Gender and Popular Culture

Producers and Consumers: Measuring the Value of Musical Content

The running debate on taste formation in the sociology of culture can be traced back to Max Horkheimer and Theodor Adorno ([1972] 2001:74, 76), who argued that the culture industry alienates consumers by forcing them to accept “cyclically recurrent and rigidly invariable types” that are formed via “a stereotyped appropriation of everything” in mass-produced culture. They also argued that producers and consumers are trapped within this system of production, as “the industry submits to the vote which it has itself inspired” ([1972] 2001:80). Since then, the exploration of taste preferences has shifted to a focus on the processes by which cultural production is embedded in social life. This includes the degree to which consumers themselves drive the production of music and musical meaning, and the degree to which popular music is determined by sources outside the control of consumers or producers. Now, it is generally understood that taste formation is a mutually constitutive process between producers and consumers in a rich extramusical context—much more complicated than the culture industry formulation first assumed. A common thread throughout this development, however, is the notion that musical preferences necessarily differ for producers and consumers.

Why? Contemporary production of culture approaches frame production processes as a dialogue with consumers, but one where the quality of musical content takes a back seat to the pursuit of profit and competition for market share, which is particularly important for generating platinum sales. The popular music market is controlled by a handful of large mega-media companies that own the production and distribution networks needed to reach mass-market consumers (Passman 2006:64). This raises the cost of competition, and the increased market share that comes with platinum sales pushes the costs up even higher as companies scramble for these scarce rewards (Frith 1981). Platinum sales are necessary to ensure the long-term success of a company, because only about 10 percent of albums make any profit for the company, yet they cannot accurately predict which albums those will be. With a platinum seller, the company can finance the other 90 percent that prove to be unsuccessful (Frith 1981: 101). A large market share increases the likelihood the next platinum seller will surface in a particular company’s ranks, providing some insurance to cover the high level of uncertainty in the company’s central profit-generating mechanism. In this framework, listener preferences are relevant only to the extent they impact a company’s control over production relative to their competitors. When a record label drops an artist that later becomes a platinum seller, for example, the situation is problematic mainly because the profits from that artist go to a competitor, threatening the label’s control over the playing field, and impinging on their ability to stumble upon the next platinum seller. Of lesser concern is whether listeners missed out on a “good” artist.

Yet contrary to a cynical mass-market approach, companies are centrally concerned with what listeners want to hear: To make a platinum seller, they need one million listeners to open their wallets. Their problem is that they have no way of reliably predicting which artists listeners will buy, so they, instead, focus on gaining control of the playing field by cornering a strategic position and capitalizing on emerging trends (Negus 1999:32). Framing the production-consumption relationship as a dialogue highlights the central role of these uncertain listener preferences—easily ignored in theories of production—and incorporates the construction of cultural meaning into a theory of production.

Listeners’ tastes are uncertain partly because they are dynamic users of popular music. Music is often used as a tool to manufacture selves and social experience: Whether or not a listener likes a particular artist depends on how they use that artist, and how the artist resonates with other aspects of everyday life. For example, M. Sagalnik, P. Dodds, and D. Watts (2006) found that the ability to predict successful songs decreased when consumers had access to information about how others had valued those songs. Music producers cannot predict the outcome of a discursive process that includes not just commercial products, but individual and collective identities and all the other contingencies this process entails. Instead, they manage this uncertainty by staying tuned in to constantly changing locally and situationally determined ways in which popular music is consumed. To do this, labels use “open” systems with relatively autonomous divisions devoted to discovering new talent (Lopes 1992) combined with loose control over the day to day operations of those subdivisions (Negus 1999). They also use multiple sources to feel the pulse of changing listener taste, such as periodic collection of “callout” research where listeners are telephoned and asked about which artists they do and do not like (Frith 1981; Negus 1999).

Efforts to cast a wide and loose net over consumer tastes, while encouraging the bottom-up surfacing of changing consumer tastes, ends up in fragmented and tenuous production processes in top-down practice. In fact, fragmented production processes require that some decisions of content are made simply to get everyone to work together and produce a finished product. In the fast-paced, scattered, and constantly changing context of popular music production, callout figures—despite their ascribed purpose—are often creatively used to confirm a producer’s preexisting beliefs about an artist (Gitlin 1983; Negus 1999). Inevitably, callout enables strategic power plays between multiple talent scouts with autonomous decision-making authority about artistic choices, but without autonomy regarding the resources necessary to back up those choices. Again, uncertain listener preferences can quickly become secondary to more immediate issues of control over production. Arguably, one consequence of open systems is that it potentially complicates production, encouraging new gulfs in the cultural valuations of promoters and listeners, rather than closing them. Thus, even with an explicit intent to focus on listeners, fragmented production processes still leave a gap between producer and listener favorites.

Musical Valuation and the Politics of Identity

There is one area, however, that music producers and consumers seem to be in agreement about: The star persona is at the center of pop music discourse. Both taste preferences and the situated use of popular music develop through a discourse of value judgments (Frith 1996:5). Following Michel De Certeau (1984), Henry Jenkins (1997) argues that listeners are “poachers” and “nomads”: They come across a particular artist they like and use that artist to answer questions of social identity and, in the process, claim “ownership” (Frith 1987:143) of that artist, then move on, appropriating and incorporating new cultural materials as they go. In this sense, audiences matter because producers know their products mean nothing if audiences do not use them in this discursive manner. Producers also know that the chances of success improve by crafting an artist worth talking about, and an identity worth poaching. Thus, production of popular music is oriented on an artist’s personal image or “persona,” which is deliberately constructed to tap into popular identity (Peterson 1997). A distinct persona type is necessary to mold a successful artist.

