Women’s revealing Halloween costumes: other-objectification and sexualization

Abstract

Women are depicted in revealing dress in the media and the depictions have costs such as objectification. Objectification theory explains that women in Westernized cultures are looked at, evaluated, and potentially objectified by others. Accordingly, objectifying gaze (by others) evokes self-objectification which has effects such as habitual body and appearance monitoring. According to the theory being objectified by others precedes self-objectification, which suggests that objectification by others could be more prevalent than self-objectification and potentially just as harmful. Researchers have found that self-objectification and other-objectification can be induced by revealing dress manipulations that vary in tightness or body coverage. We studied Halloween costumes as a site for objectification of others. In Study 1, 124 pairs of men’s and women’s Halloween costumes were content analyzed. Women’s costumes were significantly more revealing than men’s in tightness and body coverage. Since sexual objectification in the media is assessed by the presence of revealing dress in media depictions, we reasoned that women’s revealing Halloween costumes could be sexually objectifying. In Study 2, 295 participants rated women wearing revealing or non-revealing costumes in an online experiment. Women wearing revealing costumes were sexually objectified by participants. Although men rated costumed women higher on the sexually objectifying traits than women, both men and women objectified the costumed women in the revealing dress condition. Dress researchers may wish to apply objectification theory to re-interpret and explain early research on revealing dress.

Introduction

Women and girls are depicted in revealing dress in the media and the depictions and their consequences have been studied by researchers from a variety of disciplines (Aubrey, Hopper, & Mbure, 2011; Goodin, Van Denburg, Murnen, & Smolak, 2011; Graff, Murnen, & Krause, 2013). According to the report from the APA Task Force on the Sexualization of Girls (APA 2007) one such consequence is sexualization. The report holds that sexualization occurs (a) if someone’s value is based on her sexual appeal, to the exclusion of other aspects; (b) if someone is considered sexy only if she achieves a narrowly defined rigid standard of physical attractiveness; (c) if someone is sexually objectified (by others), a thing for the sexual use of others rather than being an independent decision-maker; or (d) if sexuality is forced on a person (by another person). In the report evidence is provided of cultural contributions to sexualization, which includes research that has focused on a variety of media including movies, magazines, consumer products, cosmetics, and clothing.

In fact, research demonstrates that in US culture it is common for women to be sexually objectified by others or valued as a function of their sexual appeal and this is evident in various types of media (Aubrey, 2006; Graff et al., 2013; Vandenbosch, Vervloessen, & Eggermont, 2013). Objectification is harmful to the objectifying person (Aubrey et al., 2011; Johnson, McCreary, & Mills, 2007; Zurbriggen, Ramsey, & Jaworski, 2011), as well as to the individual being objectified (Frederickson, Roberts, Noll, Quinn, & Twenge, 1998; Hebl, King, & Lin, 2004). Many researchers have studied objectification of others as a function of those others’ revealing dress (Graff, Murnen, & Smolak, 2012; Gurung & Chrouser, 2007; Loughnan et al., 2010; Loughnan, Pina, Vasquez, & Puvia, 2013; Nezlek, Krohn, Wilson, & Marusken, 2015). These researchers found that perceivers objectify others who are wearing revealing clothing. Hence, revealing clothing is implicated in objectification of others.

There are many occasions in contemporary life where women and sometimes men wear revealing clothing in public venues, such as at swimming pools or beaches, at health clubs and gyms, on New Year’s Eve, and on Halloween. Although published academic research has not investigated the revealing nature of women’s Halloween costumes, the topic has received notice in the popular press (LaBarre, n.d.; Mayer, 2014; Rosenbloom, 2006; Sharry, 2012). Mayer reports that the only choice women have when purchasing commercial Halloween costumes is sexy, which is code for short, tight, and skimpy and hence revealing. In the current research we studied Halloween costumes as a context for sexualization. The purpose of this research was (a) to investigate the extent to which Halloween costumes are gendered in terms of their revealing nature and (b) to investigate the extent to which women wearing revealing Halloween costumes are sexually objectified by others, both men and women.

Literature review

Objectification theory

Objectification theory (Fredrickson & Roberts, 1997) offers a framework for understanding the relationship between revealing dress and sexual objectification or SO. The theory was developed to outline the consequences of being female in cultures that sexually objectify women and girls. The authors explain that women in sexually saturated cultures are gazed at, evaluated, and potentially objectified by others. According to the theory, women are objectified by others (i.e., experience objectifying gaze) in three ways: in social interactions (e.g., visual inspection, experiencing catcalls), in media depictions of social interactions, and in media depictions of bodies and body parts. Furthermore, being objectified by another person (other-objectification) is thought to lead to an objectified state of consciousness (i.e., self-objectification) which influences self-perceptions and perceptions of others. In other words, objectification by others (other-objectification) is thought to lead to self-objectification.

According to the theory, women and girls are socialized to internalize an outsider’s perspective on their bodies (self-objectify) partly because they are targets of others’ objectification (i.e., the objectifying gaze). Self-objectification has consequences such as continual body and appearance monitoring and requires cognitive attention that can interfere with task performance. According to the theory being objectified by others is thought to precede self-objectification; hence objectification by others (i.e., other-objectification) may be even more widespread and perhaps more insidious, than self-objectification (Zurbriggen et al., 2011). Furthermore, Strelan and Hargreaves (2005) found that women were less likely to objectify themselves (self-objectification), than to objectify other women (other-objectification). This could suggest that other-objectification is more widespread than self-objectification. Hence, other-objectification is important to study.

