From ‘‘I Want To Hold Your Hand’’ to ‘‘Promiscuous’’: Sexual Stereotypes in Popular Music Lyrics, 1960–2008

Media content analyses indicate that gender-based differences in sexuality are common and consistent with gender stereotypes. Speciﬁcally, women are expected to focus on love and romantic relationships and have sexually objectiﬁed bodies, while men are expected to focus on sexual behavior. Although decades of research have documented the presence of these stereotypes in a broad variety of visual media, much less is known about the content of popular music lyrics. Relying on a database of 1250 songs across ﬁve decades (the top 50 songs from even-numbered years from 1960 through 2008), we documented the presence or absence of a dating relationship, the word ‘‘love’’ (and its uses), sexual activity, and sexual objectiﬁcation of females and males (separately). Analyses revealed that the vast majority of songs addressed at least one of these themes, primarily dating relationships. Although female performers were proportionally more likely to address romantic relationships than male performers, raw counts reversed this pattern because male performers substantially outnumbered female performers. Males were proportionally more likely to sing about sexual behavior and to objectify both females and males. References to romantic relationships became less common over time, while references to sexual behavior and objectiﬁed bodies became more common. Content varied across genres, with rap being the least likely to reference dating and most likely to reference sexual behavior. Implications for sexual development are discussed.

Abstract Media content analyses indicate that gender-based differences in sexuality are common and consistent with gender stereotypes. Specifically, women are expected to focus on love and romantic relationships and have sexually objectified bodies, while men are expected to focus on sexual behavior. Although decades of research have documented the presence of these stereotypes in a broad variety of visual media, much less is known about the content of popular music lyrics. Relying on a database of 1250 songs across five decades (the top 50 songs from evennumbered years from 1960 through 2008), we documented the presence or absence of a dating relationship, the word ''love'' (and its uses), sexual activity, and sexual objectification of females and males (separately). Analyses revealed that the vast majority of songs addressed at least one of these themes, primarily dating relationships. Although female performers were proportionally more likely to address romantic relationships than male performers, raw counts reversed this pattern because male performers substantially outnumbered female performers. Males were proportionally more likely to sing about sexual behavior and to objectify both females and males. References to romantic relationships became less common over time, while references to sexual behavior and objectified bodies became more common. Content varied across genres, with rap being the least likely to reference dating and most likely to reference sexual behavior. Implications for sexual development are discussed. & Andrew P. Smiler andrew.smiler@gmail.com

Introduction
Content analyses have revealed a substantial amount of sexual content in television programs, magazines, romance novels, music videos, and video games (Downs and Smith 2010;Hust et al. 2008;Kunkel et al. 2005; Romance Writers of America 2011 ;Taylor 2005;Turner 2011), including those products marketed towards and preferred by adolescents (Carpenter 1998;Jhally 1994Jhally , 2007Joshi et al. 2011;Wallis 2011;Ward 1995). Reading across studies, the analyses indicate a small, nearly ubiquitous set of messages consistent with gender stereotypes. Specifically, female characters are typically positioned as focused on dating relationships, interested in love, and ambivalent about sexual activity while male characters are typically presented as wanting sex and being uninterested in dating relationships or love. Further, sex is typically presented as occurring outside of marital relationships (and often outside of romantic relationships), almost never leads to sexually transmitted infections (STIs), and rarely leads to unplanned pregnancy (Arnett 2002;Brown and Bobkowski 2011;Clawson 2005;Downs and Smith 2010;Hust et al. 2008;Taylor 2005;Turner 2011). There is little doubt adolescents learn about dating and sexuality from the media (Brown et al. 2006;Collins et al. 2003;Epstein and Ward 2008; see reviews by Shewmaker 2015; Sutton et al. 2002;Ward 2003), which implies that understanding the full range of sexuality messages is of great importance. Theories and theorists have specified a variety of mechanisms by which this occurs, with more recent models highlighting the interaction between teens and their media (Bleakley et al. 2008;Giles and Maltby 2004;North et al. 2007;Steele and Brown 1995).
When the focus is shifted from learning to enacted behaviors, researchers have found some effects of music consumption on the behavior of adolescents. For example, in a 3 year longitudinal study of adolescents, Martino et al. (2006), found that greater levels of exposure to degrading lyrics were related to a younger age at first intercourse and also a faster progression through non-coital behaviors. In a laboratory setting, undergraduates who were exposed to 3 min of sexual lyrics rated a hypothetical job applicant as sexier than their peers who were exposed to nonsexual lyrics (Carpentier 2014). At the same time, a recent meta-analysis revealed that the relationship between adolescents' media consumption and both initiation of sex (i.e., first coitus) and risky sexual behaviors were very small, although larger for boys than girls (Ferguson et al. 2016/in press).
Although the database on media content is rather substantial, it is limited by a general absence of information regarding music (Arnett 2002;Brown and Bobkowski 2011;Roberts and Christenson 2001). This absence is particularly noteworthy because nearly all adolescents listen to music regularly. Studies indicate an average listening time of at least 5 h per day (Kistler et al. 2010;LaFerle et al. 2000;Primack et al. 2009;Ward et al. 2005; but see Rideout et al. 2010 for a lower estimate). Further, adolescents' preferences are the primary determinant of the year's top songs (Arnett 2002;Whitburn 1986) and thus impact the larger culture. Given music's ubiquity, use, and cultural importance, some reviewers have suggested that ''listening to popular music is properly seen as a natural and generally benign part of growing up in contemporary Western society'' (Roberts and Christenson 2001, p. 398).
Accordingly, we sought to determine if the content of popular music contains the same gender-based stereotypical depictions of sexuality that have been documented in visual media. Broadly speaking, we examined the presence of dating relationships, love, sexual activity (i.e., orgasm producing sexual behaviors), and sexual objectification in 1250 of the most popular songs from 1960 through 2008. We also explored variations based on performer's gender, decade, and musical genre.

