Abstract
Given the limitations of existing measures of self-objectification, the purpose of the two studies presented in this article was to develop and validate a new measure of self-objectification, the Self-Objectification Beliefs and Behaviors Scale (SOBBS). In Study 1, a total of 654 women completed an online questionnaire including a pool of items designed to measure self-objectification. The item pool was constructed through focus groups and consultation with subject matter experts. Participants also completed existing measures of self-objectification and related constructs. A subset of participants completed the item pool at a 2-week interval to allow for assessment of test–retest reliability. A 14-item, two-factor measure of self-objectification was derived through exploratory factor analysis. Data provided evidence for the SOBBS’s internal consistency and test–retest reliability as well as its convergent, discriminant, and incremental validity via its relations with interpersonal sexual objectification, body image, disordered eating, and depression. In Study 2, the factor structure of the SOBBS was reevaluated using confirmatory factor analysis and validity was again evaluated; SOBBS scores correlated with public body consciousness, private and public self-consciousness, self-esteem, and sexual functioning. We discuss implications for assessment of self-objectification as part of efforts in prevention and treatment of body image and eating disturbances. We hope the development of a new measure of self-objectification prompts further study of the topic and that increased knowledge about self-objectification allows clinicians and researchers to develop interventions that foster greater resilience against sexual objectification.
Fredrickson and Roberts (1997) developed objectification theory as a framework for explaining how the cultural practice of sexually objectifying women contributes to the disproportionate number of women faced with eating disorders, depression, and sexual dysfunction. They argued that sexual objectification, which can occur through exposure to media images depicting women as a collection of body parts or through interpersonal interactions during which a woman is viewed solely in terms of her body, socializes women to engage in self-objectification, treating their own bodies as objects capable of representing the self and viewing them as others do (Fredrickson & Roberts, 1997). A large body of literature has demonstrated that self-objectification contributes to many negative consequences for women, including poor body image (e.g., body shame, body dissatisfaction, appearance anxiety), disordered eating, depression, and sexual dysfunction (see Tiggemann, 2011, for a review).
A central challenge to studying self-objectification lies in operationalizing and measuring the construct (see Calogero, 2011, for a review). The aim of the current research was to consider the nature of self-objectification and to develop a new measure of the construct that more fully assesses the various components of self-objectification and addresses the primary limitations of existing measures. In the review that follows, we explore what it means for a woman to self-objectify. We then describe two studies related to the development and evaluation of a new measure of self-objectification, the Self-Objectification Beliefs and Behaviors Scale (SOBBS).
What Does It Mean to Self-Objectify?
Because self-objectification is thought to be a natural consequence of the constant sexual objectification to which women are subject in Westernized cultures, understanding self-objectification requires us to first consider the definition of sexual objectification. Bartky (1990) defined sexual objectification as occurring when “sexual parts or sexual functions are separated out from the rest of [a woman’s] personality and reduced to the status of mere instruments or else regarded as if they were capable of representing her” (p. 20). Drawing from Bartky’s work, Fredrickson and Roberts (1997) defined sexual objectification as “the experience of being treated as a body (or a collection of body parts) valued predominantly for its use to (or consumption by) others [emphasis in original]” (p. 174). Central to these definitions is the idea that when a woman is sexually objectified, the appearance of her body and the utility of the body to others are separated out from who she is as a person. She is no longer viewed as a whole person but instead in a piecemeal fashion where her thoughts, feelings, and other attributes do not matter and are overshadowed by the appearance of her body.
At its most basic level, Fredrickson and Roberts (1997) argued that self-objectification involves turning this sexually objectified perspective on oneself. They originally defined self-objectification as the act of “[internalizing] an observer’s perspective on self” (pp. 179–180). According to Fredrickson and Roberts, when a woman self-objectifies, she thinks about how her body might look to others, adopting a perspective that requires her to be both in her body, as its owner, and outside her body, as an observer. They described this as a “doubling” of perspectives, which creates a form of self-consciousness and undoubtedly interferes with whatever other tasks a woman might engage in. Girls and women learn to treat themselves as objects to be evaluated or sights to be appreciated by others, and especially by heterosexual men, ultimately learning that their bodies are not truly their own (Fredrickson & Roberts, 1997).
In some of the earliest empirical work related to objectification theory, Noll and Fredrickson (1998) defined self-objectification as valuing one’s own body more from a third-person perspective, focusing on observable body attributes (e.g., “How do I look?”), rather than from a first-person perspective, focusing on privileged or nonobservable body attributes (e.g., “What am I capable of? How do I feel?”; p. 624). They described self-objectification as “[a view] of the self predominated by physical appearance” (p. 633), with the assumption that greater emphasis on physical appearance than physical competence meant that one is considering their own body from an observer’s perspective. This appearance versus competence distinction forms the basis for the Self-Objectification Questionnaire (SOQ; Noll & Fredrickson, 1998), one of the two main measures of self-objectification.
Shortly before Fredrickson and Roberts (1997) published their work on objectification theory, McKinley and Hyde (1996) wrote about a concept they called objectified body consciousness, which was meant to describe women’s “experience of the body as object and the beliefs that support this experience” (p. 183). Objectified body consciousness has three components: (a) body surveillance, (b) internalization of cultural body standards manifested as body shame, and (c) beliefs about appearance control. Of the three components of objectified body consciousness, body surveillance is most similar to self-objectification. Body surveillance refers to the tendency of women to view their bodies as an outside observer does. According to McKinley and Hyde, a woman high in body surveillance “would watch her body frequently and think of her body in terms of how it looks rather than how it feels” (p. 188). This definition is the basis for the Objectified Body Consciousness Body Surveillance Scale (OBC-Surveillance; McKinley & Hyde, 1996), the other commonly used measure of self-objectification.
Conceptually, self-objectification and body surveillance are similar to one another. However, researchers disagree about the extent to which self-objectification and body surveillance are the same construct (Calogero, 2011). Some conceptualize body surveillance as a behavioral manifestation of self-objectification, using scores on the SOQ to predict OBC-Surveillance scores, given that habitual body monitoring is a theorized outcome of self-objectification (e.g., Tiggemann & Kuring, 2004; Tiggemann & Slater, 2001); however, the extent to which OBC-Surveillance reflects habitual body monitoring, rather than self-objectification, is unclear. Some conceptualize SOQ and OBC-Surveillance as parts of the same construct (e.g., Hill & Fischer, 2008; Liss, Erchull, & Ramsey, 2011), and still others use OBC-Surveillance alone as an indicator of self-objectification (e.g., Augustus-Horvath & Tylka, 2009; Lindner, Tantleff-Dunn, & Jentsch, 2012).
Consequences of Self-Objectification
One of the primary consequences of self-objectification is some form of body image disturbance, whether body shame, body dissatisfaction, or appearance anxiety (see Tiggemann, 2011, for a review). The difference between self-objectification and these body image consequences is that self-objectification is a perspective toward the body, whereas body shame, body dissatisfaction, and appearance anxiety involve negative feelings about one’s body or oneself as a result of perceptions about how the body looks or feels. One way that self-objectification leads to body image consequences is via its relation with internalization of cultural ideals of appearance (Moradi, Dirks, & Matteson, 2005). That is, when someone has adopted an observer’s perspective of her own body, the extent to which she “buys in” to cultural messages about appearance is related to how much body image disturbance she will experience. For this reason, Moradi (2010) has encouraged researchers to consider whether internalization of cultural ideals of beauty (and specifically the thin ideal) is part of self-objectification.