Complicating this picture, listeners are not an undifferentiated “mass society” but rather form orientations to practice and trajectories of life course patterned around important aspects of identity such as gender, race, sexuality, and class. The ability of listeners to make their symbolic constructions of race or gender collectively valued depends on the degree to which mass-market producers recognize and discursively amplify those constructions. For listeners with central aspects of identity already devalued in larger society, devaluation by music producers is a source of disdain for popular music, a missed opportunity for platinum sales, or both. Yet theory suggests that producers are more likely to overlook a discursively produced identity they do not already know because they are looking for the next Madonna or Lady Gaga: New representations are often recombinant clones of earlier, successful representations (Gitlin 1983), where heterosexual white male representations are overrepresented and the rest are limited and rigidly stereotyped (Entman and Rojecki 2001). A consequence of this production strategy is that mainstream artists are also “mainstream” representations of raced, classed, gendered, and sexualized identities.

Criticisms of the representations in commercially manufactured artistic images are all too common. Prince’s Purple Rain was publically criticized as “too sexual,” eventually leading to parental warnings on albums with explicit content (Nuzum 2001). Like criticisms on the grounds of talent, criticisms surrounding the social effects of artistic representations push record companies to be accountable outside organizational and profit-oriented concerns, and in so doing, implicitly assume that companies ignore the wants and needs of a wider public in the absence of such prodding.

This assumption has some empirical support. Melissa A. Milkie (2008) explored how teen magazine editors responded to well-known adolescent girls’ criticisms of the narrow visual images of female beauty in magazines, television, film, and the like. Some editors legitimized the girls’ critiques but explained that they were powerless to change the minds of the photographers who did not want to take pictures of what they perceived to be ugly girls, or advertisers who demanded their products be showcased by attractive girls, or the larger culture that perpetuated damaging stereotypical images of women. While acknowledging at least some control over the content of the images, they argued that they could not change the images because of how others might react. This explanation avoids corporate responsibility on the grounds that consumer preferences are secondary to the problems of production. Ironically, the argument is very similar to employers who fired female workers in the post–World War II era: They were worried about losing control of employees and business of clients who would protest to women on the job (Milkman 1987). Milkie’s argument is premised on the assumption that girls’ critiques of magazines are different from boys’ critiques, or magazine producer’s perspectives, because they see the images as applying to themselves. But even though producers were mostly adult men (and not adolescent girls), their valuation of the images was different because aesthetic valuation is also tied to one’s position in the institutional framework. The issue extends beyond simple diversity of identity: It involves diversity in meaning dependent on structural location.

It is easy to take a superficial reading of Milkie’s argument and assume, like magazine critics and some editors she interviewed, that these images are in fact harmful to adolescent girls. But naturalizing the good or evil of cultural productions elides the political or moral dimensions of these products that make them so socially powerful and contestable in the first place. Furthermore, censorship due to alleged corruption of youth is selective. Condemning Prince’s Purple Rain as too sexual, for example, precludes discussion of how the album challenges the notion of what it means to be a man or woman, black or white. Gender, race, and sexuality are inseparable elements of these criticisms, as they are of the representations themselves. The federal government considered N.W.A.’s (Niggaz With Attitude) “Fuck the Police” to be a threat to national security, and records were pulled off the shelves en masse, while Eric Clapton’s contemporaneous version of “I Shot the Sheriff” did not even raise an eyebrow (Nuzum 2001, see also Binder 1993). Themes are constantly silenced or amplified as collective conversations proceed, and minority voices are often silenced in the conversation. Therefore, these examples are best seen as examples of the process by which contested identities become a part of these discursively produced artistic images. Criticisms of popular music involve value judgments about race, gender, and sexuality disproportionately articulated by voices in central or hegemonic positions. Thus, while producers craft artistic personas to tap into that larger social conversation, they invariably ignore criticisms of representation in patterned ways while capitalizing on them.

Pierre Bourdieu (1993) addressed this issue by claiming cultural valuation is tied to an agent’s structural position relative to dominant social categories in the field of cultural production, and society in general. Clearly, the issue of “ownership” that is so tightly bound with aesthetic valuation will be different depending on whether an agent is marginalized according to relevant and intersecting social categories. For example, by comparing academic understanding of romance novels with the way female readers in her study interpreted romance plots, Janice Radway (1984) found readers constructed meanings of books that offered opportunities for them to deal with and escape from their own personal struggles with misogynistic attitudes toward sex and rape. This reaction to romance novels was counter to the ways many academics expected these women to react to such scenarios: They expected rejection because of the unflattering depictions that forced sex entailed. Radway’s accounting of romance novels hints at one of the central dimensions around which a listener’s attachment to cultural products is oriented. The women in her sample enjoyed reading romance novels not only because reading provided them an escape from their domestic responsibilities, but because the novels gave them an opportunity to think about and come to grips with sexism in larger society. This pattern is key to understanding the taste preferences of marginalized groups.