Music Content
A small number of lyrical reviews of the romantic and sexual content have been published, and these provide the starting point for our work. An analysis of Billboard Magazine's ''all time 100 most popular songs'' from 1958 to 1998 indexed references to love and sex, separately, per lyrical line (i.e., referent words per line of text). The researchers found that songs by female performers contained more lyrical references per line to both love and sex than male performers (Dukes et al. 2003). Mean scores indicated that love words appeared approximately three times more frequently than sex words (Dukes et al. 2003). A later and more detailed study of the Top 100 songs per year from 1959, 1969, 1979, 1989, 1999, and 2009 found that males were more likely than females to include sexual references throughout the time period, although the magnitude of the difference decreased in more recent decades (Hall et al. 2012). We note that this analysis did not explicitly assess dating or love. Another study adopted an evolutionary psychology approach to lyrical content and thus classified all references to dating relationships, love, and sexual behavior as ''reproductive messages'' (Hobbs and Gallup 2009). They found that more than 90% of popular music titles contained at least one reproductive message, but they did not examine differences based on sex of performer. Accordingly, we hypothesize that (H1) female performers will be more likely than male performers to address romantic themes, consistent with Dukes et al. (2003), and that (H2) female performers will be less likely than male performers to address sexual themes, consistent with Hall et al. (2012) and contrary to Dukes et al. (2003).
Analyses have also documented changes over time, often in interaction with other constructs. Dukes et al. documented a decrease in love words per lyrical line over time that was driven by female performers; male performers' discussion of love remained unchanged in their analysis (and was less frequent than female performers). They also reported that women's use of sex words decreased over time, mostly after 1990, while men's use of sex words increased in a linear fashion over time. Hall et al. (2012) also reported an increase in sexual content by female performers, noting levels near zero in 1959 and 1969; they reported that the rate of change was inconsistent across the subsequent 10 year gaps in their study. Accordingly, we anticipate that (H3) the number of songs with dating references will decrease over time and (H4) the number of songs with sexual references will increase over time.
Some analyses have documented differences based on the song's musical genre. We note that musical genres, and the Billboard rankings that track them, do not have static boundaries over time. Of particular importance for this discussion is rap, which gained mainstream acceptance in 1992. Dukes et al. (2003) combined rap and hip hop into a single category, while Hobbs and Gallup (2009) included rap as part of R&B (Ryhthm & Blues). Both studies followed then-current categorizations by Billboard. In the Dukes analysis, more than half of all songs were love songs, with substantial differences by genre (R&B: 96%, rock: 85%, rap/hip-hop: 59%). In the Hobbs and Gallup study, reproductive messages were more common in the R&B genre than in country or rock, which did not differ from each other. Both of these studies suggest that the rap genre is key to these differences. Several content analyses of rap lyrics and rap videos have found relatively high levels of sexual content, hypersexuality, objectification of women, sexism, and misogyny (Herd 2015;Hurt 2007;Weitzer and Kubrin 2009). Accordingly, we hypothesized that (H5) dating content would vary by genre, with relatively low levels of this content appearing in rap lyrics and (H6) sexual content would vary by genre, with relatively high levels of this content appearing in rap lyrics.
Although we do not test these issues directly due to cell size issues (see method), we also note that longitudinal changes based on genre and performer's ethnicity have also been documented. For example, Hobbs and Gallup reported that reproductive messages were more common in the R&B genre than the rock and country genres in 1999 and 2009, but not in the four prior decades; references in rock and country songs were not statistically different from each other in any decade. Hall et al. (2012) found that non-white performers were more likely to have sexualized lyrics than white performers in 1999 and 2009, but not in the four decades prior.

