Abstract
Self-sexualization refers to intentionally engaging in activities expressly to appear more sexually appealing. The purpose of our studies was to examine gender similarities and differences in how women and men conceptualize self-sexualization as well as introduce measures appropriate for large-scale data collection. In a qualitative study, we first established the existence of self-sexualization among women, although we were unable to document a similar construct among men. The Self-Sexualization Behavior Questionnaire for Women (SSBQ–W), developed over the course of three studies using samples from two small liberal arts colleges, demonstrated adequate internal consistency, stability, and construct validity. Discriminant validity for the SSBQ–W was demonstrated with the Surveillance subscale of the Objectified Body Consciousness Scale, the Sexualizing Behavior Scale, and the Enjoyment of Sexualization Scale. The SSBQ–W may especially be useful in addressing debates about the relationship of self-sexualization to sexual empowerment, agency, and oppression in young women. Additionally, our attempts to develop a parallel sound measure for men failed, and gender differences in item-by-item comparisons concluded that men are less likely to engage in specific self-sexualizing behaviors than women. These findings raise the possibility that self-sexualization is neither as meaningful nor as pervasive for men. In sum, our findings help to further refine how researchers and practitioners might best conceptualize self-sexualization, concluding that it is highly gendered activity that is largely confined to women and offering a measure for women that may promote further research.
Keywords
The report of the American Psychological Association’s (APA) Task Force on the Sexualization of Girls (2010) has intensified the already heated discussion about the images and behavioral expectations of girls and women in the United States (Lamb, 2010a, 2010b; Lerum & Dworkin, 2009; Murnen & Smolak, 2012; Peterson, 2010; Tolman & McClelland, 2011). The report argues that girls and women are exposed to messages that influence and even pressure them to adopt a heteronormative “sexy” persona, that is, to self-sexualize (looking “sexy”) rather than develop an agentic (being sexual) sexuality. Girls and women in other countries, such as Australia (Tiggemann, 2013), as well as in various American ethnic groups (Ward, Rivandeneyra, Thomas, Day, & Epstein, 2013), are subjected to this message. These messages may contribute to body image problems, depression, and sexual dysfunction among women as well as social judgments that girls and women are incompetent or immoral (Halliwell, Malson, & Tischner, 2011; Murnen & Smolak, 2013; Roberts, 2013). Our goals are to explore how college women and men conceptualize and enact their own self-sexualization and to introduce measures of self-sexualizing behaviors for use in large-scale studies. As part of this process, we also investigated gender similarities and differences in self-sexualizing behaviors.
Our emphasis on normative behaviors focused on looking “sexy” in our definition of self-sexualization is important. There is already a measure, the Sexualizing Behavior Scale (SBS; Nowatzki & Morry, 2009, p. 106) that focuses on more extreme behaviors by asking participants how likely they would be to engage in activities such as “taking a pole dancing…class,” “flashing your breasts for the Girls Gone Wild videos” or “wearing clothing…displaying the Playboy bunny symbol.” These specialized behaviors differ from everyday self-sexualizing behaviors, such as wearing low-cut shirts or make-up to look sexy, included in the current study’s conceptualization of self-sexualization. The extreme behaviors measured by the SBS are likely less common and less socially acceptable, and the SBS measures acceptance of, rather than engagement in, self-sexualizing behavior. Thus, in the present study, we aimed to develop gendered measures of everyday self-sexualization and to provide evidence that these measures yield reliable and valid scores for emerging adult women and men.
The Enjoyment of Sexualization Scale (ESS; Liss, Erchull, & Ramsey, 2011) also examines women’s attitudes about self-sexualization. This scale includes items such as “I feel proud when men compliment how I look,” “I want men to look at me,” “I love to feel sexy,” and “When I wear revealing clothing, I feel sexy and in control” (Liss et al., 2011, p. 57). This scale correlates modestly with SBS scores (r = .38; Liss et al., 2011), suggesting a possible divide between attitudes and behaviors. In addition, the construct of empowerment is integrated into some of the questions (e.g., “I feel empowered when I look beautiful”), combining assessment of self-sexualization with the perceived consequences of those attitudes (Halliwell et al., 2011; Lamb, 2010a, 2010b; Lerum & Dworkin, 2009). The self-sexualization measure we introduce in the present study, in contrast, asks only about behaviors (not their perceived consequences), and it examines behaviors in which emerging adults would be likely to engage or consider on an everyday (or at least a routine) basis.
Conceptually and in practice, self-sexualizing behaviors differ from, but are related to, self-objectification and appearance orientation. Objectification theory (Fredrickson & Roberts, 1997; Moradi & Huang, 2008; Tiggemann & Williams, 2012) suggests that when women are consistently treated like objects for the sexual pleasure of men, they ultimately internalize that objectification. Once women self-objectify, they treat their bodies like sexual objects, modifying their bodies through hair removal, weight loss techniques, make-up, and other practices in order to achieve a certain appearance (Bartky, 1990; Bordo, 1993; Mahalik et al., 2005; Smolak & Murnen, 2011). These women may also engage in self-surveillance—monitoring their appearance to be certain they are meeting cultural standards (McKinley & Hyde, 1996). Similarly, being oriented toward appearance is so normative for women that at least one measure of feminine gender roles features it as a component of American femininity (Mahalik et al., 2005).
Although the APA Task Force (2010) considers objectification as a form of sexualization, the two can be distinguished from each other. Sexualization is not objectification if the girl or woman is not directly treated as “a tool for others’ sexual use and pleasure” (Roberts & Zurbriggen, 2013, p. 4). Because children are generally not considered to be appropriate sexual objects, sexy clothing for girls (APA Task Force, 2010; Goodin, Van Denburg, Murnen, & Smolak, 2011) is a form of sexualization but not objectification. These children are “in training” to be sexual objects but are not yet fully objectified. Among adults, the separation of objectification and sexualization may be less common. However, one could monitor one’s appearance (i.e., engage in self-surveillance or worry about appearance) without explicitly engaging in behaviors to look sexy. Existing measures of objectification (McKinley & Hyde, 1996; Noll & Fredrickson, 1998) focus on the attitudinal and affective components of self-objectification. Furthermore, these measures predominantly emphasize general appearance.
In the current investigation, we focus on behaviors explicitly aimed at being sexy. Americans commonly recognize that some clothing, make-up, and hairstyles are appropriate for going to clubs or parties, but they are not for school and work. Girls and women who appear “sexy” are judged as less competent, less intelligent, and less moral than those who dress “appropriately,” albeit attractively (Graff, Murnen, & Smolak, 2012; Murnen & Smolak, 2013). In this case, the person is not necessarily an object for sexual pleasure but is defined only in terms of sexiness (APA Task Force, 2010; Roberts & Zurbriggen, 2013). Sexy also apparently differs from attractive or well-groomed, given that people react negatively to women wearing sexy clothing or make-up at work (Murnen & Smolak, 2013).
To try to clarify this distinction, we suggest that the term attractive refers to an appearance that represents good grooming (e.g., clean hair), meets social norms (e.g., of body type), and generally looks “natural.” This is the way many American women go to work (i.e., clean with light make-up and well-fitting but not tight clothing). Similarly, men might wear a dress shirt, tie, and jacket with lightly styled hair. A “sexy” appearance, on the other hand, emphasizes sexualized body parts (e.g., breasts for women, chest for men, and buttocks for both). So, when trying to look “sexy,” pants or skirts may be so tight that women cannot wear underwear for fear of “panty lines” or dresses may be so low-cut that a special bra is required. Clothing that used to be seen only in pornography, including stiletto heels, is commonly used to project this sexy appearance (Paul, 2005). Both “attractive” and “sexy” styles may represent a person’s internalization of social messages about success and desirability. The distinctions among self-sexualization, self-objectification, and appearance orientation are underscored by the empirically documented unique variance among measures of these constructs (Liss et al., 2011; Nowatzki & Morry, 2009). Self-sexualization may further reflect the belief that only a particular type of appearance is “sexy.” In sum, being sexy may be a form of attractiveness, but one can be attractive without being sexy.
