, SD = .62) than White regular characters (M = 1.35, SD = .53). A main effect for gender also emerged, F(1, 1168) = 270.30, p < .001, partial η ² = .19. Female characters were shown in clothing that was significantly more provocative (M = 1.76, SD = .73) than male characters (M = 1.20, SD = .42). A significant interaction between race and gender also emerged, F(1, 1168) = 54.97, p < .001, partial η ² = .05. Post-hoc t-tests with Bonferroni correction revealed Black females (M = 2.01, SD = .66) were shown in clothing that was significantly more provocative than Black males (M = 1.20, SD = .41). Additionally, White females (M = 1.50, SD = .72) were depicted in clothing that was significantly more provocative than White males (M = 1.20, SD = .43). Furthermore, post-hoc t-tests with Bonferroni correction revealed that Black female regular characters wore clothing that was significantly more provocative than any other regular characters (including White females) in this sample of music videos.

H2b predicted African American background characters would dress in significantly more provocative clothing than White background characters in music videos. Two separate analyses were conducted to test this hypothesis. The two analyses allowed for this hypothesis to be tested in terms of both frequency of appearance of provocative clothing as well as degree of provocative clothing on a four-point scale. For the first analysis, a two-way chi-square test was calculated to compare the frequency of African American background characters and White background characters appearing in provocative clothing. A significant difference was found between African American background characters and White background characters in frequency of provocative dress: χ² (1, N = 273) = 36.19, p < .001. At the background character level, African Americans were more than twice as likely to appear in provocative clothing than Whites and also more than twice as likely to appear in provocative clothing than nonprovocative clothing. Table 7 gives further detail about the frequency of provocative clothing among Black and White background characters.

Table 7 Crosstabulation of race of background characters by type of clothing.

For the second analysis of H2b, a two-way ANOVA was run to test the interaction of gender and race in terms of the degree of provocative clothing worn by background characters. A main effect for race emerged, as expected, F(1, 103) = 8.36, p < .05, partial η ² = .08. Black background characters were shown in clothing that was significantly more provocative (M = 1.66, SD = .63) than White background characters (M = 1.25, SD = .44). A main effect for gender also emerged, F(1, 103) = 39.17, p < .001, partial η ² = .28. Female background characters were shown in clothing that was significantly more provocative (M = 2.12, SD = .45) than male background characters (M = 1.20, SD = .41). A significant interaction between race and gender also emerged, F(1, 103) = 5.37, p < .001, partial η ² = .05. Post-hoc t-tests with Bonferroni correction revealed Black female background characters (M = 2.18, SD = .39) were shown in clothing that was significantly more provocative than Black male background characters (M = 1.22, SD = .42). However, White female background characters (M = 1.60, SD = .55) were not depicted in clothing that was significantly more provocative than White male background characters (M = 1.16, SD = .38). Finally, post-hoc t-tests with Bonferroni correction revealed that Black female background characters wore clothing that was significantly more provocative than any other type of background character in this sample of music videos.

Study II

The analyses in Study I do not tell the entire story of the presentation of sexuality in today’s African American music videos. For this reason, a second study with an additional sample of 20 videos from a program on Black Entertainment Television called BET Un:Cut was examined.

Recent research has shown that sex has grown increasingly prominent and frequent on television (Kunkel et al. 2003; Turner 2006) and moved away from sexual innuendo and suggestiveness towards more explicit sexual depictions (Arnett 2002; Brown and Newcomer 1991; Lampman et al. 2002; Truglio 1998). A recent examination of the content of two daytime music video shows on BET (Rap City and 106 & Park) and one on MTV (Sucker Free on MTV) suggests sexual content is increasing in music videos (Industry Ears 2008). In a December 2007 sample, Industry Ears found an average of 27 instances of sexual content per hour, or one instance every two minutes on the daytime music video programs. The researchers later found an even higher average of 40 instances per hour, or one instance every 90 s, in a second sample from March 2008 (Industry Ears 2008). Their report also shows that Rap City featured the highest frequency of sexual content among the three shows in both samples. Finally, in the two music video samples examined by Industry Ears there were 40 total depictions of mostly nude strippers, a sight made common on BET Un:Cut in the early 2000s (Neal 2005).

