Women have made tremendous progress entering the workplace, in large measure due to Title VII of the Civil Rights Act of 1964 that prohibits discrimination in employment “because of” sex. One of the doors that opened for women was TV news broadcasting, which journalist Barbara Walters stepped through in 1976 as the first woman to be a co-anchor on a prime time news show. Once in the door, however, Title VII is not sufficient for dealing with a more current phenomenon in TV broadcast news, namely, the sexualization of women’s dress styles in contrast to the professionalization of men’s. Today, it is common for women newscasters to be “urged,” if not required, to wear clothes that are tight, brightly colored, often sleeveless, with flowing hair styles, lots of make-up, and jewelry accessories. Male newscasters, by contrast, are required to wear monotoned dark suits, white or pale blue shirts, short haircuts, no facial hair, and no jewelry.
As is well known, courts have ruled that Title VII does not permit sex discrimination policies, defined by a reference to an immutable characteristic of a protected group or a policy that poses an undue burden on an employee (Clements, 2009). Less well known, however, is that courts have ruled that Title VII does permits trait discrimination policies. A trait is defined as “a quality that makes one person or thing different from another” (Macon, 2015, p. 47). Courts have ruled that grooming codes are permissible trait discrimination policies, even if such grooming codes do not constitute a bona fide occupational qualification (BFOQ), as long as those policies are “reasonable, work-related, and applied equally to all employees” (Bandsuch, 2009, pp. 969–971).1
The only restrictions on trait discrimination policies are that these employment directives refer to physical traits or performed behaviors that are voluntarily chosen to communicate a group identity rather than an immutable characteristic, and the policies do not impose an adverse effect on the status of employees “because of” their “race, color, religion, sex, or national origin, such as constituting an undue burden on employees (Bandsuch, 2009, pp. 969–971). Thus, for example, courts ruled that employers who fail to accommodate lactating women to breastfeed or to pump their breasts at work are not in violation of Title VII’s prohibition of sex discrimination because breastfeeding is the result of a woman’s voluntary choice, not the result of an immutable characteristic that defines all women who have been pregnant and/or have given birth (Clements, 2009).
As Mark Bandsuch (2009) notes, therefore, even though many companies establish dress codes based on “inherent prejudice and stereotyping” of people in terms of their sex and/or race, courts have failed to rule that such policies violate Title VII. Nevertheless, Bandsuch contends that trait discrimination policies are a “second generation” form of a “discriminatory policy” that harms classes of people for whom Title VII was enacted to protect, such as people identified by their race and/or sex (Bandsuch, 2009, p. 969).
Social Psychology and Dress Codes
Social psychologists have analyzed and evaluated dress codes much differently than courts. Fundamental to how social psychologists study visual cues is the concept of a diffuse status characteristic, which imparts evaluations of people on the basis of their attributes, such as sex, race, and age. These evaluations generally are more negative for women than for men. In addition, social psychologists have found that people use visual cues to make judgments about the professionalization, worth, intelligence, and competence of other people (Todorov et al., 2005; Zebrowitz & McDonald, 1991; Zebrowitz & Montepare, 2005). Specifically, they have found that sexualized clothing promotes the objectification of individuals by dividing them into “parts," which is likely to generate negativity, resulting in a less complex, more superficial evaluation of that person and of their professional competence (Haslam et al., 2005; Loughnan & Haslam, 2007). Studies find that this is the case even when women are presented as victims of violence, such a rape. Loughnan et al. (2013), for example, found that a sexualized woman who is presented as a victim of rape is perceived by people as less deserving of moral concern compared to women portrayed as rape victims in a non-sexualized way.
Physical appearance prioritized over character traits and professional competence, in fact, is a classic definition of objectification, namely, “instead of being perceived as a whole person, with feelings, personality, and character, an objectified person is perceived as a collection of body parts, displayed for the gaze of others” (Register et al., 2015). As scholars note, not only in daily experience, but also in the media, women are disproportionately depicted with a focus on their physical traits, thereby sexually objectifying women (Heflick et al., 2011). As scholars note, not only in daily experience, but also in the media, women are disproportionately depicted with a focus on their physical traits (Heflick et al., 2011). This is a problem because researchers have established that when women are objectified by directing people’s attention to women’s physical qualities rather than mental capacities, the result is that people perceive women as being more like objects than like human beings, which includes viewing women as less competent and less intelligent.
What is more, evidence of how sexual objectification turns women into objects are studies that show that though people generally can recognize objects presented in an inverted form, they cannot recognize people in an inverted form. The exception is sexualized women. When people are asked to identify an inverted sexualized image of a women, in a swimsuit, for example, they can do equally well as when the image is not inverted. The same is not true for images of non-sexualized women or for images of men, whether sexualized or not. To explain the exception of why people can recognize inverted images of sexualized women, researchers argue this is because the ubiquitous presentation in the media and elsewhere of women’s sexual objectification leads people to focus on specific parts of women’s bodies – as if women were objects. People’s focus on women as objects “impairs configural processing” (i.e., the connection of distinct features to form a coherent whole) and subsequence recognition (Heflick et al., 2011, p. 227).
Notably, a critical rationale for women’s employment, as enforced by Title VII, is the obvious premise that women are professionally competent. Our research, corroborating that of social psychologists, affirms that stereotypical dress styles for women reduce the perception of their competence. We argue that it is an obvious disadvantage in the workplace to have the perception of one’s competence reduced, especially since competence is the attribute required to be employed in the first place. We contend, therefore, that employment policies that reduce the perception of women’s competence, but not men’s, disproportionately disadvantage women as members of a protected class. Courts have defined policies that do so as constituting an adverse effect (Bornstein, 2017). While Title VII may permit trait discrimination policies that are voluntary and/or do not create an undue burden, this legislation does not permit policies that produce an adverse effect for members of a protected group. Specifically, therefore, integrating scholarship on law and social psychology reveals a new way to challenge trait discrimination policies as a violation of Title VII.