Thus, the purpose of the series of studies we present here is first to explore how college women and men talk about (conceptualize) self-sexualization and then to introduce and provide validity information for measures of self-sexualization behavior for use with emerging adult women and men. These measures should be useful both for further refining conceptualizations of self-sexualization and to future researchers attempting to delineate influences on, and outcomes of, self-sexualization. To establish discriminant validity, our measures must be shown to be distinct from self-objectification (including self-surveillance) and appearance orientation. To further examine validity, the relationship between our novel measures and existing measures of extreme self-sexualizing behaviors (SBS; Nowatzki & Morry, 2009) and attitudes about self-sexualization (ESS, Liss et al., 2011) should also be considered. Given our intention to develop separate measures for women and men, we will present data for men and women separately throughout the present studies. Doing so allows us to capture any gender differences in the definition and extent of self-sexualization behavior as suggested by objectification theory (Fredrickson & Roberts, 1997).
Study 1: Students’ Conceptualization of Self-Sexualizing Behaviors
Our first study is exploratory, intended to gather information from our target groups of college women and men that will be helpful in understanding students’ conceptualizations of heteronormative self-sexualized behavior and thus in generating questionnaire items. Focus groups help researchers understand how relevant samples translate a theoretical construct into their own experiences, beliefs, and behaviors. When theory does not clearly indicate questionnaire items and previous research has not developed relevant templates, focus groups are a good way to generate questionnaire content (O’Brien, 1993). Thus, we used focus groups with college women and men to offer suggestions regarding the behaviors in which people their age engage to look “sexy.” The emphasis was clearly on looking sexy within the socially prescribed heterosexual script. Participants were also asked to assess the appropriateness of a list of behaviors generated by the researchers. In addition to using these data to develop preliminary self-sexualization items for men and women, we probed the data for gender differences in how self-sexualization is conceptualized and experienced (Fredrickson & Roberts, 1997).
Method
Participants
A total of 25 men and 31 women participated in nine same-sex focus groups, with a minimum of four people in a group. Six groups of women and three groups of men participated. Men’s groups (M = 8.3 men) were somewhat larger than women’s (M = 5.2 women). All students were between the ages of 18 and 22 and were enrolled at an exclusively undergraduate liberal arts institution. Most of the students (>80%) were White; however, because of the contacts of one student interviewer, one of the focus groups consisted of ethnic minority women only. Although students were not asked about social class indicators, the overwhelming majority of the students attending the college are from middle- and upper-class families. Furthermore, the culture of the college is upper-middle class. All participants were informed of the purpose of the research and were assured of confidentiality of their comments during the focus groups.
Procedure
Focus group participants first discussed the issue of sexiness. Facilitators were either students or recent alumni and were matched to the gender of the group. The same man facilitated all of the male groups, whereas three different women facilitated the female groups. In introducing the topic, the facilitator specified that the discussion was not to focus on individual or personal behavior. Instead, the focus was to be on what students at this college did to be “sexy.” This was done to try to discover perceived norms of sexy behavior rather than behaviors that might be unique to a particular focus group. Furthermore, we assumed that focusing on others’ behavior would improve the comfort level and protect the privacy of participants. Because we were asking about college norms rather than individual behavior, we did not screen students for sexual orientation or sexual experience.
Students were asked seven questions, which were followed up as needed to maintain the conversation: (a) “What are the kinds of things women (men) do to look sexy to men (women)?” (b) “What kinds of things do women (men) do on a daily or regular basis to look sexy?” (c) “What kinds of things do women (men) do to look sexy when they get ready to go out to a party or club or bar?” (d) “What things are essential that women (men) do in order to look sexy?” (e) “Other than being sexually attractive to men (women), are there other rewards for women (men) to look sexy?” (f) “Are there costs to trying to look sexy?” and (g) “Are there any consequences if a woman (man) decides not to make efforts to look sexy?” Facilitators were trained to not offer their own opinions during the session. Sessions lasted about 1 hour and were audiotaped. Because the primary purpose of the focus groups was to generate potential items for the questionnaire, only responses to the first four questions are considered here.
At the end of the discussion, all participants completed a brief pencil-and-paper questionnaire. The first and second author generated a list of 14 things women might do and 13 things men might do in order to be sexy. These items were based on empirical and theoretical work (APA Task Force, 2010; Bartky, 1990; Bordo, 1993) as well as on the authors’ own experiences. Respondents were asked how many people their age they thought engaged in each behavior, using a rating scale from 1 (none) through 3 (about half) to 5 (all). Men rated men’s behavior, and women rated women’s behavior.
Following the sessions, verbatim transcripts for each focus group were generated. Two independent reviewers, a White man who was a doctoral student and the first author (a retired White female psychology professor), read all of the transcripts and identified major themes of the men’s groups and of the women’s groups. Major themes were identified as those that recurred across groups (within gender) and that were a focus of the discussion. Here we describe the themes about which the two readers agreed (verified by the two readers and by a third person—the second author who is a White middle-aged female psychology professor).
Results
What Women Do to Look Sexy
In the focus groups, participants talked about a variety of behaviors in which women routinely engage to look sexy for men. These included wearing particular clothing such as tight clothes, low-cut shirts, special underwear, and high heels. Comments specified “less covering clothes, more low-neck shirts,” “push-up bras,” “revealing clothes,” and “high heels.” They also mentioned using make-up (such as “smoky eyes, big eyes” and “mascara”), doing their hair, and working out to keep in shape.
Interestingly, women described this process as something of a transformation. They explicitly noted that women look different when going to parties or clubs than when going to classes. On weekdays, they just need to generally look good whereas on weekends they work to look sexy. Thus, their descriptions of what they do to look sexy applied to times when they are going out; they do not generally attempt to look sexy in class or in the meal halls. This differentiation underscores the distinction between attractiveness and sexiness. For example, one woman said, “There’s a big difference between everyday make-up and like ‘I’m trying to look sexy’ make-up.” Another noted that “… the act of just putting on sexier lingerie or a sexier bra and when you put on make-up, you’re not going to put on [your] shirt beforehand, because you don’t want to get make-up on the shirt so being in that outfit for a while gives you that persona of being sexy.” Yet, a third woman claimed “It’s looked down upon when you dress that nicely for class, but then if you wear that for a Friday or Saturday night, you’d be completely acceptable and so that complete differentiation of like weekday clothes and weekend party clothes.” These women further emphasized that it is important that this transformation look like you put in some effort to look good—but not too much effort: “you have to at least show that you put in some effort” while at the same time “Part of it is not looking like you’re spending a lot of time on yourself.”
What Men Do to Look Sexy
Male participants said that men do some things to look sexy for women. They do wear different clothing than they wear when just hanging out with their male friends but this usually just involves a nice shirt: “a polo,” “get dressed up…some type of collared shirt,” “bring out the big name brand stuff ” or “try to dress in clean clothes.” Hygiene is also generally important. Men mention that they should “shower, brush your teeth…comb your hair, clip my nails” or “… like a shower and then you put on a nice shirt.” Body odor is a particular concern with advice like “don’t smell bad” or “I think you need to smell nice” or “the smell thing is important too.” They also think that body shape, muscles, and fitness are part of a sexy appearance: “body shape and size, muscularity” or “I’d say physique is probably more important,” although they temper this point by shying away from an unrealistic muscularity as in “I’ve heard plenty of girls say they don’t like huge ripped guys…so Brad Pitt, he’s a sexy guy right? He’s a fit guy, he’s not the rippest ever.” In pursuit of this muscular athletic ideal, men described the gym as a place for improving one’s appearance: “I see a lot of guys hitting the weights.”
Men also viewed an aura of confidence, sometimes rooted in alcohol, as an important aspect of being sexy. They believed that “in general, women are attracted to success” and that it is important to have “a sense of sexy” and to “bring out the instinctiveness of the male.” Indeed, several of them thought that confidence was the most important aspect of sexiness. For example, one said that “I think attitude is above everything else” and another thought “confidence across everything.” Alcohol may play a role in gaining confidence: “I think that’s why a lot of men drink in general, is to sort of open themselves up. It helps you get in the zone because you can just take the edge off a little bit”; “Gives you more confidence.”
Behaviors to Look Sexy
Table 1 shows the written list of behaviors people might perform to be sexy provided by the researchers, as well as the mean and standard deviation (SD) of men’s and women’s responses to each item. The women rated 9 of the 14 items as things at least half their peers do to be sexy, with shaving legs and underarms being the most frequently named. Certain types of clothes were also perceived as commonly used. On the other hand, men rated only 2 of the 13 items as being done by a majority of men in order to be sexy: wearing scents and shaving genitals.
