• Body Attitude Test;
  • body dissatisfaction;
  • eating disorders


  1. Top of page
  2. Abstract

The Body Attitude Test (BAT) was developed by Probst et al. (1995) for female patients with eating disorders (ED). This test measures the subjective body experience and attitudes toward one's body. The present authors have developed the Japanese version of the BAT and the purpose of the present paper was to investigate its reliability and validity in control (CON, n = 599) and ED patients (n = 46). The ED patients consisted of 21 anorexia nervosa, restricting type (AN-R) patients and 25 bulimia nervosa (BN) patients. Internal consistency was determined with Cronbach's α coefficient in CON. Factor analysis was conducted on BAT ratings given by CON. Factor analysis indicated that BAT was composed of two factors. These were body dissatisfaction (factor 1) and lack of familiarity with one's body (factor 2). A comparison was made among AN-R, BN, and CON. Bulimia nervosa had a significantly higher score than the other two groups. The BAT scores of ED patients correlated significantly with the Self -rating Depression Scale, and State–Trait Anxiety Inventory. These results show that ED patients have negative feelings toward their own body, similar to the findings in the original report. On factor analysis, however, it was not possible to distinguish between negative appreciation of body size and general body dissatisfaction as described in the original report. The authors also examine influences on this difference from a cross-cultural view point.


  1. Top of page
  2. Abstract

Disturbed body image is one of the important criteria of eating disorders (ED). The definition of ‘body image’ has been argued for a long time. Cash and Brown indicated that there are two aspects to body image: one is perceptual disturbance as body size distortion, and the other is cognitive and affective disturbance as body dissatisfaction.1 Perceptual disturbance has been studied by various methods, for example, movable caliper technique,2 video distortion method3 and so on. These methods were used mainly to detect how patients estimate their body size (overestimation or underestimation). However, because cognition and affect influence the estimation of one's body size, disturbances of these aspects also should be focused on as regards body image disturbance. This is why several questionnaires were developed to aspects such as the subjective attitude towards the body.4–6 These questionnaires are the Body Shape Questionnaire (BSQ) of Cooper et al.,4 Ben-Tovim and Walker's Body Attitude Questionnaire (BAQ),5 and the Body Attitude Test (BAT) of Probst et al.6 The BSQ, developed in 1987, consists of 34 items and aims to assess the phenomenal experience of concern about body shape. In 1991 the BAQ was developed. It consists of 44 items and its subscales encompass six distinct aspects of body experience. The BAT, developed in 1995, measures the subjective body experiences and attitudes toward one's body of female ED patients. It was originally written in Dutch. It has also been translated into English, French, German, Spanish,7 Italian, and Czech. Because the BAT is used in many countries, it is suitable for the cross-cultural study of ED. The BAT is a convenient questionnaire, which consists of only 20 items. Therefore we developed the Japanese version of the BAT and examined its reliability and validity. We also studied how Japanese ED patients feel about their own bodies and examined whether BAT could be used as a screening test for ED.


  1. Top of page
  2. Abstract

Body Attitude Test

The original version of the BAT is a self-reporting questionnaire consisting of 20 items scored on a 6-point scale (0–5). It consists of the following three factors: negative appreciation of body size; lack of familiarity with one's body; and general body dissatisfaction. The maximal total score is 100 and a high score indicates that the subject has a deviated body experience. The BAT was originally written in Dutch. It has also been translated into six other languages. To date, the results of each version have been the same as the original. In the present study the Japanese version was developed as follows. The English version of BAT was first translated into Japanese, and then back-translated by a native Japanese speaker who had not seen the original version. We changed certain expressions in the draft after consulting Probst, the author of the original BAT.


The present study was performed on 46 female patients with ED who were admitted to University Hospital, Kyoto Prefectural University of Medicine and diagnosed according to Diagnostic and Statistical Manual of Mental Disorders (4th edn; DSM-IV) criteria. Twenty-one patients were diagnosed as having anorexia nervosa, restricting type (AN-R) and 25 patients were diagnosed as having bulimia nervosa (BN; this consisted of 17 with the purging type and eight with the non-purging type). We excluded anorexia nervosa, binge-eating/purging type and ED not otherwise specified (EDNOS) because of their small number. The average age, body mass index (BMI) and duration of illness of patients with AN-R were 16.6 ± 3.2 years old, 13.8 ± 1.8 and 1.7 ± 1.6 years, respectively, and those of BN were 22.1 ± 3.1 years old, 20.3 ± 1.8 and 3.2 ± 2.4 years, respectively. The patients were asked to answer self-rating questionnaires on their first visit to University Hospital, Kyoto Prefectural University of Medicine or within a week after admission.