The Sexualization of Women News Anchors
No television station has what they would call an official “edict” on what on-air women can or cannot wear. In response to numerous allegations of sexual harassment brought against former Fox News Chairman and CEO Roger Ailes in recent years, a Fox News spokesperson denied the existence of such a dress code (Terkel, 2016). The CNN website’s “Inside CNN” Frequently Asked Questions page similarly states that news anchors “have general guidelines they must follow as far as style and color but do have the flexibility to wear their own clothing” (Cable News Network, 2005). Yet, as many television viewers may have intuited, networks are investing more time and money into the appearance of women new anchors and reporters. Concurrently, many – including the very women whose appearance is in question – have observed a trend towards increasingly sexualized dress, hair, and makeup among women on air. One of the reasons behind this heightened emphasis on and scrutiny of news anchors’ appearances is the development of HD television. Wrinkles, gray hairs, and other such “flaws” that might have been obscured by lower resolution televisions are all now too obvious to viewers and news directors alike (Rhew, 2017).
Demographic changes have also led to changes in television news norms. More networks now hire younger anchors, who are cheaper than older, more established news anchors. As a result, there has been a “loosening” of more conservative dress codes. This is in part due to the on-air talent themselves as well as networks’ desire to capture the attention of millennial viewers (Rhew, 2017). Women’s dress on air has also changed, perhaps most notably, around the question of whether to bare skin. For example, Andrea Kremer recounts to the Boston Globe an experience she had covering a breaking story in Chicago in 1989 on a 103-degree day. “Seeking feedback after a great day of reporting, she asked her boss how she did. “You wore a sleeveless dress,” he said. In those days, it was a bad thing. Fast-forward almost 30 years, and it’s still all about the arms, only now they’re supposed to be on display” (Teitell, 2017). According to Adam Rhew of Racked, television stations conduct research via focus groups in order to better understand viewers’ preferences, including opinions on the physical appearance of news anchors. Networks are all but desperate to find out what will attract an audience and keep it engaged. Television stations also hire “image consultants” and “image coaches,” described as a “small but powerful group of consultants whose job it is to get us to trust, like, and watch our hometown Rob Burgundy” (Rhew, 2017). These consultants, hired by management, meet with on-air talent and advise them on their professional appearance. Not every station uses image coaches, but “their advice is widely shared at conferences and within ownership groups. News directors pass along tidbits to their staff.” News anchors who serve as leads in medium and large markets are most likely to have personal sessions with image consultants. Jedediah Bila, now a host on The View, discussed the dress code she endured during her time as a Fox News correspondent:
You get wardrobe, but it’s different than here. You used to go into a room and there were a bunch of dresses you could choose from. I was told at one point I wasn’t allowed to wear orange because Roger [Ailes] didn’t like the color orange. I didn’t see any pants. People always say, “Why didn’t you wear pants?” You notice I wear pants a lot here. I didn’t wear pants because I wasn’t given a pants option. I had to choose skirts. (Robinson, 2017)
Though Fox News may be the most blatant example of increasingly sexualized dress styles among women news anchors, this phenomenon is by no means contained to this one network. Boston’s Heather Unruh, longtime WCVB-TV anchor, resigned abruptly in the fall of 2016. She told New England Living TV that she felt, “Women are ‘encouraged’ to dress more provocatively than I feel is appropriate for delivering news” (Teitell, 2017). Similarly, a Boston-based broadcaster speaking to the Globe on the condition of anonymity said management at her station has told women to wear “tighter, smaller, shorter, more revealing clothes.” She continued, “What you don’t see on TV is that many times women have clothespins in the back to make [their clothes] tighter.” It is a “self-reinforcing situation,” she said, with less established anchors taking sartorial cues from more provocatively dressed anchors as a way of “getting attention” (Teitell, 2017). Some of the other Boston-based broadcasters who spoke to the Globe, all off the record, shared stories about “wardrobe consultants hired by station management pushing clothing that some on-air talent don’t want to wear; women crying in the makeup room because they feel pressured to dress a certain way; a modestly dressed anchor being asked to dress like a sexier new colleague who wore her skirts short and her tops unbuttoned” (Teitell, 2017).
Neither image consulting nor the general attention to physical appearance is exclusive to female talent, but the dress code policies for men do not sexualize them. Rather, according to Racked, one male anchor “…now meets once every three months or so with a [TV station] consultant who sorts through his suits and ties and tells him what to wear and what to avoid. ‘You’re pretty much banned from wearing shirts that aren’t solid light blue or white,’ he says” (Rhew, 2017). Thus, there is a tremendous amount of pressure to think about one’s appearance when working in news broadcasting. Notably, however, the policies dictated to men and women are strikingly different.
Trait Discrimination and Title VII
The imposition upon men and women in the news industry of sex stereotypical standards has drawn the attention of many legal scholars. Leslie Gielow, for example, points out that Title VII does not allow customer preference to dictate disparate treatment for men and women in the workplace. Thus, Title VII prohibits airline policies that would only hire women as flight attendants on the grounds that passengers do not prefer men in that job. As Gielow notes, policies that would allow customer preferences to determine employers’ decisions about whom to hire, promote, or fire reflect the very prejudices that Title VII was enacted to avoid in the first place (Gielow, 1985).
Yet, when it comes to customer preferences for newsroom personnel, courts have taken an interesting, if not contradictory, stand. While they have ruled that customer preference cannot determine whether men or women are hired, after men and women are on the job, courts have ruled that Title VII permits disparate treatment of male and female newscasters as based on the preferences of the public. Thus, courts have ruled that Title VII permits TV news organizations to demand that their women broadcasters appear young, attractive on the basis of stereotypical standards, and non-aggressive, because this is what the public prefers. Law scholar, Deborah Rhode defines this disparate treatment of women as the “beauty bias” – a “well-known beast that dictates that women teeter on high heels, starve on perpetual diets, fight against aging, and even submit to surgery to correct offending features” (Rhode, 2010, p. 1785).