Sexual Objectification
Sexual objectification of bodies in the media, particularly female bodies, has been the subject of several analyses of visual media. Regarding lyrics, Hobbs and Gallup (2009) assessed three messages that potentially reflect sexual objectification (genitalia references, other body part references, and sex appeal). They found each of these themes to be most common in R&B and least common in country (in 2009), while also demonstrating that each of these three themes appeared at similar rates when all songs from 2009 (combined) were compared to a selection of songs from the opera and art genres (separately). Analyses of visual media as diverse as magazines, music videos, video games, television shows, and pornography, indicates that women are routinely objectified (Downs and Smith 2010;Fouts and Burggraf 2000;Hurt 2007;Jhally 1994Jhally , 2007Picker and Sun 2008;Taylor 2005;Turner 2011;Wallis 2011). Men are also objectified in these media, although at lower rates than women (Downs and Smith 2010;Fouts and Vaughan 2002;Turner 2011;Wallis 2011).
A detailed examination of five decades' worth of cover pictures from Rolling Stone magazine illustrated this gender-based differential. Although men appeared on the cover three times more often than women, women's rate of sexual objectification was approximately four times higher in every decade and the most highly sexualized male image had a score near the median level of sexualization of female images (Hatton and Trautner 2011). The Rolling Stone analysis also demonstrated that sexualized images have become more common over time for both women and men. For men, this finding is consistent with other data indicating a greater emphasis on men's body image and appearance (Edwards et al. 2014;Pope et al. 2000). We hypothesize that objectification will increase over time for women (H7a) and men (H7b), separately, and that (H8) rates of objectification will vary by genre.
Accordingly, we sought to examine the presence of gender-differentiated sexual stereotypes in music lyrics across a 50-year period. Specifically, we expected female performers would reference romantic relationships and love more frequently than male performers, while male performers would reference sexual activity more frequently than female performers. We also expected that female bodies would be objectified more frequently than male bodies, regardless of the performer's sex. We also hypothesized that references to sexual behavior and sexual objectification would become more frequent over time, while references to romantic relationships and love would become less frequent over time. Further, we hypothesized that the rap genre would include more references to sexual behavior and (female) objectification, and fewer references to love or romantic relationships, than other genres.

Song and Artist Information
Billboard's Year-End Hot 100 chart includes the most popular songs across all genres of music as ranked by radio airplay, sales data, and, in more recent years, online streaming activity (Billboard 2009;Whitburn 1986). Using the fifty highestrated songs from the even-numbered years between 1960 and 2008, we compiled a list of 1250 songs to analyze. We recorded year, title, rank, and performer for each song.

