Introduction

Body dissatisfaction is highly prevalent among women both with and without eating disorders (ED) [1] and is an important risk factor for the development and maintenance of ED [2]. According to the tripartite influence model of body image development, three sociocultural influence factors—peers, parents and media—have an effect on body dissatisfaction. Moreover, the model proposes that this effect is exerted via the two main mechanisms of internalization of the thin ideal and appearance comparison [3]. Studies have supported the model as a useful framework [4] and in particular, a meta-analysis confirmed the impact of appearance-focused social comparison processes on the evaluation of one’s own body [5]. As a consequence, the perception and evaluation of one’s own body depend on attitudes not only towards one’s own body, but also towards others’ bodies [6].

Previous research has directly compared the evaluation of one’s own and others’ bodies. Donaghue and Smith [7] found that while women’s ratings of their own body attractiveness were higher than ratings of their body attractiveness made by others, they rated their own body with more body weight than others did, indicating self-serving attractiveness ratings and self-deprecating body size ratings. Further studies examined the influence of the presence of ED symptoms on these evaluations. Alleva et al. [8] found that women with high levels of ED symptoms rated others’ bodies as thinner than their own bodies, but as equally attractive, while women without ED symptoms rated others’ bodies as equally thin, but as less attractive than their own bodies. Likewise, in a study comparing women’s own body attractiveness ratings with ratings of them made by a panel [9], women without ED symptoms rated themselves as more attractive than did the panel. The panel rated the bodies of women with high levels of ED symptoms as less attractive than those without ED symptoms, and its evaluations were mostly in agreement with the eating-symptomatic women’s own evaluations. These results suggest that women with high levels of ED symptoms make more self-deprecating evaluations or are more objective about their body attractiveness because their bodies might actually be less attractive than other women’s bodies.

These studies suggest that women might apply different sets of principles to themselves than to others, thus engaging in double standards. The use of double standards means that two (or more) persons who objectively possess an attribute to the same extent are rated differently due to the use of different requirements to interpret that evidence [10]. Typically, the evaluated persons differ in a specific characteristic, such as age and gender [11] or skin color [12], which leads to the use of different requirements. Previous studies used the evaluation of other persons as an indicator for objective attractiveness of a body and compared this evaluation with the evaluation of the body owner [7,8,9]. As attractiveness is not a purely objective attribute that could be measured as performance, but is instead subject to personal tastes [13], it would be important to examine whether women apply their subjective standards for body attractiveness differently depending on identity. This would have implications for a woman’s body satisfaction, as it would lead to a greater discrepancy between the perceived body attractiveness of one’s own and other bodies. One previous study compared women’s subjective ratings of their own bodies with those of other females’ bodies [8]. However, as one’s own and other females’ bodies differ in objective physical characteristics and familiarity, this study did not allow the influence of identity to be disentangled from the differences in the objective characteristics of the bodies. First evidence that women’s standards for body rating change as a function of who is being evaluated was provided by a study in which women with ED and non-clinical controls watched a film presenting a woman whose body changed from thin to obese. The participants were instructed to define at which point the body was thin, normal-weight, fat, and obese, once while imagining that it was their own body and once while imagining that it was another female’s body. The results indicated that both groups made stricter ratings in the case of their own body and that women with ED made stricter ratings than non-clinical controls [14]. However, the fact that the identity manipulation was based solely on different imagination instructions and not on adopted stimulus material could have hampered identification with the body.

The present study sought to fill this gap by investigating whether double standards have an influence on women’s ratings of body attractiveness, body fat and muscle mass ratings depending on identification with the presented body. In contrast to previous studies, we did not provide the participants with pictures of their own and another body, but instead presented identical bodies in various sizes and poses, once with the subject’s own face and once with another female’s face to hold the objective characteristics of the body constant. Following Smeets [14], we expected women to rate bodies with their own face more negatively than the same bodies with another female’s face. Furthermore, based on previous research [8, 9], we hypothesized that the higher the ED symptoms, the more negatively the bodies presented with one’s own face would be rated compared to the same bodies shown with another female’s face.

Method

Participants

Inclusion criteria were female sex, age between 18 and 30 years, body mass index (BMI) of 18.5–23.5 kg/m2 and the absence of a mental disorder, which were assessed by an online survey implemented in Unipark (QuestBack GmbH; Köln, Germany). We chose these BMI criteria to ensure that all subjects’ own bodies bore the closest resemblance to our presented average-weight body. All participants had to indicate that they had no mental disorder and were not currently being treated for a mental disorder. All women who were interested in participation were provided with a link to the online survey and an appointment for the measurement. We measured N = 116 Caucasian women from a German student population. Eleven women were excluded as they did not fulfill the BMI criterion and one woman was excluded as she was undergoing psychotherapy. Thus, N = 104 participants were included in the analysis. The sample had an average age of 21.29 years (SD = 2.84) and an average BMI of 20.79 kg/m2 (SD = 1.25).