The control group (CON) was chosen from 630 female university students in Kyoto City. They were asked to answer the Eating Attitude Test−26 (EAT-26)8 and BAT. Five hundred and ninety-nine subjects who scored <20 on EAT-26, which is the cut-off point of EAT-26, were chosen as the control group. Their average age was 20.2 ± 3.2 years and their average BMI was 20.6 ± 2.4. The AN-R patients were significantly younger than the BN and CON patients (P < 0.001), and had a significantly lower bodyweight than the BN and CON patients (P < 0.001). There were no significant differences between AN-R and BN with regard to the duration of illness.

Data analysis

Factor analysis was conducted on the BAT ratings given by the control group. As in the original study, we performed principal component analysis with VARIMAX rotation. Internal consistency was determined with Cronbach's α coefficient in the CON. A comparison of a total BAT score was made among three groups: CON, BN and AN-R. The cut-off score was calculated with ED patients and 50 normal subjects selected randomly from the CON. To examine the correlation of BAT score and patients’ psychopathology, EAT, Eating Disorder Inventory (EDI),9 Self-rating Depression Scale (SDS),10 and State–Trait Anxiety Inventory (STAI)11 were used. Correlation of BAT score and age, duration of illness, and BMI were also examined. For the statistical analysis we used Kruskal–Wallis rank analysis of variance and Mann–Whitney analysis. To examine the relationship to other psychometries we used Spearman's rank correlation coefficient.

For analysis we used the statistical software spss 10.0 J for Windows (SPSS Japan, Tokyo, Japan). The present study was conducted in accordance with the ethics covenant of the Helsinki Declaration. The purpose and procedure of the study were explained to all subjects and their verbal informed consent was obtained.


  1. Top of page
  2. Abstract

Factor analysis

Three factors with an eigenvalue >1 were found (Table 1), which accounted for 46.5% of the common variance of the 20 items. The third factor had only two items (items 15 and 20) and the contents of these two items were independent to each other. Because we consider that this factor could not be extracted as any meaningful factor, we excluded this factor. Those two items of this third factor were also exactly the same items of the excluded factor in the original as ‘rest factor’.

Table 1. Variances by rotated factor analysis (n = 559)
FactorEigenvalue %% of varianceCumulative variance

Factor 1 (BAT-1): body dissatisfaction (items 1, 3, 5, 7, 8, 10, 13, 16, 18; loadings: 0.645, 0.670, 0.584, 0.677, 0.690, 0.791, 0.521, 0.475, 0.565). Factor 2 (BAT-2): lack of familiarity with one's body (items 2, 11, 12, 14, 17, 19; loadings: 0.539, 0.644, 0.781, 0.569, 0.573, 0.647). ‘Rest factor’: items 15 and 20, loadings: 0.640, 0.602.


Internal consistency was measured by Cronbach's α in CON as follows: total BAT, 0.90; factor 1, 0.91; factor 2, 0.83.

Comparison among three groups

There were significant differences among the three groups in scores for total BAT, factor 1, and factor 2 at the 0.001 probability level. Scores of total BAT, factor 1, and factor 2 increased in order of CON, AN-R and BN (Table 2). There was a significant difference between each group on total BAT, factor 1, and factor 2, but in factor 1 there was no significant difference between AN-R and CON.

Table 2. Mean score of BAT
GroupTotal BAT Mean ± SDFactor 1 Mean ± SDFactor 2 Mean ± SD
  1. BAT, Body Attitude Test; AN-R, anorexia nervosa (restricting type); BN, bulimia nervosa.

  2. Significant differences (P < 0.001) among groups are indicated by †and ‡.

  3. Significant differences (P < 0.05) among groups are indicated by §and ¶.

AN-R43.2 ± 13.3§20.0 ± 7.3 9.6 ± 5.5
BN62.6 ± 14.132.1 ± 8.313.6 ± 5.6
Control36.0 ± 15.0§19.9 ± 9.3 4.9 ± 4.5

Cut-off score

The cut-off score of 36, which was proposed by the author of the original BAT, had a sensitivity of 57.6% and a specificity of 67.5%. A cut-off score of 51 had better balance, with a sensitivity of 64.8% and a specificity of 62.7% (Table 3).