A turn towards explicit sex has been noted in Hip-Hop culture and Rap music videos (Kitwana 2004; Stephens and Phillips 2003). Specifically, Neal (2005) asserts, “BET Un:Cut [is] further evidence of the increasingly common relationship between hip-hop culture and the pornography industry” (p. 141). As music video director Nzingha Stewart puts it, “It’s almost like the other videos are like foreplay and the Un:Cut videos are the act themselves” (Neal 2005, p. 141).

The type of videos found on BET Un:Cut have become a topic of discussion and a point of contention in recent years. For example, VH-1 (VH-1 News Presents Hip Hop Videos Sexploitation on the Set) and BET (Hip-Hop vs. America Part I) have both aired programs that speak to the social implications of the sexual material and sex role stereotyping found in the Viacom music video catalog. In April 2004, Nelly, a well-known rapper, was scheduled to appear on campus at the historically Black school for women, Spelman College, in support of a bone marrow drive sponsored by his foundation 4Sho4Kids (Neal 2005). Nelly’s visit was cancelled after Asha Jennings, president of the Student Government Association at Spelman, saw Nelly’s “Tip Drill” video on BET Un:Cut. The video (which is represented in Study II’s sample) prompted Jennings to require Nelly to speak on a panel about the portrayals of women in Hip-Hop music videos (Neal 2005) if he was to be allowed to come in support of the bone marrow drive. Nelly declined and pulled his organization’s involvement. Even Hip-Hop fans are concerned about the “hypermasculinity of blackmanhood” and the related idea that “black female bodies are disrespected en masse” in current Hip-Hop videos (Hurt 2007, 2008).

What is Un:Cut?

BET Un:Cut is a one-hour music video program that aired from 3 a.m. to 4 a.m. Thursdays, Fridays, and Saturdays on BET. It was not directed towards adolescents as other music video programming typically is but was instead intended for a mature, late-night demographic (as noted in a disclaimer at the outset of the program). The videos found on this program were comprised of Rap videos performed by well-known artists like 50 Cent, Nelly, and Lil Jon as well as relative unknowns such as Twip, Filthy Rich, and 2 Sense. Un:Cut videos are noteworthy for their dependence on sexual themes, which are presented in a mostly unedited and uncensored format. Music videos on BET Un:Cut represent a distinct genre of televised popular music that has not been covered in social scientific literature.

After six years on the air Un:Cut was canceled in 2006 by Viacom, perhaps due to outside pressure and controversy surrounding the show. Despite the cancellation, Un:Cut style music videos are still being made. A recent Google search uncovered an online classified advertisement posted on March 25, 2008 that noted, “[We are] looking for a video crew & director for BET UN-CUT style video to be shot in Chicago” (Street Hop Productions 2008). Un:Cut content remains prevalent on Internet sites like YouTube and Google Video as well as in DVD releases (Daltons 2007). A brief examination of the availability of Un:Cut videos on YouTube was conducted in April, 2008. This examination revealed that Un:Cut music videos are available and quite popular online. All Un:Cut videos in the current sample were successfully located on YouTube. A comparison of five videos from the Un:Cut sample and five popular Rap videos from the regular sample revealed that Un:Cut videos are actually viewed more frequently than the regular videos. The Un:Cut videos averaged over 3 million views on YouTube while the regular videos averaged just under 1.5 million views.

