Introduction

Body image is a complex multidimensional phenomenon that can involve perceptual, cognitive, affective, or behavioral disturbances [1]. It refers to the physical, emotional, and interpersonal view of one’s body or the inner mental picture and sum of attitudes towards one’s body [2]. It is relevant in different fields such as investigations of body image disturbances with an organic etiology (e.g., post-stroke influence on the own body perception) [3], altered body image in illness related to the visible disfigurement [4], and body dysmorphic disorder related to distressing preoccupation with an imagined or slight defect in appearance [5]. The majority of research conducted in the field of body image disturbance is on eating disorders and obesity. The studies that have been conducted so far give ambiguous results that can be associated with different ways of measuring body image, different theoretical conceptualization of the construct, and the characteristics of the sample used. Among women, body dissatisfaction is observed where excess body weight negatively affects features such as the abdomen, hips and thighs. Males are divided between the desire to lose and gain weight, and wanted to increase muscle size, especially in their adolescence. As they move towards adulthood, body dissatisfaction has been reported to either decrease or remain stable [6, 7]. With the increase in the prevalence of overweight BMIs and obesity, it is logical to expect that the rates of body image dissatisfaction would similarly rise [8].

The body image concerns of obese individuals were first described in the late 1960s [9], and are related to social isolation, shame, achievement failure [10], and low self-esteem. In developed countries, satisfaction with one’s own body appearance decreases as body weight increases, and, consequently, in severely obese patients, a strongly impaired body image and high dissatisfaction with body appearance are observed [11]. Weight and body dissatisfaction could motivate unhealthy behaviors in an effort to lose weight, such as binge eating and other disturbed eating behaviors [12]. Among obese adults, body dissatisfaction may be a potential motivator for initiating healthy lifestyle changes [13]. It is important that individuals have accurate perceptions of their body shapes and weights in order for interventions aimed at obesity prevention and treatment to be successful [14]. Overweight individuals who perceive themselves as having normal weight may lack the motivation to engage in behaviors to control their weight; hence, the risk of developing obesity increases. For these reasons, it is very important to measure body image and weight dissatisfaction in overweight and obese populations.

Assessment measures for body image

There are a number of measures used in the assessment of body image. Perceptual disturbances in body image are typically measured by visual-size estimation and distorting mirrors [15]. Global questionnaires, rating scales and figural stimuli measures are designed to assess overall satisfaction with body shape and body sizes, which are subjective/attitudinal features of body image. The most commonly used measures that index weight dissatisfaction include scales such as the Eating Disorder Inventory—Body Dissatisfaction scale and the Body Shape Questionnaire [16]. These scales are ideal for use with obese samples because their focus is on the subjective rating of weight-related aspects of the body [8]. Another method involves the use of schematic figure rating methodologies. These consist of a range of figures varying in size from underweight to obese. Generally, individuals rate the figures using a protocol designed to determine their ideal and current body sizes. The discrepancy between the two ratings is used as the index of dissatisfaction (for example, the Figure Rating Scale, and the Contour Drawings Rating Scale), which has been shown to correlate highly with questionnaire measures of satisfaction [17].

A relatively small number of measures for the assessment of overall body dissatisfaction are currently available in Croatia. The most commonly used self-report measure is the Body Shape Questionnaire [16, 18], while the Physical Appearance State and Trait Anxiety Scale [19], which has also been translated and adapted to the Croatian context [20], was used to estimate the affective components of body image. The Sociocultural Attitudes Toward Appearance Questionnaire [20, 21] is used for assessing cognition and acceptance of social standards related to physical appearance. The items are arranged in two scales: awareness assessing recognition of the cultural standards for appearance and internalization measuring the degree of adoption of these standards [22]. Until now, there have been no studies on the body image of overweight and obese adults in Croatia. Previous studies of body image have been carried out mainly in adolescent populations, with samples of high school and university students, as well as those suffering from eating disorders [18, 20].