Radway’s view of consumption is complementary to Bourdieu’s conception. For Bourdieu, aesthetic taste embodied a “unique phenomenon of distance” (1982:55), and describes a process whereby the consumer puts himself in a position of privilege relative to others. Bethany Bryson (1996) latched onto this idea of consumption when she posited that consumers use cultural products to reinforce symbolic boundaries between themselves and categories of people they do not like, such as disliking music styles associated with uneducated people (heavy meal, country, or rap) because—crudely put—those consumers do not like uneducated people. Under Radway’s conceptualization of the same processes, a poor man who likes opera risks the accusation that he is buying into a classist system, just as preferring misogynistic fiction risks the accusation that a woman supports a sexist social system. But this poor man may like opera because it gives him status within a classist society where he perceives a lack of individual control. Both popular and academic concepts of aesthetic preference as “taste” implicitly assume listeners come from a privileged position. Not all listeners do. For the same reasons that social theory overlooks marginalized experiences with media, producers are more likely to miss emerging trends associated with listeners from less privileged structural positions. While differences between and within promoter and listener preferences will likely exist along identity lines, and exploration of these differences will provide insight into the areas of conflict and contradiction around which privilege and disadvantage are oriented, they cannot be fully understood without accounting for differential ways taste preferences are formed for structurally advantaged and disadvantaged groups.

Richard A. Peterson (1997) argued that nostalgic representations of “the way we were” and “how we ought to be” form the core of artistic personas, what Pierre Bourdieu (1990) termed the ever-present dialectic between heretical or new ideas and the orthodoxy of old tradition. Peterson (1997) characterized this as the running dichotomy of hard core versus soft shell country music. In the context of the argument here, the dialectic of identity can be reframed as the orthodoxy of the dominant versus cultural critiques of the marginalized. Cultural producers aim to provide cultural products that tap into this running conversation, so consumers can battle over the heresy or orthodoxy of a particular politics of representation while they orient toward popular music. Importantly, however, cultural producers are steeped in the processes of production, with strategies oriented toward control of market share, so they may not always tap into specific aspects of identity around which their consumers circumnavigate. But as the above argument shows, marginalized positions are not equally represented in this dialectic. Producers tap into this conversation from their own structural positions, and more often than not, it is a privileged position. Magazine editors, for example, discounted girls’ critiques partly because they were not themselves adolescent girls, making it easier to discount the point of view of a structural position that did not apply to them. Music promoters, also known as Artist & Repertoire (A&R) men, and top-level record executives are still overwhelmingly dominated by white, upper-middle class, often heterosexual men, within an institutional culture that is white, upper-middle class, male, and heterosexual (Horkheimer and Adorno [1972] 2001).

Hypotheses

This paper explores whether or not promoters and listeners value the same artistic personas, with a focus on the degree to which the gender of artists and listeners defines any emerging differences. While the theory I developed above suggests differences across multiple dimensions, available data only allowed for comparisons along gendered lines. Specifically, I compare promoter preferences with the preferences of male and female listeners via artist types defined by gender. I expect the following:

These expectations draw from two main sources. First, I expect differences between promoters and listeners regardless of gender, because demands of the production process and competition for market share draw producers away from simply catering to listener tastes. Second, I expect promoters and male listeners will be more likely to favor stereotyped representations of gender because these are the types that are most readily available on the market (Gitlin 1983), and because unflattering depictions of gender do not apply to them, they will be less likely to look beyond them (Milkie 2008). I also expect that some female listeners will like these types as well (Radway 1984), but that they will also be more likely to prefer gender-nonconforming artists (Milkie 2008).

Data and Method

Analytic Strategy and Dependent Variables

Figure 1 presents the path diagram used to guide analysis. Following standard notation, observed variables are represented using squares, and latent variables are represented using circles, with arrows representing regression relationships between the observed and latent variables. The three dependent variables are located in the center of Figure 1. The central question concerns differences between the choices of male listeners, female listeners, and label promoters for specific artist types. Accordingly, the basic analytical model compares the relationship of artist types with these three dependent variables. The regression relationships are represented by the three arrows leading from artist type to the three dependent variables, and asterisks represent parameters of interest (*A,*B,*C) described in the results below. For each artist type, I compare these regression relationships and report significant differences in regression coefficients.

OPEN IN VIEWER

Figure 1.

Path diagram and plan for analysis for songs appearing in callout research, June 2002–October 2004, N = 922.

Promotional support

Promotional support is a latent dependent variable, represented by a circle in the center left of Figure 1, measured by regressing the four separate indicators of promotional support onto the latent variable, represented by the four arrows leading to advertisements, payola, videos, and fourth quarter releases. A latent variable is necessary in this case because promotional support cannot be measured directly, and measurement through each of the four available indicators alone would only capture a small portion of the level of promotional support. Furthermore, the measurement model includes terms for error, and, thus, the resulting latent variable is estimated while accounting for measurement error in these four indicators.

Each indicator of promotion is measured relative to artists, such that multiple songs by the same artists are coded as receiving the same level of promotion. The first measure of promotion is a manually coded dichotomous indicator that measures whether or not payola payments were made on an artist’s behalf. Specifically, it indicates whether or not an artist appeared in the exhibits to the payola settlements of 2005 (see Rossman, Chiu, and Mol 2008). Payola is a valid indicator of those records that are overproduced or heavily promoted. The fact that a record company will pay large amounts of money to induce airplay is concrete evidence that the company expects large revenues from that particular album. The second indicator measures promotional videos made for songs released during the sample period. A count indicator was coded as positive for each song if a promotional video was listed on AMG or included in databases available at mtv.com, vh1.com, rapvideos.com, or cmt.com. The third measure indicates the number of the artist’s songs first played on the radio in the fourth fiscal quarter of the year. Most labels will hold their most promising releases until this time, to cash in on the holiday shopping season. The fourth indicator measures advertisements appearing in Rolling Stone Magazine (2007) during the sample period. ³ Advertisements were typically those for a promotional tour, a new release, or a consumer product featuring a particular artist. This variable logs the number of advertisements appearing in the magazine for each artist during the study period.