Hypotheses and Research Questions – Study II

The main research question driving Study II asks: Are Un:Cut videos more dependent on sexual content than music videos not found on Un:Cut? Previous researchers (Kitwana 2004; Stephens and Phillips 2003) have argued that in general Rap music videos have recently taken a turn toward explicit sex. Recent empirical research supports this claim (see Industry Ears 2008). Additionally, Neal (2005) argues that Un:Cut’s programming acts as evidence for a growing relationship between Hip-Hop culture and the pornography industry. Based on the previous literature the following hypotheses were tested in the current study:

• H1Un:Cut music videos will depict significantly more sexual content than current music videos recorded at other times on the five major music video channels (including BET).

• H2Un:Cut music videos will be significantly more likely to depict background characters in provocative clothing than music videos recorded at other times on the five major music video channels (including BET).

It is important to clarify why only background characters’ (and not regular characters’) clothing was analyzed for H2. As will be covered in further detail in the Results section, almost all regular characters in the Un:Cut sample were coded as fully clothed voyeuristic African American males who could be seen watching naked or nearly naked African American female background characters engage in sexual behaviors. For this reason it made little conceptual sense to analyze the clothing of the regular characters in the Un:Cut sample in relation to the regular characters seen on the five major video channels.

Quoting a music video director, Neal (2005) points out that music videos found on the five major channels merely depict sexual foreplay when compared to Un:Cut videos, which depict the sex acts themselves. Based on this claim the following research question was examined in the current study:

• RQ1 Will there be content differences between Un:Cut videos and videos recorded at other times on the five major music video channels (including BET) with respect to discouraged sexual behavior, nonsexual behavior, and homosexual behavior?

Method

Sample

A sample of music videos was recorded on videotape (Thursday, Friday, and Saturday only) between January 7 and March 2, 2005 from 3 a.m. to 4 a.m. on BET. Due to frequent nonairings of the Un:Cut program, it took nearly two months to get 10 h of music video programming or one-fifth of the total hours (50) recorded in Study I. The ten hours of sampled programming from Un:Cut rendered 46 recorded videos of which 28 were unique. Of those 28 unique videos, 20 were randomly selected for analysis. This rendered a randomized sample that was one-fifth the number of videos in Study I. As in Study I, duplicate videos were accounted for statistically during data analysis. Each video was weighted according to the number of times that video appeared in the original 10-hour sampling frame. All results reported below are derived from the weighted sample.

Content Categories

As in Study I, there were two units of analysis in this study: (a) the video, and (b) the characters. Largely the same procedures and variables were used in Study II that were used in Study I; exceptions to this rule are discussed below.

The Video

Genres

In Study II, genres were not coded as only one genre (Rap) appears on Un:Cut.

Sexual Content

As in Study I, Silverman’s (1979) sexuality scale was employed to assess the level of sexual content in Un:Cut videos. As in the first study, the Sexual Behavior subscale was used during data analysis. The Sexual Behavior subscale was used to sum and assess the frequency of instances of sexual behavior. Two additional sexuality subscales were created for data analysis in the second study. These scales were not used in Study I. Like the Sexual Behavior subscale, the two additional subscales collapse a number of variables into one, allowing for a more graspable examination of the frequency of a variety of similar behaviors. The first additional subscale, Nonsexual Behaviors, included only the most innocuous depictions of intimacy as coded in the first part of the Silverman Scale: (a) Kissing, (b) Hugging, and (c) Affectionate Touching. The second subscale, Discouraged Sexual Behaviors, included 10 of the 11 sexual behaviors coded in the second part of the Silverman scale (Pedophilia, Aggressive Sex, etc). Homosexuality was excluded from this subscale and examined independently.

The Characters

Characters in Study II were coded using the identical procedures carried out in Study I. For that reason, only one slight difference between the two studies warrants coverage here. Whereas not all videos in Study I contained background characters, all 20 of the Un:Cut videos in Study II displayed this second-level of characterization.