Research provides support for objectification theory and much of the early research focused on self-objectification and its negative effects (Budesheim, 2011). However, people observe and objectify others (i.e., other-objectification) and this is associated with negative effects for the objectifier. For example, Johnson et al. (2007) studied other-objectification in the context of advertising. They exposed university men to ads of women or men in revealing dress. In the revealing female condition, slender women were shown in swimsuits or lingerie. In the revealing male condition, muscular men were shown with their chests and torsos exposed. Ads for the control condition depicted cars and consumer products. Participants who viewed the revealing images of women reported greater anxiety and greater hostility than those who viewed the revealing images of men or the control images. Aubrey et al. (2011) studied the effects of sexual objectification of women in music videos on undergraduate men. Videos were considered sexually objectifying in part if the artist had a high degree of body exposure (i.e., skin exposure). As compared to participants who saw videos of female artists low in SO, those who saw videos of female artists high in SO reported more adversarial sexual beliefs and were more accepting of interpersonal violence.

Zurbriggen et al. (2011) surveyed undergraduate men and women to study whether consumption of mass media (i.e., TV, films, magazines) was associated with other-objectification in the context of romantic relationships. They found that consumption of mass media was positively related to partner objectification (other-objectification), which was negatively related to relationship satisfaction. Thus, when men or women objectified their partner (i.e., other-objectification) they experienced less relationship satisfaction and, for men only, less sexual satisfaction. Strelan and Hargreaves (2005) found that self-objectification among undergraduate women and men was positively related to other-objectification, objectification of other women and men. Lindner, Tantleff-Dunn, and Jentsch (2012) surveyed female undergraduates and like Strelan and Hargreaves found evidence of a relationship between other-objectification and self-objectification. To summarize, negative outcomes are associated with the objectification of others. Men who objectify others experience more anxiety, hostility, adversarial sexual beliefs, acceptance of interpersonal violence, and less sexual satisfaction. Both women and men who objectify others experience self-objectification and less relationship satisfaction. Thus, it is important to study other-objectification.

Dress versus costume

In experimental research on other-objectification (Graff et al., 2012; Gurung & Chrouser, 2007; Loughnan et al., 2010, 2013; Nezlek et al., 2015) and on self-objectification (Fredrickson et al., 1998; Hebl et al., 2004) scholars have employed dress to manipulate objectification. But how exactly is dress defined beyond the commonly accepted usage?

Scholars studying dress have defined it as all body supplements and all modifications to the body (Roach-Higgins & Eicher, 1992). A special type of dress that is not worn for everyday activities but is reserved for theater, festivals, rituals, and ceremonies is called costume. Examples of body supplements include shoes, clothing, jewelry, hearing aids, braces, and glasses. Body modifications include tattoos, cosmetics, tanning, piercing, use of perfumes, and weight loss. Two items of dress (clothing and cosmetics) were identified in the APA report as products that contribute to the sexualization of girls and women. Since costume is a special type of dress, costume could also contribute to the sexualization of girls and women.

Revealing dress

Researchers have investigated objectification of others as a function of those others’ revealing dress (Graff et al., 2012; Gurung & Chrouser, 2007; Loughnan et al., 2010, 2013; Nezlek et al., 2015). Yet, what makes dress revealing is not always clear in the literature. Dress that reveals the body and the consequences of wearing it have been studied, although researchers have not been consistent in how dress manipulations were described or operationally defined. For example, Edmonds and Cahoon (1986) developed non-sexy (slacks and a blouse) and sexy (dress that revealed breasts and legs) conditions. While it appears that Edmonds and Cahoon were varying the extent to which the dress revealed the body, they do not explain why their manipulations are sexy or non-sexy. On the other hand, Lewis and Johnson (1989) describe their dress manipulation as varying in provocativeness based on a pretest assessing the amount of body exposure of each outfit (i.e., the extent to which the dress revealed the body). In that study it is clear what variable is manipulated. We adopt the term revealing dress because it reflects what is manipulated (dress that reveals the body) and also is often used in the objectification literature (Goodin et al., 2011; Graff et al., 2013; Prichard & Tiggemann, 2005; Tiggemann & Andrew, 2012).

Researchers studying women’s revealing dress found that it was associated with judgments of responsibility for sexual harassment (Johnson & Workman, 1992; 1994) and sexual assault (Edmonds & Cahoon, 1986; Lewis & Johnson, 1989). Results of these studies showed that when women wore revealing dress (skimpy, see-through, or short) they were often assigned responsibility for their own sexual harassment and sexual assaults.

Revealing dress and sexualization in the media

More recently, as previously demonstrated researchers studying sexual objectification (SO) in media depictions (e.g., magazines, MTV videos, television) of women and girls have measured SO in part by the presence of revealing dress in those depictions (Aubrey et al., 2011; Johnson et al., 2007; Zurbriggen et al., 2011). In applying content analysis to various media (music videos, clothing on retail websites, girls’ magazines, video game characters), other researchers have operationally defined SO in the media in part as the extent to which women’s bodies are exposed to the gaze of others (Goodin et al., 2011; Graff et al., 2013; Vandenbosch et al., 2013). Since women’s (and girls’) revealing clothing exposes their bodies to the gaze of others, revealing clothing is implicated in SO and sexualization. Hence, an examination of women’s revealing dress in their depictions is a way to identify SO or sexualization.