Law scholar Leslie Gielow (1985) terms this policy of placing sex “stereotypical expectations,” that is, sex segregated trait discrimination policies, on top of sex neutral hiring policies, as a “sex plus” policy. She, along with Rhode and many other legal scholars, finds it appalling that courts have interpreted Title VII to allow the imposition of trait discrimination policies that sexualize women, but not men (Gielow, 1985, p. 444). As Gielow argues, sex plus discrimination policies affect not only the particular woman upon whom they are imposed, but rather all women as a group, “due to the fact that trait discrimination based on stereotypes reinforces those stereotypes for all of society, employers, and women themselves (Gielow, 1985, p. 444). Thus, as Nadine Taub notes, “a woman forced to conform to role expectations is ‘doomed to play a part in reinforcing the vitality of such criteria in her own eyes, the eyes of other women, and in the eyes of her employer and co-workers’” (as cited in Gielow, 1985, p. 444).2
Many constitutional and employment practice law scholars, therefore, have sought to expand Title VII to prohibit trait discrimination policies that violate protected group identities, such as race, class and sex. Yet, this has been all but impossible because courts categorize most grooming requirements as permissible trait discrimination, rather than unlawful sex discrimination. Trait discrimination, such as sexualized dress styles, are defined by physical traits or “volitional” performed behaviors that can be voluntarily chosen to communicate a group identity. By contrast, courts generally rule that traits that invoke an immutable characteristic of a group, violate a fundamental constitutional right, or constitute an adverse effect, such as posing an undue burden, are unlawful forms of sex discrimination (Clements, 2009, pp. 171–173). An African American woman airline attendant, Renee Rogers, for example, challenged an American Airlines’ grooming policy that prohibited braided hair styles, claiming that this policy was a form of race discrimination in violation of Title VII. However, a district court dismissed her claim on the grounds that she did not demonstrate that cornrow-braided hair styles are characteristic of only, or, of even predominantly, African Americans. What is more, the court ruled that all employees were prohibited from cornrow braids, not just African Americans. Furthermore, rather than an immutable characteristic of a racial group, wearing a cornrow braid was judged to be an "easily changed characteristic" (Clements, 2009, pp. 172–173).
Typically, therefore, courts have ruled that employers’ grooming requirements that require women, for example, to wear more make-up than men, more complicated hairstyles than men, and more elaborate clothing and jewelry accessories than men do not constitute a form of sex discrimination that violates Title VII because such requirements do not pose an undue burden on women. The Ninth Circuit held in Jespersen v. Harrah's Operating Co, for example, that "’sex-based difference[s] in appearance standards alone, without any further showing of disparate effects,’ do not establish a prima facie case of discrimination’” (as cited in Bandsuch, 2009, p. 982). As law scholar, Mark Bandsuch (2009), notes, in this case, the court “reasoned that the costs, time, and hassles of putting on make-up (including the emotional hardships associated with complying unwillingly with gender stereotypes) were not an unequal burden” on women (Bandsuch, 2009, p. 982).
Social Psychology and the Perception of Competence
Social psychologists, however, raise a different question: what affects the perception of competence? One of their most important answers focuses on how diffuse status characteristics are socially constructed and how they affect the perception of competence (Jackson et al., 1995). A diffuse status characteristic is “an initially non-valued characteristic that can acquire status value and generalized expectation states from its association with already established and valued status elements” (Berger & Fişek, 2006). As Wagner and Berger (1997) explain, diffuse status characteristics generate specific and more general expectations. The former refers to expectations in a clearly defined situation, such as one’s competence in performing on a math test, and the latter to perceptions of general intelligence.
Significantly, the components that constitute the structure of a diffuse status characteristic include gender as a “socially significant characteristic” that can partition populations into groups and attribute to individuals in those groups significantly different conceptions of their capacities, social worth, honor, prestige, and physical attractiveness (Berger & Fişek, 2006; Jackson et al., 1995; Wagner & Berger, 1997; Webster & Driskell, 1983). As Berger and Fişek put it, “gender may be a diffuse status characteristic for a population at a given time if the members of that population hold differential status evaluations for male and female, with, say, males being more highly esteemed, honored, and in general more socially valued than women, and with the members of the population commonly assuming that men are in general superior and more capable than women on a wide (and typically unspecified) range of valued tasks and activities” (Berger & Fişek, 2006, p. 1039).
Jackson et al. (1995) note that status generalization theory “evolved from expectation states theory and status characteristics theory,” and it has been important for explaining “how external status characteristics influence interaction and outcomes” in task-oriented groups (p. 108). Along with gender, social psychologists contend that physical attractiveness is a diffuse status characteristic in American society. Significantly, physical attractiveness affects people’s “cognitions and behavior,” whether physical attractiveness is relevant for performing a task or not. People who are perceived to be physically attractive, for example, are also perceived to be “more intellectually competent than unattractive people” (Jackson et al., 1995, p. 108).
Not surprisingly, social psychologists have studied the interaction between gender and physical attractiveness as diffuse status characteristics. Since males are generally perceived to be more competent than females, and since physically attractive people are also perceived more generally to be more competent than less physically attractive people, the result is that physically attractive males will be perceived as the most competent compared to physically attractive females or physically unattractive males or females (Jackson et al., 1995, p. 108). What is more, some scholars have found that “attractiveness can be detrimental to women in certain employment contexts,” such as occupations perceived as masculine sex-type jobs” (Johnson et al., 2010, p. 300).
Social psychologists rarely, if ever, use the term “trait discrimination policies.” However, in effect, they have been studying trait discrimination policies concurrently with law scholars for a very long time, under the rubric of “objectification” and “sexual objectification” in particular. What is more, the phrase that “we are what we wear” has been shown to signify one’s nationality (Hakoköngäs & Sakki, 2016), social class (Mann & Loren, 2001), race (Smith, 2003), religion (Eller, 2000), political positions (Boris, 2017), incarceration (Gubar, 1981), modernity (Wilson, 2003), social change (Zurbriggen, 2013) and, notably, one’s sex identity, the latter being a precursor to sexual objectification. Law scholar Katharine Bartlett (1994), for example, recounts a disturbing anecdote about how clothing – even a simple barrette – defines sex, even among children as young those in nursery school. As she tells the story:
[My son, Jeremy,] naively decided to wear barrettes to nursery school. Several times that day, another little boy insisted that Jeremy must be a girl because "only girls wear barrettes." After repeatedly insisting that "wearing barrettes doesn't matter; being a boy means having a penis and testicles," Jeremy finally pulled down his pants to make his point more convincingly. The other boy was not impressed. He simply said, "Everybody [boy] has a penis; only girls wear barrettes" (Bartlett, 1994, p. 135).