Norms About College Peers Who Engage in Behaviors to be Sexy.
Note. N/A = not available. SD = standard deviation. Only women (n = 31) rated items for women; only men (n = 25) rated items for men. The scale used was 1 (none or almost none), 2 (some, less than half), 3 (about half), 4 (majority), 5 (all).
Table 1 also shows gender comparisons for the 12 items that were answered by both men and women. Nine of these comparisons were significant, indicating that women thought a higher proportion of women engaged in these behaviors than men thought their peers did. A higher proportion of women were believed to style their hair, dye their hair, shave their legs (chest for men), shave their underarms, wear tight clothes, wear short skirts (shorts for men), wear bikinis (speedos for men), wear low-cut shirts, and get genital piercings. Some effect sizes were small (styling hair; wearing tight clothes), some moderate (wearing a low-cut shirt), and some large (shaving under arms and wearing a bikini/speedo). Of these 12 behaviors, the men’s scores were less than 2 on the 5-point scale (indicating that fewer than “some” men engaged in these behaviors) for 7 of the items whereas for women, the mean was less than 2 for only 1 item.
Discussion
The focus group results first indicate that, as sexualization theorists predict (Lamb, 2010a, 2010b; Lerum & Dworkin, 2009; Peterson, 2010), women make a greater effort than men to be sexy through their behaviors. Women reported that women their age engage in specific behaviors—such as wearing low-cut and tight clothing, special hairstyles, special underwear, and jewelry, as well as removing body hair—to be sexy. Men focused primarily on hygiene, smelling good, and some body hair removal, with some attention to a confident attitude. The men’s behaviors did not seem as focused on sexiness as distinct from attractiveness. Furthermore, of the 12 self-sexualization behaviors considered in gender comparisons, women were perceived as engaging in 9 of the 12 more frequently than men did. Women, therefore, presented as “sexy objects” whereas men were less invested and more agentic in their “sexy” behaviors.
One interesting finding in the focus groups is the context-dependent nature of self-sexualization. The women discussed that becoming “sexy” involved a substantial transformation from their “everyday” appearance. On a daily basis, the women strove to simply be attractive; they believed that a truly “sexy” appearance would be viewed as “too much” and inappropriate for attending classes or meetings and so on. This distinction that the women raised may exemplify the difference between daily objectification and sexualization and is consistent with the argument that self-sexualization may represent a narrow and specific definition of sexiness (APA Task Force, 2010; Roberts & Zurbriggen, 2013). Women are always expected to be attentive to their appearance; sexiness is not necessarily a daily expectation but needs to be “put on” when it is required. Men, on the other hand, did not perceive a need to be attentive to attractiveness on a daily basis but did need to at least pay attention to hygiene when they wanted to be sexy. Although these actions may appear to be “self-beautification” or daily grooming as much as self-sexualization behaviors, it is important to note that participants in the focus groups specifically identified these behaviors as things they did to look “sexy” to the other sex, separate from just being attractive or looking good. Indeed, research indicates that behaviors such as hair removal may serve multiple purposes, including hygiene and sexiness (Smolak & Murnen, 2011; Tiggemann & Hodgson, 2008). Therefore, although these behaviors may sometimes be grooming habits, participants, especially women, were clear in emphasizing that these behaviors were related to the process of “putting on” sexiness.
Thus, the focus groups clearly indicated at least two important points. First, college-age men and women are fully aware that their peers engage in behaviors specifically intended to be “sexy” to the other gender. They strongly indicated that these sexy behaviors are recognizably distinct from daily grooming and attractiveness. The sexy behaviors are context-dependent—reserved for parties, clubs, and other heterosocial environments. Second, both men and women claimed that there was more pressure on women to be “sexy.” The questionnaire data underscore this point. Women rated self-sexualizing behaviors as substantially more common among women than men rated these for men. Indeed, being sexy was virtually a norm for women when they go out (as opposed to everyday attractiveness).
Study 2: Development and Validation of Self-Sexualization Behavior Questionnaires
With our second study, we aimed to develop separate measures for college women and men and then to provide psychometric data for these measures of self-sexualizating behavior, including construct and incremental validity. In developing the measures, we used the qualitative and quantitative data from Study 1 to generate items for preliminary self-sexualization behavior questionnaires: 19 items for women and 18 items for men. Five items were added to the women’s and men’s preliminary questionnaires because of comments made during the focus groups: take a shower, psych yourself up, be assertive, wear special jewelry, and drink. All of these, except “wear special jewelry,” were added because of men’s comments. Fourteen questionnaire items were identical or similar enough to facilitate gender comparisons.
In terms of validity, we sought to evaluate differences between self-surveillance and self-objectification in predicting sexual assertiveness, sexual consciousness, and sexual monitoring. Both self-surveillance and self-sexualization have been hypothesized to affect sexual functioning (Calogero & Thompson, 2009; Fredrickson & Roberts, 1997; Tolman, 2012). It has also been suggested that both may affect body image so we also considered body shame and positive body image as dependent variables. Furthermore, self-sexualization may increase the salience of sex and hence trigger more traditional gender roles within sexual relationships (Sanchez, Fetterolf, & Rudman, 2012). Therefore, we also included benevolent sexism as a dependent variable in the incremental validity analyses.
Thus, Study 2 had four goals: (a) develop a scale to measure self-sexualization behaviors among emerging adult women that demonstrates adequate internal consistency, stability, and construct validity (Goal 1); (b) develop a similar questionnaire for emerging adult men (Goal 2); (c) demonstrate that self-sexualization differs from self-surveillance (as an indicator of self-objectification) in its relationship to sexual attitudes, body image, and benevolent sexism (Goal 3); and (d) assess gender differences in self-sexualization behaviors by comparing the structure and meaning of the men’s and women’s scales (Goal 4).
Method
Participants
Participants were 564 students at two liberal arts colleges, one in the Midwestern (n = 408) and the other in the Southeastern (n = 156) United States. Students at these colleges are typically between the ages of 18 and 23 years. In the current sample, the largest group of participants was first-year students (n = 199, 35%), with 139 (25%) sophomores, 107 (19%) juniors, 110 (20%) seniors, and 9 (2%) failing to identify their class year. The majority of the participants were women, with 403 (71.5%) women and 155 men. There were 403 (72%) participants who self-identified as White, whereas 41 (7%) self-identified as Black, 24 (4%) as Asian, 16 (3%) as Latina/o, 28 (5%) as bi- or multi-cultural, 3 as American Indian, and 2 as Arab Americans (there was no answer by 47, 8% students). Students were also asked to indicate their sexual orientation: 31 (5%) checked bisexual, 462 (82%) heterosexual, 23 (4%) homosexual, 7 pansexual, 13 queer, and 16 questioning (with 12 skipping this question).
Procedure
The project was approved by institutional review boards (IRBs) at both participating colleges. Participants were recruited via e-mail and were offered course credit. Students were provided with two computer links: one to a consent form and another to the questionnaire. Both forms were accessed through SurveyMonkey. Data were downloaded in aggregate form. Individual students completed the consent form and questionnaire at their convenience from any computer with Internet access. Only the questionnaire measures used in the present study are described here; they were ordered as listed subsequently. For each measure, the items were summed to create a total score.
Demographics and Context Questions
Participants were asked to provide their year in school, their gender (woman, man, transgender), their sexual orientation (bisexual, heterosexual, homosexual, pansexual, queer, questioning), and their race/ethnicity (an open-ended item). Participants were asked how important on a daily basis it was to them to look sexy, rated on a scale from 1 (not at all important) to 5 (very important). Using this same rating scale, participants were also asked how important it was to them to look sexy when going to a bar, club, dance, or event.
Self-Sexualization Behavior Questionnaires
Using the focus group and survey data from Study 1, we generated separate lists of sexualizing behaviors for women and for men. The women’s list consisted of 19 items (see Appendix) and the men’s list of 18 items. 1 In order to facilitate gender comparisons, items overlapped as much as possible, yielding 14 overlapping items. Students were asked, “How often do you do the following things specifically in order to look sexy.” For each listed behavior, a 5-point scale ranging from 1 (never) to 5 (always) was provided.