Goodin et al. (2011) studied girls’ clothing at the retail websites of 15 retailers. Clothing was coded as sexualizing if it revealed or emphasized a sexualized body part (e.g., chest, waist, buttocks, and legs), had sexy characteristics (e.g., slinky fabric), or had sexualized writing on it. Graff et al. (2013) studied depictions of girls in girls’ magazines and used similar categories as Goodin et al.; they coded characteristics of girls’ appearance as sexualized if it revealed, emphasized, or enhanced sexualized body parts. In a content analysis of Belgian music videos, Vandenbosch et al. (2013) coded sexualization as manifest in part by whether the character wore clothing that focused on sexual body parts or had sexually suggestive slogans. In a study of video game characters, Miller and Summers (2007) also used revealing dress in their coding system; they relied on skin exposure to assess the revealing nature of clothing. Thus, there is empirical support for using revealing dress to assess SO in media depictions.

How is SO of others manifest?

Based on the definition of the APA Task Force (2007), sexual objectification (condition c of the report) is subsumed by sexualization or put another way, sexual objectification is a special case of sexualization. This also agrees with how Frederickson and Roberts explained sexualization and sexual objectification (1997, p. 175). However, other scholars who have investigated SO have assessed it differently.

Researchers do not agree with respect to how other-objectification manifests. Loughnan et al. (2013) have argued that people who are objectified by others will be denied moral standing, mind, and humanity and also that they will be judged less competent than someone who is not objectified by others. In their revealing condition the stimulus person wore a bikini; in the non-revealing condition she wore jeans with a white top. The stimulus person was attributed less mind, less moral concern, and less competence in the revealing condition than in the non-revealing condition; these differences were taken as evidence of objectification (of others).

Heflick and Goldenberg (2009) did not use a revealing dress condition. They argued that objectification (induced by focusing on a woman’s appearance) leads to lower perceptions of humanness and competence than focusing on a woman as a person. Since the appearance focus led to lower perceived competence and human essence scores, this was interpreted to demonstrate objectification (of others). Heflick, Goldenberg, Cooper, and Puvia (2011) argued that objectification would reduce perceptions of warmth, morality, and competence, thus appearance focus manipulations that led to such reductions were taken to demonstrate objectification (of others). Budesheim (2011) challenged such definitions and called for more clarity in measuring and manipulating objectification. Budesheim suggested that objectification (of others) is a process of “focusing only on women’s appearance and thinking of them solely in terms of their sexual appeal” (p. 169), which is consistent with what the report from the APA Task Force (APA 2007) calls sexualization (condition a).

Graff et al. (2013) argue that if sexualizing lowers a person’s status, that person should be seen as less desirable, specifically in terms of morality and self-respect. Gurung and Chrouser (2007) claim that their stimulus person in revealing clothing was objectified (by others) because she was rated higher on the traits that suggested she was viewed as a sexual object (attractive, sexually experienced, and desirable) and lower on capability traits (strength, determined, capable). This view is consistent with Nezlek et al. (2015), who claim that objectification (of others) occurs when someone’s capabilities are rated lower and the person is perceived mostly as a sexual object. Thus, there is disagreement in definitions of SO of others and in how it is manifest experimentally. For our research, we followed Budesheim (2011) and defined SO of others as viewing people primarily in terms of their sexual appeal (i.e., as sex objects) and operationally defined it as higher ratings of sexualizing traits and lower ratings of positive traits.

Halloween

In daily life in western cultures, there are many contexts in which people wear revealing dress. As a function of that revealing dress they may be susceptible to other-objectification. One such context in which women and men may wear revealing dress in the US is Halloween. However, scholars have not studied the extent to which men’s and women’s Halloween costumes are revealing.

Halloween is an important US holiday for retailers. In 1990 Belk reported that adult Halloween costumes were a growing part of business for costume shops. In 2015 $2.53 billion was spent in the US for Halloween costumes (Annual Halloween expenditure, 2015). Another source (Halloween Statistics, 2015) put the figure higher and reported that spending for children’s costumes in the US reached $1,240,000,000 and spending for adult costumes reached $1,550,000,000 in 2015. The annual NRF Consumer Spending survey conducted prior to Halloween had predicted spending of $1.2 billion on adult costumes, $950 million on kid’s costumes and $350 million on pet costumes (Reynolds, 2015).

Two main themes emerge from extant literature focusing on Halloween costumes and Halloween. Researchers have studied Halloween as an event on college campuses (Belk, 1990; Miller, Jasper, & Hill, 1991; Mueller, Dirks, & Picca, 2007) and gender stereotyping of children’s Halloween costumes (Belk, 1990; Nelson, 2000). Miller et al. surveyed over 1200 mostly college students about Halloween and dressing in costume. About 71 % wore costumes. A major part of the appeal of wearing costumes was the ability to adopt a different role. Similarly, Belk noted that people wearing costumes may suspend other identities in favor of the one indicated by the costume. Mueller et al.’s (2007) undergraduate respondents saw Halloween as a holiday that provided freedom from social norms. Together these results suggest that dressing in costume on Halloween may provide license to violate social norms.