Performer Information
For each song, the performer's sex, ethnicity, and primary musical genre were recorded from allmusic.com. There were 919 unique performers across all years. Omitting guest performers (e.g., ''Usher featuring Lil Jon and Ludacris''), there were 895 unique primary artists (e.g., Usher). Sex and ethnicity were coded for all listed performers. The majority of performers (n = 721, 80.5%) had exactly one song in this analysis. The five performers (and their genres) who appeared most often in our analysis were Mariah Carey (R&B), The Beatles (Rock), Janet Jackson (R&B), Madonna (R&B), and Usher (R&B).
Genre was coded using solely the primary artist. For genre information, we chose allmusic.com instead of popular retailers (e.g., iTunes, Amazon.com) for several reasons. First and foremost, allmusic.com provides substantive, detailed descriptions of major genres and their subgenres that identifies the primary musical characteristics by which each genre is defined; this information had previously been available in print (e.g., All Music Guide, 1997). The commercial services do not provide this information, nor any information about how artists are categorized. As researchers (and fans), we are confused by iTunes' categorization of Frank Sinatra, Beyonce, and Lady Gaga as ''Pop'' artists and believe Allmusic.com's categorization of these artists as Vocal, R&B, and Pop, respectively, more accurately reflects their musical approach. Second, allmusic.com lists performers as having multiple genres and subgenres (where relevant), and provides this information based on their recordings, not alphabetically. iTunes does not appear to allow multiple genres and Amazon.com does not appear to provide any type of genre information. Third, Allmusic.com identifies rap as its own genre, distinct from R&B, mirroring current recording industry distinctions as well as researchers' treatment of rap as a distinct genre.
Of the 1250 songs, 328 were performed by a female solo artist or all female group (e.g., Diana Ross and the Supremes, The Pussycat Dolls), 827 were performed by a male solo artist or all male group (e.g., The Rolling Stones, Backstreet Boys), and 95 were performed by a mixed-sex group. Because mixed-sex groups include both female and male singers, these songs were excluded from  250 songs per decade, 1250 songs total a Multi-ethnic includes multi-ethnic individuals as well as groups whose members are from different ethnic groups analyses examining performer's gender. Of the remaining 1155 songs, women performed 28.4% (Table 1). The percentage of female (vs. male) performers varied substantially and significantly across decades, with female artists constituting no more than 20% of performers in the 1960s and 1970s and at least 25% of performers in subsequent decades, v 2 (4, n = 1155) = 61.82, p \ .001. Most songs were performed by individuals of European-American (n = 720, 57.6%) or African-American descent (n = 430, 34.4%). Multi-ethnic individuals and groups were the next largest group (n = 76, 6.1%) and the remaining performers did not have identifiable ethnic groups (\2%) ( Table 1). The percentage of songs performed by European-Americans dropped after the 1980s and was mirrored by an increase in the number of songs performed by individuals of African-American descent, as well as mixed-ethnicity groups, v 2 (12, n = 1250) = 120.29, p \ .001.
Most songs (qua performers) were identified as belonging to the rock/pop genre (''rock''; n = 650, 52.0%), with a substantial number in the Rhythm and Blues (''R&B''; n = 373, 29.8%) and rap genres (n = 87, 7.0%) (  (100) 140 (100) 1250 (100) From ''I Want To Hold Your Hand'' to ''Promiscuous'': Sexual… genre (e.g., Country) accounted for as much as 3.5% of the songs, so the remainder were grouped together as ''miscellaneous'' genres (n = 140, 11.2%). The number of songs from each genre appearing in the Top 50 varied over time, v 2 (12, n = 1250) = 230.63, p \ .001. Rock songs dominated the charts from the 1960s through the 1980s and they constituted more than half of the songs in the analysis. Rap music was not widely known until the 1980s and no rap song made the charts until the 1990s. This coincides with the acquisition of smaller record labels that produced rap music by the major music companies in the early 1990s, which enabled both greater production and greater distribution (Hurt 2007;Jhally 2007;Weitzer and Kubrin 2009). Performers' sex, ethnicity, and genre interacted in systematic ways; each twovariable Chi squared test was significant v 2 (dfs = 3-9, n = 1155-1250) C 22.9, ps \ .001 (Table 2). Males outnumbered females among all ethnic groups, with particularly large differences among those of European-American (76.4%) and African-American (66.2%) descent. Similarly, males outnumbered females among all genres, with particularly large margins among rap (93.2%), rock (77.8%), and the miscellaneous genres (70.9%). European-Americans dominated the rock (91.8%) genre and were more common among the miscellaneous genre (66.4%); African-Americans dominated the rap (83.9%) and R&B (82.8%) genres. Given the variability in ethnic composition of performers within each genre, as well as the focus on genre in the literature, we chose to highlight genre, not ethnicity (but see Hall et al. 2012, for a focus on performer's ethnicity).