Using each of these four indicators, I used structural equation modeling in Mplus to estimate a latent indicator of overall promotional support. The program progresses through iterative maximum likelihood estimations of four simultaneous regression equations, changing the value of promotional support each time, until an optimal solution is reached that captures the most variance in the four indicators through this new latent variable. As a result, the latent variable is an estimation that captures only the variation in these indicators that is shared. Doing so cancels out many of the idiosyncratic elements of each of these four indicators. Coefficients for the regressions of promotional support onto each indicator are reported next to the regression arrows in Figure 1, with standard errors in parentheses. The coefficient for payola was the lowest of the four indicators, and this indicator also has a high level of known sampling error: Payola was documented in a federal investigation looking for evidence of the widespread existence of pay-for-play practices, which did not require evidence of all artists receiving payola. Estimated promotional support is a standardized continuous latent factor with a mean of 0 and a standard deviation of 1, such that levels of promotion are always interpreted relative to this overall mean.

Listener preferences

Listener preferences, represented by the two squares (male and female listeners) in the center middle and center right of Figure 1, are measured as continuous variables reflecting the percentage of listeners who knew a particular song reporting that they liked that song. Listener preferences for songs appearing in the dataset two or more times were averaged together, resulting in a dataset with one observation per song reporting the average callout scores for that song. Using an average measure of preference decreases the contribution of measurement error, as a song’s callout can vary considerably from week to week.

Artist type

Each of the three above dependent variables was regressed onto artist type. In Figure 1, artist type is represented by a circle inside of a square, to signify the unique measurement of this indicator: an assignment based upon a categorical latent variable. I used latent class analysis to empirically identify latent artist types through analysis of patterns of interrelationships found in artist moods descriptors available through AMG. To use an oversimplified example, in an analysis that specifies two classes of artists, suppose statistical estimation places artists categorized as “sexy,” “calm,” and “fun” in one category and “sexy,” “aggressive,” and “fun” in another category. It is then the researcher’s task to apply meaning to these categorizations via labeling. Thus, the labels are merely identifiers—specific to the analytic task—and should not be taken as reified categories. Information on artist type was not available for mixed-gender groups. As a result, 20 songs by mixed artists were deleted from the sample, ⁴ resulting in a final sample size of 922 (767 male and 155 female) songs.

Artists fell into 10 male and six female classes: Table 1 provides illustrative examples of well-known members of each class to give the reader an opportunity to apply meaning to the classification scheme used in this paper. Table 1 shows that artists are not simply male or female, but gendered through discrete artistic personas. Common, a “sophisticated singer-songwriter” known for challenging the messages of black masculinity in popular music, is a different type than Ginuwine, a sexy “soul singer” known for his popularity with women, or 50 Cent, a “thug” known for his gangsta image. Yet all three are men—specifically, black men. Artist types are different measures of masculinity or femininity as much as they are measures of different artistic personas or racialized identities. Categorization here is structured to emphasize gender differences, thus, the resulting artist types emphasize the degree to which they are defined by gender, and can be distinguished from each other as different representations of gender. For example, “leaders” and “sophisticated singer-songwriters” reflect personas that emphasize the successful negotiation of masculinity, while “thugs,” “antiheroes,” and “absurd extremists” represent masculine struggle, depicted through displays of hypermasculinity (domination/aggression). Preferences for particular artist personas, then, are in part preferences for particular representations of gender. It is important to note that these types transcend race or genre (e.g., Billy Idol is also a “thug”). For a platinum-focused market, multiple versions of hit songs are made to maximize airplay across radio formats, making persona a more stable descriptor for these genre-spanning mega-stars. More detailed information on the types shown in Table 1 can be found in Donze (2010), which details the analytic process used to classify artists into these discrete types.

OPEN IN VIEWER

Table 1.

Illustrative Examples of Artist Types, Classified by “Moods,” N = 1,986 (Male), N = 522 (Female).