Given the power of clothing to define identity, scholars seek to understand the impact of clothes on perceptions of character, likeability, and professional competence. Some researchers focus on how our brains globally or locally process visual information. It is well documented that when people perceive the world around them, they are more likely to use global processes that focus on the entire global visual field rather than local processes that focus on visual parts or local details (Gervais et al., 2012). However, numerous studies have suggested that global and local processing, and the interplay between the two, underlie person versus object recognition.
Global processing is used to recognize people, because in order to recognize a face or body, perceivers must connect information about specific body parts and their relation to each other. Local processing, on the other hand, underlies the recognition of objects, because in order to recognize objects, perceivers only require information about the parts of the object, rather than the context or information about the relations between different parts (Gervais et al., 2012). For example, a study conducted on person versus object processing used this whole versus parts recognition paradigm. It found that a door on a house was recognized similarly regardless of whether it was presented in the context of the “whole” house or in isolation as part of a house. Extending this existing evidence to the sexual objectification of women, Gervais et al. (2012) suggest that women’s sexual body parts are processed similarly to objects through local, rather than global, processing. In other words, women’s sexual body parts are recognized in isolation and do not require the context of an entire body to be processed by a perceiver.
In the Gervais et al. (2012) study, male and female participants were first shown one image (the “original image," which was a body part), then two images side-by-side (the original image, and a slightly modified version of the original image). They were then asked to identify which of the two images was the one they had first seen. The results confirmed the hypothesis that women’s bodies are reduced to their sexual body parts, both by male and female perceivers: female body part recognition was better than male body part recognition, and female body part recognition was also better than female whole-body recognition. The “sexual body part recognition bias” is thus defined as the phenomenon in which perceivers recognize women’s sexual body parts in isolation without requiring spatial information and context provided by the entire body (in line with local processing). The numerous studies and examples of the sexual objectification of women cited by Fredrickson and Roberts (1997) show how pervasive and normalized it has become, both in interpersonal interactions as well as in visual media. For example, a 1990 study found that women are gazed at more than men (Hall, 1990). Women are also more likely to feel “looked at” in interpersonal encounters. Multiple studies (Fromme & Beam, 1974; Gardner, 1980; Henley, 1977) have found that men direct more non-reciprocated gaze toward women than the other way around, particularly in spaces that are public or open, and that this gaze is often accompanied by sexually evaluative commentary.
The concept of the “male gaze,” first coined by feminist film critic Laura Mulvey, has entered the mainstream vocabulary as a way of understanding the experience of living as a woman today. Yet, despite the extensive literature on the social, political, and historical roots of sexual objectification, or its implications on both recipients and perceivers of the gaze, little research has been done to understand exactly how sexual objectification might take place. The study by Gervais and colleagues (2012) is one of the first attempts at an empirical study investigating these mechanisms. While we are not the first to suggest a connection between clothing and objectification (Bartlett, 1994; Dellinger, 2002; Goodin et al., 2011; Gurung & Chrouser, 2007), we argue that sex-segregated clothing, as a set of norms both formal and informal, may facilitate the negative consequences of objectification, particularly for women.
Other recipient-focused effects of objectification include increased shame, anxiety, stress, and lowered intellectual performance. Interestingly, in many of these studies, it is clothing that is manipulated as a proxy for objectification. For example, Fredrickson et al. (1998) found that women who wore a swimsuit performed more poorly on a math test than women who did not. The same was not true for men. The authors also note that certain kinds of necklines and hemlines require constant vigilance for women, creating opportunities for anxiety and regular body monitoring, which can lead to many of the negative repercussions of objectification that the authors outline.
Partisanship, Women in Politics, and Perceptions of Competence
Women running for political office have long faced the trial of “what to wear?” (Ridgeway, 2009). As Hillary Clinton, the first woman to compete for the presidential nomination of a major political party, put it when preparing for one of her primary debates: “You know . . . I’ve got no idea [what to wear] you have suggestions, I’m open to them” (as cited in Funny Or Die, 2016). As Olivia Aylmer notes, women politicians find themselves in a bind. Should they take the conservative road and wear pants suits or should they don dresses and skirts. Either way, women candidates for political office are bound to end up scrutinized and criticized by the media for what sartorial choices they make, if not also analyzed in terms of how their dress styles do or don’t sexualize them (Alymer, 2015). There is nothing comparable for men in politics when they must decide how to present themselves as candidates for political office. When Hillary was asked by a commentator, for example, what she thought Donald Trump would wear to his 2016 primary debate, she responded: “I assume he’ll wear, you know, that red power tie” – i.e., all he needs to establish his masculine credentials as an acceptable candidate (as cited in Funny Or Die, 2016).
Political scientists also analyze perceptions of competence, most often in the context of political campaigns (Bauer, 2013). They find that gender stereotypes are more commonly applied to women than to men candidates running for political office.3 Voters assume that women candidates are more caring, warm, and supportive of social policy issues, such as health care, welfare provision, and public funding of daycare and family leave policies (Bauer & Carpinella, 2018; Fridkin & Kenney, 2009). Using the case of the Republican presidential campaign of John McCain and his running mate, Sarah Palin, in 2008, Heflick and Goldenberg (2009), for example, build a strong empirical case suggesting that Sarah Palin was sexually objectified by the American public. Social psychologists have found that a pernicious consequence of sexual objectification is the tendency of the “objectified” to internalize the perspective of “observers” as if they are their own. This process is termed “self-objectification” (Register et al., 2015). Self-objectification often results in “broad and negative influences, such as appearance anxiety and body shame in the case of body image (Register et al., 2015). What is more, self-objection based on the internalization of the sexual objectification of women tends to make women behave more as if they really are objects rather than human beings who have mental abilities (Heflick & Goldenberg, 2014). Research suggests, therefore, that the focus on Palin’s appearance, a classic component of sexual objectification, likely undermined perceptions of Palin’s competence, warmth and morality (Loughnan et al., 2017). In addition, it may also have increased Palin’s focus on her own appearance, which, “consistent with research on self- objectification, likely impaired the competency of her actual performance” (Heflick & Goldenberg, 2009).