Coding Scheme
The coding team consisted of ten undergraduates and one doctoral faculty member. An initial training round used twenty-six off-year songs (e.g., 1975) split into two segments. During training, all coders scored all songs for all variables. The initial coding scheme was clarified and altered, as needed, and overall intercoder agreement was in excess of 80% across all coders and all variables.
For the focal coding, songs were segmented into ''packets'' of 10 songs with consecutive rankings from each year (e.g., 1-10, 11-20, etc.), creating 125 packets (5 per year, 25 years). The packets were randomized so that any ''drift'' in coding standards would not be confounded with the year in which the song was released. Packets were then randomly assigned to members of the coding team.
Working independently, two coders assessed each song. Inter-rater reliability was greater than 80% for every variable in this analysis. Krippendorff's alpha (Hayes and Krippendorff 2007) was also computed for each of variable. Only dating (ka = .72) exceeded the conventional standard of .70; sex (ka = .63) and presence of love (ka = .67) were relatively close to the threshold, while explicitness (ka = .51) and objectification (ka = .41) were much worse. In an assessment of reliability statistics used in content analyses, Oleinik et al. (2014) reported that Krippendorff's alpha returned values substantially lower than Pearson's r and that Krippendorff's alpha is better suited for stylistic or literal coding (e.g., presence or absence of dating) than for rhetorical or interpretive coding (e.g., explicitness). In order to improve the fidelity of the data set, all coding disagreements were reconciled in two to three person teams; reconciliation required consensus among all members of the reconciliation team.
Coding samples are provided in Table 3. To assess dating, coders determined if the song described a romantic relationship that was expected to persist for months or years and presumed to be monogamous (Bogle 2008;Garcia et al. 2012). The relationship could be desired (future), current, or past and could be referenced directly, colloquially, or clearly implied; coders were trained to not assume that a dating relationship existed simply because a song also referenced sexual behavior. Similar parameters were used to assess sex, defined here as any activity leading to orgasm (including masturbation); sex could be desired, current, or past, and could be referenced directly, colloquially (e.g., ''hit it''), or clearly implied. Sexual activity was not taken as a sign of dating relationship. Coders also indicated if the word ''love'' appeared in the song and, if so, whether it was used romantically (e.g., ''in love''), as a sexual reference (e.g., ''make love''), both romantically and sexually, or in some other form, such as admiration or love for a friend. Coders also determined the explicitness of the sexual reference, recording if it was through metaphor, via the word ''love'' (e.g., ''make love''), or directly (including slang, such as ''hit it'').
To assess objectification, coders determined if the song referred to females or males (separately) as sexual objects. References to the singer or an unnamed target (''you'') were scored based on the performer's sex and assumed that songs were written for a heterosexual audience. Songs by mixed-sex groups were omitted from the objectification analyses because this level of detail was beyond the scope of our coding scheme.
A small number of songs (n = 15, 1.2%) were instrumental and thus did not include any lyrics. Because our goal was to examine how frequently specific types of content appear, we retained the instrumental songs in our analysis. Instrumental songs appeared primarily in the 1960s (n = 9) and into the 1970s (n = 5), with only one instrumental present in the 1980s and none later. This decade-based pattern was not expected by chance, v 2 (4, n = 1250) = 20.92, p \ .001.

Analytic Approach
Before proceeding, we draw attention to the differential between the descriptive raw counts (Table 4) and the proportional values ( Table 5) that form the basis of inferential comparisons. Songs by male performers outnumber songs by female performers approximately 2.5-1. Figure 1a (raw counts) and b (proportions) illustrate a concern regarding our results by illustrating the findings for dating songs. As indicated in Fig. 1a and Table 4, dating songs by male performers substantially outnumber those by female performers from the 1960s through the 1980s, with the numbers becoming more equal in the 1990s and 2000s. However, as indicated in Fig. 1b and Table 5, the proportional values indicate that female performers sang  about dating more frequently and the percentages were more similar across all five decades. Accordingly, we address both descriptive and inferential statistics throughout the ''Results'' section, following Hatton and Trautner's (2011) approach to a similar disparity in female and male representation on Rolling Stone covers. Throughout our study, we used analysis of variance (ANOVA) to examine lyrical outcomes. Because our variables were scored 0/1, the means are equivalent to proportions (cf. ter Bogt et al. 2010). The use of ANOVAs allowed us to examine linear, quadratic, cubic, and 4th order patterns of change across decades, as well as interactions between primary variables (e.g., performer's sex x decade). Although Chi Squared is typically considered more appropriate for categorical data (e.g., Hall et al. 2012), it does not allow direct assessment of the type of change over decades (e.g., linear, quadratic) or the interactional patterns we sought to identify.

Descriptive Overview
The majority of popular songs addressed a dating relationship (71%) for the whole sample and in every decade (Table 5, top panel). A similar pattern was found for use of the word love (57%), which appeared in a majority of songs for every decade except the 2000s (49%); love was most frequently used as a romantic reference, not a sexual reference (or both). A substantial minority of songs contained sexual references (22%), and these occurred most frequently via metaphor (11% of all songs). Sexual objectification appeared in a minority of songs (14%), with sexual objectification of female bodies (13%) occurring more frequently than objectification of male bodies (4%).

Analyses
Our analyses began with an examination of differences in song content based on performer's sex and decade. Because the rap genre did not appear in the rankings until the two most recent decades, genre was assessed in a separate analysis (see below). For each coded variable, we computed a series of 2 (performer's sex) 9 5 (decade) ANOVAs. Descriptive information, as well as the results of significance tests, are provided in Table 5. The top panel provides means (or proportions) for all songs in each decade; the middle and bottom panels provide information for female and male performers, respectively.