Class	Examples	Class	Examples
Men
Leader	Beatles, Rod Steward, Ray Charles, Temptations, REO Speedwagon, Barry Manilow, Bob Dylan, Harry Belafonte, Hall & Oates, Neil Diamond, Phil Collins, Elton John, Dave Matthews Band	Sophisticated Singer-Songwriter	Al Green, Enrique Iglesias, Kenny G, Kenny Loggins, Kenny Rogers, Art Garfunkel, Ruben Studdard, Common, Air, John Mayer, Marc Anthony, Ziggy Marley, Al Jarreau, Peter Gabriel
Emotype	Bruce Springsteen, Josh Groban, Used, Creed, Depeche Mode, Fallout Boy, Leonard Cohen, Johnny Cash, Echo and the Bunnymen, Elvis Costello, Silverchair, Jimmy Eat World, Dashboard Confessional, Starsailor	Antihero	Alice in Chains, Black Sabbath, Korn, Marilyn Manson, Megadeth, Metallica, NIN, Offspring, Three Days Grace, Celtic Frost, Dropkick Murphys, Nickelback, Shinedown, Social Distortion, Cypress Hill, Eminem, Public Enemy
Romantic Singer-Songwriter	Clay Aiken, Michael Bolton, Jackson Browne, Coldplay, Five for Fighting, Journey, Paul Simon, Simple Minds, Elliott Smith, Tears for Fears, Wallflowers, Trace Adkins, Clint Black, Garth Brooks, Leonard Cohen, Randy Travis	Thug	2Pac, 50 Cent, Busta Rhymes, Fabolous, Geto Boys, Ice Cube, Lil Wayne, Snoop Dogg, Mos Def, Naughty by Nature, Method Man, AC/DC, GNR, Billy Idol, Motorhead, Iggy Pop
Soul Singer	Babyface, Ginuwine, Lenny Kravitz, Michael Jackson, Seal, R. Kelly, John Legend, Marc Anthony, Maroon 5, Usher	Absurd Extremist	Aerosmith, Def Leppard, Stray Cats, Poison, Gwar, Alice Cooper, Kiss, Van Halen, Motley Crue, Prince, Kid Rock, Good Charlotte, Ludacris, Beastie Boys, Frank Black, Blink 182, Daft Punk, Dr. Dre, Ozzy Osbourne
Summer Party Jammer	Medeski Martin Wood, Jimmy Buffett, Phish, String Cheese Incident	Other	Beck, ATB, Unknown, Pink Floyd, Weird Al Yankovic, Spyro Gyra, David Byrne
Women
Underground Feminist	Indigo Girls, Kelly Clarkson, Fiona Apple, Ani DiFranco, Enya, PJ Harvey, FeFe Dobson, Alanis Morissette, Sinead O’Connor, Michelle Shocked, Patti Smith	Pop Princess	Britney Spears, Beyonce, Mariah Carey, Christina Aguilera, Ashanti, Aaliyah, Destiny
Femme Fatale	Joan Jett, Courtney Love, Lil’ Kim, Lindsay Lohan, MC Lyte, Da Brat, Pink, Chicks on Speed, Queen Latifah	Models of Femininity	Beu Sisters, Sheryl Crow, Alison Krauss, Celine Dion, Dolly Parton, Jessica Simpson, Deniece Williams
Chanteuse Singer-Songwriter	Tracy Chapman, Diana Krall, Barbara Streisand, Sarah McLachlan, Joni Mitchell, Norah Jones, Toby Lightman, Carly Simon, Suzanne Vega, Alicia Keys, Rosanne Cash	Divas	Gloria Estefan, Shakira, Pussycat Dolls, Janet Jackson, Whitney Houston, Jennifer Lopez, Pat Benatar, Dream, Alicia Keys, Joss Stone, Lucy Woodward, Madonna

Note. Moods = aggressive, angry/angst-ridden, amiable, bittersweet, boisterous/brash, bright/cheerful, confident, calm/gentle, energetic, fun/carefree, intimate/romantic, provocative, reflective, sensual, sexy, street smart, stylish/sophisticated, thuggish.

Results and Discussion

The first research question addresses whether male and female listeners prefer different artist types. The answer is yes: Significant differences were found between male and female listeners, Wald = 15.076, 1 degree of freedom (df), p = .000. This necessitates consideration of male and female listeners separately in subsequent analysis. Male and female preferences were significantly correlated with each other, however, and the curve (Figure 1) connecting male and female preferences captures the significant correlation between these two variables, Z = 12.9, p = .000. Figure 1 reports the estimated correlation between these two dependent variables in the measurement model: at .507, about 25 percent of the variation is shared. Interestingly, there was no correlation between promoter and listener preferences, Z = .554, p = .579 (men); Z = .289, p = .773 (women).

Before turning to the second research question, it is useful to first look at only artist gender, rather than gendered artist type, to establish a baseline for comparison. Here, there were no significant differences between promoter and listener preferences, Wald = 0.248, 1 df, p = .619. Thus, at first glance, it appears that promoters and listeners prefer male and female artists to roughly the same degree. Disaggregating by listener gender produces slightly different results. There were no significant differences between promoter preferences and male listeners, yet female listeners preferred female artists marginally more than promoters, Wald = 3.557, 1 df, p = .059. Comparing promoters to listeners as a group masked differences concerning female listeners and female artists. This is important to remember given that Mediabase does not gather information about listener gender in most radio formats.

But more interesting are the ways promoters differ from listeners in terms of artist type, in response to the second research question. When promoter preferences for gendered artist persona (rather than just gender) are compared with male and female listeners, the results were significant, Wald = 72.113, 14 df, p = .000 (men); Wald = 127.982, 14 df, p = .000 (women). This means that promoters and listeners do not value the same artist personas to the same degree—even promoters compared with male listeners. These results reinforce the conclusion that one must look at gendered persona to fully grasp gendered differences. This brings analysis to the last research question: Exploration of how promoter and listener preferences differ by exploring relative preferences for gendered artist types. I expect promoters and male listeners will like artists fitting hypermasculine or hyperfeminine types more than women, whereas female listeners will like gender-nonconforming artists more than men and promoters. Results for these comparisons are reported in Table 2.

OPEN IN VIEWER

Table 2.

Promotional Support and Male and Female Listener Preferences on Latent Artist Persona Types, for Artist-songs ^a Appearing in Callout Research 2002–2004, N = 922.