The adoption of a third-person perspective toward the body leads to consequences for one’s consciousness or awareness—that is, where their attention is placed. Theoretically speaking, those who self-objectify should be less aware of internal bodily states, something termed “interoceptive awareness” or “private body consciousness,” but literature on this topic is mixed (Tiggemann, 2011). However, by virtue of being aware of how the body looks to others, women are keenly aware of the body as a social object, something termed “public body consciousness” (Miller, Murphy, & Buss, 1981). This differs from, but is related to, self-consciousness in that self-consciousness refers to awareness of internal thoughts and feelings (“private self-consciousness”) and awareness of the self as a social object (“public self-consciousness”; Fenigstein, Scheier, & Buss, 1975). Self-objectification has been described as a form of self-consciousness and is therefore likely related to public and private self-consciousness. Consistent with these ideas, Calogero and Watson (2009) found that self-objectification, as measured by OBC-Surveillance, was related to public self-consciousness and public body consciousness in a sample of U.S. college students (they did not measure private self-consciousness or private body consciousness).
Both directly and indirectly via its association with various negative body image states, self-objectification has also been linked to poorer mental health. Disordered eating, including caloric restriction and bulimic behaviors, are well-established by-products of self-objectification (e.g., Calogero, 2009; Moradi et al., 2005). Self-objectification is also related to increased depressive symptoms (e.g., Muehlenkamp & Saris-Baglama, 2002; Tiggemann & Kuring, 2004) and lower self-esteem (e.g., Choma et al., 2010; Liss et al., 2011). Fredrickson and Roberts (1997) also proposed that self-objectification would lead to sexual dysfunction among women. Consistent with their proposition, an emerging body of literature now points toward poorer sexual functioning and decreased sexual satisfaction among women who self-objectify (e.g., Calogero & Thompson, 2009; Tiggemann & Williams, 2012).
Advancing the Study of Self-Objectification
Over the last 20 years, researchers have been able to learn a considerable amount about how gender socialization affects women’s mental health through the lens of objectification theory. However, both Calogero (2011) and Moradi (2010) recommended that research on self-objectification would advance best with improvement in the operationalization and measurement of the construct, each of which informs the other. From an operationalization perspective, it is clear that the internalization of a third-person observer’s perspective on the body is a fundamental aspect of self-objectification. Determining how to best measure internalization of this perspective is more challenging.
Both the SOQ and OBC-Surveillance have been used to measure self-objectification since the inception of objectification theory. The SOQ relies on the assumption that valuing the body for what it looks like over what it can do reflects internalization of the observer’s perspective (Noll & Fredrickson, 1998). OBC-Surveillance is based on the premise that self-objectification includes thinking about how the body looks to others and valuing what the body looks like over how it feels (McKinley & Hyde, 1996). While each measure has a body of evidence supporting its use, both measures also have limitations. Briefly, the rank-order format of the SOQ yields ipsative data for which measures of internal consistency reliability cannot be calculated. Recommended methods for addressing data loss (i.e., multiple imputation) cannot be used with this format, which is problematic, given that several authors have reported problems with data loss because participants often complete the scale incorrectly (e.g., Lindner et al., 2012; Myers & Crowther, 2007). In addition, there are concerns about the extent to which the SOQ adequately captures the experience of self-objectification of people from diverse backgrounds with respect to race, ethnicity, sexual orientation, or gender (see Calogero, 2011, for a detailed review). There is evidence for the reliability and validity of OBC-Surveillance, including a broader evidence base regarding its use with men and people from racially and ethnically diverse backgrounds (e.g., Breitkopf, Littleton, & Berenson, 2007; Chen & Russo, 2010). However, some of the items on the scale appear to reflect different but related constructs (e.g., social comparison), and, as noted earlier, there is disagreement in the field about the relation between OBC-Surveillance and the SOQ.
Bearing in mind the questions in the field about the conceptualization of self-objectification as well as the limitations of current measures of the construct, the goal of the current research was to develop a new measure of self-objectification. This new measure was designed with the hopes of capturing a broader range of women’s self-objectification experiences, thus allowing for advancement of our understanding of how self-objectification translates into mental health risks for women. We chose to focus our scale development efforts on young adult women, given that objectification theory posits that sexual objectification and the resulting self-objectification are gendered experiences, though we also recognize an emerging body of literature showing that men also experience self-objectification (e.g., Calogero, 2009; Martins, Tiggemann, & Kirkbride, 2007). The decision to focus on young adult or college-aged women was guided by evidence that self-objectification changes across the life span (Tiggemann & Lynch, 2001). In our conceptualization of self-objectification, we sought to integrate the conceptual and operational definitions of self-objectification and sexual objectification presented in the literature, by focusing on three areas: (a) thinking about the body as an outside observer would (as was done in OBC-Surveillance and as is assumed as part of the logic that underlies the SOQ), (b) valuing the body more for what it looks like than what it can do (as is addressed by both the SOQ and OBC-Surveillance), and (c) treating the body as if it is capable of representing the self as a person (as is highlighted by the definition of sexual objectification offered by Bartky, 1990). We sought to bring together these three conceptual definitions of self-objectification into a single instrument, thus allowing for assessment and exploration of multiple aspects of self-objectification, improving our ability to both understand the nature of self-objectification as a construct and better predict related outcomes.
Study 1
The purpose of Study 1 was to develop an initial pool of items for the new measure of self-objectification, to derive a final scale via exploratory factor analysis (EFA), and to examine the reliability and validity of the newly derived scale. Specific hypotheses regarding the scale and its reliability and validity were as follows: (1) EFA will yield a three-factor measure of self-objectification reflecting each of the three components of self-objectification articulated in the literature: the internalization of an observer’s perspective, valuing appearance of the body over competence, and viewing the body as if it is capable of representing the self. (2) Scores on the subscales of the newly developed measure will be positively correlated with scores on the SOQ and OBC-Surveillance, reflecting the measure’s construct validity. (3) Consistent with the objectification theory framework, scores on the newly developed measure will be correlated with sexual objectification experiences, body shame, appearance anxiety, eating disorder symptoms, and depressive symptoms and will be negatively correlated with body satisfaction, providing evidence for the scale’s convergent validity. (4) Scores on the newly developed measure will be unrelated to narcissism, providing evidence for the scale’s discriminant validity. (5) Scores on the newly developed measure will predict body image, disordered eating, and depression above and beyond SOQ and OBC-Surveillance scores, providing evidence of the scale’s incremental validity. (6) Scores on the newly developed measure will demonstrate stability over a 2-week test–retest interval.
Method
Participants
We analyzed data from 654 participants, all female undergraduate psychology students at a large Southeastern university. Participants ranged in age from 18 to 30 (M = 20.43, SD = 2.60) and had a mean body mass index (BMI) of 23.37 (SD = 4.66, range = 15.64–45.19). A majority of participants identified as Caucasian (58.72%, n = 384), 18.35% identified as Hispanic/Latina (n = 120), 10.86% as Black/African American (n = 71), 4.43% as Asian/Pacific Islander, 2.45% (n = 29) as Biracial (n = 16), 1.22% as Native American (n = 8), and 3.98% as Other (n = 26). Over 90% of participants identified as heterosexual (90.90%, n = 595), 5.05% identified as bisexual (n = 33), and 2.91% as gay or lesbian (n = 19). Based on the guidelines for EFA by Comrey and Lee (1992), the sample size was over 300 and therefore adequate for EFA.
A subsample of participants was invited to complete measures of self-objectification 2 weeks after their initial participation. Of the 273 participants who received the invitation, 56 provided Time 2 data (20.51%). One participant who completed Time 2 measures at much longer intervals than other participants (82 days) was excluded from analyses, yielding a final test–retest sample of 55 participants. These participants ranged in age from 18 to 27 (M = 20.27, SD = 2.24). Mean BMI was 22.83 (SD = 4.17, range = 16.44–36.61). Two thirds of participants who completed test–retest measures identified as Caucasian (67.27%, n = 37), followed by Hispanic/Latina (12.73%, n = 7), Black/African American (10.91%, n = 6), Asian/Pacific Islander (5.45%, n = 3), Biracial (1.82%, n = 1), and Other (1.82%, n = 1). Nearly all participants in the test–retest sample identified as heterosexual (94.55%, n = 52). There were no statistically significant differences on demographic or other study variables (i.e., self-objectification, sexual objectification, body image, and mental health risks) between participants who completed Time 2 measures and those who received invitations but did not participate.