Coding Procedures

Forty percent of the 20 sampled Un:Cut videos (8 videos) were randomly selected and independently coded by a second coder. As in Study I, all data analysis is based on the coding of the first coder. The first and second coders were the same as in Study I. Before coding any Un:Cut videos, the second coder received instruction on how to properly use all elements of the recording instrument with a subset of Un:Cut videos not included in the Study II sample. Additional training was necessary because Un:Cut video content required the incorporation of numerous parts of the Silverman Sexuality scale (1979) that went mostly unused in Study I. After training and coding were completed, intercoder reliability was estimated using Krippendorff’s (1980) alpha. Reliability coefficients reported in Study I are a combination of intercoder data from both Study I (24 videos) and Study II (8 videos).

Results

Sexual content of some kind occurred in 95.0% of the Un:Cut music videos as compared to 58.5% in videos sampled in Study I. H1 in Study II predicted Un:Cut videos would depict a significantly higher number of the sexual acts included on the Sexual Behavior subscale than music videos recorded at other times on the five major music video channels (including BET). To test this hypothesis, an independent-samples t-test comparing the mean number of sexual acts per music video from Study I and the mean number of sexual acts per video in the Un:Cut sample was conducted. This analysis revealed a significant difference between the two groups in the predicted direction: t (442) = 17.00, p < .001.The mean number of sexual acts in the Un:Cut videos was significantly higher (M = 49.22, SD = 43.49) than the mean in the Study I videos (M = 6.39, SD = 8.17).

In terms of the character-level analyses, the Un:Cut weighted sample was made up 105 characters, who were almost entirely African American (92.4%) males (94.3%). Background characters were made up entirely of females (51.4%) and mixed males/females (48.6%). H2 in Study II predicted Un:Cut videos would be significantly more likely to depict background characters in provocative clothing than videos recorded at other times on the five major music video channels (including BET). To test this hypothesis a two-way chi-square test was calculated to compare the frequency of African American background characters and White background characters appearing in provocative clothing. A significant difference was found between background characters in the Un:Cut videos and background characters in the videos from the five major video channels in terms of frequency of provocative dress: χ² (1, n = 404) = 27.66, p < .001. Un:Cut videos displayed significantly more background characters (97.3%) in sexy clothing than regular videos (52.3%).

The first (and only) research question in Study II asked whether there would be content differences between Un:Cut videos and videos recorded at other times on the five major music video channels (including BET) with respect to discouraged sexual behavior, nonsexual behavior, and homosexual behavior. This research question was examined in three ways. The first examination called upon the Discouraged Sexual Behaviors subscale to compare the mean number of socially discouraged sexual behaviors (such as Prostitution, Aggressive Sex, and Exhibitionism) depicted in the Un:Cut videos with the mean number of socially discouraged sexual behaviors depicted in the videos collected from the five major video channels. An independent-samples t-test revealed a significant difference between the two groups: t (442) = 16.72, p < .001. The mean number of discouraged sexual acts in the Un:Cut videos was significantly higher (M = 3.59, SD = 3.76) than the mean from the videos collected on the five major channels (M = .11, SD = .59).

The second examination of the first research question called upon the Nonsexual Behaviors subscale to compare the mean number of nonsexual behaviors (such as kissing, hugging, and intimate touching) depicted in the Un:Cut videos with the mean number of nonsexual behaviors depicted in the videos collected from the five major video channels. An independent-samples t-test revealed a significant difference between the two groups: t (442) = 3.45, p = .001. The mean number of nonsexual behaviors in the videos collected on the five major channels was significantly higher (M = 4.06, SD = 6.77) than the mean in the Un:Cut videos (M = .22, SD = .48).

The third examination of the first research question compared the mean number of homosexual behaviors depicted in the Un:Cut videos with the mean number of homosexual behaviors in videos collected from the five major channels. An independent-samples t-test revealed a significant difference between the two groups: t (442) = 13.35, p < .001. The mean number of homosexual behaviors depicted in the Un:Cut videos was significantly higher (M = 7.54, SD = 11.47) than the mean in the videos collected from the five major channels (M = .03, SD = .16).