Artist persona (n)	Percent liking artist b			Standardized preferences			Standardized preferences
	Men	Women	p c	Promoters	Men	p	Promoters	Women	p
Soul Singer (54)	65.8	69.3	.000	.491	.169	.888	.491	0.479	.471
Leader (108)	69.1	70.6	.082	.618	.572	.516	.618	0.644	.518
Romantic S/S (66)	66.2	66.9	.552	.289	.217	.581	.289	0.158	.312
Thug (95)	69.5	66.6	.007	.537	.622	.321	.536	0.131	.051
Antihero (165)	61.1	61.3	.775	.471	−.412	.042	.470	−0.57	.000
Absurd Extremist (124)	64.1	63.8	.739	.402	−.039	.632	.402	−0.238	.004
Sophisticated S/S (21)	65.4	71.3	.000	−.093	.117	.177	−.093	0.737	.074
Emotype (63)	59.5	62.1	.022	.668	−.606	.004	.668	−0.456	.000
Summer Partier (44)	64.5	66.7	.201	.116	.005	.710	.115	0.131	.600
Other Male Artists (27)	63.8	65.2	.318	.134	−.082	.879	.133	−0.051	.324
Femme Fatale (12)	65.8	68.2	.362	.679	.169	.646	.678	0.326	.187
Models of Femininity (35)	61.1	64.6	.009	.387	−.406	.154	.387	−0.145	.026
Pop Princess (63)	63.4	66.3	.010	.320	−.127	.643	.319	0.085	.204
Chanteuse S/S (14)	64.2	71.2	.008	.472	−.031	.618	.472	0.728	.915
Underground Feminist (0)	—	—	—	—	—	—	—	—	—
Divas (31)	62.2	67.4	.000	.000	−.274	.147	.000	0.220	.231

a

Observations consist of 922 songs by artists singing/performing via 16 discrete categories of artist persona. A large majority of artists have more than one song in the dataset. Thus, standard errors are adjusted to account for the clustering of variation by artist.

b

Coefficients are reported as mean percentages (for men and women) and as z scores. (All dependent variables are continuous. Promotion is distributed as exponential decay and listener preferences are normally distributed.)

c

All p values are from Wald statistics testing the equality of men’s and women’s artist preferences (first panel) and the equality of coefficients for promotional support and male listener preferences (second panel) and female listener preferences (third panel). Results significant at the .05 level (two-tailed) are flagged in bold. Also, note that p values refer to a “hypothetical” general population. Because callout is not a representative sample of artists, these significance values do not generalize outside the sample.

The 10 male and six female artist types are reported in the leftmost column of Table 2. The number of each artist type included in callout data is reported in parentheses. All of the following were overrepresented in callout: leaders, thugs, antiheroes, pop princesses, and models of femininity. Thus, the types most researched in callout are hypermasculine or submissive-feminine, in line with the expectation that promoters would be most interested in these types. Of the 768 songs by male artists, 48 percent of them were songs performed by “leaders” (n = 108), “thugs” (n = 95), and “antiheroes” (n = 165). Before limiting the sample to songs in callout, these types were only 35 percent of male artists. Of the female types, 64 percent of the songs researched in callout were performed by “pop princesses” (n = 63) and “models of femininity” (n = 35). Yet these types were only 37 percent of the female artists in the original Mediabase sample. This is a reminder that, by using a sample that was partially chosen by promoters, data cannot equally represent promoter and listener preferences.

Over- and underrepresentation was more extreme for the female artists. For example, “underground feminists” were nonexistent in the callout sample. This could be an artifact of the urban, country, or alternative formats available, as they usually found airplay at college radio stations. However, some of these artists—such as PJ Harvey—are classified as alternative in record stores and on AMG. It could also mean that these artists are simply not researched: In other words, promoters are not as concerned with whether anyone likes “underground feminists.” For men, the “other” category was slightly underrepresented as well, but this is less cause for concern. This was a catchall category that included artists who did not fit neatly into other categories, such as Weird Al Yankovic, and, thus, serves more as a reference rather than a distinct artist type. In sum, these issues of sample representativeness underscore gendered differences outlined here so far, and lend some support to Hypotheses 2 and 3: Promoters differ from female listeners partly because they ignore gender-nonconforming types that women might prefer by not researching them. This is important, as promoters have eliminated these artists from the rest of the analysis by doing so.

The first panel of Table 2 reports the differences in the mean percent of men and women reporting they like songs by a particular artist type, corresponding to the equality of the parameters labeled *B and *C in Figure 1. Artists are generally well-liked, which is to be expected given that songs are chosen for callout if promoters expect they will do well. The overall differences are consistent with Hypothesis 3: Men like the three most hypermasculine types more than women (“thugs,” “antiheroes,” and “absurd extremists”), and women like the three most nonconforming types more than men (“chanteuse singer-songwriters,” “divas,” and “emotypes”). Wald tests between listener groups were significant for half of the artist types. For the male artists, women’s relative preferences for “soul singers,” “sophisticated singer-songwriters,” and “emotypes” and men’s for “thugs” reached statistical significance. All of these results make sense given what is known about these types. For example, promoters target a presumptively heterosexual and female audience with “soul singers,” and the listener preferences reported here suggest these promotional strategies are effective. These results are also consistent with Hypothesis 3: Lacking bravado and aggression, women liked male artists who emphasize emotionality and vulnerability more than men. While “thug” artists—a hypermasculine and aggressive type—are not targeted to exclusively male listeners, controversies surrounding the misogynistic representations of women by these artists might explain why they are less popular with women. For the female types, women liked all female types significantly more than men except for songs by the “femme fatale” type.

It is possible that some of these differences are driven by the fact that women are more agreeable in callout research. It is well known that women are more likely to respond to surveys, thus, they might also be more likely to report liking a song. If so, then gender comparisons will overestimate types women like more than men, but underestimate types men like more than women. Women’s scores hover about 2.5 percent above men’s. If this difference were the result of such agreeableness, significance levels would change even though the above-described pattern would remain the same. Thus, it cannot be said the results are driven by such bias.