Measures
Self-objectification
To develop the initial pool of items, we conducted two focus groups with female college students to learn more about how self-objectification manifests itself in people’s daily lives. Eight women ranging in age from 18 to 22 (M = 19.38, SD = 1.51) participated. Two participants identified as Caucasian (25.0%), three as African American (37.5%), two as Latina/Hispanic (25.0%), and one as Biracial (12.5%). Focus groups (one with three participants and one with five participants) were facilitated by the first author (D.L.) using a facilitator script to ensure that discussions were guided by the same prompts. At the start of the focus group, participants were provided with a conceptual definition of sexual objectification and some examples (e.g., catcalling, media depictions of body parts). After group members discussed their impressions of sexual objectification, the facilitator introduced the term self-objectification. Group members were provided with conceptual definitions of self-objectification based on the literature: “the extent to which a woman takes on an observer’s perspective when thinking about her own body” (paraphrase based on Fredrickson, Roberts, Noll, Quinn, & Twenge, 1998, p. 270); “the tendency to introject an objectifying third-person perspective on one’s own body, evaluating it in terms of its value and attractiveness to others, rather than its value and function for the self” (Fredrickson, Hendler, Nilsen, O’Barr, & Roberts, 2011, p. 690); and “valuing the body for what it looks like over what it can do” (a paraphrase based on the logic underlying the SOQ; Noll & Fredrickson, 1998). Group participants were then asked a series of questions about how they saw self-objectification manifested in their daily lives. Each group discussion was transcribed and the content, including participants’ statements about aspects of appearance, situations, behaviors, and attitudes, generated a pool of 61 items.
Three researchers who study self-objectification and have an interest in its measurement served as subject matter experts and provided feedback regarding the item pool. Subject matter experts included two women and one man; two identified as Caucasian and one as a member of an ethnic minority group. They were provided with the same conceptual definitions of self-objectification the focus group participants had been given but were also instructed to use their training and expertise in this area to help them rate the items. Each expert rated the extent to which each of the 61 potential items reflected the construct of self-objectification (1 = characteristic of someone who engages in a very low level of self-objectification, 4 = characteristic of someone who engages in neither a very high nor very low level of self-objectification, 7 = characteristic of someone who engages in a very high level of self-objectification) via online survey. They were also given the opportunity to offer written qualitative feedback after rating the items. Items that did not discriminate (i.e., had mean ratings close to 4.0 and/or large standard deviations suggestive of disagreement among subject matter experts) were eliminated or modified based on qualitative feedback.
We administered 55 items to participants. All items were worded in a positive direction, as the potential benefit of being able to detect non-content-based responding via negatively worded items was outweighed by the potential for factor analysis to yield method factors (i.e., one factor with positively worded items and a second with negatively worded items), rather than factors reflective of underlying constructs (DeVellis, 2012). Participants were asked to rate their level of agreement with each item using a 5-point Likert-type scale (1 = strongly disagree, 3 = neither agree nor disagree, and 5 = strongly agree), with higher scores indicating greater levels of self-objectification. The items reflected the conceptual definitions of self-objectification that are described in the research literature: (a) taking on an observer’s perspective about one’s body (39 items, many of which were similar in structure but asked about different body parts or situations; e.g., “I frequently think about how the proportions of my body look to others,” and others which reflected behaviors in which someone who self-objectifies might engage; e.g., “I choose specific clothing or accessories based on how they make my body appear to others”); (b) valuing the appearance of one’s body over competence (5 items; e.g., “My physical appearance is a more important contributor to my self-worth than my level of physical fitness”); and (c) viewing one’s body as if it is capable of representing the self (11 items; e.g., “How slim my body looks says something about who I am as a person”). Development of the final version of the new self-objectification measure is further described in the Results section.
In addition to completing the item pool for the SOBBS, participants completed the SOQ (Noll & Fredrickson, 1998) and the OBC-Surveillance (McKinley & Hyde, 1996). The SOQ consists of 10 items (5 appearance based and 5 competence based). Participants are asked to rate the items in order of importance to their physical self-concept. The sum of the competence items is subtracted from the sum of the appearance items, yielding scores that range from −25 to +25. Higher scores reflect greater self-objectification. Cronbach’s α cannot be calculated for the SOQ. It has been used to measure self-objectification with a variety of samples, including American, Australian, British, and Canadian women (e.g., Calogero, 2009; Grippo & Hill, 2008; Morry & Staska, 2001; Tiggemann & Kuring, 2004; Tiggemann & Lynch, 2001); college-age men (Daniel & Bridges, 2010; Grieve & Helmick, 2008); heterosexual and lesbian college students (Hill & Fischer, 2008); and women of different racial and ethnic backgrounds (e.g., Buchanan, Fischer, Tokar, & Yoder, 2008; Grabe & Jackson, 2009; Mitchell & Mazzeo, 2009).
OBC-Surveillance measures the extent to which people take on an observer’s perspective toward their own bodies. The scale has 8 items, each of which is rated on a 7-point Likert-type scale. A mean score for all items is obtained, with higher scores reflecting greater self-objectification. Since its development, the Body Surveillance subscale of the OBCS has been used to measure self-objectification in a wide range of samples, including adult women from the United States, Australia, and the United Kingdom (e.g., Calogero & Thompson, 2009; McKinley, 2006; Tiggemann & Kuring, 2004; Tiggemann & Lynch, 2001; Tylka & Hill, 2004); women of diverse ethnic backgrounds (Breitkopf et al., 2007; Crawford et al., 2009); heterosexual and lesbian women (Kozee & Tylka, 2006); and heterosexual and gay men (e.g., Martins et al., 2007; McKinley, 1998; Tiggemann & Kuring, 2004). Cronbach’s α in the present study was .82.
Sexual objectification
Participants completed the Interpersonal Sexual Objectification Scale (ISOS; Kozee, Tylka, Augustus-Horvath, & Denchik, 2007), a 15-item measure of the frequency of sexual objectification that one experienced within the past year (1 = never to 5 = almost always). Items assess the extent to which participants feel “looked at” in interpersonal interactions, the frequency of sexually objectifying comments, and the frequency with which respondents experience unwanted sexual advances. Data collected from female undergraduate students provide evidence for internal consistency reliability, test–retest reliability, and construct validity of the ISOS (Kozee et al., 2007). The internal consistency estimate of the scale in the present study was .93.
Body image
Body image was operationalized as body shame, body satisfaction, and appearance anxiety. Body shame was assessed using the Objectified Body Consciousness Body Shame Scale (McKinley & Hyde, 1996), which consists of 8 items (e.g., “During the day, I think about how I look many times”) that are rated on a 7-point Likert-type scale, with higher scores reflecting greater body shame. This subscale demonstrated adequate construct validity, internal consistency, and test–retest reliability in a female undergraduate sample (McKinley & Hyde, 1996). Internal consistency in the present study was .82 for women.
Body satisfaction was measured using the Appearance Evaluation Subscale of the Multidimensional Body-Self Relations Questionnaire (MBSRQ-AE; Cash, 2000). Each of the scale’s 7 items (e.g., “I like my looks just the way they are”) is rated on a 5-point Likert-type scale and the mean score for all items is obtained, with higher scores reflecting greater body satisfaction. The MBSRQ-AE is widely utilized, especially in nonclinical samples. Scores on the MBSRQ-AE are related to scores on other measures of body satisfaction, body image investment, and body image affect (e.g., Brown, Cash, & Mikulka, 1990). The scale’s 1-month test–retest reliability was .91 in a sample of college women (Cash, 2000). In the present study, internal consistency was .91 for women.