In the last decade, one of the commonly used measures for assessing the body image of obese people is the Body Uneasiness Test (BUT-A), which was developed and validated in the Italian language. Although there are other measures of body image, BUT-A has good psychometric properties and possibilities of use as a diagnostic and prognostic test, in studies of the clinical course and outcomes of eating disorders, and in evaluations of the changes in negative body image in weight reduction programs [23]. The measure not only examines dissatisfaction with specific body parts, shapes, and functions but also includes some more general and less-specific aspects that are often very difficult to describe. The BUT-A has been used and validated on samples of men and women with normal body weight, with clinical samples of patients with anorexia and bulimia nervosa, patients with binge eating disorders, and obese patients seeking treatment [23, 24]. However, there is a lack of research on obese individuals in the general population that would allow the generalization of the obtained results and comparisons between different clinical and non-clinical groups. Therefore, the aim of this study was to investigate the factor structure and the internal consistency of the Body Uneasiness Test (BUT-A) in a general sample of overweight and obese subjects and to evaluate the differences in the body uneasiness structure between men and women. The second aim was to examine the relation of the body uneasiness (BUT-A) with body dissatisfaction measured by the Figure Rating Scale (FRS) [25].

Methods

Participants

The sample consisted of 320 participants (156 women and 164 men) aged 20–71 years with a mean age of 46.49 (SD = 11.32). The participants’ BMIs ranged from 24.00 to 48.83 with a mean of 31.26 (SD = 4.87). This was a non-clinical sample of overweight and obese adults. The data were collected in a primary care provider’s office, and the questionnaires were given during a visit to the physician’s office. Informed consent was obtained from the participants prior to completion of the questionnaire.

Measures

The Body Uneasiness Test (BUT-A) [24, 26] was used as part of a larger study to assess body image uneasiness [27, 28]. To assure the quality of the translation, the questionnaire was translated from Italian to Croatian by two native speakers and was back-translated to Italian. The 34 BUT-A items measure several areas of body dissatisfaction: weight phobia, body image concerns, avoidance, compulsive self-monitoring, and depersonalization. Answers are given on a six-point likert-type scale from 0 (never) to 5 (always). Higher ratings indicate higher levels of body uneasiness.

The Figure Rating Scale (FRS) [25] was used to measure participants’ dissatisfaction with their body images. The scale consists of nine female or male silhouettes ranked by increasing weight from the thinnest to the widest, and numbered from 10 to 90. Participants are asked to make two ratings: first to choose the picture that reflects their current body appearance and then the picture that represents their ideal body appearance. A body size dissatisfaction variable was created for each participant by subtracting the number of the figure selected as the ideal body size from the number of the figure selected as the self body size. Greater dissatisfaction is represented by a great difference between these two measurements.

Procedures

The questionnaires were administered individually in the physician’s office. Apart from that, a physician measured the participants’ weight on a medical scales and height on a stadiometer while they were in their underwear and without shoes.

Results

All of the analyses were carried out with the aid of the packages “lavaan” [29], “sem” [30] and “semTools” [31] for the “R” language and environment for statistical computing [32]. The main idea was to test the original five-factor model with orthogonal dimensions [24], the updated model with correlated dimensions [23], and the one-factor model originally discarded by the same authors [26] and to test for eventual model modifications and localized areas of model strain. Furthermore, given the typically reported gender asymmetries regarding body image [6–9, 33], an additional problem was to test for model invariance across genders. The distribution of the measured variables was quite positively skewed (which is quite understandable given that the sample was not a clinical one). The average mean response for the single items ranged from 0.77 to 2.38 signaling occurrences in the lower part of the theoretical distribution. The models were fitted to log-transformed and unaltered data. Given the similarities between them, only the models on original (unaltered) data are presented to retain the original scale metric and to ease model comparisons. The models were fitted using the maximum likelihood method with robust standard errors, and using Satorra and Bentler correction [34]. The estimated five-factor latent structure models consisted of five factors: weight phobia (WP), body image concerns (BIC), avoidance (A), compulsive self-monitoring (CSM) and depersonalization (D). All of the latent variables have multiple indicators (5-9). In the one-factor model, all of the variables are indicators of the general factor (34 of them). The variances of all of the latent dimensions were fixed to one (standardized values). The main model parameters are presented in Table 1.

Table 1 Fit indices for the three competing models
Full size table

None of the fitted models reproduced the data well enough in a formal way (the model implied variance/covariance matrix differs significantly from the observed one). The five-factor model with orthogonal dimensions was the worst fitting model (it had a significantly lower fit indices and higher error: presented in Table 1). The other two competing models (five correlated factors and one general dimension) fitted the data almost equally well. Generally, the five-factor model with correlated dimensions provided the best fit, even if the fit indices (presented in Table 1) are somewhat lower than suggested [35]. The correlations between the fitted latent dimensions were quite high (above .90), suggesting that the five correlated factor model should be used and interpreted with caution. The simple one-factor model did not differ significantly from the more complex model (χ2 = 16.97, df = 10, p > .05) and should be regarded as a more parsimonious, robust and adequate model.

