Body Appearance (Dis)satisfaction Scale applied to Brazilian university students

Objective: This study aimed to estimate the psychometric properties of the Body Area Scale (BAS) applied to a sample of Brazilian university students to support the validity and reliability of the obtained data. Methods: We performed confirmatory factor analysis (CFA) using the Comparative Fit index (CFI), Tucker-Lewis index (TLI), and Root Mean Square Error of Approximation (RMSEA). We calculated the average variance extracted (AVE), composite reliability (CR), and ordinal alpha coefficient (α). All estimates were calculated separately for women and men. A thousand students participated (women = 64.0%, mean [ M ] age = 20.92, standard deviation [ SD ] = 2.40 years; body mass index (BMI) M = 23.24, SD = 3.97 kg/m 2 ). Results: Considering the CFA results, the original unifactorial model (24 items) did not fit data for women and men. Instead, the two-factorial model with 19 items had adequate fit for male and female samples (CFI = 0.922-0.958; TLI = 0.912-0.952; RMSEA = 0.090-0.096), as well as good convergent validity (AVE = 0.536-0.668) and reliability (CR = 0.920-0.952; α = 0.916-0.948). Once BAS was proposed to evaluate satisfaction/dissatisfaction with one’s own body, and considering the reformulation of the factorial model, we proposed a new name for the instrument: Body Appearance (Dis)Satisfaction Scale (BAS-R). Conclusion: Professionals can now include the BAS-R in future protocols to evaluate satisfaction/dissatisfaction with the body in Brazilian students.


INTRODUCTION
Satisfaction/dissatisfaction with one's overall body appearance involves positive or negative self-attributions in the perception of the real versus ideal body image 1 .The satisfaction/dissatisfaction concept contributes to the evaluation of body image.
Body dissatisfaction is one characteristic that can predispose individuals to developing eating disorders or body dysmorphic disorder 2,3 .Dissatisfaction mainly occurs because of an internalized ideal body image that has been established culturally as necessary to self-perceive body adequacy 4 .The need to achieve the ideal body has been noted by other researchers in this area 1 ; however, as this ideal is often unattainable, this standard can generate dissatisfaction 5 .Body perception dissatisfaction is common among vulnerable individuals (80%), such as young people undergoing constant physical and psychological changes 5,6 .Furthermore, some studies 5,7,8 have suggested that body satisfaction and dissatisfaction can differ between females and males.
For the evaluation of satisfaction/dissatisfaction with general body appearance, it is common to administer selfrating scales containing body figures, or psychometric scales 5 .Among these, we can highlight one presented by Rosen and Ross 9 and adopted by Lerner et al. 10 .In 2009, Conti et al. 6 named this scale the Body Area Scale (BAS) and presented the BAS Portuguese version.It was developed as a 24-item instrument that evaluates individuals' satisfaction/ dissatisfaction with weight and specific parts of their body (e.g., eyes, hips, and arms).The BAS proposal corroborates with the "beauty standards" presented by Tiggermann 11 , which refer not only to weight, but also to eyes, height, legs, and skin color.Thus, despite the existence of other instruments evaluating body satisfaction, most are focused on weight and body shape 12 , while the BAS encompasses specific body areas that can be relevant for the assessment of body satisfaction/dissatisfaction.
Despite the availability of the BAS in Portuguese, its psychometric properties have not been fully presented in the literature.Conti et al. 6 reported adequate internal consistency of the BAS for a sample of Brazilian adolescents; however, the authors did not perform an analysis to verify the factorial validity.It should be highlighted that evaluating psychometric properties, including validity estimates, is important 13 .
Regarding the BAS factorial model, the original proposal is unifactorial, that is, considering all body areas simultaneously 10 to compose the concept of general satisfaction/dissatisfaction.However, it is noted that the content of the items considers body parts and face components separately (e.g., waist, arms, shoulder width versus eyes, mouth, and nose).Cash and Smolak 5 reported that satisfaction with appearance can be related to different characteristics (e.g., arms, legs, face, and hair).This difference can occur in the context of social interaction, visual attention 14 , and body exposure; thus, the evaluation of face and body parts separately may be relevant to understanding one's satisfaction/dissatisfaction.In this way, as an alternative to the unifactorial model, a two-factor model could also be investigated.
University students have been highlighted as vulnerable populations for negative body image 15 .Laus et al. 12 reported that most instruments used to assess body image in the Brazilian context were evaluated in samples of university students.BAS was evaluated only in adolescents 6,12 .It should be mentioned that the transition between high school and university is generally marked by new challenges as the academic demands and socialization needs.At this stage, individuals may have inappropriate eating behaviors (e.g., fast-food overconsumption and excessive alcohol intake) 15 that can influence body image.This fact associated with the exposition of an individual to "beauty standards" 11 may trigger feelings of inadequacy, possibly resulting in body dissatisfaction.
Thus, the aim of this study was to investigate the psychometric properties of the BAS Portuguese version when applied to a sample of Brazilian university students.