Self-Surveillance
The Objectified Body Consciousness (OBC) Scale (McKinley & Hyde, 1996) contains three subscales measuring aspects of the internalization of objectification. The Surveillance subscale contains 8 items (e.g., “During the day I think about how I look many times”) that measure the tendency to think about one’s body as an outside observer. McKinley and Hyde (1996) found the internal consistency coefficient of this subscale across multiple samples to range from .76 to .89. For validity, they found that scores were positively correlated with public self-consciousness and appearance orientation, but they were not significantly correlated with private self-consciousness, as expected. In the present study items were rated on a 7-point Likert-type scale from 1 (strongly disagree) to 7 (strongly agree), and scores were coded such that a higher summed score indicated greater surveillance. Cronbach’s αs in the present study were .83 for women and .85 for men.
Body Shame
Body image was measured using the OBC Body Shame Scale (McKinley & Hyde, 1996). The OBC Body Shame Scale consists of 8 items such as “I feel like a bad person when I don’t look as good as I could.” Like the self-surveillance scale, it is rated on a 7-point scale ranging from 1 (strongly disagree) to 7 (strongly agree), with higher scores indicating more body shame. McKinley and Hyde (1996) have demonstrated internal consistency (αs ranging from .75 to .84), test–retest stability, and construct validity (body shame scores were negatively correlated with body esteem and positively correlated with restrictive eating). In the present samples, αs were .83 for women and was .78 for men.
Sexual Attitudes and Behaviors
Two subscales of the Sexual Awareness Questionnaire (Snell, Fisher, & Miller, 1991) were administered. Sexual consciousness refers to the idea that some people focus more attention on the sexual aspects of their lives and this reflects awareness of one’s sexual needs and preferences. Agreement with 6 items (e.g., “I am aware of my sexual feelings”) measures this construct. Snell, Fisher, and Miller (1991) reported internal consistency coefficients of .83 for men and .86 for women, and they showed evidence of construct validity (e.g., significant correlations of this subscale with erotophilia and internal locus of control). Sexual monitoring refers to the concern one has that others are evaluating their sexuality. It is measured by agreement with 9 items including, “I’m concerned about the sexual appearance of my body.” Snell et al. reported internal consistency coefficients of .80 for men and .82 for women, and responses were positively correlated with both erotophilia and depression. In our study, the αs for sexual monitoring were .82 for women and .76 for men; for sexual consciousness, .87 for women and .85 for men.
Sexual assertiveness was measured by Rosenthal, Moore, and Flynn’s (1991) Sexual Self-Efficacy Scale, which has a subscale to measure sexual assertiveness. Participants rate their degree of confidence in engaging in nine sexually assertive behaviors such as, “Tell your partner how to treat you sexually.” This subscale was derived through factor analysis, and scores were found to be moderately correlated with scores on two other subscales (ability to “say no”; ability to take sexual precautions) and negatively related to sexual risk taking (Rosenthal, Moore, & Flynn, 1991). Rosenthal and colleagues reported a coefficient α of .77 in their sample, and in our study, it was .70 for women and .67 for men.
Body Appreciation
The Body Appreciation Scale (BAS; Avalos, Tylka, & Wood-Barcalow, 2005), which consists of 16 items focusing on acceptance of and respect for one’s body, was also used to measure body image. For example, one of the questions is, “Despite its flaws, I accept my body for what it is.” Avalos, Tylka, and Wood-Barcalow (2005) have presented evidence for reliability and validity of the scores on the BAS with women through four studies they conducted in developing the scale. Through factor analysis, they found an acceptable one-factor solution in more than one sample and high internal consistency coefficients among responses (e.g., α = .94). Scores were negatively related to eating disorder symptomology and predicted variation in well-being measures beyond other scales that measure body esteem. In addition, Swami, Stieger, Haubner, and Voracek (2008) have presented evidence for the reliability and validity of scores on the BAS with both men and women. Through factor analysis, they found an acceptable unidimensional factor structure for the scale with both men and women, as well as high internal consistency coefficients for both men (α = .85) and women (α = .90). In the present samples, internal consistency was excellent with αs of .90 for women and .88 for men.
Benevolent Sexism
Glick and Fiske (1996) defined benevolent sexism as behaviors and attitudes that reflect sexist stereotypes but are positive in affective tone and may be perceived to be helpful or intended to promote intimacy. This is normally an 11-item scale but, due to an error, 1 item was omitted (“A good woman should be set on a pedestal”). An example of an item that was on the scale is, “In a disaster, women ought to be rescued before men.” Each question is rated on a 6-point scale ranging from 1 (strongly disagree) to 6 (strongly agree). Although it was found to be a multi-dimensional scale through factor analysis, questions can be summed to create a total score, with a higher score indicating greater endorsement of benevolent sexism. Glick and Fiske presented discriminant, convergent, and predictive validity for scores on this scale with both men and women. For example, across six samples they found the internal consistency coefficient to range from .73 to .85, and they found scores on the scale to relate to other measures of sexism. In the present sample, internal consistency was good with αs of .88 for women and .84 for men.
Results
Goal 1: Questionnaire Development for Women (SSBQ–W)
Because there was no theory or empirical evidence (Nowatzki & Morry, 2009) of self-sexualization behaviors to suggest a multifactor structure, we constructed the Self-Sexualization Behavior Questionnaire for Women (SSBQ–W) with the intention of tapping a single set of general behaviors. Because we also wanted to include the fewest number of items possible, we based our decisions to include items in the final scale statistically. First, we examined the mean, median, and SDs of the 19 items. Three items (take a shower, remove leg hair, and remove underarm hair) had medians of 5, meaning that the more than half the women answered “always” (for 2 of the 3 items, more than 70% gave this answer). The means, therefore, were quite high and possible answers within 1 SD of the mean fell outside the scale’s possible answers. Two additional items (dying hair and piercing sexual body parts) had median values of 1 and thus had a similar problem at the low end of the scale. Because these items had truncated response ranges which might affect correlation-based analyses (e.g., regression and factor analysis), they were dropped from further analyses.
We used exploratory factor analysis (EFA) to evaluate the structure of the remaining 14 items. Principal axis factoring with quartimax rotation was employed. The sampling variance was adequate for factor analysis, as indicated by Kaiser-Meyer-Olin (KMO) = .88 and Bartlett’s Test of Sphericity, χ2(91) = 1,303.64, p < .001. Three factors with eigenvalues greater than 1 were extracted without rotation. Given that identifying three factors is consistent with expectations for a 14-item factor analysis (Tabachnick & Fidell, 2007), three, and only three, factors were extracted and rotated. Items with factor loadings
Factor Loadings for Rotated SSBQ–W Items in Exploratory Factor Analysis (Study 2 Women Sample).
Note. SSBQ–W = Self-Sexualization Behavior Questionnaire for Women. Bolded items are considered part of the factors with loadings ≥.40 if the item does not load ≥.35 on any other factor. N/A indicates that the item was not included in that gender’s questionnaire and/or factor analysis. Bolded items in Factor 1 for women constitute the SSBQ–Women and bolded items for Factor 1 for men constitute the scale we tested for men.
A completely new sample of women, from the same Midwestern college, was used to assess stability. The women were generally 18–22 years old, with 80% being first-year and sophomore students. Participants were given course credit for their participation. The women in the sample completed the questionnaires twice, 2–3 weeks apart. Stability was excellent (r = .93, p < .001; n = 58). Because more women participated at Time 1 (n = 140) than at Time 2 (n = 87), the internal consistency of the self-sexualization scale was assessed with Time 1 data and proved good (α = .82, N = 125).
Goal 2: Questionnaire Development for Men (SSBQ–M)
Again, because there was no theory and even fewer extant data than with women to guide a factor analysis, we derived the structure of the SSBQ for Men (SSBQ–M) empirically by using an EFA. Prior to the EFA, we examined frequency distributions for each of the 18 items. One item (taking a shower) had a median score of 5, with nearly 79% of respondents saying they “always” showered to look sexy. In addition, 6 items (dying hair, removing chest hair, removing underarm hair, wearing a speedo, piercing sexual body parts, and wearing special jewelry) had median scores of 1, with 63% to 98% of the respondents saying they never engaged in the behavior in order to look sexy. As was done in the women’s analysis, these truncated items were dropped from the EFA, leaving 12 items.