The intercepts are generally low, which is understandable due to the positive asymmetries, and several items are less well covered by the general factor (e.g., items 1, 3, 21, and 32 all have communalities lower than 0.30). The mentioned items have lower communalities even in the initial validation studies [23]. The measurement invariance across gender groups was tested on several levels. The first model was the unconstrained configural invariance model (with the same configuration of the two groups). The second model also constrained the factor loadings to be equal across groups (weak invariance). The third model constrained the factor intercepts to be equal between gender groups (strong invariance), and finally, the strict invariance model fixed the residuals to be equal between gender groups. The fit indices for the single group models (for males and females separately) and the test of the increasingly constrained invariance models are presented in Table 2.

Table 2 Invariance test between gender groups for the one-factor solution
Full size table

Both models fit the data in the male and female subsets equally well. Although it is not possible to say that the model invariance holds (all models significantly differ from one another as model invariance constrains increase), it is possible to look at the differences in fit indices and to compare the impact of parameter constrainment. The model with equal loadings fit the variance–covariance matrix worse than the unconstrained model, but the addition of the intercept constrainment reduced the model fit considerably (ΔCFI for the model with equal intercepts is twice the size of the ΔCFI of the equal loadings). The average intercept and slope estimates of the latent variable for the entire BUT-A items are presented in Table 3.

Table 3 Average statistics for the factor loadings on 34 BUT items by gender
Full size table

Even though all of the estimates in the female subsample are higher, the slope estimates are roughly comparable between genders (the mean difference is 0.15 and the confidence intervals overlap). Greater differences are present in the intercept estimates (the mean difference is 0.65, and the confidence intervals did not overlap).

Given that the BUT-A scales showed acceptable structural characteristics as a one-factor measure, the reliability of the scale was calculated as Cronbach’s α = 0.97, while the average inter-item correlation was 0.52. The general BUT-A score correlated significantly, albeit to a modest extent, with BMI, current body appearance, ideal appearance, and body dissatisfaction (measured by FRS; Table 4).

Table 4 Body image measures intercorrelations
Full size table

To test for the BUT-A predicitivity of body dissatisfaction, a regression model with appearance discrepancy as a dependent variable and body mass index, BUT-A and gender as predictors was performed. All of the predictor variables were grand mean centered to ease interpretation. The BUT-A scale was also averaged (mean of all items). The results are presented in Table 5.

Table 5 Regression analysis results: predictors of body dissatisfaction measured with Figure Rating Scale
Full size table

The first model fitted the data reasonably well (R 2 = 0.20). There are small gender differences in body dissatisfaction, while BMI is a significant body dissatisfaction predictor (Table 5). Males show a smaller body appearance discrepancy than females, while increasing BMI is associated with greater body dissatisfaction. The second model included the BUT-A general score in addition to the two present in the first model. The addition of the BUT-A general score resulted in a significant increase in the model fit (ΔR 2 = 0.05).

Discussion

We had several goals for this study. Our first aim was to verify the factor structure and the internal consistency of the BUT-A scale in a general sample of overweight and obese subjects and to test for the structural invariance among male and female subsamples. The obtained results showed that the BUT-A items could be described with a general one-factor structure almost equally well as with the originally proposed five-factor structure. The more parsimonious one-factor model suggests possible problems in the interpretation of the originally proposed structure [23, 24, 26]. The scale is certainly adequate to measure general body uneasiness, but the specific uneasiness facets could be a problem, especially in non-clinical samples such as this one. It is possible that the structure of the BUT-A measure is of a hierarchical type, with five specific factors and a general uneasiness factor. The complexity of the hierarchical model and a relatively small sample size debilitates such estimations even though the obtained results point in that direction. Some items are poorly covered by the general uneasiness factor. Those items have small communalities even in the original validation studies [23, 24, 26], specifically with regards to body evaluation practices (e.g., body checking in the mirror or avoiding mirrors, body shame, and anxiety in the presence of others). It is possible that those items are more appropriate for measuring uneasiness in younger individuals, hence the differences between the results observed in the mature subjects in this study and previous research on younger people. The mentioned effect is in line with the previous findings of body image importance across the lifespan [36].