Study design
This was a cross-sectional observational study designed with a non-probabilistic convenience sample.Initially, we calculated the minimum sample size under the assumption of 5-10 individuals per parameter 16 of the BAS original model 10 .As the scale applied was originally developed as a unifactorial model composed of 24 items (48 parameters = 24 items + 24 errors), the minimum sample size was 240-480 individuals.Accounting for a dropout rate of 20%, we corrected the minimum sample size for participant recruitment to be 300-600 participants.

Participants
We invited students from a public university (São Paulo, Brazil) to participate in the study.Those who completed all scale items and signed the informed consent form were 18-30 years old.Participants provided data on sex, age, major area of the course (human sciences: public administration, economic sciences, social sciences, letters, education; exact sciences: bioprocess engineering and biotechnology; health/biological sciences: pharmacy), year of study, weight, and height.In addition, we estimated the participants' economic level using the Brazilian Economic Classification Criteria 17 .Self-reported measures of weight and height were used to calculate body mass index (BMI) and to classify anthropometric nutritional status 18 .A total of 1,159 students participated in the study, and 159 (13.7%) were excluded because they did not complete all the items of the instrument, leaving a final participant total of 1,000 students.The students completed the questionnaire during class hours with the teachers' authorization to collect the data.The researchers emphasized in class that participation was voluntary, and we included only those who gave informed consent.The study protocol was approved by the Human Research Ethics Committee of the School of Pharmaceutical Sciences of São Paulo State University (CAAE: 63553516.4.0000.5426).

Measure
We estimated the students' satisfaction and dissatisfaction with the general appearance of their body using the Portuguese version of the BAS 6 .The BAS was originally proposed in English as a unifactorial model with 24 items related to specific parts of the body and the body's general appearance 10 .The instrument was suited for both men and women 10 .A 5-point Likert-type scale was used to respond to the items (1 = very satisfied, 5 = very dissatisfied).Conti et al. 6 found Cronbach's alpha (α) = 0.88 and 0.90 for female and male samples, respectively.In addition, girls completed BAS in 2.1 (standard deviation [SD] = 0.8) minutes and boys in 2.0 (SD = 0.8) minutes.

Psychometric analysis
The psychometric sensitivity of items was investigated using summary measures as mean, median, and standard deviation, and measures of form of distribution as skewness and kurtosis.Adequacy of psychometric sensitivity was considered when absolute values of skewness were < 3 and kurtosis < 7 13 .The samples of women and men were evaluated separately.
The original psychometric properties of the scale were based on Lerner et al.'s 10 unifactorial model.We conducted confirmatory factor analysis (CFA) with the robust weighted least squares estimation method adjusted for mean and variance and employed the Comparative Fit index (CFI), the Tucker-Lewis index (TLI), and the Root Mean Square Error of Approximation (RMSEA) with a 90% confidence interval (90% CI).The fit of the model to the data was considered acceptable when factor loadings (λ) were ≥ 0.50, CFI and TLI ≥ 0.90, and RMSEA ≤ 0.10 13 .For model fit, we considered the modification indices (MI ≥ 11) calculated from Lagrange multipliers 13 .As the unifactorial model did not fit the data, we tested an alternative model (with two factors) based on theoretical associations between the items that could compose appearance satisfaction 5 .In this stage, the items that presented very high MI (>250) with other items and/or factors were deleted stepwise (i.e., from highest to lowest).All analyses were performed separately for women and men.
We evaluated convergent validity via the average variance extracted (AVE), and values > 0.50 were assumed to denote adequate convergent validity 19 .Moreover, the discriminant validity was also investigated.Discriminant validity was adequate when the AVE values of both factors were greater than the square of the correlation between these factors 19 .
We evaluated reliability via composite reliability (CR) according to the proposal of Fornell and Larcker 19 and the ordinal alpha coefficient (α) according to the proposal of Gadermann, Guhn, and Zumbo 20 .Adequate reliability was assumed at CR and α ≥ 0.70.