Appearance anxiety was assessed using the Appearance Anxiety Scale (AAS; Dion, Dion, & Keelan, 1990), a 14-item measure of apprehension or anxiety about one’s physical appearance and the extent to which it will be evaluated by others (e.g., “I feel nervous about aspects of my physical appearance,” “I worry about how others are evaluating how I look”). Participants rate each item using a 5-point rating scale (ranging from 1 = never to 5 = almost always). A mean score is obtained and higher scores reflect greater appearance anxiety. In a sample of college women, scores on the AAS were negatively correlated with self-esteem and positively correlated with public self-consciousness. Two-week test–retest reliability was .89 (Dion et al., 1990). The internal consistency estimate for the scale was .93 in the present study.
Eating disorder symptoms
The Eating Attitudes Test (EAT-26; Garner, Olmstead, Bohr, & Garfinkel, 1982) was used to assess eating disorder symptoms including dieting, preoccupation with food, and bulimic behavior. Respondents answer each of the scale’s 26 items using a 6-point rating scale (1 = always to 6 = never). The measure was scored continuously in this study and higher scores reflected greater eating disorder symptomatology. The EAT-26 is appropriate for clinical and nonclinical samples (Mintz & O’Halloran, 2000), and it has demonstrated internal consistency and construct validity in samples of college students (Engeln-Maddox, Miller, & Doyle, 2011; Garner et al., 1982). Internal consistency in the present study was estimated at .91.
Depression
The Zung Self-Rating Depression Scale (Zung, 1965) is a reliable and valid measure of affective, somatic, and cognitive symptoms of depression. This scale consists of 20 items rated on a 4-point scale (1 = none or a little of the time to 4 = most or all of the time). A mean score is obtained for all items, with higher scores indicating greater depressive symptoms. The measure has been validated as a screening tool in general medical practice and in a sample of undergraduate students (Szymanski & Henning, 2007; Zung, 1990). Internal consistency in the present study was .85.
Narcissism
The Narcissistic Personality Inventory-16 (NPI-16; Ames, Rose, & Anderson, 2006) is a measure of subclinical narcissism consisting of 16 forced-choice pairs of contradictory items (e.g., “I like to be the center of attention vs. I prefer to blend in with the crowd”). The NPI is scored by computing the total proportion of responses consistent with narcissism, with total scores ranging from 0 to 1. There is evidence of the scale’s convergent and discriminant validity and test–retest reliability in undergraduate students (Ames et al., 2006). Cronbach’s α for the present study was .67.
Procedure
Following institutional review board approval, over four semesters we recruited participants from the psychology department at a large Southeastern university via Sona Systems. Female students ages 18 and older were eligible for participation. Study measures were presented in a counterbalanced order via the online survey platform Qualtrics. All data were treated anonymously. Upon completion of the questionnaire, we provided participants with psychoeducational information about self-objectification and its relation to mental health risks as well as contact information for the campus counseling center. During two semesters of recruitment, debriefing information also served to inform participants that they may be invited by the researchers to participate in an optional follow-up survey in 2 weeks’ time. To recruit participants for the test–retest portion of the study, the first author (D.L.) contacted participants via Sona Systems 2 weeks after their initial participation with an invitation to participate in a follow-up study. The invitation encouraged participants to complete the follow-up questionnaire within 3 days. Only the measures of self-objectification (i.e., the SOBBS item pool, the SOQ, and OBC-Surveillance) were administered at Time 2 and this was done in a fixed order. As with initial data collection, the survey was administered via Qualtrics. Contacting participants for participation in the test–retest study via Sona Systems preserved participants’ anonymity. Following completion of the questionnaire, participants were again provided with psychoeducational information about self-objectification and contact information for the counseling center.
Results
Data were analyzed using IBM SPSS Statistics 22 (IBM, 2013b) and IBM SPSS AMOS 22 (IBM, 2013a). A total of 799 women completed the questionnaires. Participants who missed at least 1 of the 2 items designed to detect non-content-based responding were dropped from the analyses (n = 68, 8.51%). Participants were also excluded from analyses if they were missing more than 20% of data or 5 items on a single scale, whichever was lower (n = 10, 1.25%). Of the remaining 721 participants, 219 (30.37%) had at least one missing data point. Little’s (1988) test of missingness indicated that data for the women were missing completely at random, χ2(df = 5, 132) = 5,020.89, p = .86. Given that no pattern could be detected to the missing data, the expectation maximization algorithm was used to replace missing values, avoiding substantial reduction in sample size and statistical power. Data were also screened for univariate and multivariate normality prior to proceeding with analyses. Skewness for all variables was less than 3 and kurtosis for all variables was less than 10, indicating that no transformations were necessary (Kline, 2011). Eight multivariate outliers based on Mahalanobis’s distance were dropped from the analyses leaving 713 participants. Last, given concerns about changes in self-objectification and other body image variables across the life span (see Tiggemann & Lynch, 2001), we conducted initial descriptive and correlational analyses both with and without participants above age 30. The inclusion of participants above age 30 resulted in some changes in the magnitude of correlation coefficients, particularly among correlations between self-objectification and other body image variables. In addition, participants over 30 scored significantly lower than those under 30 on OBC-Surveillance, t(711) = 2.45, p = .02, while also scoring lower on the MBSRQ, t(66.19) = 2.37, p = .02, thus lending support to the idea that relations among study variables may be different for older women. Thus, participants above age 30 were excluded from the analyses (n = 59), yielding the final sample size of 654 young adult women.
As has been the case in other research (e.g., Grippo & Hill, 2008; Lindner et al., 2012; Myers & Crowther, 2007), there were problems with the completion of the SOQ. Of participants whose data were retained for analysis, 119 (18.20%) assigned the same rating to more than one appearance attribute despite explicit instructions not to do so. There were no significant differences in age, BMI, or other study variables between participants who completed the SOQ correctly and those who did not. These participants were retained, as excluding them would have resulted in a substantial reduction in sample size and statistical power. Analyses involving the SOQ were completed using only those participants who completed the measure properly.
Hypothesis 1
EFA was used to derive the scale from the item pool. Factors were extracted using maximum likelihood extraction with direct oblimin rotation. Delta weight was set at zero to allow for moderate correlation among factors. Parallel analysis (Horn, 1965) was used in conjunction with the scree plot to determine the number of factors to retain. Items with a primary factor loading greater than .50 and no cross loadings greater than .30 were retained (Tabachnick & Fidell, 2007). Results of EFA appear in Table 1.
SOBBS Items and Factor Loadings for Study 1 and Study 2.
Note. Study 1, n = 654; correlation between Factor 1 and Factor 2 = .55. Study 2, n = 175; correlation between Factor 1 and Factor 2 = .54. Numbers above correspond to the item’s place in the item pool. Bolded factor loadings denote assignment to a primary factor. SOBBS = Self-Objectification Beliefs and Behaviors Scale; EFA = exploratory factor analysis; CFA = confirmatory factor analysis.
All 55 items were entered into the first EFA. The Kaiser–Meyer–Olkin measure of sampling adequacy was .97, and Bartlett’s test of sphericity was statistically significant, χ2(1,485) = 21,644.42, p < .001, indicating that the data were appropriate for factor analysis. Two factors were retained. Factor 1 contained items about internalizing an observer’s perspective of the body, and Factor 2 contained items about equating the body to who one is as a person and valuing physical appearance above other attributes. Thirty-one items were retained, with 24 items on Factor 1 and 7 items on Factor 2. Among the 31 items, the two-factor solution accounted for 50.43% variance. Items on each factor were examined for redundancy at the conceptual and statistical level. A cluster of 4 items on Factor 1 had similar content and inter-item correlations greater than .70; the 1 item with the simplest wording was retained in this case (i.e., “I often think about how my body must look to others”), resulting in the deletion of 3 items.