Building upon an earlier 2009 study, which examined the effects of objectification on the perceived competence of both Sarah Palin and Angelina Jolie, Heflick and colleagues expand their female targets in a later study to include Michelle Obama, CNN morning show host Robin Meade, and several unknown, female weather forecasters. Matching the results of the 2009 study, they find that each female target is perceived as less competent when respondents are instructed to focus on their appearance (Heflick et al., 2011). Several other studies also explore the objectification of female political candidates, agreeing with the overall findings of this study (Heldman & Wade, 2011; Lundell & Ekstrom, 2008; Wasburn & Wasburn, 2011).
This real-world case and the research around it call for further exploration of dress. Specifically, can we replicate some of these findings in a controlled lab environment? Is it possible to isolate the effects of dress? We bridge the gap between the experiences of women in the newsroom and politics (which are similarly dependent on the transmission of consumable information) by performing an experiment involving actors hired to play the role of newscasters reading the news.
Our study contributes to this literature in two ways. Methodologically, we use an experimental design with a larger sample size to further investigate how perceived competence is affected by dress style and gender. This is an area scholars are increasingly interested in exploring further (Chiao et al., 2008; Ditonto, 2017; Oh et al., 2019; Praino et al., 2014; Todorov et al., 2015). In studies of political leadership, however, only two studies on electoral success and perceived suitability or competence for office have taken clothing into account. Using an experimental design, Hayes, Lawless, and Baitinger (2014) manipulate descriptions of a male and female candidate’s appearance as “sloppy” or “stylish,” concluding that negative coverage of a female candidate’s appearance does not negatively affect her electoral success. While important to establish, this does not focus specifically on the critical markers of difference between men and women’s fashion, namely the sexualization of women’s dress styles, but not of men’s. In addition, Laustsen (2014) finds that facial competence predicts electoral success both as a mediator (for facial attractiveness) and as a variable by itself, controlling for non-facial factors, such as whether clothing (in photos) was formal or informal. This limited categorization, however, does not test how the sex-segregated nature of dress may affect electability. Second, to the authors’ knowledge, this is the first large-N experimental work that incorporates video into measurements of perceived competence, thus simulating real world conditions more closely than do photos and/or voice alone (Coffé, 2018; Klofstad, 2017).
Below we present our expectations, which we evaluate using the results of the experiment in the next section.
Previous research has shown that perceptions of competence vary for women based on context. When most people rate an “average” woman, they consider the paternalistic default: traditional homemaker (Haddock & Zanna, 1994). While this subgroup is perceived as less competent and warmer on average, there is variation based on status and competition— for example, “sexy women” are not seen as incompetent, but warm in one sample (Fiske et al., 2002). However, the effect of sexualized dress, we know, is moderated by occupational status (see also earlier discussion on how women’s perceived competence is lowered when respondents are instructed to focus on her appearance), something that may not need cuing for certain kinds of outfits—for example, Heflick et al., 2011). Glick and colleagues (2005, replicated by Wookey et al., 2009) find that sexualized dress only reduces perceived competence for women with higher status occupations (managerial or high skill), but not for women with lower status occupations. Based on the occupational status scale of Bukodi et al. (2011), we posit that the female target in our experiment will be perceived as a high status or high skill worker (Bukodi et al., 2011). For this reason, we expect that her perceived competence will be negatively impacted by sexualized dress. For the male target, we expect to see less variation across treatment conditions for perceived competence, though we note that attention from researchers on perceptions of male fashion is virtually non-existent (Molloy & Larner, 2010).
Hypothesis 1: A woman with a high status occupation who is wearing a sexualized outfit will be perceived as less competent than the same woman wearing a professionalized outfit.
While there is no research on perceived competence that specifically addresses gender, occupational status, partisanship, and dress, the general literature on gender attitudes by party and/or ideology provides some insight (Dolan, 2004). Experimental studies find that Democrats have a stronger preference for female political candidates (a high status group) than their Republican counterparts (Kirkland & Coppock, 2018). In addition, Democrats generally favor gender equality more than do Republicans, leading some to say that political “conservatism and feminism . . . [are] . . . mutually exclusive” (Campbell & Erzeel, 2018, p. 81). In addition, political scientists have found that left-wing parties have historically supported gender equality, while rightwing parties have been more regressive or conservative on such issues (Elder & Greene, 2012; Köttig et al., 2017; Sanbonmatsu, 2004; Wolbrecht, 2010).
Others analyze three types of ideological feminism: hostile sexism (anti-feminists and those who adamantly reject principles of sex equality), modern sexism (lack of support for sex equality demands in societal roles and in political arenas), and traditional sexism (viewing women’s “proper place” to be in the home as wives and mothers (Long et al., 2021). Republicans in general are more supportive of traditional gender roles and Democrats of modern sexism (Long et al., 2021, pp. 10, 22). Interestingly, however, Democratic support for women political candidates stems in part from the application of sex stereotypes to those women, such as having a greater capacity for addressing social issues, including poverty reduction and health care (Dolan, 2004, p. 207). Consequently, as Jennifer Lawless’ research and that of others establishes, people's gender and their partisanship affect their evaluations of political candidates (Lawless, 2004; Lefkofridi et al., 2019). We extend that finding to the supposition that so, too, will the evaluation of newscasters be affected by the gender and partisanship of viewers.