RESULTS
Among the students, 64% were women.The mean age of the total sample was 20.92 years (SD = 2.40).The mean age of men was 21.09 years (SD = 2.51), and the mean age of women was 22.80 years (SD = 2.34).The mean BMI was 23.24 kg/m 2 (SD = 3.97 kg/m²), with men at 24.05 kg/m 2 (SD = 3.94) and women at 22.80 kg/m 2 (SD = 3.93).Table 1 shows the participants' characteristics.
The participants predominantly studied in the field of human sciences attending the first years of their course, and they were classified in anthropometric nutritional status as normal range.Moreover, most students were at economic level B (i.e., people with average purchasing power).
The descriptive statistics of the BAS items for women and men samples are presented in Table 2.We did not find extreme violations of the normality assumption, which indicates adequate psychometric sensitivity of the items.
To fit the model to the male sample, we permitted a correlation between the error values of Items 9 and 13.For the female sample, the model was adequate without any additional fit.As for the convergent validity and reliability of the fitted model, all values of AVE, CR, and α were adequate for both the female and male samples.Discriminant validity was not adequate only for the female sample.

DISCUSSION
This study presents, for the first time to the best of our knowledge, the psychometric properties of a previously used self-rating scale for the assessment of satisfaction/ dissatisfaction with body appearance in a sample of Brazilian university students.The original unifactorial model of the BAS had not been subjected to confirmatory analysis in previous studies.In the present work, this model was not adequate for the samples.Therefore, we developed and evaluated a new model based on the theoretical associations between items.In earlier literature, we found only data on the reliability of the scale 21 .The reliability data were interesting but should be calculated only after assessing instrument validity 13 .Thus, we performed a broad evaluation of the scale's psychometric properties to help researchers in this subject area use it comfortably.We hope that our results will substantially contribute to the literature on body image by providing needed information about a good tool for evaluating satisfaction/dissatisfaction with the general appearance of the body.
The new two-factorial model for the BAS designed in this study (BAS-R) is composed of 19 items and two factors, and this structure yielded an adequate fit for both male and female samples.First, we removed Item 11 (legs), due to its high correlation with Item 23 (thighs), possibly because participants regarded both the upper (thighs) and lower (knee) parts as "legs."It will be interesting in further studies to specify exactly which body part the item refers to.We removed Items 5 (weight), 7 (height), 8 (ankle), and 24 (face) because they highly correlated with both scale factors, were not specific, and undesirably reflected both concepts simultaneously.
Regarding the scale's configurational structure (i.e., the number of items and factors), it proved similar between male and female samples, except for a correlation between Item 9 (waist) and Item 13 (hip) that was found in the male sample.This result can be ascribed to differences in body perception between men and women 22 .We believe that the high correlation observed between Items 9 and 13 could have occurred because the "waist" and "hip" are not generally the parts of the body that men care about most.
The comparison of our results with past literature is difficult because there are no prior data regarding the factorial validity of the scale we used.In addition, the name of the scale was altered in this study to reflect the new factorial model.In the original study, the authors 10 did not propose name for the scale; they merely reported that the instrument was developed to assess characteristics of satisfaction with the body.Subsequent studies 21,23 have described the scale as "Lerner's Body Image Scale" or, in a Brazilian study 6 , the "Body Area Scale (BAS)."Discrepancies in the name of the scale interfere with its use through difficulty tracing the instrument in literature analyses.Accordingly, we propose the name "Body Appearance (Dis)Satisfaction Scale (BAS-R)" to fully consider the revised structure of the new model.Some limitations of this study include the use of a nonprobabilistic sample, preventing broad generalization of these results to all Brazilian students or young adults.Our restrictions on the age of these participants prevent confident use or comparisons among various age groups.In addition, the size of our samples did not allow testing of the factorial invariance using independent subsamples.Thus, future studies using the BAS-R should be carried out to assess factorial invariance across men and women and independent samples to provide evidence about the external validity of the instrument.Despite these limitations, our study provides psychometric support for this means of evaluating satisfaction/dissatisfaction with one's own body appearance.

CONCLUSION
The unifactorial model of the scale evaluated was not adequate for samples of Brazilian students.Thus, we developed a two-factorial model and re-named the instrument Body Appearance (Dis)Satisfaction Scale (BAS-R).The BAS-R now shows adequate validity and reliability among both male and female Brazilian samples, making it possible for investigators to use this tool to evaluate individuals with characteristics similar to those of the participants of this study.

Table 1 .
Characteristics of the sample of university students * Note.Brazilian Economic Classification Criteria 2019.Brazilian reals (BRL) were converted into American dollars (exchange rate in October 2019 -1 BRL = 4.13 dollars).

Table 2 .
Descriptive statistics of the responses given to Body Area Scale (BAS) items (Women n = 647; Men n = 353) SD: standard deviation.