To study gender stereotyping in Halloween costumes Nelson (2000) content analyzed children’s costumes and their descriptions. She found gender stereotyping such that costumes for boys tended to be named by occupational roles (e.g., Policeman), while costumes for girls were named by referring to appearance and/or relationships (e.g., Beautiful Bride). Nelson also found a stereotypic active-masculine/passive-feminine dichotomy (e.g., Superman/Beauty Queen). Although she did not analyze them, Nelson noted a trend for women’s and girls’ costumes to be hypersexualized, which is consistent with popular press reports (Mayer, 2014; Rosenbloom, 2006; Sharry, 2012).

Alexander (2014) also studied Halloween costumes and performed a content analysis of men’s costumes. Although she did not contrast men’s and women’s costumes, she did code men’s costumes according to how sexualized they were. Costumes were classified as sexualized if they were skimpy and revealed muscularity; only 4 % of the costumes were so classified. While specific information was not provided, Alexander may have based her coding on the extent to which the costumes revealed the body. The existing research demonstrates a gap in the literature on Halloween costumes, namely, the lack of work focusing on gender-differentiated revealing nature of adult costumes. If women’s Halloween costumes are more revealing than men’s and since the presence of revealing dress is a way to assess SO in the media, it follows that women’s Halloween costumes are potentially sexually objectifying.

Research questions

As previously argued, the existing research on Halloween demonstrates a gap such that there is no research that has focused on the gender-differentiated revealing nature of adult costumes. To address this gap, the first research question was developed. RQ1: Do men’s and women’s Halloween costumes differ in terms of revealing the body? As the literature review demonstrates, the presence of revealing dress is a way to assess SO in media. If women’s Halloween costumes are found to be revealing as the popular press suggests, then it is possible that women wearing revealing Halloween costumes will be sexually objectified by others. Therefore, the following research question was developed. RQ2: Will women wearing revealing Halloween costumes be sexually objectified by others?

As proposed, Objectification Theory explained women’s experiences and mental health risks as a function of SO and therefore most of the research has focused on women. However, media researchers have demonstrated SO of men in Western media (Elliott & Elliott, 2005; Leit, Gray, & Pope, 2002). In studies of other-objectification, results have shown that men objectify others, both men and women, whereas women objectify other women but not men (Bernard, Loughnan, Marchal, Godart, and Klein, 2015; Loughnan et al., 2013; Nezlek et al., 2015). Given this focus on men and objectification, the following research question was developed. RQ3: Will men and women differ in the extent to which they sexually objectify other women wearing Halloween costumes?

To investigate these research questions we conducted two studies. In Study 1 we determined whether or not women’s Halloween costumes were more revealing than men’s. In so doing, we also isolated stimuli that were then used in Study 2. In Study 2 we investigated the extent to which women wearing revealing Halloween costumes are objectified by others, a type of sexualization. Furthermore, in Study 2 we assessed whether or not both men and women objectify other women wearing revealing Halloween costumes.

Study 1

Method

Images of Halloween costumes for men and women (e.g., male magician, female magician) were sourced from 15 websites (e.g., partycity.com, halloweencostumes.com, costumeexpress.com) which resulted from a Google search using search term “Halloween costume.” Images were freely available on multiple Halloween retail websites. Children’s costume websites such as Disney store, Carters, and Chasing-fireflies were excluded. Sourcing for new images stopped when no more matched costume pairs were found, resulting in 124 different pairs of costumes that were used in the content analysis. Because we were investigating gender differences in the costumes, we selected costumes for which there were male and female versions (male–female pairs). We wanted to limit differences between the men’s and women’s costumes to the fact that one was manufactured for men and the other was for women within each costume pair. Because Halloween costumes are sold by “occupation” we were able to find pairs of costumes that were comparable that way (e.g., male pirate, female pirate; male clown, female clown).

Based on Prichard and Tiggemann’s (2005) comment that tight and revealing clothing place women in the “objectification limelight” (i.e., facilitate objectification), we rated the costumes on tightness. We also reasoned that clothing can reveal by exposing the body due to its sheerness or because it uncovers the body (i.e., shows skin). Other researchers have also measured sexualizing characteristics of clothing according to the extent to which it uncovers the body and is tight (e.g., Graff et al. 2013). Accordingly, costumes were rated on the extent to which they were tight, sheer, or uncovered the body; each of these terms was used to rate one aspect of the revealing nature of the costumes. By using all three terms in the ratings we were able to get a more fine-grained assessment of the revealing nature of the costumes. To facilitate rating, each costume image was divided into four parts: upper upper torso (neckline to bust/chest), lower upper torso (bust/chest to waist), lower torso (waist to hips) and legs (hips to ankles). Each of these four parts was rated so that twelve ratings were made for each costume in a costume pair. Hence, there were three ratings for each of four parts of the body per costume. Likert scales (1 = strongly disagree, 5 = strongly agree) were used in ratings. Thus, coders indicated their agreement that the costume was sheer, tight, or uncovered the body in each of the four body areas rated. See Fig. 1.