To the extent that a woman dressed in a sexualized manner may reflect less conservative (or more liberal) ideals about women, we expect heterogeneous effects based on party ID, with Republicans overall being the least supportive. More specifically, we anticipate an impact on perceived competence. For Republicans/Conservatives, facial femininity is associated with lower competence ratings; on the other hand, for Democrats/Liberals, it is connected to higher competence ratings (Carpinella & Johnson, 2013; see also Ditonto, 2017). Thus, given that full make-up and revealing dress enhance the feminine features of the female face and body (discussion in next section), we predict the following:
Hypothesis 2: Republicans more than Democrats perceive a woman with a high status occupation who is wearing a sexualized outfit as less competent than the same woman wearing a professionalized outfit.
We also investigate the relationship between objectification and content encoding (retention of political information), which is an important purpose of both watching the news and engaging in politics, something scholars highlight as necessary for a functioning democracy. There are two major elements of limited capacity information processing that scholars have identified: (1) viewers are information processors who make important decisions about perceiving, decoding, processing, and encoding visual messages and (2) humans are limited by how much information they can process. Thus, cognitive resources are distributed across various aspects of processing (Lang, 1995). When images and video incite strong emotions, particularly in the form of negative ones, individuals are more likely to divert cognitive resources to processing and storing that information—thus, encoding is greater in such contexts (Lang et al., 1996; Newhagen & Reeves, 1992; see also Hogan & Brashers, 2015, for general theory on information-seeking behavior and anxiety).
Similar effects occur for arousing content, since this increases the allocation of processing resources to the source of the stimulus (Lang et al., 1999). However, the relationship between this kind of content and encoding may depend on other factors, such as gender. For example, Grabe and Samson (2011) recruit a female news anchor to test whether dress has an effect on respondent memory (n = 386 university students). They find that women encode more verbal information when the news anchor is dressed in a sexualized outfit than when she is not, while the opposite is true for men. It is possible that women have more of a negative reaction to other women dressed in a sexualized style (Muggleton et al., 2019; Vaillancourt & Sharma, 2011), and so they are more likely to engage in information-seeking behavior.
Hypothesis 3: When the female target’s outfit is sexualized, more information is encoded by female respondents compared to male respondents.
Data and Method
We test our hypotheses using three randomized survey experiments conducted between 2017-2018 in the United States. Table 1 provides information on duration and sample size. The Digital Lab for the Social Sciences is a volunteer social science lab which is comparable to, if not an improvement on, other online convenience samples such as MTurk (Strange et al., 2019).
|DLABSS Woman||Aug 2017||465||Female Newscaster, Professional versus Sexualized Outfits
|DLABSS Man||Sept 2017||367||Male Newscaster, Professional versus Sexualized Outfits
|DLABSS Cycling||Feb 2018||653||Male and Female Newscasters in Cycling Outfits
Note. Conducted with sample information, dates, and other details.
The experiment involves no deception; participants were provided with basic information about the study, though we avoid the use of the word “competence” so that we do not prime respondents. In all versions, respondents are asked to watch a short one minute video featuring the newscaster (either male or female) who is wearing a particular outfit style (Professional, Sexualized, or Cycling). No background for any of the treatment conditions was given, but participants were aware that the study involved the topic of dress style. Following the short video, respondents are asked to rate the competence of the newscaster they just saw on a 0–10 scale (“Please use the sliding scale to indicate how competent you think the newscaster is”), and to answer five content based questions regarding the material presented by the newscaster in the video. Please see the Supplementary Materials for further details, including survey statistics, video transcript, questions, and perceived competence scale.
These six video clips were filmed at a university film studio in May 2017: three with the female actress and three with the male actor (see Figure 1 for images). A university media expert provided editing services so that the quality of the videos would mimic actual TV news (including banners, backdrop, etc.). The two actors were selected for this survey experiment based in part on their similar physical appearance in terms of their age, ethnicity, English language as their primary language, nationality, height, body structure, complexion, and acting experience. Social psychological scholarship emphasizes the way tasks, not just individuals, acquire gender specific attributes. Given that, as of the today, fifty percent of television newscasters are women, we contend that newscasting is not a gendered task, but rather is what Wegner and Berger refer to as a neutral task (Wagner & Berger, 1997).
Furthermore, the intonation and gestures of the two actors were directed by a trained media expert based in NYC to be as identical as possible across all six versions of the video. Thus, we endeavored to minimize differences between the actors’ performances in terms of what Wagner and Berger refer to as nonverbal cues, such as eye-contact and hand motions, as well as “non-fluencies” in speech, such as hesitancies in speaking and/or garbled words (Wagner & Berger, 1997).
While there is not a very specific or detailed formal literature on what might be considered comparable ‘sexy’ outfits for men and women, we used the work of several scholars to curate the looks and outfits we selected for each sartorial condition (please see Supplementary Materials p. 4 for more on this). We note that the red dress worn by our female target in the sexualized treatment condition may seem particularly exaggerated; however, increasingly, female news anchors themselves have complained about feeling forced to dress in a more sexualized style (Chambers et al., 2004), leading Katie Couric to note that their outfits seem more appropriate ‘for clubbing’. We believe the red dress on our female target may be particularly sexualized, but not out of the realm of possibility, given similar examples we have seen on networks like FOX (see Supplementary Materials for further details).
We also filmed a set of videos in which each actor wore the exact same outfit: a red cycling uniform. This is a baseline comparison to rule out potential confounders that are independent of dress. The results of this robustness check are presented in the next section. Our analysis focuses on within-gender comparisons, given variation in dress style. We are not claiming that the sexualized style we curated for the male newscaster can be directly compared to the sexualized style we selected for the female newscaster. Indeed, it is extremely difficult to curate ‘sexy’ looks for men and women that are comparable (across genders) without introducing another factor that could bias such a comparison, namely sexual orientation (Schofield & Schmidt, 2005).