Materials and procedure

ED pathology was measured with the Eating Disorder Examination Questionnaire (EDE-Q) [15, 16]. The EDE-Q consists of 22 items assessing eating disorder symptoms on the four subscales Eating concern, Weight concern, Shape concern and Restraint. Items refer to the last 28 days and are measured on a seven-point Likert scale from no days/not at all (0) to every day/markedly (6). Various studies found good to excellent internal consistencies for the subscales, with alphas ranging from 0.73 to 0.93 in community samples of adult women [17]. In the sample of the current study, alphas ranged from 0.79 to 0.89, underpinning good reliability.

In total, we presented 25 body stimuli with the participant’s face and the identical 25 body stimuli with another female’s face in a randomized order. Five types of bodies with different builds, i.e., thin, average-weight, overweight, athletic, and hypermuscular, were shown in five different poses. All body stimuli were created with the software DAZ studio 4.6 (DAZ Productions, Inc.; Salt Lake City, Utah, USA). To manipulate identification, we used the face as the most distinctive self-referential stimulus, as is often applied in self-identification research [18]. To create the own face pictures, we photographed the heads of all participants in frontal view and with a neutral facial expression. We then cut out the face including hair and mirrored it to create the view of looking in a mirror using the software GIMP. Using MATLAB 2008 (MathWorks; Natick, Massachusetts, USA), we placed the faces on the bodies and grey-scaled the whole picture. For the other female’s face, we selected an averagely attractive face (labeled “neutral_y_f_25”) [19] from the CAL/PAL Face Database [20]. To maximize the coherence of bodies and faces, we applied a comic filter by AKVIS ArtWork 8.1 (AKVIS; Perm, Russia), which reduced the tone of the picture to create a homogenous image similar to a cartoon (see Fig. 1 for body stimuli).

Fig. 1
figure 1

Demonstration of face manipulation (the left face is that of one author, representing the participant’s face; the right face is the other female’s face) on the thin, average-weight, overweight, athletic and hypermuscular body (a) and on the average-weight body in the five different poses (b). Arrows indicate that bodies were presented with the own face as well as with the other female’s face on top of them

Full size image

Before the assessment, participants were presented with written instructions about the rating scales and the procedure, i.e., they were told that their own face would be depicted on the bodies, which should help them to imagine that these were their own bodies. After each 3-s presentation of each 1400  ×  1050 px body stimulus, participants rated it according to body attractiveness (not face attractiveness), body fat and muscle mass. All ratings were made on a nine-point Likert scale anchored by the labels very unattractive–very attractive, very little body fat–very much body fat, very little muscle mass–very much muscle mass. As stimulus validation, at the end, we asked the subjects one question to judge how coherent the body stimuli looked overall, thus how well bodies and heads match (very poor match–very good match). Presentation and ratings were provided by the experimental software E-Prime® 2.0 (Psychology Software Tools, Inc.; Sharpsburg, Pennsylvania, USA). The study protocol was approved by the local ethics committee of the University Osnabrück. Participants gave written informed consent and were reimbursed with student participant credit or 7 Euros.

Statistical analysis

We ran 2 × 5 repeated-measures MANOVA with the factors Identity (Self, Other) and Build (Thin, Average-weight, Overweight, Athletic, Hypermuscular) using SPSS Statistics (IBM; Armonk, New York, USA). The dependent variables were ratings of body attractiveness, body fat and muscle mass. For post hoc ANOVAs, we applied the Greenhouse–Geisser correction by default. For post hoc t tests, the Bonferroni correction was used. As effect size measures, we report eta-squared η2 for analysis of variance and Cohen’s d for t tests [21, 22]. To test whether body evaluation and double standards are associated with ED pathology, we calculated correlations between the four EDE-Q subscale scores and the averaged body evaluations over the two faces and between the four EDE-Q subscale scores and the differences between the evaluation of each body with one’s own and the other female’s face. We used these difference scores as a measure of the extent of the application of double standards. A large negative difference between the ratings of a body depending on the face reveals a higher rating of the dependent variable for the other-face condition and a large positive difference between the rating of a body depending on the face reveals a higher rating of the dependent variable for the own face condition. The greater these differences, the greater were the extent of the application of double standards.

Results

The average rating of the coherence of the body stimuli was M = 5.76 (SD = 1.92). Means and standard deviations of body evaluations are presented in Table 1 and Fig. 2. The MANOVA revealed a main effect of the factor Build, Pillai’s trace = 0.98, F(12, 92) = 419.47, p < .001. Post hoc ANOVAs yielded significant effects for body attractiveness ratings, F(2.55, 263.06) = 164.69, p < .001, η2 = .567, body fat ratings, F(3.23, 332.30) = 1477.76, p < .001, η2 = .914, and muscle mass ratings, F(3.02, 310.99) = 1196.42, p < .001, η2 = .906.