Performer's Sex
Given the disparity in number of songs by female versus male performers (328 vs. 827), there should be no surprise that raw counts for male performers were higher for every variable in the analysis. The female:male ratio ranged from approximately 1:2 to as high as 1:7, with the largest disparities occurring in the areas that female performers were least likely to sing about (e.g., objectification of female bodies, using the word ''love'' to refer only to sex). Dating references occurred in a greater percentage of songs by female performers (F values are provided in Table 5, middle panel). Specifically, women were more likely than men to sing about dating (.80 vs. .67), use the word love (.66 vs. .53), and use love strictly in its romantic sense (.56 vs. .44), consistent with H1. Sexual references occurred in a substantial minority of songs, appearing less frequently for female than male performers (.17 vs. .22). This was also true for explicit (.16 vs. .22) and direct (.02 vs. .05) references to sex, consistent with H2. Female performers were less likely to include objectifying lyrics (.08 vs. .16), especially lyrics objectifying women (.05 vs. .15). There were no differences between female and male performers in the use of love to refer to sex, the use of love to refer to both romance and sex, metaphoric references to sex, or the objectification of males.

Decade
Raw counts indicated some general trends over time. Broadly speaking, references to romantic relationships and uses of the word love were relatively stable from the 1960 s through the 1990s, but were less frequent in the 2000s. We note a relative highpoint in romantic songs during the 1980s. Use of the word love to refer to romantic relationships became less frequent over time, dropping by approximately one-third. References to sexual activity increased fourfold over time, both as metaphor and directly; use of the term ''making love'' peaked in the 1970s. Objectification also showed a fourfold increase over time.
The proportion of songs that addressed dating did not change significantly across decades, although most other lyrical components did exhibit temporal change (F values are provided in Table 5, top panel). Although overall use of the word love did not change over time, there was a linear decrease in romantic uses of the word love and a quadratic change in uses of the word love to refer to both romance and sex, described by an inverted-U that peaked in the 1980s and 1990s. These findings are mostly contrary to H3, which predicted a general decrease in use of the word love.
Across decades, there were linear increases in sexual content and references to sex (overall), as well as metaphorical references to sex, consistent with H4. Direct references to sex and the objectification of women demonstrated linear and quadratic increases over time, with the rate of increase accelerating from the 1980s into the 1990s and then accelerating further into the 2000s; for objectification of women, the shift is from a relatively stable 6-7% of songs from the 1960s through the 1980s to 12% of songs in the 1990s and 31% of songs in the 2000s. We note that objectification of men also increased over time, but was consistently smaller than and increased at a slower rate than objectification of women, consistent with H7a and H7b. Performer's Sex 3 Decade As noted earlier, songs by female performers were much less common than songs by male performers and this pattern was replicated for all coded variables. The 1990s provided a few exceptions, and we note that this decade had the smallest female:male disparity (1:1.25). Here, women's lyrical references to dating  relationships, the word love, use of love in its romantic sense, and use of love to refer to both love and sex, outnumbered male performer's lyrical references. Some other cells had equal or similar numbers, although not in any discernible pattern. Among our outcome variables, the only significant interactions between performer's sex and decade occurred for explicit references to sex, as well as direct references to sex (F values in Table 4, bottom panel). To decompose these interactions, univariate ANOVAs within each sex were computed. For direct references to sex, women demonstrated no change across decades and men demonstrated both linear and quadratic increases.

Genre
Finally, we examined differences across the rap, R&B, rock, and other genres. Rap songs do not appear in the Top 50 until the 1990s, so we computed the analyses twice, once with all decades and once with just the 1990s and 2000s. The pattern of results were highly similar across analyses. We present results for the 1990s and 2000s (Table 6).
Inferential analyses relied on a series of One-Way ANOVAs with genre as the independent variable. Results revealed differences between rap and all other genres. In particular, rap songs were significantly less likely than songs from all other genres to refer to a dating relationship, include the word love, or use the word love to refer to romance, consistent with H5. Rap songs were more likely to directly address sexual behavior, reference sexual behavior explicitly (i.e., not via metaphor or ''love''), objectify women, and objectify men, consistent with H6 and H8. There were also several differences between the rock and R&B genres. R&B songs were more likely to reference dating relationships, use the word love, address sexual behavior, use metaphors to describe sexual behavior, objectify women, and objectify men.