We first present our results related to the perceived competence of the newscasters. In line with our first expectation, we do find a negative effect associated with sexualized dress for the female newscaster. Table 2 presents the results (numbers in the cells are group averages by treatment condition) for this first test: a comparison between the female newscaster in the professional outfit versus the sexualized outfit (right side of table), as well as a comparison between the male newscaster in the professional outfit versus the sexualized outfit (left side of table). Differences in group means between the outfit styles (professional versus sexualized), holding gender constant, can be interpreted as the average treatment effect (ATE) of the sexualized outfit style on perceived competence for both the female and male newscaster, respectively.
||ATE [95% CI]||p||Male
||ATE [95% CI]||p|
|Professional Dress||Sexualized Dress||Professional Dress||Sexualized Dress|
|Perceived Competence||6.397||5.595||0.802 [0.283, 1.322]||.0026||5.841||5.382||0.459 [-0.206, 1.124]||.175|
Note. N = 832 respondents. ATEs with significance p < .05 have been bolded.
Table 2 reveals two key findings. First, the negative effect of sexualized dress is significant for the female newscaster. The effect size is 0.802 on the 10-point competence rating scale; this difference is approximately one third of a standard deviation (denoting a small to moderate effect size; Cohen, 1992). Second, for the male newscaster, there does not appear to be an effect based on outfit style, which is also in line with what we expect, though there is no explicit literature of which we are aware that tests or theorizes how male clothing styles might impact (male) perceived competence.
Turning to our second hypothesis, we next consider whether there might be heterogeneous effects based on party ID. We theorized that Republicans may be less supportive of high status women wearing sexualized outfit styles, perceiving them as less competent compared to Democrats, due to their historical support for more regressive or traditional gender standards. Table 3 shows the conditional average treatment effect for Democrats and Republicans based on outfit style (sexualized or Professional), holding the gender of the newscaster constant.
||CATE [95% CI]||p||Male
||CATE [95% CI]||p|
|Professional Dress||Sexualized Dress||Professional Dress||Sexualized Dress|
|Democrats||7.41||5.8||1.61 [0.704, 2.516]||< .001||6.064||6.172||-0.108 [-1.188, 0.972]||.843|
|Republicans||5.441||4.902||0.539 [-0.553, 1.632]||.329||5.378||4.521||0.857 [-0.663, 2.377]||.263|
Note. N = 267 Democrats and N = 216 Republicans. CATEs with significance p < .05 have been bolded.
Contrary to our expectations, we find that Democrats are more likely to penalize the female newscaster when she is wearing the sexualized outfit compared to the professional style. The CATE in this case is 1.61 and this difference is approximately three quarters of a standard deviation (0.773), which is considered very close to a large effect size (of 0.80, see Cohen, 1992). A similar effect among Democrats is not present for the male newscaster, given the treatment condition in his case.
For Republicans, the sexualized style of clothing also does not seem to have a significant effect for either the male or female newscaster, though the conditional average treatment effect is negative in both cases. When we interact the treatment (outfit style) with respondent party ID, we find that the difference in effect sizes between Democrats and Republicans fails to reach conventional levels of statistical significance (p = .14), though it is the case that, more generally, Republicans rate the female newscaster as less competent than Democrats do (p < .001). A similar interaction in the case of the male newscaster yields something similar (p = .29 for the interaction between treatment and party ID), though, here, there is not a general effect of partisanship (p = .30 for Republicans). Thus, we consider the large negative effect size among Democrats for the female newscaster as suggestive in the next section.
Finally, we consider our last hypothesis: do female respondents encode content better than male respondents when the female newscaster is wearing a sexualized style? We have null findings for this. Table 4 below presents group means (by respondent gender) and conditional average treatment effects (CATE) by respondent gender and treatment condition, respectively.
||CATE [95% CI]||p||Male
||CATE [95% CI]||p|
|Professional Dress||Sexualized Dress||Professional Dress||Sexualized Dress|
|Men||2.00||2.045||0.045 [-0.371, 0.282]||.788||1.636||1.87||0.234 [-0.635, 0.168]||.252|
|Women||1.879||1.784||-0.096 [-0.296, 0.487]||.63||1.688||1.573||-0.115 [-0.295, 0.525]||.579|
Note. N = 392 male respondents and N = 354 female respondents. CATEs with significance p < .05 have been bolded.
The content encoding score is a simple tally of the total number of correct answers per respondent (see Supplementary Materials for a review of the content questions asked). Scores could range from 0 to 5 (average score is 1.807, SD = 1.306). Though we are most interested in heterogeneous effects given respondent gender, we checked to see if there are any overall effects, which we do not find (see Supplementary Materials p. 6). Table 4 shows that none of the conditional average treatment effects by respondent gender and dress style is significant.
Using a very similar experimental design (Grabe & Samson, 2011; also replicated in Samson, 2018), we note that at least two studies did find that female respondents encoded more information given a sexualized female. We contend that further research is needed to investigate how motivational mechanisms and selective visual attention that have evolved to provide preferential access to certain stimuli over others operate in the context of media and politics, where information is both visual and verbal. Teasing apart how attention is directed between the two sources (visual and verbal), and whether there is variation based on gender could be an important part of explaining longstanding puzzles in political science, like why women tend to have less political knowledge than men (see Dassonneville & McAllister, 2018).
We performed robustness checks to test our main results, as well as our treatment manipulation. To check whether the negative effect of the sexualized outfit for the female target generalized beyond the comparison to a professional outfit, we ran a third survey with videos of both targets dressed in a cycling uniform. We chose this outfit because, for the woman, it reveals a comparable amount of skin as the sexualized outfit does and is also tight-fitted. Respondents either (randomly) saw the male or female target in the cycling uniform (with the same news transcript and follow up questions as the other videos). In line with our main finding, we still find a significant negative effect associated with the sexualized outfit for the female newscaster even when compared to the cycling uniform (see Table A3, Supplementary Materials). The effect size is slightly smaller (0.627 compared to 0.802), but it is significant at p = .016. The cycling uniform also allows us to compare between the male and female target given that they both are wearing the exact same outfit. In the cycling treatment condition, there is no significant difference in perceived competence between the male and female target, though the female target does have a slightly higher average rating (see Figure A6, Supplementary Materials).