Table 1 Means, standard deviations, t test results and Cohen’s d for the five types of body dependent on the face placed on the bodies: Self vs. Other
Full size table
Fig. 2
figure 2

Ratings of body attractiveness (a), body fat (b) and muscle mass (c) dependent on the face placed on the bodies (self vs. other) and the five different builds. Error bars represent standard errors. Asterisks highlight significant post hoc discrepancies between the different builds and between the different faces in the case of the overweight body

Full size image

The average-weight body was rated as most attractive (M = 6.70, SD = 1.11) followed by the athletic (M = 3.88, SD = 1.33), the thin (M = 3.72, SD = 1.68), the hypermuscular (M = 3.37, SD = 1.29) and the overweight body (M = 2.76, SD = 1.02). All bodies differed significantly in body attractiveness rating, all p < .001, except for the differences between the thin and athletic and the thin and hypermuscular body, both p > .083. The overweight body was rated with most body fat (M = 7.41, SD = 0.69) followed by the average-weight (M = 3.72, SD = 0.87), the athletic (M = 2.13, SD = 0.73), the hypermuscular (M = 1.91, SD = 0.65) and the thin body (M = 1.76, SD = 0.67). All bodies differed significantly in body fat rating, all p < .001, except for the difference between the thin and hypermuscular body, p = .059. The hypermuscular body was rated with most muscle mass (M = 8.18, SD = 0.51) followed by the athletic (M = 7.81, SD = 0.57), the average-weight (M = 4.65, SD = 1.00), the overweight (M = 2.54, SD = 1.09) and the thin body (M = 2.44, SD = 1.12). Ratings for muscle mass differed significantly between the bodies, all p < .001, except for the difference between the thin and overweight body, p = .497. In line with our stimulus construction, participants validated that they perceived the bodies as overweight, average-weight, thin, muscular and hypermuscular.

The MANOVA also yielded a main effect of the factor Identity, Pillai’s trace = 0.12, F(3, 101) = 4.37, p = .006. Post hoc ANOVAs showed a significant effect for body attractiveness ratings, F(1, 103) = 9.75, p = .002, η2 = .002, but not for body fat ratings, F(1, 103) = 1.83, p = .179, η2 < .001, or muscle mass ratings, F(1, 103) = 2.51, p = .116, η2 < .001. Women rated bodies with their own face as less attractive (M = 4.00, SD = 0.77) than bodies with the other female’s face (M = 4.17, SD = 0.80).

In the MANOVA, the interaction between the factors Build and Identity was also significant, Pillai’s trace = 0.36, F(12, 92) = 5.00, p < .001. Post hoc ANOVAs showed significant effects for body attractiveness ratings, F(3.36, 345.71) = 5.74, p < .001, η2 = .003, body fat ratings, F(3.81, 392.33) = 11.07, p < .001, η2 = .002, and muscle mass ratings, F(3.62, 372.78) = 5.19, p = .001, η2 = .001. Post hoc t tests (see Table 1) revealed that participants rated the overweight body as more unattractive, higher in body fat and lower in muscle mass when it had their own face compared to another female’s face. No further significant self-other differences were found, all p > .012 (Bonferroni-corrected α = .01).

Regarding ED symptoms, our sample revealed an average score of 0.89 (SD = 1.02) on the Restraint scale, an average score of 0.53 (SD = 0.68) on the Eating concern scale, an average score of 1.17 (SD = 1.10) on the Weight Concern Scale and an average score of 1.53 (SD = 1.10) on the Shape concern scale. Correlation analysis revealed that the higher the Weight Concern Score, the more attractive subjects rated the thin body to be (r = .339, p < .001). Moreover, a higher Weight Concern Score (r = − .300, p = .002) as well as a higher Shape concern score (r = − .360, p < .001) significantly correlated with ratings of the overweight body as less attractive. After Bonferroni correction (α = .0025), the correlation of a higher Shape concern score and the rating of the overweight body as having more body fat became marginally significant (r = .293, p = .003). No further significant correlations between the EDE-Q subscale scores and the averaged body evaluations over the two faces or between the EDE-Q subscale scores and the self-other differences in body evaluation were found, all p > .0044 (Bonferroni-corrected α = .0025).