Comparison of listeners by most and least favorite artist types places in perspective some of the significant gender differences found above. Table 3 reports artist types rank ordered by preference for men, women, and promoters, respectively. Like correlations in the measurement model, the correlation between rankings only reached significance for men compared with women, r_S = .654, p < .05 (men with women); r_S = .186, p > .05 (men with promoters); r_S = −.058, p > .05 (women with promoters). Although preferences for “emotypes” differed significantly between men and women, Table 3 shows that the magnitude of that difference was relatively small compared with the difference between listeners and promoters: It was among the least favorite for men and women yet it was a favorite for promoters. Overall, men reserved the most and least praise for male artist types. Their least favorite was the “emotype” (59.5 percent liked songs by this type) and their most favorite was the “thug” (69.5 percent). Most female types appeared in the bottom half of the men’s ranked list of favorites. Female listeners, however, had male and female types dispersed equally throughout their list of favorites. This is consistent with the claim that listeners claim ownership of their favorite artists, and that they use these artists to define themselves as gendered beings. For men, masculinity is defined as the exclusion of femininity and of women, whereas femininity is partially defined through attachment to men. The favorite male type for women, for example, was the “sophisticated singer-songwriter”: confident, successful, and not aggressive. Promoters are aware of this: They sell male artists to women as imaginary boyfriends, but female artists to men as sex objects (Dickerson 2005). Indeed, the favorite female type for men was the “femme fatale,” the most sexualized female persona. The “femme fatale” was also popular with women, but I suggest that male and female listeners might like this type for different reasons, in accordance with the way these artists are marketed differently to male and female consumers. Like the women’s favorite female type—the “chanteuse singer-songwriter”—the “femme fatale” presents an image of strength, independence, and control.

OPEN IN VIEWER

Table 3.

Artist Types Ranked by Promoter and Listener Preferences, N = 942.

Promoters	Male listeners	Female listeners
Femme Fatale	Thug	Sophisticated S/S
Emotype	Leader	Chanteuse S/S
Leader	Romantic S/S	Leader
Thug	Soul Singer	Soul Singer
Soul Singer	Femme Fatale	Femme Fatale
Chanteuse S/S	Sophisticated S/S	Diva
Antihero	Summer Partier	Romantic S/S
Absurd Extremist	Chanteuse S/S	Thug
Models of Femininity	Absurd Extremist	Summer Partier
Pop Princess	Other Male Artist	Pop Princess
Romantic S/S	Pop Princess	Other Male Artist
Other Male Artist	Diva	Models of Femininity
Summer Partier	Models of Femininity	Absurd Extremist
Diva	Antihero	Emotype
Sophisticated S/S	Emotype	Antihero

Comparisons with promoter preferences add some perspective to these results, to which we now turn. The second panel of Table 2 compares promoter preferences to standardized male listener preferences, corresponding to the equality of parameters labeled *A and *B in Figure 1. Comparisons between promoters and female listeners are reported in the third panel of Table 2, corresponding to the equality of the parameters labeled *A and *C in Figure 1. To compare listener preferences with promotional support, listener preferences are standardized with a mean of 0 and a standard deviation equal to 1. For all comparisons, “divas” formed the reference category, with a mean set to 0. Looking at promoter preferences, there were no huge differences in the range of promotional support across all artist types. The most favored artist class was the “femme fatale,” which received promotion .679 standard deviations above the mean, while the least favored class was the male “sophisticated singer-songwriter,” associated with promotional support .093 standard deviations below the mean.

Men had a much greater variety to their responses, and the large number of negative coefficients for male listeners shows that, in general, men were also more discriminating than women or promoters. Men were more discriminating partially due to their lack of enthusiasm for “emotypes,” but also because of their lack of enthusiasm for female artists in general. Men preferred “thugs” more than promoters, but for all other artist types, men’s preferences were either below or significantly below promoter preferences. For “emotypes,” “antiheroes,” and “models of femininity,” male listener preferences were roughly one standard deviation lower than promoter relative preferences for these types.

Comparisons between promoters and female listeners shed some light on these differences. For some types—such as the “thug,” “diva,” and the “chanteuse”—promoter preferences fell in the middle of these two gendered poles. This suggests that differences between listeners and promoters could reflect different tastes of male and female consumers, rather than a simple lack of connection between promoters and listeners. But otherwise promoters seem to miss the mark. They overestimate the relative preferences of female listeners for “emotypes,” “antiheroes,” and “absurd extremists” just as they do with male listeners. Likewise, they underestimate preferences of all listeners for “sophisticated singer-songwriters,” “romantic singer-songwriters,” and “summer party jammers.” And, overall, promoters differed from female more than male listeners: Female listeners liked “sophisticated singer-songwriters” the most, yet promoters liked them the least. This cannot be explained by male preferences, because men liked songs by this artist type as well.

In sum, these results show gender disparity in several ways. Promoters put less energy into female artists in general. Similar to Milkie’s (2008) argument about photographers, promoter’s actions could be understood as catering to male listeners, who generally do not like female artists. Similarly, greater differences between promoters and female listeners can be explained by the fact that female listeners prefer male and female artist types, whereas male listeners generally prefer only male artist types. Thus, promoters catering to a mixed-gender audience will form preferences more similar to male listeners: The results of the three groups regarding the “femme fatale” exemplify this pattern, as it is the only female type favored with men. Third, promoters are not putting resources into the female types that female listeners want to hear. They focus on the hypersexualized “femme fatale” (as well as the demure “pop princesses” and “models of femininity”) instead of alternative versions of womanhood—such as the “chanteuse”—that women prefer the most. These promoter preferences cannot be entirely explained by comparison with male preferences either, because the latter two female types were not favorites for male listeners. This could be due to overreliance on recombinant versions of existing submissive and hypersexualized female types (Gitlin 1983). A similar pattern is found with male artist types: “emotypes,” “antiheroes,” and “absurd extremists” were overpromoted and “sophisticated singer-songwriters,” “romantic singer-songwriters,” and “summer party jammers” were underpromoted relative to listener preferences. With the exception of “emotypes,” these results are consistent with Hypothesis 3, as the overpromoted types emphasize hypermasculinity, whereas the underpromoted types were all nonaggressive.