The portrayal of homosexual behaviors requires a brief qualitative analysis in addition to the above quantitative analysis. Homosexual behaviors were qualitatively different in the videos found in regular rotation on the five major video channels when compared to the Un:Cut videos’ treatment of homosexuality. In the Un:Cut videos, behaviors coded as homosexual acts were in all cases two or more females enacting sexualized lesbian behaviors often for the entertainment and titillation of male onlookers present in the video. For example, in one of the Un:Cut videos two women who were dressed in lingerie were shown pouring chocolate syrup on one other while kissing and rubbing their bodies together. During this scene male onlookers were shown enjoying the performance and encouraging the two women to continue.

On the other hand, in the sample pulled from the five major video channels, only 3 of the 120 unique videos depicted homosexual behaviors of any kind (compared to 11 of the 20 unique videos in the Un:Cut sample). Of those three videos in the major channel sample, two videos showed realistic depictions of two males engaging in intimate homosexual behaviors. In both cases, the depictions of homosexual behaviors were included as part of a serious/dramatic narrative being carried out in the video’s storyline, not for purposes of titillation or voyeurism by other characters in the video. The third video containing a homosexual act in the Study I sample, portrayed two women engaging in lesbian sexual behaviors for the benefit of male onlookers in a similar but less graphic fashion as seen in the Un:Cut videos.

General Discussion

The results of Study I are clear: African American music videos are more sexualized than White music videos both in terms of provocative clothing and frequency of depictions of sexual behaviors. Videos featuring traditionally African American genres displayed significantly more sexual content than genres that typically feature White performers. Study I also shows that videos performed by Black musical artists feature significantly more sexual content than videos performed by White artists. This finding clarifies that frequency of sexual content is not merely an artifact of musical genre but is specifically related to the race of the performing artist(s). When one adds to this that African American regular characters and background characters were significantly more likely to be portrayed in provocative clothing than White characters, it becomes clear that sexuality is deployed at unequal rates along racial lines in music videos.

Inequities occurred not just along racial lines in Study I but were manifest along gender lines as well. The current results support previous research that found women to be underrepresented in music videos (Gow 1996; Seidman 1992; Sommers-Flanagan et al. 1993). In the Study I sample, male characters outnumbered female characters by a rate of nearly 3-to-1. Furthermore, female characters (at both the regular and background levels) were significantly more likely to appear in provocative clothing than males. Women, then, are less likely to appear in music videos than men and more likely to be sexualized by their clothing.

This reality might very well point a gender bias that exists in Western society generally and mass mediated messages specifically. Alternatively, these findings might simply be an artifact of the coding process, which is necessarily linked to coders’ understandings biological realities (i.e., that female bodies can arguably be more readily exploited by provocative clothing than males’). This biological difference, then, can perhaps lead viewers (including coders) to consider female clothing as more provocative than similar male clothing. However, this alternative biologically-based reading is less applicable when looking at the interaction of race and gender in terms of provocative apparel. Specifically, results from Study I also reveal that African American female characters (at both the regular and background levels) were significantly more likely to appear in provocative clothing than any other character type, including White females.

Music videos are but one determinant of a young person’s understanding of sexuality. A young person’s social surroundings and her/his internal cognitive processes always mediate and moderate media influences (Bandura 2001). That being said, social cognitive theory suggests the structure of music videos makes them an influential media form that can contribute to deleterious sex role development by skewing adolescents’ understanding of the possible results of their actions as well as their understanding of their own masculinity and femininity.