Discussion

The aim of the present study was to examine whether women apply double standards in body evaluation depending on identity and whether the extent of these double standards depends on the extent of ED symptoms. The results showed that participants judged the overweight body as less attractive and as higher in body fat and lower in muscle mass when it was presented with one’s own face compared to another female’s face, i.e., they showed self-deprecating judgments. Obesity of another woman therefore does not seem to be judged as negatively as obesity of oneself, which classifies as a double standard [10]. This finding might be explained by similar mechanisms to those contributing to interpretation biases [23], which are typically described in the context of ambiguous scenario evaluations. For example, women with higher levels of ED psychopathology were found to rate the outcome of ambiguous self-involved scenarios more negatively than women with lower levels [24]. According to the cognitive-behavioral theory of eating disorders by Williamson et al. [23], these findings might be due to the activation of a negative self-schema in people with ED, leading them to shine a more negative light on the scenarios. With regard to body judgment, the identification with a body might lead to an even stronger activation of a self-referential body schema, which in turn influences the evaluation of the presented body. In our society, overweight bodies are not just seen as unattractive, but overweight people are also stigmatized and confronted with negative stereotypes such as being lazy, unmotivated or less competent [25]. Thus, negative conditional beliefs (e.g., “If I am overweight, I will be ugly and disliked”, “I must not be overweight”) encompassed by the body schema are activated when identifying with an overweight body. In turn, the activation of this self-referential body schema might lead to a more negative and strict view of the overweight body. The rating of other women’s overweight bodies might not be so strict as obesity in other women does not lead to feared disadvantages for oneself. Correspondingly, having a higher degree of weight and shape concerns was accompanied by stricter ratings of the overweight body and more positive ratings of the thin body, which might reflect a greater condemnation of body fat. However, in contrast to our hypothesis, ED pathology did not correlate with the application of double standards. A possible explanation for this might be that most of our subjects had low EDE-Q subscale scores from 0.53 to 1.53 and low standard deviations from 0.68 to 1.10, which are actually below the scores and standard deviations of a control group from an EDE-Q evaluation study [16]. Thus, our sample had a low ED pathology and low variance in ED pathology. Samples with a higher ED pathology might conceivably give stricter ratings not only to overweight bodies, but also to other bodies when they identify themselves with these bodies.

In contrast, our subjects did not rate the non-overweight bodies differently depending on identity. Thus, in the case of the average-weight bodies, which were rated as most attractive and with which their own bodies bore the closest resemblance, the subjects applied the same standard for themselves and other women. Therefore, it might be assumed that previous findings that non-ED women rate their own bodies as more attractive than those of other people [7, 8] might not be due to a less strict standard with respect to their own bodies than to those of others (i.e., they did not see themselves “through rose-colored glasses”). Instead, they may have learned to especially like their own bodies’ idiosyncratic characteristics.

This study is the first to examine double standards in body evaluation with regard to body attractiveness, body fat and muscle mass dependent on identification with the bodies, fostered by the use of participants’ own faces. However, some limitations should also be mentioned. The fact that we presented digitally produced cartoon-like bodies might have hampered the identification with these bodies. These body stimuli were used to ensure that the bodies looked objectively the same, without idiosyncratic characteristics, to mask skin color differences between faces and bodies, and thus to present coherent body stimuli. Thus, to achieve high internal validity, we conducted a standardized artificial task. However, this artificial stimulus material might reduce the ecological validity. Furthermore, there were some small effect sizes for the effect of identity. It is important to note that identity is one of the various factors that influence body rating and could not solely explain it. As mentioned above, the present study comprised females without ED. As we did not measure women with ED diagnoses or severe body concerns, our results can therefore not be transferred to these clinical populations.

In sum, the present study provided a first hint for self-deprecating double standards for overweight bodies and no self-serving double standards in average-weight women. In light of our findings, we suggest that a body-related identity bias should be considered in the cognitive-behavioral models of body image. Williamson et al. [23] postulated that a person’s body schema, which includes beliefs and memories related to body size and shape, can be activated by body-related information, and influences cognitive biases, behavior and emotions. We presume that a woman’s body schema could contain different standards for one’s own and others’ bodies, which are activated by body-related information and influence cognitions, behavior and emotions differently. For average-weight women without ED symptoms, self-referential standards might thus be activated when imagining being or becoming overweight (e.g., when the weighing scale indicates weight gain). This, in turn, could aggravate emotions like fear or disgust and maintain negative attitudes to an overweight female body. In contrast, if they see an overweight woman, they might apply less strict standards, implying that they accept overweight in other women more readily than overweight in themselves. In women with ED symptoms and high in body dissatisfaction, it is conceivable that self-referential standards are also activated by other self-referential body information, leading to a greater discrepancy between ratings of other women’s bodies and one’s own body. It can therefore be concluded that the investigation of double standards in body evaluation, which has been neglected in body image research so far, seems to be a promising area to shed light on the development and maintenance of body image disturbance in women.