Again, we used principal axis factoring with quartimax rotation. The sample showed sufficient variance for an EFA as indicated by KMO = .72 and Bartlett’s Test of Sphericity, χ2(66) = 360.21, p < .001. Although the initial unrotated extraction indicated three factors with eigenvalues greater than 1, the three factors could not be extracted and rotated because of communality problems. We attempted a second analysis limiting the extraction to two factors but got the same results. Finally, we restricted the analysis to one factor, which was extracted but not rotated (because there must be at least two factors to rotate). There were 7 items that loaded .40 or higher on this factor and hence constitute the preliminary SSBQ–M (see Table 2). Six of the items overlap with those in the SSBQ–W. Alpha for our sample was .70.
An entirely different sample of men, from the same Midwestern college, was used to assess stability. The men were all of college age, with 86% being first-year and sophomore students (approximately 18–20 years old). These men completed the questionnaire twice within a 2- to 3-week period. Stability was very good, r = .83, p < .001, n = 23. Again, because the sample was larger at Time 1 (n = 53) than at Time 2 (n = 37), this group was used to evaluate internal consistency, which was low (α = .48, N = 50).
Goal 3: Construct Validity
Bivariate Pearson correlations for women and for men among all of the study variables are shown in Table 3. Because of the large number of correlations, α is set at .01. In terms of the construct validity of scores on the self-sexualization behavior scales, it is particularly relevant that SSBQ scores were significantly correlated with both importance of looking sexy on a daily basis and when going out to a club or party for both men and women. Furthermore, as expected, self-sexualization behaviors were correlated with self-surveillance. These correlations were moderate to strong (r = .50, p < .001 for men and r = .42, p < .001 for women). However, these correlations indicate substantial unshared variance between self-sexualization and self-surveillance (r 2 < .25) and hence demonstrate discriminant validity. The relationship of self-sexualization behaviors to body shame, sexual consciousness, sexual monitoring, sexual assertiveness, and benevolent sexism was similar for men and women.
Correlations Among Study Variables: Study 2.
Note. Correlations for women (n = 334–391) are presented below the diagonal; for men (n = 117–144) above the diagonal. All tests are two tailed. Daily sexy = “On a daily basis, how important is it to you to look sexy?” Party sexy = “When you are going out to a dance, bar, club, or event, how important is it to you to look sexy” Sexual assertive = sexual assertiveness; Sexual conscious = sexual consciousness; Sexual monitor = sexual self-monitoring; BS = benevolent sexism.
**p < .01, *** p < .001.
In order to further ascertain whether self-surveillance and self-sexualization were distinct constructs, a series of six regressions were run for women and for men. Alpha was set at .01 to reduce the likelihood of Type I error. For all analyses, the predictor variables were school from which participants were recruited, OBC Surveillance, and total SSBQs. School was included because several variables showed significant differences by school. Because objectification is a form of sexualization (APA Task Force, 2010) and self-surveillance is an outcome of objectification (McKinley & Hyde, 1996), we were particularly interested in whether self-sexualization was associated with variance in the outcome variables beyond the variance explained by self-surveillance. Criterion variables were OBC body shame, body appreciation, sexual assertiveness, sexual consciousness, sexual monitoring, and benevolent sexism.
Table 4 shows the results of these analyses for women. The regressions for body shame and body appreciation were significant, with self-surveillance as the only significant predictor variable. OBC–surveillance was positively related to body shame but negatively correlated with body appreciation. The regressions for sexual assertiveness, sexual consciousness, and sexual monitoring were all significant, with all three predictor variables making unique contributions to the equations. In the sexual assertiveness and sexual consciousness regressions, OBC–surveillance was negatively and SSBQ–W positively correlated with the outcome variable. On the other hand, higher levels of both OBC–surveillance and SSBQ–W were associated with higher sexual self-monitoring. Finally, school and SSBQ–W contributed to the benevolent sexism equation, with higher scores on SSBQ–W associated with more endorsement of benevolent sexism.
Incremental Regressions to Assess Variance Associated With Self-Surveillance Versus Self-Sexualization: Study 2.
Note. **p < .01, ***p < .001.
The results for men are also shown in Table 4. As was true for women, only OBC–surveillance was associated with body shame; the relationship was positive. The equation for sexual self-monitoring indicated only a positive relationship between OBC–surveillance and the outcome. On the other hand, only SSBQ–M was positively related to sexual assertiveness, sexual consciousness, and benevolent sexism. Finally, surveillance was negatively, and SSBQ–M positively, related to body appreciation.
Goal 4: Gender Comparisons
Of the items used in the EFA analyses and originally considered for inclusion in the self-sexualization scales, 16 of them were the same for women and men. To assess gender differences in the frequency of performing these actions, we conducted a series of t-tests. Again, α was set at .01 to reduce the likelihood of Type I error without dramatically increasing the chance of Type II error. Results for these gender comparisons are shown in Table 5.
Gender Differences in Self-Reported Self-Sexualizing Behaviors: Study 2.
Note. df = degrees of freedom; SD = standard deviation.
In general, women were more likely than men to engage in self-sexualizing behaviors. In order to look sexy, women were more likely to wear cologne or scents, to style their hair, to dye their hair, to shave or trim underarm hair, to shave or remove pubic hair, to wear tight or fitted clothes, to wear shorts (or short skirts for women), to wear bikinis or speedos, to wear low-cut clothes or clothes that expose part of their chest, and to wear special jewelry. There were no gender differences in showering, drinking, psyching yourself up, wearing dressy clothing, or piercing sexual body parts. As in Study 1, men were more likely than women to report low frequencies of individual behaviors. They reported mean frequencies of less than 2 (where 2 indicates seldom) for six behaviors, whereas women’s frequency was less than 2 for two behaviors (see Table 5).
Discussion
The main goals of Study 2 were to identify items for our questionnaires targeting college women and men and then to establish the validity and reliability of scores on these self-sexualization behavior questionnaires. The SSBQ–W, a 10-item scale for women, correlated with women’s interest in looking sexy on a daily basis as well as when going out to a bar, club, dance, or event. This pattern suggests that women rated SSBQ–W items in terms of being sexy, not simply being attractive or well-groomed on a daily basis. The results of Study 2 further indicated adequate internal consistency of the SSBQ–W across two different samples, test–retest stability, and discriminant validity from scores on OBC–surveillance. The SSBQ–W was also related to sexual assertiveness and sexual self-monitoring supporting our contention that the SSBQ–W is tapping behaviors that are related to feeling “sexy”. Because the SSBQ–W was not correlated with body appreciation and had only a small correlation with body shame, the SSBQ–W does not seem to be primarily assessing general appearance. The relationship between the SSBQ–W and benevolent sexism is similar to the findings of Liss, Erchull, and Ramsey (2011) for the relationship between the ESS and benevolent sexism, suggesting the SSBQ–W is assessing a different aspect of the same construct of self-sexualization. The relationship between the SSBQ–W and benevolent sexism is also similar to the relationship found by Nowatzki and Morry (2009) for the relationship between their scale and hyperfeminity. Future studies should examine the relationship between the SSBQ–W and hyperfeminity as well.
As with the SSBQ–W, scores from the men’s scale, the SSBQ–M, correlated with sexual assertiveness and sexual consciousness, indicating that the scale is, again, tapping a construct related to being “sexy.” Furthermore, stability and divergent validity from scores on self-surveillance were demonstrated for the men’s scale, the SSBQ–M. However, internal consistency reliability was not clearly evidenced. Furthermore, the results of the EFA were equivocal. Overall, it remains questionable whether a psychometrically sound scale of self-sexualization for men can be devised. This failure to develop a clearly valid men’s scale is consistent with theory indicating that sexualization is much more part of the socially prescribed feminine than masculine role (Bartky, 1990). Additionally, there are substantial gender differences in how frequently emerging adults engage in self-sexualization behaviors. Women were significantly more likely to engage in 11 of the 16 compared behaviors, again supporting the contention that self-sexualization is part of being feminine among White, middle-class American young adults.
Study 3: Relationships Among Sexualization Questionnaires for Women
The first goal of this third study is to replicate the factor structure of the SSBQ–W using confirmatory factor analysis (CFA). In addition, important convergent and divergent validity questions remain regarding women’s self-sexualization, and the relationships of the SSBQ–W to the extant measures of self-sexualization require investigation. Specifically, we consider the relationship of the SSBQ–W to the two other measures of self-sexualization: the SBS (Nowatzki & Morry, 2009) and the ESS (Liss et al., 2011).