The attentional processes in SCT suggest a compounding effect wherein prestigeful and attractive characters draw a viewer’s attention to behaviors (sexual or otherwise) in a video and in turn the sexual content increases attention to and liking of the video generally. The motivational processes of SCT are useful to isolate the group of consumers most likely to model the compounded sexualized gender role complex found in music videos. Specifically, a model’s similarity to an attendee in terms of race and gender increases the likelihood that the learned behavior will be modeled. SCT and results from Study I suggest that when compared to all other audience members, young Black women are most likely to attend to the relatively frequent sexual content and sexualized apparel choices in music videos featuring African American genres and performers and are most likely to be motivated to act out these behaviors in their own lives. That being said, results in previous research show that sexualized media messages tend to exert less impact on young Black peoples’ sexual behavior when compared to their White counterparts (Somers and Tynan 2006; Brown et al. 2006). The possible theorized effects of sexual content in African American music videos, which are based on SCT and the current content-based findings, then, should be interpreted in light of these two previous studies.

Un:Cut videos, which are now available on video-sharing websites and DVDs, are made up entirely of Rap videos performed almost exclusively by African American male musical artists. The Un:Cut videos in Study II averaged more than seven times as many sexual acts as videos in Study I. Results also show that the type of sexuality in the Un:Cut videos depict a raw version of sex noteworthy for reduced depictions of innocuous intimate behaviors like hugging and kissing. Additionally, discouraged behaviors such as voyeurism (mostly males watching female strippers perform), group sex, and lesbian sex performed for the gaze of male onlookers occurred significantly more often in the Un:Cut sample than in the regular sample. The only gender role performed by Black women in Un:Cut videos was a sexual one used for the entertainment and stimulation of Black men. All other gender roles are conspicuously absent. While African American men dominated the visual foreground in Un:Cut videos, African American women were relegated to the background and were almost always portrayed in revealing bathing suits, lingerie or nothing at all. African American women, then, are visually exiled to a secondary gender status and given no opportunity to enact complex and fulfilling gender roles that go beyond what the women offer sexually.

The above results have implications for young people’s understanding of gender roles beyond purely sexualized ones. In a recent content analysis of occupational diversity and occupational prestige in relation to a casts’ racial diversity on network primetime television programs, Signorielli (2009) found “women, particularly Black women in the mostly minority programs have the least diversity and prestige in terms of the jobs in which they are cast” (p. 347). This arrangement seems to also bare out in the Un:Cut videos, which have the least racial diversity of any of the videos examined in either study, wherein Black women only have access to occupations that lack prestige, stripping and the other sexual professions. Brown and L’Engle (2009) found that exposure to sexually explicit content not only predicted early adolescents’ sexual attitudes and behavior but also predicted less progressive gender role attitudes in both young men and young women. In short, explicit Un:Cut content as well as less explicit content from more innocuous music videos might impact young people’s understanding of masculinity and feminity in the working world and other arenas beyond sexual and relational ones.

Current results from the Un:Cut sample also signal a growing relationship between Hip-Hop and pornography (Kitwana 2004; Neal 2005; Stephens and Phillips 2003) and might be an indication of movement towards more explicit sexual depictions in music videos. When one considers that Un:Cut style videos are well represented and frequently viewed on the Internet and that these types of videos are still being produced for DVD and Internet release, the content sampled from Un:Cut remains wholly relevant. These videos are no longer limited to late nights on the weekend because they are now available at any time to anyone with a computer and an Internet connection. In this way, the movement of Un:Cut video content can be viewed as emblematic of a larger movement of fringe television content taking a more central role in our mediated lives as it is displaced to our computer screens.

Social prompting effects of SCT (Bandura 2001) allow for a more expansive interpretation of what sexual content in music videos (Un:Cut or otherwise) might mean for young consumers regardless of where the videos are viewed. Johnson et al. (2000) found that exposure to violent Rap music “produced more negative responses to Black than to White target persons on stereotype-related dimensions” (p. 247) for both Black and White subjects. Johnson and his colleagues note their results “suggest a unique vulnerability of Blacks” (p. 248), wherein Blacks respond in stereotypic ways to media portrayals of Blacks as much as Whites do. Similarly, Dixon and Linz (1997) found that a racially mixed group of college students interpreted a highly sexually explicit Rap song as more patently offensive than an equally sexually explicit non-Rap song. SCT’s social prompting effects together with the aforementioned studies suggest stereotypical gender roles in Rap music support and reinforce beliefs about a demonized Black sexuality (Dixon and Linz 1997) already common in the U.S.