The results show that the gender invariance of the one-factor structure was only marginal. It can be seen that the relative importance of the items is roughly comparable between genders, although women tend to give higher responses on all the items. These gender differences replicate the findings of previous studies using this measure [24, 26] as well as with studies using other similar measures [37]. Our findings suggest that the differences in body dissatisfaction between men and women [e.g., 38] that are often discussed could be a problem of intensity (the different mean levels for men and women) and not necessarily a problem of differences in body uneasiness idiosyncratic conceptualization (differences in factor structure). The literature suggests that women give greater importance to body appearance than men do, a result that was confirmed in our study as well: women are less satisfied with their bodies and feel greater body uneasiness which is a common finding in the field [39, 40]. Women also seem to attach more importance to body weight as body mass increases [41–45].

The final aim was to verify the relationship structure between the objective measures of body mass and the subjective evaluation of body dissatisfaction (BUT-A and FRS) among overweight and obese subjects. The results on the BUT-A scale are moderately positively related with body dissatisfaction and the evaluation of current body appearance, as well as with BMI. The BUT-A scale estimates more complex aspects of body dissatisfaction (for example, cognitive: body image concern; affective: weight phobia; and behavioral: avoidance behavior or compulsive self-monitoring) that are connected, to a lesser extent, with the subjective assessment of general body dissatisfaction, as well as with objective measures of body weight status as measured by BMI.

Body mass index and body dissatisfaction have an intricate and complex relationship, especially when studied longitudinally (see [46] for a review). Given the tendency for BMI to increase with age [37, 47], it may play a crucial role in the development of poor body image. Taking into account the variability of reported association magnitude across researches [e.g. 8, 9, 33, 48], it is possible that BMI and body uneasiness might have a nonlinear relation. Therefore, the lower obtained correlation in this research may not come as a surprise and is in line with previous research [49]. Overall appearance satisfaction may also focus on non-weight-related features, such as body parts or muscularity, or may include items that assess an individual’s conception of how he/she looks in clothes or appears to other people [17]. Adami et al. [50] hypothesized that some psychological aspects are related to body weight and shape. The preoccupation with body weight and shape, body dissatisfaction, and self-blame might reflect inner feelings, which in obese patients, become partially independent from somatic morphology and do not normalize with the stabilization of body weight and shape [50].

The assessment of ideal appearance is virtually independent of the results obtained on the BUT-A scale. Obese patients express dissatisfaction with their body by selecting the ideal figure, which is, on average, two units thinner than their evaluation of their current body figure, which corresponds to the findings of previous research [51]. For overweight and obese men, body dissatisfaction is lower than that in women, as many previous studies have shown [52]. It is interesting to note that BMI is highly correlated with the perception of current body appearance in overweight and obese patients, which indicates that participants are relatively accurate in assessing their body shape.

The BUT-A score contributes significantly to the prediction of general body dissatisfaction, even after controlling for gender and BMI. However, the level of explained variance is relatively low (5 %) but still significant. Similarly, results of previous studies, i.e. [46, 49], showed that the BUT-A score and general body dissatisfaction share some common conceptual components unrelated to objective parameters such as BMI. The multifaceted nature of the BUT-A might be the basis of its relatively low contribution to the explanation of general body dissatisfaction.

Although the study reports some limits, for example, the relatively small sample size and the absence of normal weight people, the results show that the misperception of body image and dissatisfaction with physical appearance are moderately associated with BMI. The strong point of this study is that it is conducted in a sample of overweight and obese individuals in the general population that were not included in previous research.

We can conclude that the Body Uneasiness Test is a good and reliable measure of body dissatisfaction, but we suggest that it can be treated as a one-factor measure. In previous studies, the BUT-A proved to be a good measure of body dissatisfaction with a five-factor structure. This study shows that the one-factor solution is parsimonious and a psychometrically more justifiable option. This finding suggests that the five-factor solution is probably more appropriate in clinical samples, whereas in samples of the general population, the one-factor structure of this questionnaire has to be considered as the more stable model. It is possible that structures of the construct, as well as the BUT-A questionnaire, are hierarchical, with general dissatisfaction as an overlying factor. The different facets of the uneasiness construct may be more apparent at an earlier age, but the general evaluation without diversification is more justifiable in middle age. Female participants showed higher results, implying higher body uneasiness, but the importance of the items is roughly comparable between genders. The BUT-A score significantly contributes to the prediction of general body dissatisfaction, even after gender and BMI are controlled.