Fig. 1

Four parts of the body coded in Study 1

The three authors first rated 20 pairs of costumes. Discrepancies between raters were discussed and the 20 pairs of costumes were re-rated. This training process continued until ratings agreed, at which point two authors who were doctoral students at the time rated the remaining 104 costume pairs. Cohen’s Kappa was calculated for all twelve sets of ratings. Cohen’s Kappa is used when two raters code nominal or ordinal data and is a measure of the strength of agreement over and above chance (Sim & Wright, 2005). Agreement over and above chance was substantial or greater for the ratings of (a) the extent to which the costume uncovered the body (revealed skin) and (b) tightness; actual scores varied from 64.6 % to 91.9 % (see Table 1 for all Kappas, their interpretation, and significance). Agreement over and above chance varied regarding how sheer the costumes were from slight (19.6 %) to almost perfect agreement (82.4 %). Due to poor agreement across the body areas, the sheerness of men’s and women’s costumes was not considered in the following analysis.

Table 1 Cohen’s Kappa for twelve sets of ratings of men’s and women’s Halloween costumes on eight coded body areas

Results and discussion

MANOVA was used to test for differences between men’s and women’s costumes on the eight coded body areas. The overall multivariate effect for gender was significant, approximate multivariate F (8, 239) = 40.55, p < .0005. Univariate ANOVAS showed that for all of the coded body areas, mean ratings of tightness in women’s costumes were greater than in men’s costumes (ps < .0005) (see Table 2 for main effects, means, and standard deviations). For upper upper torso and legs, univariate ANOVAS found significant gender differences in the extent to which the costume uncovered the body. No gender differences were found in the extent to which the lower upper torso and lower torso were uncovered. Men’s costumes were never rated to uncover more of the body than women’s costumes. Thus, our analysis demonstrates that men’s and women’s Halloween costumes do differ in terms of the extent to which they reveal the body for six of the eight body areas analyzed. Furthermore, averaging the ratings of six areas of the body for which men’s and women’s costumes differed in tightness and the extent to which the costume uncovered the body, the average rating for women was 3.8. Based on a five-point scale, this average demonstrates that raters tended to agree that women’s Halloween costumes were revealing. The comparable average for men’s costumes was 1.98. This means that raters tended to agree that that men’s costumes were not revealing.

Table 2 Means, standard deviations, and F-values for women’s and men’s costumes

The purpose of Study 1 was to investigate the extent to which Halloween costumes were gendered in terms of their revealing nature. Accordingly, a content analysis of men’s and women’s Halloween costumes (124 pairs) was performed to compare the revealing nature of the costumes by gender. To do so, costumes were rated on sheerness, tightness, and the extent to which they uncovered the body. The reliability of ratings of sheerness of the costumes was poor so those ratings were not analyzed further. Sheerness may be a quality that is indeterminate when working from an image and actual costumes may need to be analyzed to achieve adequate reliability.

Results of the content analysis found that the costumes were gendered in terms of two aspects of their revealing nature (tightness and the extent to which they uncover the body). Like Nelson’s (2000) research that found gender differences in children’s Halloween costumes, our results found gender differences in adult Halloween costumes. So even though our study and Nelson’s focused on different aspects of gender differences, the results are consistent in finding gender differences in the costumes. Furthermore, in our study raters tended to agree that women’s costumes were revealing and that men’s costumes were non-revealing. This finding is consistent with Alexander’s (2014) research which also found that men’s Halloween costumes tended not to be revealing.

For both men’s and women’s costumes, raters agreed that the middle portion of the costume covered the body (lower torso, lower upper torso), such that there were no gender differences for those areas. This may be related to current fashion trends, which currently do not feature uncovered mid-torsos. Assuming that Halloween costume designs are influenced by fashion, we might have found gender differences had we conducted this research 15–20 years ago, when low-slung jeans, exposed midriffs, and thong underwear were popular for women (Kuczynski, 2004).

Study 2

Method

In Study 1 we found evidence that Halloween costumes are gendered in terms of their revealing nature, such that women’s costumes are more revealing than men’s costumes. In Study 2, we extend the research on revealing dress in two ways. First, we examine whether a revealing dress manipulation in the context of women’s Halloween costumes causes observers to sexually objectify (i.e., judge the costumed women in terms of their sexual appeal). Second, we examine sex differences in judging the costumed women in terms of their sexual appeal.

Participants were recruited from Amazon Mechanical Turk (AMT), an online crowdsourcing marketplace from Amazon, Inc. Researchers have studied the characteristics of AMT workers and their data. Paolacci, Chandler, and Ipeirotis (2010) found that about 65 % of U.S. AMT workers are women, their average age is 36, and they tend to be better educated than the general population. Buhrmester, Kwang, and Gosling (2011) found that AMT workers are more diverse than Internet samples or US undergraduate samples, and that AMT data are as reliable as data collected via traditional methods. Berinsky, Huber, and Lenz (2012) found that AMT workers are often more representative of the general population than other convenience samples.

Experimental stimuli were selected based on results of Study 1. Tightness ratings and body coverage ratings from all four parts of the body (from Study 1) were averaged together for each woman’s costume. The three women’s costumes that were rated least tight and least uncovered were used in the non-revealing condition (${\bar{\text{x}}}$ = 1.33) and the three women’s costumes that were rated most tight and most uncovered were used in the revealing condition (${\bar{\text{x}}}$ = 3.56). Multiple images were used in each condition for stimulus sampling purposes (Wells & Windschitl, 1999). Each participant was exposed to all three stimuli in one condition.