Specifically, we see the perpetuation of the hypermasculine Black male buck (see Neal 2001) and the objectified and overly sexualized one-dimensional Black female who is reduced to “decorative eye candy” (Emerson 2002, p. 123). To better understand these sexualized social prompting scripts it is important to examine how they formed within a racialized and sexualized socioeconomic context (Stephens and Phillips 2003). Drawing attention to Rap and R&B music videos as distinctly and overtly sexualized without looking at socioeconomic antecedents in the music industry would fall short of a complete analysis of the current state of affairs in music videos.

The concept of hegemony has been used to examine the way song lyrics and popular music videos are a continuation of dominant conceptions of sexuality, gender, and race (see Abrams 1995). “Within the U.S. culture, racist and sexist ideologies permeate the social structure to such a degree that they become hegemonic, namely, seen as natural, normal, and inevitable” (Collins 2000, p. 5). Through this normalization process existing power relations and social realities are concealed and naturalized (Shuker 2002). The media play a key role in this process and popular cultural forms like popular music songs and videos are able to contribute to and reinforce this theorized macro-social-consciousness (Frith and McRobbie 1990; Shuker 2002).

Through a White-controlled music industry, record executives have for some time been able to exploit Black culture through images that effectively dominate Black bodies (Kelley 2002). Currently, the music industry is led by five major media conglomerates, all White-owned, that control as much as 85% to 90% of the recording, production, and distribution of popular music in the U.S. (Kelley 2002). In terms of consumption, it is estimated that White males purchase about 70% of Rap music in the U.S. (Kelley 2002). Stephens and Phillips (2003) argue that the sexualized African American woman’s body has been formed against the backdrop of neocolonialism by media elites who sell Black youth culture to White mainstream consumers. Chapple and Garofalo (1977) point out that “the [music] business is a particularly interesting and glaring example of this pervasive racism, since it has grown fat off the creative impulses and culture of black people. Through the years black music has been stolen, covered, and pilfered in more subtle ways, while its artists have been effectively suppressed and denied their due” (p. 267). Because of limited African American involvement in the decision-making process of the music industry, Black musicians have been denied agency and access to the public sphere (Pough 2004). For example, in the early 1980s, MTV was criticized in both the social scientific literature (Brown and Campbell 1986; Peterson-Lewis and Chennault 1986; Williams 2003) and in the popular press (Garofalo 2002a) for exclusively airing videos performed by White artists.

Pough (2004) argues, “For a historically marginalized and invisible group, the spectacle is what allows them a point of entry into a public space that has proved to be violent and exclusionary” (p. 29). It is asserted here that sexual content is used to create a successful spectacle as an attention-getting device in order to garner increased liking of music videos. The concern is that White-controlled Viacom video channels might be more willing to play Rap and R&B videos that perpetuate racial inequities and sexual stereotypes and less willing to play counter-hegemonic songs that might bring about positive outcomes for the African American community (Bynoe 2002; Hurt 2007). To overcome their own exclusion from music video channels, young Black performers have called upon the spectacle of sex and as a result act in ways that conform to the recording industry’s conception of Blackness. Data from the current study suggests that Blacks no longer struggle for a place on MTV, BET, and the like (36% of the 376 characters analyzed in this study were African American), but their increased prominence may be due to an unspoken agreement to rely on sexual spectacle to sell music videos.