Discussion
Our analysis of lyrical content of the Top 50 songs from even numbered years between 1960 and 2008 found that dating and sexual content is quite common and is partly consistent with cultural notions of gender-differentiated sexual activity. We found that references to romantic relationships appear in the vast majority of songs, and the word ''love'' appears in slightly more than half of all songs and is most typically used to refer to romantic love (i.e., being in love), while references to intercourse (and other orgasm producing activities) and sexual objectification appeared in a sizable minority of songs. Content varied by performer's sex, decade, and genre, with sexual (vs. dating) content proportionally more common among male performers, in more recent decades, and in the rap genre. However, we note that male performers outnumbered female performers by a substantial margin (2.5:1), so raw counts for males were higher for almost all cells in the analysis, even when inferential tests indicated the content was more common in female performers' lyrics. Broadly speaking, gender differences in our results suggest the portrayal of dating and sexuality in popular music lyrics is quite similar to the portrayal in other media formats (Clawson 2005;Herd 2015;Kunkel et al. 2005;Taylor 2005;Ward 1995) and consistent with cultural expectations and stereotypes (Arnett 2002;Smiler 2013;Tolman 2002). In particular, women were more likely to sing about dating and love, and men were more likely to objectify others, particularly women. Men were also more likely to sing about sex; this difference did not reach statistical significance but given the substantially greater number of songs by male performers, the raw counts are notably different. Women and men did not systematically vary in how they used the word love or the explicitness of their sexual references.
Raw counts told a story about men that is contrary to cultural stereotypes. Our data showed that men sang about dating in two-thirds of the songs we analyzed and love in half of our sample songs, more than doubling the number of times women addressed these topics. Moreover, male performers referenced dating relationships approximately three times more often than they referenced sex. The emphasis on dating relationships and love more so than sex are consistent with the most common reasons male and female youth endorse for both dating and intercourse (Meston and Buss 2007;Smiler 2008;Smiler and Heasley 2016). The pattern by which a minority of men emphasize sex in way that creates a male versus female difference is also found in studies of desired (Schmitt et al. 2003) and actual partners (Dariotis et al. 2008;Humblet et al. 2003) and appears to contribute to our cultural stereotypes about male sexuality (see Smiler 2011Smiler , 2013 for discussion).
The nearly 2.5-1 ratio of songs by male to female performers in this data set may be said to distort the differences we identified. For example, while males were more likely to sing about sex and female were more likely to use love romantically, the two appeared with identical frequency (sexual references for males, n = 185; romantic love references for females, n = 184). Thus, men could be said to sing about sex as often as women sing about romantic love, although the claim relies on men's overrepresentation in the dataset. The differing trends based on raw counts and sex-based comparisons suggest that it might be difficult for an individual adolescent to identify the ways in which lyrics simultaneously support and challenge stereotypes.
Over these five decades, relational content became somewhat less common and sexual content, including objectification of women, became more common; these trends were slightly more apparent in songs performed by males than females. The changes seem to be driven by the rap genre, which did not place a song in the Top 50 until the 1990s. Unlike pop/rock, rhythm and blues, and other genres, rap songs were much less likely to address romance and love, and much more likely to include sex or objectifying lyrics. These findings are mostly consistent with both Dukes et al. (2003) and Hall et al. (2012) Our analysis also revealed that the lyrical content of rap music is notably different from the other genres we assessed. Whereas rock, R&B, and other miscellaneous genres tended to emphasize dating and use the word ''love'', these themes appeared in less than half of all rap songs. By contrast, the content that was infrequent in rock, R&B, and other genres-sex and objectification of womenoccurred in more than half of all rap songs. Rap songs were notably less likely than other songs to use the word love to refer to romantic love, were more likely to use love to refer to both romantic love and sex within the same song, and favored explicit sexual references over metaphoric ones (for additional discussion of lyrics and videos, see Herd 2015;Hurt 2007, Weitzer andKubrin 2009).
For several analyses, R&B appeared to inhabit a middle position between rock and rap, with mean scores that were much more similar to rock than rap. This finding may be a function of our classification of artists; Turner (2011), for example, explicitly categorized some artists as belonging to a blended genre ''R&B/Rap'' that did not appear in our database.