Conclusion

While music might not brainwash listeners like Strange Brew or Clockwork Orange, it structures thought and experience. Choosing energetic music while exercising is one example of how people mindfully use music to do this (DeNora 2000). Adorno was more worried about people using music to escape from experience: Repetitive drum beats with lyrics that implore listeners to “just dance” encourage listeners to be mindless, which opens the door for music to be used to control people (DeNora 2003:21). Music is one of the most powerful tools that adolescents use to construct an adult sense of self. The mindless consumption of recombinant and stereotyped artistic identities by listeners engaged in such important work is cause for concern. It could lead to precarious constructions of self, especially for marginalized listeners.

This project asked whether or not the record industry provides listeners with musical tools they find useful. The production of culture perspective suggests that there will be differences between the artists labels choose to promote and the artists listeners actually want to hear. Production necessarily introduces a gulf between promoters and listeners because promoters focus on market share, profit, and recombinant types reminiscent of previous platinum sellers rather than newer and more creative representations that listeners favor more. The focus on platinum sales—which means promoting genre-spanning artists that can attract the largest possible spending audience—means that labels rely on easily recognizable and stereotyped persona types.

A striking difference between promoter and listener preferences as a whole concerned the “antihero” and the “emotype.” Promoters put a great deal of energy into these types of artists, yet they were among the least favorite types for both male and female listeners. Central artists in the “antihero” category include Black Sabbath, Megadeth, Marilyn Manson, Eminem, and in the “emotype” category, they include Bruce Springsteen, Johnny Cash, Depeche Mode, Elvis Costello. Looking at some of the defining artists in these categories, this difference is most likely caused by promoters pushing established artists with a large back catalogue of greatest hits and earlier albums promoters are hoping to sell. These artists have releases spanning decades. Labels push their new releases or rereleases because they hope to hook a new generation of listeners. But these artists are also valuable to promoters because they fit well within existing organizational structures. As Todd Gitlin (1983) argued, new types are often recombinant “spinoffs” of established types, because framing new artists in terms of the established personas of their predecessors makes production predictable and efficient.

A similar argument explains the difference between promoters and listeners regarding the female artist types. Promoters would be better off putting more energy into the “chanteuse singer-songwriter”: Just like the male “sophisticated singer-songwriter,” this category includes more recent artists that listeners tend to like more, such as Alicia Keys or Erykah Badu. But promoters put a lot of energy into “models of femininity” and “pop princesses”: This includes artists such as Britney Spears, Christina Aguilera, Dolly Parton, Celine Dion, and Jessica Simpson. As with the male types referenced above, promoters are favoring artists with well-established personas and large back catalogues. This also means they are putting energy into representations of traditional femininity—passive, submissive, and sexualized—when listeners favor different gender representations by newer artists.

Notably, these bubbly and positive images of traditional femininity are racialized as predominantly white images. The degree to which (especially nonwhite) women are transforming hypersexualized and objectified derivative images into powerful statements of femininity (Keyes 2004), and the degree to which the nonconforming images of male or female sexuality can be constructively used to grapple with sexism in larger society (Radway 1984), cannot be fully understood without simultaneously looking at how images are racialized. My data are consistent with the contention that the “femme fatale” is performing that role. But without information about how these gendered preferences are also racialized, the picture is necessarily incomplete. Further research should analyze promoter and listener preferences for raced, gendered, and sexualized personas.

This paper also showed that differences between promoters and listeners are more pronounced with female listeners and female artist types, which could be partially explained with reference to male listener preferences. With the exception of songs sung by the “femme fatale,” male listeners generally preferred songs sung by men. Artists in the “femme fatale” category include Joan Jett, Courtney Love, Pink, and Queen Latifah. Producers know they cannot produce a platinum seller without the support of male and female listeners, and, thus, the “femme fatale” is the most heavily promoted, and the most sexualized female persona. Producers who encourage new artists to adopt the “femme fatale” identity do so partly because of the structural limitations posed by the gender hierarchy in larger society: Producers are catering to an audience that, overall, prefers male artists. Much like Ruth Milkman (1987) and Milkie (2008), I found that the responsibility for gender preferences is not just borne by the cultural producers, but also shared by the diffuse and ever-present effects of culturally institutionalized sexism.

Culturally institutionalized sexism also explains differences in same-gender listener preferences. Representations of struggle were distinctly gendered. Male listeners preferred hypermasculine representations of struggle—domination of women, violent aggression, and anger. This suggests that, in their social construction of self, male listeners are using music to draw boundaries between themselves and others (Bryson 1996). Female listeners, however, preferred representations of nonconformity: confidence, anger, street smarts, and artists defying social and gender conventions. The data are limited on this point, however, as the most nonconforming female type—the “underground feminist”—was eliminated from the sample because none of these artists were researched in weekly callout. These artists were energetic, aggressive, angry, and reflective, and included artists such as PJ Harvey, Tori Amos, and Avril Lavigne. Further research should gauge preferences using a different data source—one that is more representative of the artists available on the market today—so that levels of promotion or listener preference can be generalized outside the sample population. Doing so would give a more complete picture of the artist types that promoters and listeners prefer.