Once again, because of the hypothesized relationship between self-sexualization and self-objectification, we assess discriminant validity between scores on self-sexualization and self-objectification measures. In addition, because the items in the SSBQ–W and ESS particularly focus on feminine, sexy appearance, discriminant validity between scores on these scales with measures of feminine appearance concerns and hyperfeminity were also investigated. Mahalik and colleagues’ (2005) Investment in Appearance Scale assesses commitment to society’s appearance norms for women. It should be the case that SSBQ–W scores are correlated with an investment in appearance. The hyperfemininity scale measures the extent to which women “accept” a sexually objectified role as part of their self-definition (Murnen & Byrne, 1991). Therefore, hyperfemininity should also be positively correlated with the SSBQ–W. Because of the problems in identifying items for the SSBQ–M and because the SBS and the ESS have only been validated on women, Study 3 investigated these relationships using an exclusively female sample.
Thus, we propose five hypotheses. First, the CFA will support the single-factor solution for the SSBQ–W. Second, the behaviors included in the SSBQ–W should be endorsed more frequently than those listed in the SBQ because SSBQ–W behaviors are believed to be behaviors in which many young women might engage daily, rather than the extreme behaviors represented on the SBQ. Third, although the three self-sexualization scales (the SBS, ESS, and our newly developed SSBQ–W) should be intercorrelated, these correlations are expected to be modest because the three scales appear to investigate different aspects of self-sexualization. Fourth, although scores on each of the three self-sexualization scales should be positively correlated with self-objectification, these correlations again are expected to be small to moderate, indicating that the scales are not all measuring a single construct. Finally, all three sexualization scales are predicted to demonstrate small-to-moderate correlations with investment in feminine appearance and hyperfemininity.
Method
Participants
Participants were 93 different women recruited from the same two liberal arts U.S. colleges (one Midwestern and the other Southeastern). All four class years were represented, with 27 (29%) first-year students, 28 (30%) sophomores, 20 (22%) juniors, and 18 (19%) seniors. The mean age was 20.68 years (SD = 2.64, range = 18.3–40). The majority of the participants (n = 64; 72%) self-identified as White, with 3 identifying as Black, 13 as Asian, and 9 as multiethnic.
Procedure
The IRBs of both institutions approved our project. Students were awarded course credit for their participation. Informed consent information, consent form, all surveys, and debriefing were available on SurveyMonkey. Students were sent a link to the survey and could access it through any computer. In the informed consent information, students were urged to complete the questionnaire by themselves and in a private setting. Students first provided demographic information including school attended, age, class year, and ethnicity. Three scales (the SSBQ–W, OBC Scale–Surveillance, and the OBC Scale–Body Shame subscales) were also used in Study 2 and are described there. In Study 3, the αs for these scales were .84 for SSBQ–W, .82 for OBC–Surveillance, and .82 for OBC–Body Shame. These were the final three questionnaires in the packet. The remaining questionnaire measures were presented in the following order.
Conformity to Feminine Norms Inventory-Appearance
The Conformity to Feminine Norms Inventory (CNFI)-Appearance Scale (Mahalik et al., 2005) taps investment in appearance and includes items such as “I regularly wear make-up” and “It is important to look physically attractive in public.” Its 7 items are rated on a 4-point scale from 1 (strongly disagree) to 4 (strongly agree), with higher summed scores indicating greater investment in appearance. Mahalik and colleagues (2005) reported that scores on this subscale were significantly correlated with scores on the Thinness subscale of the CNFI as well as with measures of eating disordered attitudes and body dissatisfaction. Further, women had significantly higher appearance scores than men. Internal consistency in Mahalik and colleagues’ study was .82; α in the current sample was .76.
SBS
The SBS (Nowatzki & Morry, 2009) measures acceptance of engaging in sexualizing behaviors such as “taking pole dancing or strip aerobics class” or “dancing provocatively at a dance club with female friends.” The 10 behaviors are rated on a scale from 1 (not very likely) to 5 (very likely) so that a higher summed score indicates greater acceptance of self-sexualization. Nowatzki and Morry (2009) present a CFA indicating that these 10 items represented a single factor distinct from other measures (that were not intended to measure self-sexualization) such as benevolent sexism. In the current study, α was .75.
Hyperfemininity Scale
This 26-item scale assesses attitudes toward women’s role in heterosexual relationships with emphases on physical attractiveness, using sexuality to manipulate men, and the importance of being in heterosexual relationships (Murnen & Byrne, 1991). Examples of items include “I like to flirt with men” and “I like to play hard-to-get.” Murnen and Byrne (1991) reported that hyperfemininity scores were correlated with traditional attitudes about women and a belief that marriage was more important for women than having a career. The internal consistency coefficient was .76, using a forced-choice response format, and a test–retest correlation was .89. In the present study, respondents rated items on a scale from 1 (strongly disagree) to 5 (strongly agree), a response format consistent with Murnen (1998). A mean score is reported here, with a higher score indicative of greater hyperfemininity. Alpha in the current sample was .85.
ESS
The ESS (Liss et al., 2011) aims to measure how much women enjoy sexualization experiences. The attitudinal measure consists of 8 items such as “I feel proud when men compliment the way I look” and “I love to feel sexy.” The items are rated on a 6-point scale from 1 (disagree strongly) to 6 (agree strongly) and were summed to yield a single score. A higher summed score indicates greater enjoyment of sexualization. Liss et al. (2011) report good internal consistency (α = .85) and construct validity indicated by positive correlations with self-objectification and engaging in self-sexualizing behaviors. They demonstrated discriminant validity between scores on the ESS and measures of self-objectification (Noll & Fredrickson, 1998) and surveillance (McKinley & Hyde, 1996) as well as the SBS (Nowatzki & Morry, 2009). In the present study, Cronbach’s α was .85.
Self-Objectification Questionnaire
The Self-Objectification Questionnaire (SOQ; Noll & Fredrickson, 1998) lists 10 physical attributes: half concern physical functioning (e.g., health) and half are about appearance (e.g., physical attractiveness, sex appeal). Respondents are asked to rank the importance of these qualities to their “physical self-concept.” Scores can range from −25 to 25 and are calculated such that a higher summed score indicates greater self-objectification (i.e., greater investment in physical appearance than physical function). Noll and Fredrickson (1998) presented construct validity for scores on the SOQ by reporting correlations with appearance anxiety and body dissatisfaction. Many studies do not report internal consistency for the SOQ because its structure precludes the use of standard internal consistency statistics. However, Hill and Fischer (2008) have argued that a high negative correlation between the rankings of the appearance and the function items indicates internal consistency. In the current study, this correlation was −1.00.
Results
Hypothesis 1: CFA
We used STATA 12 to run a CFA using the current sample combined with the female sample from the Time 1 test–retest analysis in Study 2. This yielded a large enough sample (N = 238) to effectively use in a CFA. The original model specified a single factor using the 10 items identified in Study 2 as part of the SSBQ–W. These items were use cologne, style hair, remove genital hair, wear tight clothes, wear dressy clothes, wear short skirt/shorts, wear a low-cut blouse, wear a special bra, wear high heels, and wear special jewelry. Maximum Likelihood Estimation was used. The model was not a good fit for the data, χ2(35) = 125.95, p < .0001, root mean square error of approximation (RMSEA) = 0.10, and comparative fit index (CFI) = 0.89. However, the item coefficients were all statistically significant (p < .05), and STATA identified several instances of error covariance. Because items were selected to attempt to measure a single construct, item error covariation was expected.
Using 5 significant error covariances identified by STATA and the original 10 SSBQ–W items, we re-ran the CFA, again using Maximum Likelihood Estimation. The model fit was only marginally adequate, χ2(29) = 60.64, p = .001, RMSEA = 0.07, and CFI = 0.95. Four more significant error covariances identified by STATA were added to the model, and we re-ran the CFA, again using Maximum Likelihood Estimation. All item coefficients, as well as the coefficients for the error covariances, were statistically significant. Coefficients for this model are shown in Figure 1. This model fit the data well, χ2(29) = 29.31, p = 0.25, RMSEA = 0.03, and CFI = 0.99.
Confirmatory factor analysis for the Self-Sexualization Behavior Questionnaire for Women (SSBQ–W). All standardized coefficients between the scale items and self-sexualization are significant at p < .001. All error c-variances are significant at p < .05. N = 238.