Qualtrics survey software was used to develop and host an online experiment. The university Institutional Review Board approved the research for use with human participants. Participants were provided a web link through AMT which took them to a page explaining the research procedure. Based on a branching question participants were linked to three models wearing revealing costumes or to three models wearing non-revealing costumes. All models were judged by the coding team to represent culturally ‘attractive and thin’ females.

With each costumed image participants were provided a context: “This woman is on her way to a Halloween party. Please rate her personal characteristics using the following system. She has/is…” These instructions were followed by the dependent variable adjective list. After exposure to each costumed image, all participants completed the same set of dependent variables and manipulation check items. Finally, demographic information was collected and participants were thanked. Upon completion of the experimental tasks, a code was provided which could be used to receive an incentive ($.50) paid by AMT. Participation was restricted to US citizens of age 18 or older.

Like others (Graff et al. 2013; Gurung & Chrouser, 2007; Heflick & Goldenberg, 2009) who assessed other-objectification, the dependent variables were the composite variables sexualizing traits and competency, as well as a set of positive traits (considerate, faithful, feminine, moral, self-respecting, sincere). Seven-point Likert scales were used for all ratings unless otherwise noted. Sexualizing traits included 8 items (attractive, desirable, flirtatious, promiscuous, seductive, sexually appealing, sexually experienced, sexy) drawn from several studies (e.g., Gurung & Chrouser, 2007) that confirm that the stimulus person is perceived as a sexual object. Competency consisted of three items (capable, competent, intelligent); Heflick and Goldenberg (2009) reported α = .88. Considerate, faithful, feminine, moral, self-respecting, sincere were all single item dependent variables.

We reasoned that if someone is valued primarily for her sexual appeal, it is reasonable to expect that person to be rated higher on sexualizing traits and lower on other positive traits (i.e., considerate, faithful, feminine, moral, self-respecting, sincere). We operationally defined SO as higher ratings on the sexualizing traits variable and lower ratings on positive traits. These items were followed by a manipulation check, consisting of four items. Participants were instructed to consider all the costumes they had seen and rate them according to the extent to which they (1) were NOT revealing, (2) accentuated the women’s bodies, (3) showed a lot of skin, and (4) were tight. Finally, participants responded to demographic items: ethnicity (closed-ended), education (closed-ended), age (open-ended), sex (closed-ended), citizenship (closed-ended), and college student status (closed-ended).

Results and discussion

There were 319 participants who finished the experimental tasks. The survey used filtering items to determine if participants were actually reading the items (e.g., there are 8 days in a week). Three respondents were eliminated based on the filtering items, two identified their location as outside the US and were deleted, and 19 were also eliminated due to large blocks of missing data. After cleaning the data, there were 295 participants with usable data. The sample consisted of more men than women (F = 133, M = 162), who were primarily Caucasian American (84.8 %), in their middle 30 s (${\bar{\text{x}}}$ = 34.59), and had not completed a college degree (91.9 %). Few participants (10.7 %) self-identified as college students.

The composite dependent variables had excellent reliability: α = .918 (competency) and α = .944 (sexualizing traits). Items within each composite variable were averaged and used in the main analyses. The dependent variables considerate, faithful, feminine, moral, self-respecting, and sincere were entered as single item trait measures.

To check the manipulation, the four manipulation check items were averaged after reverse scoring as needed and entered into ANOVA as the dependent variable with Dress as the independent variable. Results yielded F(1, 324) = 1306.591, p < .0005. Inspection of the cell means showed that those in the revealing dress condition rated the costumes significantly more revealing (more revealing, more tight, more uncovered, and as showing more skin) than those in the non-revealing condition (${\bar{\text{x}}}_{\text{rev}}$ = 6.09 vs. ${\bar{\text{x}}}_{\text{non}}$ = 2.98). Thus, the manipulation was successful.

The experiment used a 2 (Dress: revealing, non-revealing) by 2 (Sex of participant: men, women) between subjects factorial design. Data were analyzed with MANCOVA. Dress and sex were the independent variables, age was a covariate, and the dependent variables were: competency, considerate, faithful, feminine, morality, self-respecting, sexualizing traits, and sincere. The overall multivariate main effects for dress [F(8, 283) = 55.42, p < .0005] and sex [F(8, 283) = 3.61, p < .001)] were significant. The multivariate effect for the covariate (age) and the multivariate Dress by Sex interaction were not significant.

ANCOVAs also showed that Dress had effects on the extent to which the stimulus persons were rated as considerate, [F(1, 290) = 28.87, p < .0005); η ² = .085], faithful, [F(1, 290) = 28.91, p < .0005); η ² = .091], feminine, [F(1, 290) = 11.63, p < .001); η ² = .039], moral, [F(1, 290) = 34.53, p < .0005); η ² = .106], self-respecting, [F(1, 290) = 11.51, p < .001); η ² = .038], sexualized [F(1, 290) = 283.09, p < .0005); η ² = .494], and sincere, [F(1, 290) = 10.61, p < .001); η ² = .035]. The Dress effect for competency was not significant.