The distinction between rap and R&B, and rap and other genres more generally, is very important. Analysts have repeatedly observed that rap videos present wellmuscled, well-monied, conspicuously consuming black male performers and hypersexualized, scantily clad black women (Herd 2015;Jhally 2007;Turner 2011) with one commentator calling the imagery ''regressive'' (Hurt 2007). Our data suggests that this image of women is embedded within the lyrics; anecdotally (but uncoded), we believe this is also true of the men's imagery. Although rap is routinely portrayed as black music (Hurt 2007;Jhally 2007) and is dominated by African-American artists, R&B was similarly dominated by African-Americans (83.9 vs. 82.8%) but presented a notably different message. Given that an artist must reach a broad cross-section of the American audience in order to have a song on the annual Top 100 list, and that audience is mostly White, we join others in speculating why this particular image of Black-ness is being sold to a mostly White audience (Hurt 2007;Kimmel 2008;Turner 2011). Rap had the most skewed sex ratio (93% male) of any genre in our analysis, which contributes to notions that the answer may be related to (re)claiming men's power over women (Hurt 2007;Kimmel 2008; see also Denski and Sholle 1992). By comparison, the other ''black'' genre, R&B, was the least gender skewed (56% male) and had a very different lyrical profile.
Our analysis explicitly examined sexual objectification (cf. Fredrickson and Roberts 1997) in music lyrics, which we believe has not previously been assessed. Consistent with cultural images, women's bodies were objectified at higher rates than men's bodies (e.g., Hatton and Trautner 2011); this effect is due to the higher rates of objectification of women by male performers (particularly rap performers) and is furthered by the 2.5:1 disparity between male and female performers. Like Hatton and Trautner, we note that objectification of female and male bodies is not parallel or equal. Descriptions of female bodies tended to focus more on specific body parts, while descriptions of male bodies tended to be more wholistic and often referred to a general style; this disparity was evident within songs, as indicated by the coding examples for ''What's Luv'' and ''Lollipop'' in Table 3. The extant research suggests that exposure to objectifying terms can increase the likelihood an individual will experience appearance anxiety (Zurbriggen et al. 2011) or initiate intercourse at a younger age (Martino et al. 2006).
Our content analysis included assessment of dating relationships. Although very important for most teens (Collins 2003), these relationships are often left out of content analyses (e.g., Dukes et al. 2003;Hall et al. 2012; see also Ferguson et al. 2016) in lieu of a focus on sexual behavior. Our results revealed that the majority of popular songs referenced dating relationships, with female and male performers referencing these themes in more than half of their songs (until the 2000s, for men). This was true for every genre except the male-dominated and male-preferred rap genre. These patterns mimic gender stereotypes that emphasize relationships for females and sexual activity for men (Crawford and Popp 2003;Smiler 2013;Tolman 2002).
The patterns observed in our analyses suggest that media literacy training may be especially important for adolescents engaging with music (American Psychological Association Task Force on the Sexualization of Girls 2007; Brown and Bobkowski 2011;Potter 2010;Shewmaker 2015). Media literacy activities provide adolescents with tools to identify messages that are being promoted by the media that they consume, compare those messages with their own values, and evaluate whether or not they choose to accept those messages as true (e.g., Pinkleton et al. 2008;Scull et al. 2014). The contradictory findings in several areas might provide rich areas of constructive conversation for adolescents. For example, both men and women sing about love and romance, in contrast to the stereotype that men are primarily interested in sexual relationships. The differences in the objectification of women and references to explicit sex across genres is a topic that might lead to a conversations about when these patterns emerged, and why they are higher in some areas than others.
As always, there are limits to our analyses. We focused strictly on music lyrics, not music videos or the overlap between the two, and thus results should be applied accordingly. Regarding the content analysis, the use of a 0/1 coding scheme provides a coarse description of lyrical content. Cohort effects may have impacted our results; the coding team was comprised almost entirely of individuals who were approximately 20 years old and thus may have failed to interpret double entendres and metaphors for the older songs in our analysis. Because the data were naturally occurring and not equally distributed, there were no rap songs for the first three decades of our analysis. As a result, we could not compute a full factorial analysis to explore all potential interactions (performer's sex x decade x genre). Finally, the exclusion of 95 songs performed by mixed-gender groups from some analyses places limits on the generalizability of our gender-based findings to this set of performers.
Our results indicate that the content of popular music has changed over the past five decades, with lyrics gradually shifting away from dating relationships and to more explicit sexual content and sexual objectification. Male and female performers emphasized somewhat different themes, and the cross-sex comparisons were generally consistent with gender stereotypes, even though male performers sang about all topics more frequently than female performers. Our analysis also found that the content of rap lyrics is substantially different from that of other genres, including R&B, which is also dominated by performers of African-American descent.

Compliance with Ethical Standards
Conflict of interest The authors report no conflicts of interest regarding the production of this paper. No grant or award money was used to support development of this paper in any way.
Human and Animal Rights The authors affirm that the rights of humans and animals were protected during the research and writing phases of this paper.
Informed Consent This paper relied on information available in the public domain. No data or personal information was collected directly from any individual during the development of this paper. Accordingly, consent was neither requested or obtained.