Hypothesis 2: SSBQ–W Behaviors More Common Than SBQ Behaviors
Acceptance of the behaviors in the SBS was quite uncommon, with 8 of the 10 items showing an average rating of less than 2 (not likely). The ESS items were rated as being more common, with all 8 items averaging at least a rating of 2, whereas four of the questions averaged 4 or higher on a 6-point scale. The SSBQ–W items were also endorsed as being fairly common, with 9 of the 12 items averaging a rating of 3 (sometimes) or more.
Hypotheses 3–5: Intercorrelations of Scales
Pearson correlations were used to examine intercorrelations among the scales. Alpha was set at .01 to reduce the likelihood of Type I error. However, because the conservative α increases the risk of a Type II error, the r 2 effect size is also provided to assist in interpretation of the relationships. The SBS is not significantly correlated with the SSBQ–W, r = .26, p = .02, r 2 = .07, or with the ESS, r = .19, p = .08, r 2 = .04. The ESS does correlate with the SSBQ–W, r = .50, p < .0001, r 2 = .25 although 75% of the variance of the two scales do not overlap. The SSBQ–W is positively correlated with both the Self-Objectification Scale, r = .29, p =.01, r 2 = .07, and the OBC–Surveillance scale, r = .41, p = .0001, r 2 = .17. The SSBQ–W is not significantly correlated with Hyperfeminity, r = .22, p = .051, r 2 = .05. The SSBQ–W shows a moderate-to-strong correlation with the CNFI-Appearance Scale, r = .48, p < .0001, r 2 = .23, although there still is 77% nonoverlapping variance. None of the three sexualization scales correlates with the Body Shame Scale.
Discussion
In support of the first hypothesis, the CFA indicated that a single-factor solution for SSBQ-W fits the data. As suggested in Hypothesis 3, the SSBQ–W is significantly correlated with an attitudinal measure of self-sexualization, the ESS (Liss et al., 2011). This finding suggests that there is an underlying construct of self-sexualization that is tapped by these two measures. However, the correlation suggests that only 25% of the variance between the ESS and the SSBQ–W scores is shared, thereby indicating discriminant validity for scores on the SSBQ–W. The ESS measures attitudes rather than behaviors so it is clearly assessing a different component of self-sexualization than the SSBQ–W is. The SSBQ–W is not significantly correlated with the SBS, although there was 7% overlapping variance suggesting some relationship. The behaviors assessed in the SSBQ–W are considerably more common than those in the SBS, as argued in Hypothesis 2, which further suggests that the SBS and SSBQ–W are measuring different components of self-sexualizing behaviors. This finding also indicates that the behaviors measured by the SSBQ–W are behaviors that are more relevant for college women. Given that the SBS does not correlate with either the ESS or the SSBQ–W, it is possible that the SBS is tapping a construct other than self-sexualization. It is also possible that the SBS is simply measuring extreme behavior, whereas the ESS and SSBQ-W measure common attitudes and behaviors. Consistent with Hypothesis 4 and Study 2, the SSBQ–W showed adequate discriminant validity from the measures of self-objectification. Finally, this third study demonstrated that the SSBQ–W is not simply measuring investment in appearance or hyperfemininity. Although there was substantial shared variability between the SSBQ–W and appearance investment, the majority of variability (75%) was non-shared.
General Discussion
The primary purposes of our research were to further conceptualize the construct of self-sexualization for college women and men and to develop questionnaire measures of self-sexualizing behaviors. Previous measures for women emphasized attitudes toward sexualization (Liss et al., 2011) or fairly extreme forms of self-sexualization (Nowatzki & Morry, 2009). Our measures focus on self-sexualizing behaviors that were more commonly endorsed than the SBS items. Both an EFA and CFA seemed to suggest a single factor composed of 10 items for women, although additional research is needed to confirm this. The measure for women showed acceptable internal consistency in three different samples as well as good stability.
We also attempted to develop a self-sexualization measure for men; to our knowledge, ours was the first study to attempt to do so. An EFA indicated a 7-item scale for men. However, the EFA was unable to extract or rotate factors because of communality problems, so the results must be interpreted cautiously. This caution is bolstered by marginal and poor internal consistency findings for two different samples. Thus, it may be argued that the measure for men is not tapping a single, uniform construct. Indeed, there may not be a construct of self-sexualization that actually applies to American college men at this time. The measure for men did show good test–retest stability. This means that more research will be required to definitively establish the viability of a measure for men. Some of this research should examine whether there is in fact a constellation of behaviors in which men engage to appear sexy. The qualitative data in our research, as well as the relatively low means and medians on most of the self-sexualization items for men, suggest that there may not be. Indeed, one potential conclusion from our studies is that self-sexualization itself is highly gendered by confining most of its pressures to women.
There are two other available measures of self-sexualization in women. One of these, the ESS (Liss et al., 2011), assesses attitudes and so is clearly tapping a different component of self-sexualization than the SSBQ–W is. However, the SSBQ–W and the ESS show a moderate correlation, indicating that they do measure a shared underlying construct. We suggest that future research includes both the ESS and SSBQ–W in order to get a fuller picture of self-sexualization. The other measure, the SBS (Nowatzki & Morry, 2009), asks about the acceptance of more extreme behaviors than those included in the SSBQ–W. Arguably, it is not particularly appropriate for use with emerging adults who are not likely to even have the opportunity to participate in several of the activities. The SSBQ–W, on the other hand, asks about behaviors that might occur in college students and perhaps even in adolescents. Future research does need to evaluate the validity of scores on the SSBQ–W with adolescents, however.
Importantly, the data indicate that self-objectification (as measured by OBC Surveillance and the SOQ) and self-sexualization are distinct constructs for both genders. This finding is consistent with arguments that “sexy” is a specifically and narrowly defined role that is distinct from everyday appearance (Sanchez et al., 2012). In the qualitative data, women distinguished between looking attractive for everyday activities and looking sexy to go out. They clearly recognized self-sexualization as a unique, definable phenomenon. In Study 2’s quantitative data for the women, OBC–surveillance and self-sexualization made unique contributions to all three measures of sexual functioning. For sexual assertiveness and sexual consciousness, self-sexualization was positively associated with the outcome variable once self-surveillance’s negative contribution was controlled. There are at least two possible explanations for this pattern. First, it is possible that self-surveillance and self-sexualization are intertwined but that, at least statistically, removing the effects of self-surveillance results in self-sexualization appearing to be a positive influence. This argument suggests that, in reality, the two may be inextricable. On the other hand, it is possible that there are women who are high on both and also women who are high on self-sexualization but low on self-surveillance. Only future research will be able to disentangle the relationships between self-surveillance and self-sexualization. Furthermore, only prospective research will allow us to ascertain whether one or both actually contribute to the development of sexual assertiveness and sexual consciousness.
For women’s sexual self-monitoring, both self-surveillance and self-sexualization demonstrated positive associations. These findings raise several interesting questions for future research. First, the finding for sexual self-monitoring (concerns about being evaluated as sexy) differs from those for sexual consciousness (awareness of sexual needs and preferences) and for sexual assertiveness (requesting or expecting that one’s wishes be respected in sexual situations). This suggests that various aspects of sexual functioning may not be similarly related to self-sexualization. Self-sexualization is associated with more self-monitoring (a negative outcome) and also with more self-assertiveness and self-consciousness in sexual situations (both positive outcomes). Second, the finding concerning benevolent sexism indicates that women’s self-sexualization may be associated with gender role beliefs and expectations, a finding consistent with arguments that any sexualization cues may increase gender role adherence (Sanchez et al., 2012). However, it is also possible that endorsement of traditional gender roles actually contributes to self-sexualization. Directionality is an issue that will be key to debates about the meaning of self-sexualization and so is worthy of much more research attention, particularly because other researchers have also reported correlations between self-sexualization and benevolent sexism (Liss et al., 2011; Nowatzki & Morry, 2009).
For the other criterion variables, it was sometimes self-surveillance and sometimes self-sexualization that made significant contributions to the equations. This suggests a different pattern of relationships and, potentially, influences for the two variables. Self-sexualization and self-surveillance do co-vary, but they are clearly not interchangeable constructs. Much research will be needed to disentangle this relationship.