Inspection of the cell means showed that when wearing revealing dress (${\bar{\text{x}}}$ = 4.61), the costumed women were rated as less considerate than when wearing non-revealing dress (${\bar{\text{x}}}$ = 5.20). Also when wearing revealing dress (${\bar{\text{x}}}$ = 4.26), women were rated less faithful than when wearing non-revealing dress (${\bar{\text{x}}}$ = 4.84). When wearing revealing dress (${\bar{\text{x}}}$ = 5.92), women were judged to be more feminine than when wearing non-revealing dress (${\bar{\text{x}}}$ = 5.58). Women were rated less moral when wearing revealing dress (${\bar{\text{x}}}$ = 4.29), as compared to women wearing non-revealing dress (${\bar{\text{x}}}$ = 4.94). Ratings of self-respect were lower in the revealing dress condition (${\bar{\text{x}}}$ = 4.60) than in the non-revealing condition (${\bar{\text{x}}}$ = 5.03). Stimulus persons were judged to be less sincere in the revealing dress condition (${\bar{\text{x}}}$ = 4.62) than in the non-revealing condition (${\bar{\text{x}}}$ = 4.98). Finally and importantly, stimulus persons were rated as more sexualized in the revealing dress condition (${\bar{\text{x}}}$ = 5.71) as compared to the non-revealing condition (${\bar{\text{x}}}$ = 4.38).

Sex of respondent also affected judgments of sexualizing traits [F(1, 290) = 20.12, p < .0005); η ² = .065]. Inspection of the cell means revealed that men rated the stimulus persons as more sexualized (${\bar{\text{x}}}$ = 5.22) than women (${\bar{\text{x}}}$ = 4.87).

The purpose of Study 2 was to determine the extent to which women wearing revealing Halloween costumes are sexually objectified by others, both men and women. Or put another way, does a revealing dress manipulation in the context of women’s Halloween costumes cause others to sexually objectify the women (i.e., judge the costumed women in terms of their sexual appeal)? We also examined sex differences in other-objectification; or put another way do men and women differ in the extent to which they objectify Halloween-costumed women (judge the costumed women in terms of their sexual appeal)?

Results showed that revealing dress affected judgments of all dependent variables except competency. As compared to stimulus persons wearing non-revealing dress, stimulus persons wearing revealing dress were perceived to be less considerate, less faithful, less moral, less self-respecting, and less sincere. These results are consistent with others (Graff et al., 2013; Gurung & Chrouser, 2007) who found that women wearing revealing dress were rated lower on positive traits than women wearing non-revealing dress. For example, Graff et al. found revealing dress was related to lower ratings of morality and self-respect; they argue that if sexualizing someone lowers that person’s status then that person “should be seen as less desirable in some way, perhaps less moral and self-respecting” (p. 766).

In addition, women wearing revealing Halloween costumes were rated more feminine than women wearing non-revealing costumes which may suggest that wearing revealing dress is part of a female gender role. This is also consistent with Gurung and Chrouser (2007); their stimulus persons were rated more feminine when dressed in revealing dress as compared to when dressed in athletic sports outfits. Finally and most importantly revealing dress affected judgments of sexualizing traits and this was a moderate effect. Thus, women wearing revealing Halloween costumes were sexualized more than those wearing non-revealing costumes. This is consistent with results of Gurung and Chrouser; in their revealing dress condition the stimulus person was rated higher on traits selected to measure sexuality (attractive, sexually experienced, desirable) than in the non-revealing condition.

Some researchers have interpreted lower ratings of competency or capabilities as a function of a revealing dress manipulation together with attribution of sexualized traits to demonstrate objectification (Graff et al., 2013; Gurung & Chrouser, 2007; Nezlek et al., 2015). Our results did not demonstrate a competency effect which is inconsistent with those previous studies. This non-significant effect could be due to sample differences. Most experimental studies of other-objectification have relied on undergraduate samples (Graff et al., 2013; Gurung & Chrouser, 2007; Heflick & Goldenberg, 2009; Heflick et al., 2011; Loughnan et al., 2010; 2013; Nezlek et al., 2015), whereas our sample was more diverse in terms of age and education. Undergraduate women and men are at a stage of development when SO is likely to be salient, which may not be true of our sample.

A second possible reason for the lack of support for judging competency as a function of revealing Halloween costumes is the notion that Halloween is a holiday that provides freedom from social norms for a day. The women wearing the revealing costumes in our study may have been given a “pass” on ratings of competency because respondents knew the women were headed to a Halloween party, because it is fairly common knowledge that women’s commercial costumes are revealing, and because consumer choice is limited. Finally, because costumes are not worn for everyday dress, participants may have been reluctant to judge the women’s competency based on the revealing costumes.

Objectification has been assessed in several ways and there is disagreement regarding how to assess it (Budesheim, 2011). Our results do show that as compared to women wearing non-revealing costumes, women dressed in revealing costumes were seen as less desirable in terms of consideration, faithfulness, morality, self-respect, and sincerity and were rated higher on traits suggesting they were viewed as sexual objects. The main effect for sex (of participant) on sexually objectifying traits was due to the fact that men rated the costumed women higher in sexually objectifying traits than women did. The main effect for dress demonstrates that men and women were both likely to sexually objectify the costumed women. These results extend research on revealing dress. A revealing manipulation of Halloween costumes led to SO and this was true for men and women.