Twenty-eight items were entered into a second EFA. The two-factor solution accounted for 48.70% variance. Seeking a more parsimonious explanation that could also perhaps account for more variance, we examined item factor loadings and communalities. Five items were removed because their communalities were below .40. Three of these items were about specific body parts (i.e., thinking about how breasts, arms, or butt look to others), 1 was about clothing, and 1 was about a specific situation (i.e., thinking about how one looks to others during sexual activity). Two additional items, both specific body parts (i.e., proportions and legs), were removed because, conceptually, it did not make sense to value these specific areas of the body over others that were not retained. This left 21 items; 14 items on Factor 1 and 7 items on Factor 2.
The remaining 21 items were entered into a third EFA. The two-factor solution accounted for 51.76% variance. Further examination of items revealed that the communality for 1 item (i.e., “I think about how my body will look to others if I assume a particular pose/position”) dropped below .40. This item was removed. Communalities for 5 items about objectification in specific contexts (e.g., with friends, at work) were approximately .40. These items were removed to examine whether there was substantial improvement in the amount of variance accounted for by the factor solution. The 15-item solution accounted for 53.93% variance. One additional item with the lowest communality and factor loading was removed from Factor 1, as achieving an equal number of items for each factor likely makes calculation and interpretation of total scale scores simpler for users.
The remaining 14 items, 7 items on each of two factors, were entered into EFA. This solution accounted for 54.86% variance, and the two factors were moderately correlated, r = .55. Factor 1 had an internal consistency of .89 and inter-item correlations from .43 to .65. Factor 2 had an internal consistency of .88 and inter-item correlations from .42 to .68. Cronbach’s α for the full 14-item scale was .91 and inter-item correlations ranged from .17 to .68. Although a three-factor solution was originally hypothesized, the two-factor solution that emerged offered the best fit for the data. Results that follow are presented for both factor scores and the SOBBS Total score.
Hypotheses 2–6
Descriptive statistics and correlations among study variables appear in Table 2. SOBBS scores are presented for each factor as well as in the form of a total score (the average of the two factor scores). As predicted in Hypothesis 2, factor and total scores on the SOBBS were positively correlated with scores on the SOQ and OBC-Surveillance. This provides evidence of the measure’s convergent validity. Mean scores on SOQ and OBC-Surveillance were consistent with those found in other samples of college-age women (e.g., Calogero, 2009; Lindner et al., 2012; McKinley & Hyde, 1996; Noll & Fredrickson, 1998). Mean scores on SOBBS Factor 1 fell just above the scale midpoint, and mean scores on SOBBS Factor 2 and SOBBS Total scores fell just below the scale midpoint.
Descriptive Statistics and Correlations Among Variables for Study 1.
Note. SOBBS Factor 1 = Self-Objectification Beliefs and Behaviors Scale–Observer’s Perspective; SOBBS Factor 2 = Self-Objectification Beliefs and Behaviors Scale–Body as Self; SOBBS Total = Self-Objectification Beliefs and Behaviors Scale Total Score; SOQ = Self-Objectification Questionnaire; OBC-Surv = Objectified Body Consciousness Body Surveillance Scale; ISOS = Interpersonal Sexual Objectification Scale; OBC-Shame = Objectified Body Consciousness Body Shame Scale; MBSRQ = Multidimensional Body-Self Relations Questionnaire Appearance Evaluation Scale; AAS = Appearance Anxiety Scale; EAT = Eating Attitudes Test-26; Zung = Zung Self-Rating Depression Scale; NPI = Narcissistic Personality Inventory.
*p < .05. **p < .01. ***p < .001.
Table 2 shows that SOBBS scores were positively correlated with sexual objectification experiences, body shame, appearance anxiety, disordered eating, and depressive symptoms and negatively correlated with body satisfaction. Correlations were consistent with Hypothesis 3 and provided additional evidence for the measure’s convergent validity. The OBC-Surveillance Scale performed similarly to the SOBBS in this regard. SOQ scores were positively correlated with body shame, appearance anxiety, disordered eating, and depression and were negatively correlated with body satisfaction. Counter to the tenets of objectification theory, SOQ scores were uncorrelated with sexual objectification experiences.
Consistent with Hypothesis 4, Table 2 shows that SOBBS scores were uncorrelated with NPI scores. Scores on the SOQ and OBC-Surveillance also were uncorrelated with NPI scores. These data provide evidence for the discriminant validity of all three measures.
Consistent with Hypothesis 5, hierarchical multiple regressions demonstrated that one or both SOBBS factors predicted outcome variables above and beyond existing measures of self-objectification. Table 3 shows the results for hierarchical multiple regressions evaluating SOBBS Factor 1 and Factor 2 against the SOQ and OBC-Surveillance. We applied a Bonferroni correction given the number of statistical tests performed. To elucidate the contribution of each measure of self-objectification, the SOQ was entered in the first step of each regression, followed by OBC-Surveillance in the second step, and SOBBS Factors 1 and 2 were entered in the third step. Body shame, body satisfaction, appearance anxiety, disordered eating, and depression were entered as dependent variables. Consistent with objectification theory, SOQ scores predicted all outcome variables in Step 1 of the hierarchical regressions. In the second step, OBC-Surveillance predicted all outcome variables while the SOQ did not. The addition of SOBBS Factor 1 and Factor 2 in the third step of the regressions resulted in a significant R2 change for all five dependent variables. SOBBS Factor 1 and Factor 2 were statistically significant predictors of body shame and appearance anxiety. OBC-Surveillance was also a statistically significant predictor, though the relation decreased from Step 2 to Step 3. Unexpectedly, SOBBS Factor 1 predicted body satisfaction, but SOBBS Factor 2 was not statistically significant at the p < .01 level (p = .04). As with the other body image variables, the relation between OBC-Surveillance and body satisfaction decreased from Step 2 to Step 3. For disordered eating and depression, SOBBS Factors 1 and 2 were the only statistically significant predictors in Step 3; OBC-Surveillance was no longer statistically significant. Table 4 shows the incremental validity of SOBBS Total scores relative to the SOQ and OBC-Surveillance. A similar pattern of results emerged for Total scores, with the SOQ predicting outcome variables in Step 1, but not when considered along with OBC-Surveillance in Step 2. The relation between OBC-Surveillance and outcome variables diminished for body shame, body satisfaction, and appearance anxiety and disappeared for disordered eating and depression once SOBBS scores were entered into the equation in Step 3. Overall, the addition of the SOBBS to the regression model, whether in the form of Factor 1 and 2 scores or a Total score, resulted in a significant improvement in the prediction of outcome variables.
Incremental Validity of the SOBBS Factors Relative to the SOQ and OBC-Surveillance Scale.
Note. Degrees of freedom for comparisons: Model 1 (1, 533); Model 2 (1, 532); Model 3 (2, 530). A Bonferroni correction was applied given the number of statistical tests performed (.05/5). SOQ = Self-Objectification Questionnaire; OBC-Surv = Objectified Body Consciousness Body Surveillance Scale; SOBBS Factor 1 = Self-Objectification Beliefs and Behaviors Scale–Observer’s Perspective; SOBBS Factor 2 = Self-Objectification Beliefs and Behaviors Scale–Body as Self; OBC-Shame = Objectified Body Consciousness Body Shame Scale; MBSRQ = Multidimensional Body-Self Relations Questionnaire Appearance Evaluation Scale; AAS = Appearance Anxiety Scale; Zung = Zung Self-Rating Depression Scale.
*p < .01.
Incremental Validity of SOBBS Total Scores Relative to the SOQ and OBC-Surveillance Scale.