The men’s incremental regression analyses also indicated a distinction between self-surveillance and self-sexualization, although the findings were a bit less clear than for women. Body appreciation was the only criterion variable for which both self-surveillance and self-sexualization were significant contributors. Otherwise, there were different patterns of relationships for the two variables. The bivariate correlations suggest that these differential patterns do not simply reflect multicollinearity in that the variable (whether self-surveillance or self-sexualization) that was not related to the regression outcome did not typically show a bivariate relationship to that criterion variable either. Additionally, some equations show different patterns for men than women. For sexual assertiveness, sexual consciousness, and benevolent sexism, only self-sexualization was a significant contributor to the equations. This suggests that, for men, self-sexualization may actually be viewed as a positive masculine characteristic associated with positive sexual functioning. Given the tenuous factor structure and internal consistency of the attempted measure for men, these findings must be interpreted cautiously. Again, only future research, preferably of a prospective nature, can sort out these findings.
For women, body shame and body appreciation were only related to OBC–surveillance in the regressions. This finding could be interpreted as indicating that women’s self-sexualization is not related to body image. We believe that such an interpretation is premature. There are small but statistically significant correlations between the body image variables and SSBQ–W. It is particularly interesting that self-sexualization is negatively related to body appreciation. There may be other elements of body image that are more strongly related to self-sexualization. We suggest that future research investigate other aspects of body image, including behaviorally based constructs such as body checking (Reas, Whisenhunt, Netermeyer, & Williamson, 2002). Again, prospective research will be needed to distinguish causal versus correlational relationships as well as the direction of any causal influences. We also suggest that researchers investigate possible moderating relationships in the relationships of self-sexualization and body image. However, the limited relationships of SSBQ–W to body shame, body appreciation, hyperfemininity, and CFNI-appearance do suggest that the SSBQ–W is measuring a sexiness construct rather than simply assessing general appearance concerns.
We examined gendered patterns of self-sexualizing behavior in Studies 1 and 2. The results were striking and consistent. Men do engage in self-sexualization behaviors, and these behaviors appear to be related to sexual functioning. However, women engage in substantially more self-sexualizing behaviors than men do. For example, the women’s scale consists of 10 inter-related items, whereas the one we explored for men had only 7. Furthermore, women engage in individual self-sexualizing behaviors more frequently than men do, with some differences being quite large. Indeed, although men report engaging in self-sexualizing behaviors, they also indicate that some of these behaviors are uncommon (or only a minority of men engage in them at all). Even in the qualitative data, both men and women acknowledged the expectation that women will self-sexualize more than men and that women will invest greater effort, time, and money in self-sexualization behaviors. We believe that these findings support feminist arguments about the social construction of female versus male sexuality in the United States (Bartky, 1990; Bordo, 1993; Kim et al., 2007). However, again, research is necessary to evaluate feminist theoretical arguments as opposed to evolutionary psychology arguments, for example, that suggest that women naturally use sex appeal to attract men, whereas men use power indicators to attract women (Swami & Furnham, 2008). Indeed, the SSBQ–W, and self-sexualization in general, confounds women as both actors and targets of self-sexualizing behaviors, leaving future researchers to disentangle these and further expose the sources of these pressures on women. The SSBQ-W can be used to investigate influences and outcomes of self-sexualization.
Limitations
Although our research presents evidence of a valid and unique behavioral measure of self-sexualization for emerging adult women, it does have several limitations. First, it is important to acknowledge that self-sexualization likely is a developmental process (Else-Quest & Hyde, 2009; Murnen & Smolak, 2012; Smolak & Murnen, 2011). It will be important to find ways to evaluate the meaning of these changing behaviors. Thus, although both an elementary school girl and a young adult woman might wear body-revealing clothing (e.g., midriff tops), their intentions may differ, as may their understandings of the rewards and consequences of such behaviors. Researchers will need to be sensitive to such differences and will need to develop measures that are appropriate for different ages while still allowing for outlining the developmental changes and processes involved in self-sexualization.
Feminists argue that female sexuality, including self-sexualization, is socially constructed (Bartkey, 1990; Bordo, 1993). If this is the case, then we should expect ethnic group, social class, and cultural differences in self-sexualization. The samples in our study were overwhelmingly White and middle-to-upper class Americans; however, we did have samples from two different schools. Interestingly, there were school-based differences in the frequencies of a number of self-sexualizing behaviors as well as in the outcome variables. These findings underscore the importance of evaluating the applicability of the self-sexualization measures in various groups. It will be important not only to assess internal consistency, which given the number of items in the SSBQ–W is likely to often be adequate, but also factor structures as well as relationships to other variables, including self-surveillance. The CFA for the SSBQ–W only reached good levels of goodness of fit indices when several error covariance terms were included. The inclusion of these terms may have artificially inflated the goodness of fit. Thus, the CFA in particular should be replicated.
Validity questions were not included in either the SSBQ–W or the attempted scale for men. It is possible that the students misunderstood or failed to carefully read or simply randomly answered the questions. Future research may want to evaluate this possibility. Furthermore, it is possible that the students did not fully read or understand the instructions to answer the questions in terms of sexiness rather than attractiveness. Although our qualitative finding—as well as the relationships of particularly the SSBQ–W to appearance investment, self-objectification, and the ESS tends to argue against this possibility, future researchers might want to emphasize the words “in order to look sexy” in the instructions by using boldface or capital letters. Future researchers may also want to consider whether results would be different if the SSBQ-W was given with directions to consider what participants do “in order to look attractive” instead of sexy. In addition, group comparisons reported here are tentative until tests of measurement of invariance are performed.
The purposes of our research were to explore constructs and to develop measures, not to evaluate theoretical positions. Of course, theory guided our choices of criterion variables. Our goal here was simply to demonstrate discriminant validity. We provided evidence to support validity and reliability of scores associated with the SSBQ–W, although not for a measure for men. We do not claim to have tested or evaluated theoretical models or debates about women’s self-sexualization, clearly possibilities for future research. Indeed, an essential value of the new SSBQ–W is its potential to help address the debates concerning the relationship of self-sexualization to empowerment and oppression in women’s sexuality and other realms.
Practice Implications
Self-sexualization is a concept that has been linked by the APA Task Force (2010) and other researchers to body image, depression, and sexual dysfunction problems among women as well as social judgments that girls and women are incompetent or immoral (Graff et al., 2012; Murnen & Smolak, 2013; Roberts, 2013). The results of the current study suggest that treatment programs might address the concept of self-sexualization, perhaps as a maladaptive behavior or a form of body checking to be monitored and changed during the course of therapy. The results particularly underscore a role for self-sexualization in sexual behavior and monitoring, although there is some evidence for relationships between self-sexualization and body image. Therefore, the SSBQ–W could be utilized to examine the frequency of self-sexualizing behaviors in a clinical setting and to open a dialog about the sociocultural influences encouraging participants to engage in these behaviors as well as the potential effects of self-sexualization. Furthermore, SSBQ–W scores were also related to benevolent sexism. Benevolent sexism may increase susceptibility to some of the problems associated with traditional gender roles. Self-sexualization may provide a mechanism to discuss benevolent sexism and how it may negatively impact opportunities.
Conclusion
Theorists, researchers, and advocates have all argued that self-sexualization is an issue for women, possibly impacting body image, depression, and sexual functioning (APA Task Force, 2010; Lamb, 2010b; Murnen & Smolak, 2012). Although there has been some debate concerning the meaning and effects of self-sexualization, there has been widespread acceptance that self-sexualization primarily is a concern for girls and women (Lamb & Peterson, 2012; Lerum & Dworkin, 2009). In the current study, we successfully developed a questionnaire measure, SSBQ–W that can be used for large-scale studies to evaluate these claims and debates. The internal consistency, reliability, and validity were demonstrated for the SSBQ–W in multiple samples of emerging adult women. Although more work is necessary to validate the SSBQ–W’s usefulness in samples of various ages, ethnicities, and nationalities, it appears to be ready to be used in research.
On the other hand, the attempt to develop a parallel measure for men was not successful. Scale items and internal consistency could not be reliably demonstrated. Furthermore, qualitative data indicated that both men and women believed self-sexualization was more expected of and applicable to women. These data support the contention that women are more affected by self-sexualization, although more research is needed. Additionally, our exploration of the construct of self-sexualization and how it overlaps with and is different from less everyday forms of sexualized behavior and self-objectification is provocative. The current study helps to lay the groundwork for further refinements of this likely body image pressure in the lives of college women.
Footnotes
Appendix
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Note
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