Limitations

The samples in both Study I and Study II were limited in a number of ways. First, neither sample was randomly selected, which can affect the generalizability of the results included in these studies. That said, a purposive sample allowed for the selection of time slots and music channels that were most likely to contain music videos (rather than other forms of programming) watched by young people. Specifically, the sample in Study I was limited to weekday mornings. This decision was made in part because this was the only time all five channels (MTV, MTV2, BET, VH-1, and CMT) aired music videos. Additionally, the selection of weekdays is not completely without warrant. Sun and Lull (1986) reported that the high-school-aged MTV viewers they surveyed watched videos almost as much on weekdays (124 min per day) as they did on weekends (132 min per day). Together these factors suggest a weekday morning sample was appropriate. Additionally, by recording only two hours of programming each day, the Study I sample was collected over five weeks, which provides greater generalizability than would a sample taken over the course of one day or one week.

The sample in Study I was further limited because it was confined to five channels, all of which are owned and operated by the same parent company, Viacom (Viacom: Brand Index 2008). A pretest, however, indicated that the Viacom channels were more popular among the young people sampled than those channels not owed by Viacom. The current sample, then, is strong with respect to describing content most likely to be modeled and socially learned.

The current sample was also limited by the fact that not all the recorded videos were analyzed. A more complete picture of the music video milieu would have been garnered had all 200 unique videos (172 in Study I and 28 in Study II) been coded. On the other hand, use of random selection increased the likelihood that the most frequently played videos in the current sample were analyzed. Also, the use of a weighted sample increased the number of videos statistically analyzed from 140 to 444 (407 in Study I and 37 in Study II). Finally, the sub-sample of the BET Un:Cut videos was limited by its size. Despite this being the first look at an alternative form of music videos, a larger sample of Un:Cut programming would have enabled a more complete understanding of the ways sexuality was treated by the video producers.

This research project’s methodology was also limited during the coding process. The author and main researcher was the first coder. A second coder analyzed 20% of the videos in the Study I and 40% of the videos in study II. Acceptable intercoder reliability scores using Krippendorff’s (1980) alpha and percent agreement suggest that the results in this study are not based solely on the human biases of the main author but are instead based on explicit operational definitions, which controlled assignment of content to definitive categories (see Riffe et al. 1998). Use of a measure (Silverman’s sexuality scale) that has consistently been reliable in previous music video research should further reduce researcher bias. However, researcher bias could have been further reduced by increased numbers of independent coders. Future research on music videos should strive to utilize multiple independent coders. Coding in the current research was further limited because reliability analyses were calculated and reported as combined scores between Study I and Study II. Keeping these scores separate would have allowed for differentiation between coding inconsistencies that might have occurred in one study but not the other. Finally, the coding process in the current research was limited by the fact that the genre variable did not include an “Other” category. Although this did not appear to greatly impact the reliability of the genre variable (alpha = .87), to increase the face validity of the variable and to ensure flexible and accurate coding of musical genres, future studies should include an “Other” category.

Conclusions

Methodological limitations aside, the results from Study I and Study II clearly show that an unequal sexual reality between African Americans and Whites and an unequal gender role arrangement between men and women is depicted in music videos. These findings, when examined under the purview of the social cognitive theory, indicate that African American women are most likely to model the deleterious behaviors portrayed in music videos. The probability of this notion deserves further attention because AIDS diagnoses for African American women are 24 times the rate for White women and African Americans have the highest rates of sexually transmitted diseases among any ethnic group in the U.S. (Centers for Disease Control and Prevention 2007). More research is needed to look at possible relationships between sexualized music videos and sexual health risk behaviors. Future research should also examine how social and cognitive influences mediate and moderate the observational learning effects that SCT purports music videos have on young audience members. Furthermore, future research should implement longitudinal analyses to determine whether or not sexual content in music videos is actually increasing in explicitness as Study II’s results suggest. Lastly, both critical and social scientific researchers should continue to examine how hegemonic and economic realities in the music industry contribute to the production of particularly sexualized Black music videos.