Note. Degrees of freedom for comparisons: Model 1 (1, 533); Model 2 (1, 532); Model 3 (1, 531). A Bonferroni correction was applied given the number of statistical tests performed (.05/5). SOQ = Self-Objectification Questionnaire; OBC-Surv = Objectified Body Consciousness Body Surveillance Scale; SOBBS = Self-Objectification Beliefs and Behaviors Scale Total Score; OBC-Shame = Objectified Body Consciousness Body Shame Scale; MBSRQ = Multidimensional Body-Self Relations Questionnaire Appearance Evaluation Scale; AAS = Appearance Anxiety Scale; EAT-26 = Eating Attitudes Test-26; Zung = Zung Self-Rating Depression Scale.
*p < .01.
Participants in the test–retest sample (n = 55) completed Time 2 measures 14–33 days after Time 1 (M = 17.78, SD = 4.95). According to the criteria published by Cicchetti (1994), the intraclass correlation coefficient, ICC (3, 1) indicated that the temporal stability of SOBBS Factor 1 (.89), SOBBS Factor 2 (.73), and the SOBBS Total Scale (.89) were excellent. The test–retest reliability of the SOQ was good (.67) and the test–retest reliability of OBC-Surveillance was excellent (.81; Cicchetti, 1994).
Study 2
The purpose of Study 2 was to reevaluate the factor structure of the SOBBS via confimatory factor analysis (CFA) and to further evaluate the measure’s validity, including its relations with sexual functioning, public and private body consciousness, public and private self-consciousness, self-esteem, and socially desirable responding. Specific hypotheses for Study 2 were as follows: (1) Consistent with the tenets of objectification theory, SOBBS scores will be negatively correlated with sexual functioning, and SOBBS scores will predict sexual functioning above and beyond existing measures of self-objectification. (2) SOBBS scores will be positively correlated with public body consciousness and negatively correlated with private body consciousness. (3) SOBBS scores will be positively correlated with public and private self-consciousness. (4) SOBBS scores will be negatively correlated with self-esteem. (5) SOBBS scores will be uncorrelated with socially desirable responding.
Method
Participants
We analyzed data from 175 women recruited from the United States via Amazon Mechanical Turk. Participants ranged in age from 19 to 30 (M = 25.85, SD = 3.08) and had a mean BMI of 26.35 (SD = 7.15, range = 14.78–56.49). One hundred thirty (74.29%) participants identified as Caucasian, 15 as Black (8.57%), 15 as Hispanic (8.57%), 6 as Asian/Pacific Islander (3.43%), 3 as American Indian/Alaska Native (1.71%), and 3 as Other (3.43%). Seventy-two participants were employed full-time (41.14%), 40 participants were students who may be employed part-time (22.86%), 32 were employed part-time (18.29%), 25 were unemployed (14.29%), and 6 listed some other employment status (3.43%). Nearly three quarters of participants identified as heterosexual (n = 127, 72.57%), 18.86% as bisexual (n = 33), 4.57% as gay or lesbian (n = 8), and 7.00% as other (n = 7). Most participants reported that they were in a committed relationship of some kind (78.86%, n = 138), with others identifying as single (20.57%, n = 36), and one participant describing their relationship status as other (.57%). The CFA model evaluated in this study had 76 degrees of freedom. The sample size of 175 was sufficient for the analyses based on sample size guidelines published by MacCallum, Browne, and Sugawara (1996).
Measures
Participants completed the same measures of self-objectification as in Study 1. Internal consistency estimates of SOBBS Factor 1 and Factor 2 were .91 and. 92, respectively, and internal consistency for the full scale was .92. Internal consistency of OBC-Surveillance was .86. Participants also completed a number of other measures described below.
Sexual functioning was assessed using the PROMIS Sexual Function and Satisfaction Measures Brief Profile (PSxFBP; Flynn et al., 2013). The scale contains 11 items that assess several domains of sexual functioning over the past 30 days: interest in sexual activity, orgasm, global satisfaction with sex life, vaginal discomfort, and lubrication. Anchors vary depending upon the individual item, but higher scores reflect better sexual functioning. Scores on PSxFBP domains correlate well with the Female Sexual Function Index (Rosen et al., 2000), a lengthier established measure of sexual functioning for women, and test–retest coefficients are greater than .65 for all domains. For the purposes of this study, raw scores for each domain and a single raw total score were used. Cronbach’s α for the full scale was .82.
The Public and Private Body Consciousness Scales of Miller, Murphy, and Buss’s (1981) Body Consciousness Questionnaire were used to assess participants’ awareness of observable aspects of the body (public body consciousness; 6 items) and awareness of internal bodily sensations (private body consciousness; 5 items). Items are rated on a scale of 0 = extremely uncharacteristic to 4 = extremely characteristic, and sum scores for each scale are calculated. Test–retest reliability for public body consciousness and private body consciousness were .73 and .69, respectively. In the present study, Cronbach’s α was .71 for public body consciousness and .75 for private body consciousness.
The Public Self-Consciousness and Private Self-Consciousness Scales of Fenigstein, Scheier, and Buss’s (1975) Self-Consciousness Scale were used to measure general self-focused attention. The Public Self-Consciousness Scale contains 7 items that assess the respondent’s “awareness of the self as a social object that has an effect on others” (Fenigstein et al., 1975, p. 523). The Private Self-Consciousness Scale contains 10 items that assess a respondent’s concern with, or awareness of, their inner thoughts or feelings. Items are rated on a scale of 0 = extremely uncharacteristic to 4 = extremely characteristic. Two items on the private self-consciousness scale are reverse-scored, and the sum of items on each scale is obtained. Test–retest reliability was .84 and .79 for public and private self-consciousness, respectively. Cronbach’s α in the present study was .82 for public self-consciousness and .72 for private self-consciousness.
The Rosenberg Self-Esteem Scale (RSES; Rosenberg, 1965) was used to measure participants’ global self-evaluations. The RSES contains 10 items rated on a 4-point Likert-type scale ranging from 0 = strongly disagree to 3 = strongly agree. After reverse scoring 5 items, a sum score for all items is obtained. Scores range from 0 to 30, with higher scores reflecting more positive self-evaluations. The measure has demonstrated satisfactory reliability and validity with adolescent and adult samples (Blascovich & Tomaka, 1991). Internal consistency was .93 in the present study.
We used the Impression Management Scale of the Balanced Inventory of Desirable Responding (BIDR; Paulhus, 1991) to assess the extent to which participants may be representing themselves in a more favorable light. The scale contains 20 items rated on a 7-point scale (1 = not true to 7 = very true). Ten of the scale’s items (e.g., “I sometimes tell lies if I have to”) are reverse-scored, and then participants are given one point for each item with a response of 6 or 7 and a sum is obtained. Higher scores reflect higher levels of socially desirable responding. The measure has demonstrated satisfactory reliability and validity in college student samples (Paulhus, 1991). Cronbach’s α for the present study was .81.
Procedure
Following institutional review board approval, we recruited participants using Amazon Mechanical Turk. Women between the age of 18 and 30, who spoke English, lived in the United States, and who were sexually active (defined as having engaged in any form of sexual activity in the last 30 days; e.g., caressing; foreplay; masturbation; and/or oral, anal, or vaginal sex), were eligible to participate in the study. Participants completed all study measures anonymously online and were compensated US$0.50 for study completion. Measures were presented in counterbalanced order via SurveyGizmo. Upon completion of the online questionnaire, we provided participants with psychoeducational information about self-objectification and contact information for mental health resources that are available nationally.
Results
Validation of the SOBBS’s Factor Structure
CFA was conducted using AMOS Version 22 (IBM, 2013a). Model fit was evaluated using the recommendations of Hu and Bentler (1999), with good model fit being reflected by comparative fit index (CFI) values of around .95 or greater, standardized root mean square residual (SRMR) values of around .08 or lower, and root mean square error of approximation (RMSEA) values of around .06 or lower. Each of the scale’s 14 items were specified to load only their latent factor, and the relation between the two latent factors (observer’s perspective and body as self) was also estimated. The model provided excellent fit for the data, CFI = .97, SRMR = .06, RMSEA = .06, χ2(76) = 122.19, p = .001. Factor loadings appear in Table 1.