Integrated discussion

In two studies, the revealing nature of Halloween costumes was investigated. In Study 1 124 pairs of men’s and women’s Halloween costumes were content analyzed. Analysis showed that women’s costumes were significantly more revealing than men’s in terms of tightness and the extent to which they uncovered the body. Researchers interested in studying SO in media presentations have often relied on revealing dress in those depictions as evidence for SO. Since women’s Halloween costumes were found to be more revealing than men’s in Study 1 and that SO in the media is assessed by the presence of revealing dress in media depictions, we reasoned that women’s revealing Halloween costumes could be sexually objectifying. In Study 2, we conducted an online experiment; participants from AMT viewed and rated three images of women wearing revealing or non-revealing Halloween costumes. Participants rated women wearing the revealing costumes higher on sexually objectifying traits (perceived women in terms of their sexual appeal) and lower on a set of positive traits than women wearing the non-revealing costumes. Thus, the women in the revealing costumes were sexually objectified (i.e., they were judged primarily for their sexual appeal). Although men rated women higher on the sexually objectifying traits, both men and women were affected by the revealing dress condition.

Implications

These results demonstrate that revealing Halloween costumes contribute to sexualization of women; hence, Halloween parties may be sexually objectifying experiences. Since both men and women sexualized the women in the revealing costumes, it is important for all involved to assess and reflect on their beliefs about revealing dress. Since dressing in costume on Halloween is popular on university campuses, the degree of sexualization found in this research may be of concern to university administrators and counselors. In fact, Halloween parties have been associated with sexual assault on some college campuses (Halloween night rape, 2014; Rhodan, 2014; Soley-Cerro & Burch, 2015). Dealing with Halloween parties as potential sexually objectifying experiences requires more than policing women’s dress at the parties.

To address sexually objectifying experiences on campus, myths and stereotypes surrounding rape culture must be confronted; this will require re-education of the campus community. Programs to increase campus awareness about Halloween parties as sexually objectifying experiences could be added to or combined with programs that many US universities already have in place as a result of Title IX¹ investigations. For example, Harvard Law School has recently developed programs to provide training for staff and to educate students regarding policies and procedures relating to complaints of sexual assault, harassment, or misconduct (Harvard Law School found, 2014). These types programs should be provided for the entire campus community.

Limitations

In Study 1 the raters did not agree on the extent to which the costumes were sheer; that variable had to be dropped from further analyses. Furthermore, other clothing characteristics such as fabric hand (i.e., slinkiness) and garment color have been used in analyses of sexually objectifying depictions in the media and would be good to investigate in future Halloween costume research. In Study 2, some participants had missing data, which could reflect AMT workers’ attempt to complete tasks quickly. Our sample was large enough to allow adequate statistical power after deleting responses with missing data. However, as with all non-random samples, results cannot be generalized.

Due to stimulus sampling, we did not investigate individual costumes. It is possible that all costumes in a condition are not equal in terms of the impressions conveyed. In our design, this variation was averaged over and became part of the error term in the analysis. This could contribute to the non-significance of the competency variable. Competency was not affected by the revealing dress manipulation. It is possible that judgments made of people wearing ordinary everyday clothing do not always generalize to special dress like Halloween costumes which are seldom worn.

Future research

Future researchers may wish to determine the extent to which academic research findings on dress extend to academic research findings on costume. In other words, if dress manipulations in research are replaced with costume manipulations, will similar results be found?

Our research focused on the extent to which adults sexualized other adult women wearing Halloween costumes. Future researchers should investigate the age at which other-objectification begins. It seems likely that high school students sexualize other high schoolers wearing Halloween costumes, but do grade school students sexualize other grade schoolers?

Some costumes marketed for children are offensive; some because they are kid versions of sexy adult costumes (Horton, 2015), which are banned by some schools (“Some schools implement,” n.d.). A content analysis of children’s Halloween costumes could investigate the extent to which they are revealing and if there are gendered differences in the revealing nature of the costumes.

Parents of school children should be surveyed regarding their attitudes and beliefs regarding revealing Halloween costumes. Where they shop for their children’s Halloween costumes and their preferences for their children’s costumes should also be investigated. Grade school teachers should also be surveyed to determine their perceptions of the Halloween costumes their students wear and could also provide useful information.

Finally, in the 1980s and 1990s when early research on revealing dress (Johnson & Workman, 1992, 1994; Lewis & Johnson, 1989) was conducted, revealing dress was not characterized as sexually objectifying and sexual assault and sexual harassment were not labeled sexually objectifying experiences. Future dress researchers may wish to revisit revealing dress research from the perspective of objectification theory.

Conclusions

These studies extend research on objectification theory. Since most objectification research has centered on women, the current research extends objectification theory to include men. In addition, most objectification research has studied self-objectification, so these results contribute to building knowledge related to other-objectification. Finally, these results extend dress research. Research on revealing dress can now be extended to include Halloween costumes. Furthermore, early research on revealing dress (Johnson & Workman, 1992, 1994; Lewis & Johnson, 1989) can now be interpreted and understood from the vantage point of objectification theory.