To test whether our treatment had its intended effect, we fielded two separate surveys on DLABSS targeting respondents who did not take any of the three main surveys. Our sample size for these two additional surveys totaled 899 respondents. Respondents were randomized into one of the two surveys, which asked respondents to rate on a 0 to 10 scale how sexually attractive they found the outfits in the treatment conditions. We used a head to toe still shot of either the female target in the sexualized outfit or the cycling outfit, keeping the gesture and facial expression the same across the two still shots. Analysis revealed a 1.954 average difference between the two outfits on the sexualization scale, with the sexualized outfit being rated as more sexually attractive than the cycling one, and this difference is significant at p < .001.
Discussion and Conclusion
In this study, we evaluate how respondents perceive the competence of and remember information presented by a male or female target wearing either a professional, sexualized, or cycling outfit in the context of a newsroom. In line with previous research, we find that all respondents are likely to penalize the female target when she is wearing a sexualized style compared to a professional one. We also find results by respondent party affiliation contrary to what we expected—Democrats, but not Republicans, dock the female target’s competence rating when she is wearing a more sexualized style of dress.
What might explain this discrepancy given our hypotheses? One explanation could be related to the fact that Democrats may especially value professional sex equality. Thus, they may prefer that professional women dress in a similar way or equal way compared to professional men. Hence, in a professional setting, Democrats may prefer women to be dressed in professional attire like a subdued suit, rather than in a sexualized style. Research by Kaiser and colleagues (2006) also points to prejudice expectations, which moderate attention to cues that threaten social identity. Here, a possible explanation would relate to Democrats’ higher sensitivity to sexism (particularly following the 2016 U.S. presidential election, see Zucker et al., 2019), leading them to focus more on cues likes dress style, especially sexualized ones.
Another possibility relates to Republican preferences for sex typicality. Researchers have found that female Republican candidates who appear relatively more feminine were more likely to win their election (Carpinella et al., 2016). While this study only looked at facial cues, to the extent that fashion, particularly sexualized styles, highlights the feminine features of the face and body (make-up, hair, skin, and shape; see Supplementary Materials p. 4 for more), then it is possible that Republicans were less negatively affected by a less conservative dress style, so long as it was reinforcing the typical characteristics associated with each gender. The same researchers (Carpinella et al., 2016) also found that relatively masculine looking Democrats, regardless of sex, were more likely to be selected in a hypothetical vote choice task. This could provide some explanation for why Democrats penalized the female newscaster when she was wearing the sexualized outfit compared to the professional one, since the suit style may have attenuated, instead of highlighted, her femininity. More research is needed to examine how other factors, like signals regarding sexual orientation, may have played a role for Democrats who saw the male newscaster in the sexualized outfit.
We expected, but did not find, that female respondents would be more likely to pay attention to a sexualized female target. We note that at least two studies did find that female respondents encoded more information, given a sexualized female target and in a very similar experimental design (Grabe & Samson, 2011; also replicated in Samson, 2018), so we believe that further research is needed. Teasing apart how attention is directed between visual and verbal sources, and whether there is variation based on gender could be an important part of explaining longstanding puzzles in political science, like why women tend to have less political knowledge than men (see Dassonneville & McAllister, 2018).
The result that, overall, dress styles seem to matter more for women is corroborated by the anecdotal experiences of female professionals, particularly in the world of politics (for a historical overview, see North, 2018).4 In news, networks are investing increasingly more time, money, and energy into the appearance of women news anchors (Nitz et al., 2007), leading many—including the very women whose appearance is in question—to note a trend towards increasingly sexualized dress, hair, and make-up on air (Barnes, 2005; Chambers et al., 2004; Perryman & Theiss, 2014). As dress styles for all women become increasingly more sexualized (for review, see Fasoli et al., 2018), the potential for costly judgments that might negatively impact women in professional fields also increases. Our paper provides part of a solution. By increasing awareness of these negative effects, it is possible to correct biases that differentially impact women more than men.
This work is limited in its external validity due to its experimental focus. A logical next step in this line of research is to collect observational data on female politicians and candidates to see if partisan differences exist in practice. According to our findings, it may be the case that female politicians on the left face more of a balancing act when it comes to what they wear. We would expect them generally to wear more conservative, less feminine clothing than their Republican counterparts. Anecdotal news coverage already appears to have picked up on this difference between Republican and Democratic female politicians (Lubitz, 2016). Another interesting area for future scholarship lies at the intersection of clothing, sexual orientation, transgendered identities, and partisanship. How might dress cues that connote sexual orientation impact perceived competence? And, is this effect moderated by partisanship and/or gender? As marginalized groups enter politics in higher numbers, understanding how clothes communicate information and to whom constitutes a crucial component of their success. In addition, it would be productive and interesting to expand this study to include cross-national comparisons. Is the sexualization of the dress styles of American women who are TV newscasters, for example, the norm for their counterparts in other regions like Europe, the Middle East, Asia, South America, or Africa? Based on anecdotal observations, we expect tremendous cross-national variations, thereby raising the question: why?
Finally, the findings of this study open the door to further analysis, if not the use, of Title VII for generating workplace equality between men and women. This is because to date, courts have ruled that dress codes are simply trait requirements that do not violate Title VII. Thus, so far, it has been permissible for employers to require sex-segregated clothing styles for male and female employees, including newscasters. Significantly, however, to date, we are not aware of any research or court cases that have analyzed or focused on the consequences of sexualized dress codes for women and professionalized ones for men in terms of how such sex-segregated clothing styles affect the perception of competence in the workplace. Our research, however, does analyze those consequences and affirms that sexualized dress styles for women do reduce the perception of women’s competence. We contend that this reduction in the perception of women’s competence disproportionately disadvantages members of a protected class: women. Significantly, courts have ruled that doing so constitutes an adverse effect, and, notably, courts have ruled that Title VII prohibits policies that impose adverse effects. Thus, by integrating legal standards with social psychological scholarship, this study presents a new foundation for the claim many legal scholars have sought to make, namely, why at least some trait discrimination policies, such as sex-stereotypical dress styles for women, violate Title VII.