Hypotheses Tested
Descriptive statistics and correlations among Study 2 variables appear in Table 5. Consistent with Hypothesis 1, SOBBS Factor 1, Factor 2, and Total scores were negatively correlated with overall sexual functioning. SOQ scores and OBC-Surveillance scores were uncorrelated with sexual functioning across all domains. Given the absence of correlations between SOQ or OBC-Surveillance scores and sexual functioning, the incremental validity of the SOBBS relative to these measures was not explored.
Descriptive Statistics and Correlations Among Variables for Study 2.
Note. SOBBS Factor 1 = Self-Objectification Beliefs and Behaviors Scale–Observer’s Perspective; SOBBS Factor 2 = Self-Objectification Beliefs and Behaviors Scale–Body as Self; SOBBS Total = Self-Objectification Beliefs and Behaviors Scale–Total Score; SOQ = Self-Objectification Questionnaire; OBC-Surv = Objectified Body Consciousness Body Surveillance Scale; PSxFBP = PROMIS Sexual Function and Satisfaction Measures Brief Profile; Private BC = Private Body Consciousness; Public BC = Public Body Consciousness; Private SC = Private Self-Consciousness; Public SC = Public Self-Consciousness; RSES = Rosenberg Self-Esteem Scale; BIDR = Balanced Inventory of Desirable Responding.
*p < .05. **p < .01. ***p < .001.
Hypothesis 2 was partially supported. The SOBBS Total score was positively correlated with public body consciousness and, unexpectedly, was uncorrelated with private body consciousness. This same pattern was observed for SOBBS Factor 1. SOBBS Factor 2 exhibited a small negative correlation with public body consciousness and was uncorrelated with private body consciousness. By comparison, SOQ scores were uncorrelated with public and private body consciousness, and OBC-Surveillance scores were positively correlated with public body consciousness and uncorrelated with private body consciousness.
The SOBBS Total score was positively correlated with public and private self-consciousness; this pattern also held true for SOBBS Factor 1. SOBBS Factor 2 was uncorrelated with private self-consciousness but was positively correlated with public self-consciousness. SOQ and OBC-Surveillance scores were positively correlated with both public and private self-consciousness. Consistent with Hypothesis 4, SOBBS Factor and Total scores were negatively correlated with self-esteem. SOQ scores were uncorrelated with self-esteem and OBC-Surveillance scores were negatively correlated with self-esteem. Contrary to Hypothesis 5, scores on all measures of self-objectification (including the SOBBS Factor 1, Factor 2, and Total scores, the SOQ, and OBC-Surveillance) exhibited small but statistically significant negative correlations with socially desirable responding.
Discussion
The primary goal of the current research was to explore the construct of self-objectification and to develop a new measure of self-objectification that addressed some of the problems with existing measures, including data loss and concerns about validity (Calogero, 2011). The resulting measure, the SOBBS, consists of 14 items which loaded onto two factors. The first factor represents the internalization of the observer’s perspective on the body, and the second represents treating the body as if it is capable of representing the self, including placing greater emphasis on what the body looks like than on what it can do or how a person thinks or feels. This two-factor structure of the measure integrates the existing conceptual definitions of self-objectification and offers a new tool for its measurement, answering the call of Moradi (2010) and Calogero (2011) for clarification of the construct and refinement of assessment.
The two studies presented in this article provide preliminary evidence regarding the psychometric properties of the SOBBS. In Study 1, the SOBBS demonstrated excellent internal consistency and test–retest reliability. With regard to validity, the SOBBS Factor 1, Factor 2, and Total scores were positively correlated with interpersonal sexual objectification, body shame, appearance anxiety, disordered eating, and depression; negatively correlated with body satisfaction; and uncorrelated with narcissism. The SOBBS also allowed for substantial improvements in the prediction of body image, disordered eating, and depressive symptoms beyond the existing measures of self-objectification. Study 2 provided additional evidence regarding the SOBBS’s validity, as SOBBS Total scores and Factor 1 scores were positively correlated with public body consciousness, private self-consciousness, and public self-consciousness and negatively correlated with self-esteem and sexual functioning. This pattern held true for SOBBS Factor 2 scores; however, SOBBS Factor 2 was uncorrelated with private self-consciousness. Reconsidering the definition of private self-consciousness, which relates to awareness of one’s own thoughts and feelings, the absence of a relation makes sense, given that Factor 2 reflects valuing of appearance over other attributes (including thoughts and feelings) but not necessarily awareness of them. An unexpected finding in Study 2 was that self-objectification was uncorrelated with private body consciousness regardless of how self-objectification was measured. This finding points to the need to clarify the difference between private body consciousness and interoceptive awareness, which were previously thought to be similar and were expected to be negatively correlated with self-objectification. Despite a few unexpected findings, the results of Study 1 and Study 2 provide evidence for the use of the SOBBS as a measure of self-objectification in future research.
The current study provides stronger evidence for the reliability and validity of the SOBBS than for the SOQ. Internal consistency reliability cannot be calculated for the SOQ; the measure’s test–retest reliability was lower than SOBBS Factor 1, Factor 2, and Total scores; and SOQ scores were not consistently correlated with those variables theoretically related to self-objectification. These results are consistent with the concerns summarized by Calogero (2011) about the reliability and validity of the measure. Thus, from a psychometric perspective, the SOBBS is an improvement over the SOQ.
From a practical perspective, there was significant data loss in Study 1 due to incorrect completion of the SOQ, a problem that has been documented elsewhere (e.g., Lindner et al., 2012; Myers & Crowther, 2007). Missing data were not a problem in Study 2. This is likely due to adjustments to the format of the survey in online software (i.e., setting the question so that each rank could only be used once) and, due to the nature of Amazon Mechanical Turk’s compensation structure, where workers are aware that their work can be rejected if they do not complete the task. Researchers should be aware of the potential for missing data in their use of the SOQ and should (a) plan for data loss when planning sample size, (b) be attentive to structuring of the questionnaire during administration, or (c) select a different measure like the SOBBS that is less subject to data loss.
Across Studies 1 and 2, the SOBBS performed as well or better than OBC-Surveillance. The biggest area of overlap between the two measures is the consideration of SOBBS Factor 1 and OBC-Surveillance. These two indices were highly correlated across both studies. There were low-to-moderate correlations between SOBBS Factor 2 and OBC-Surveillance and moderate-to-high correlations between SOBBS Total scores and OBC Surveillance. Both measures were correlated with most theoretically related variables in the expected direction and the magnitude of the correlation coefficients was generally similar, with the exception of sexual functioning, where SOBBS scores were negatively correlated with sexual functioning and OBC-Surveillance scores were not. OBC-Surveillance has been related to lower sexual self-esteem and higher self-consciousness during sex in previous studies but not with more physiologically oriented indicators of sexual functioning (e.g., Calogero & Thompson, 2009; Steer & Tiggemann, 2008). The measure of sexual functioning used in the present study (i.e., attitudinal dimensions, like interest in and satisfaction with sexual activity, as well as physiological dimensions, like orgasm, vaginal discomfort, and lubrication) is consistent with the types of sexual dysfunction described by Fredrickson and Roberts (1997) and, thus, the relation between SOBBS scores and scores on the PSxFBP supports convergent validity in the current sample. Examination of incremental validity of the SOBBS relative to OBC-Surveillance in the prediction of body shame, appearance anxiety, body satisfaction, disordered eating, and depression showed that the two SOBBS factors offered improvements over OBC-Surveillance. In the prediction of body image variables, OBC-Surveillance remained a statistically significant predictor in the second step of hierarchical linear regressions, suggesting that the SOBBS and OBC-Surveillance each measure something unique. It may be that OBC-Surveillance is more representative of the kind of habitual monitoring said to arise as a result of self-objectification rather than of self-objectification itself, as some have previously argued (e.g., Steer & Tiggemann, 2008; Tiggemann & Kuring, 2004).
Given the current state of the literature, we recommend using the SOBBS when the goal is to measure the two aspects of self-objectification represented by the SOBBS factors and use OBC-Surveillance when the goal is to measure habitual self-monitoring; this follows Calogero’s (2011) recommendation to be specific about terminology in communicating study results. Further exploration of the relation between the SOBBS and OBC-Surveillance and clarification of the construct represented by OBC-Surveillance are needed. When utilizing the SOBBS, researchers who want to measure a specific aspect of self-objectification represented by one of the factor scores may wish to consider factor scores separately, whereas researchers interested in a holistic measure of self-objectification that integrates both components may wish to use all the items.
Theoretical Implications
The development of the SOBBS facilitates additional study in the area of objectification theory. First, the process of developing the SOBBS extends discussion about how to define self-objectification. This study offered an opportunity to explore how the definitions in the literature fit together and which aspects are most essential to the construct. The two-factor structure of the measure is consistent with the definition of self-objectification offered by Fredrickson, Hendler, Nilsen, O’Barr, and Roberts (2011) in their retrospective on the development of objectification theory; the SOBBS includes the internalization of an observer’s perspective on the body plus a valuing of the appearance of the body over physical abilities and what one thinks and feels.
The two-factor solution is different from the hypothesized three-factor structure. The hypothesized three-factor structure separated out the valuing of appearance over competence (the operationalization underlying the SOQ) from the valuing of appearance over other aspects of who one is as a person. The two-factor structure could have emerged for several reasons. From a theoretical perspective, it is possible that the appearance versus competence distinction is simply part of valuing the physical appearance of the body over other non-observable, non-appearance-related aspects of the self. This is consistent with the content of items on Factor 2 in the final version of the questionnaire, which includes valuing physical appearance over physical abilities, personality, and intellect. It is also possible that physical ability gets confounded with appearance in the way that some participants think about their bodies. For instance, someone who is physically strong likely has well-defined muscles and a toned, athletic appearance. Last, we also acknowledge that the item pool was unbalanced in that we started with a small number of items assessing the appearance versus competence dimension. While the two-factor solution could have been a statistical artifact for this reason, four of the five appearance versus competence items were consistently cross loaded on multiple factors in each of the models we tested. When we evaluated models in which three factors were retained (as was suggested by parallel analysis alone), these items did not form a separate appearance versus competence factor.
Practice Implications
Continued study of self-objectification and its relation to mental health risks has important clinical implications. Developing the SOBBS provided us with an opportunity to determine which behaviors and attitudes are the essential elements of self-objectification. This improved our understanding of potential targets for intervention, including helping people to become more aware of their tendencies to think about how others view their bodies and to cultivate an appreciation for aspects of themselves beyond physical appearance, as has been suggested by Calogero and Tylka (2014) and Tylka and Augustus-Horvath (2011). Improved measurement of self-objectification may also aid in the identification of at-risk groups or individuals and, perhaps most important, objectification research highlights the relation between body image and aspects of mental health beyond eating disorders. One of the criticisms of eating disorder prevention efforts is that they can be resource intensive but beneficial to a relatively small proportion of the population (Levine & Smolak, 2006). Effectively targeting self-objectification and risk factors for the development of negative body image in prevention efforts may decrease the risk of eating disorders as well as other adverse mental health outcomes. New tools for measuring self-objectification may facilitate evaluation of these prevention efforts, particularly given that the SOBBS predicted mental health outcomes better than existing measures of self-objectification.
Limitations and Future Directions
Although there are many strengths of the study, including the development of a scale based on a broader conceptualization of self-objectification, there are limitations that result in the need for further study. First, the studies presented here relied on data from female participants between the ages of 18 and 30, most of whom identified as Caucasian. Given that objectification theory is a feminist framework and we sought to operationalize self-objectification, it made sense to do so with a sample of women. Given evidence that self-objectification varies across the life span (e.g., Tiggemann & Lynch, 2001), we chose to focus only on young adult women. Consequently, it is important to acknowledge that the results cannot be generalized beyond similar samples. Thus, an important next step is to examine the psychometric properties of this scale (particularly in relation to existing measures of self-objectification) in samples of racial and ethnic minority women, lesbian women, women at different ages, and men.
The SOBBS allows researchers to measure two aspects of self-objectification that correspond with conceptual definitions of self-objectification and its precursor, sexual objectification: the internalization of an observer’s perspective on the body and treatment of the body as if it can represent the self. Given that a separate appearance versus competence domain did not emerge, it will be important to further explore the distinction between physical appearance and physical competence and its place in understanding self-objectification and other aspects of body image. Further, it will be important for future studies to consider how internalization of cultural ideals about appearance relate to self-objectification. In the earliest work on objectification theory, body shame was defined as feeling bad about oneself as a result of realizing that one does not live up to cultural standards of beauty (Noll & Fredrickson, 1998). Informed by this definition, and other models of body image development, Moradi (2010) has suggested that internalization of cultural ideals about appearance should be considered when studying self-objectification.
Future psychometric evaluations of the SOBBS should carefully attend to the issue of discriminant validity. In Study 1, SOBBS Factor 1, Factor 2, and Total scores were uncorrelated with narcissism, providing evidence for discriminant validity. However, we acknowledge that the internal consistency of this scale (α = .67) was somewhat low. To address this issue, we explored the relation between SOBBS scores and socially desirable responses in Study 2. Unexpectedly, SOBBS Factor 1, Factor 2 and Total scores and the SOQ and OBC-Surveillance had small but statistically significant negative correlations with social desirability. While unexpected, this information may be interpreted in light of the relations among self-objectification, self-esteem, and depression. That is, this finding may have emerged because those high in self-objectification are lower in self-esteem and higher in depression, and thus not particularly motivated to portray themselves in a positive light. An alternate interpretation is that high face validity of the set of items on the BIDR may alert participants to the fact that it is a social desirability measure and, as such, they may downplay socially desirable traits and behaviors. However, the mean score on the measure was similar to mean scores obtained in other samples (e.g., Paulhus, 1991), suggesting that the former explanation is more plausible than the latter.
Another area for exploration in future research relates to quantifying self-objectification in various populations. Because the SOQ and OBC-Surveillance have been used in research for nearly two decades, we now know that women typically have positive scores on the SOQ, while men typically have negative scores (e.g., Calogero, 2009; Noll & Fredrickson, 1998). Women’s scores typically fall above the midpoint on OBC-Surveillance (e.g., Lindner et al., 2012; McKinley & Hyde, 1996). SOBBS Factor 1 scores fell just above the midpoint in Studies 1 and 2, and Factor 2 and Total scores fell below the midpoint. In the absence of other samples with which to compare, it is difficult to draw conclusions about the extent to which this reflects characteristics unique to the two samples or whether young adult women generally report what would be considered a low-to-moderate level of valuing appearance over other attributes. It is important to note, however, that in both samples, SOBBS Factor 1, Factor 2, and Total scores spanned the full range of possible scores (from 1 to 5).
Conclusions
Although not without its limitations, this work makes a valuable contribution to the literature on self-objectification. Our primary contribution is the development of an additional tool for measuring self-objectification that, in this preliminary investigation, demonstrated sufficient reliability and validity and assesses a broader range of thoughts, feelings, and behaviors related to self-objectification than existing instruments. This broadening of the scope of assessment led to a meaningful improvement in the ability to predict mental health outcomes. A second, related contribution of this study is the exploration, at both a theoretical and empirical level, of the meaning of self-objectification, a process that we hope researchers continue as they use the scale in other studies.
Footnotes
Authors’ Note
Data for Study 1 were collected as part of the first author’s dissertation research at the University of Central Florida, Orlando, FL. Data can be obtained by contacting the first author.
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.