Table 3. Cut-off point
 Eating disorder patients (n = 46)Control group (n = 50)
Score (>36)3457.62542.4
Score (<36)1232.42567.5
Score (>51)2464.81335.1
Score (<51)2237.33762.7


Table 4 shows the correlation coefficient between BAT and other questionnaires in all ED patients. We used the same classification12 as the original to evaluate the size of the effect of intercorrelations: 0–39 = low; 40–69 = moderate to substantial; and 70–100 = high to very high.

Table 4.  Correlation coefficient between BAT and other questionnaires (n = 46)
 Total BATBAT-1BAT-2
  1. BAT, Body Attitude Test; EAT, Eating Attitude Test; EDI, Eating Disorder Inventory; SDS, Self-rating Depression Scale; STAI, State–Trait Anxiety Inventory; BMI, body mass index.

  2. Spearman rank correlation coefficient: **P < 0.01.

EAT26 0.336 0.28 0.312
EAT-1 dieting 0.365 0.331 0.311
EAT-2 bulimia and food preoccupation 0.574** 0.507** 0.486**
EAT-3 oral control−0.289−0.336−0.154
Total EDI 0.682** 0.679** 0.503**
EDI-1 drive for thinness 0.555** 0.517** 0.475**
EDI-2 bulimia 0.617** 0.613** 0.453**
EDI-3 body dissatisfaction 0.725** 0.788** 0.492**
EDI-4 ineffectiveness 0.527** 0.485** 0.383
EDI-5 perfectionism 0.228 0.306 0.106
EDI-6 interpersonal distrust 0.318 0.307 0.27
EDI-7 interoceptive awareness 0.441** 0.382 0.353
EDI-8 maturity fears 0.175 0.178 0.056
SDS 0.511** 0.384 0.513**
STAI: state 0.218 0.206 0.151
STAI: trait 0.512** 0.482** 0.347
Age 0.419** 0.482** 0.299
Duration of illness 0.245 0.268 0.239
BMI 0.526** 0.630** 0.312
Eating Attitude Test

On the EAT subscale ‘bulimia and food preoccupation’ correlated with total BAT, factor 1, and factor 2.

Eating Disorder Inventory

Total EDI, EDI subscale ‘drive for thinness’, ‘bulimia’, and ‘body dissatisfaction’ correlated with total BAT, factor 1, and factor 2. The EDI subscale ‘ineffectiveness’ correlated with total BAT and factor 1. The EDI subscale ‘interoceptive awareness’ correlated with total BAT.

Self-rating Depression Scale

This correlated with total BAT and factor 2.

State–Trait Anxiety Inventory

Trait anxiety correlated with total BAT and factor 1.


Age correlated moderately with total BAT and factor 1.

Duration of illness

There was no correlation detected.

Body Mass Index

The BMI correlated with total BAT and factor 1. In addition, there was no correlation found in the control group.


  1. Top of page
  2. Abstract

First of all, with regard to the factor analysis, it was based on the ED patients in the original study. In the present study, however, it was conducted on the control group because of the small number of ED patients. Because BAT is the questionnaire to detect one's feeling toward their own body, the structure of factors should be consistent from ED patients to normal (healthy) subjects. Therefore we consider that the factors extracted from our results would represent some of the Japanese characters toward body image. The EAT-26 was chosen to exclude subjects with AN in the control group but subjects with binge eating/purging could not be excluded. There is a certain limitation in the control group. Therefore we should not compare these results directly with the orginal study or the Spanish study, but we could recognize some similarities and differences.

Here we would like to summarize the studies conducted using the original version, which Probst et al. developed,6 and the Spanish version by Gila et al.7 Both studies extracted three factors and those are: factor 1, negative appreciation of body size; factor 2, lack of familiarity with one's body; and factor 3, general body dissatisfaction. In the original study the items related to each factor are: factor 1, items 3, 5, 6, 10, 11, 13 and 16; factor 2, items 2, 4, 9, 12, 14, 17 and 19; factor 3, items 1, 7, 8 and 18.6 Although the Spanish study had some differences on the items in each factor,7 they suggest that the sample's age, culture or prevalence of individual ED may have caused these differences between these two studies

In the original study ED patients had significantly higher BAT scores than controls. The lack of familiarity with one's body had the greatest significant difference among the three factors. Probst et al. compared the BAT score of four types of ED: AN-R; mixed AN (AN-M); BN; and EDEDNOS.6 Anorexia nervosa-R scored the lowest BAT and BN the highest. With regard to each factor, AN-R had a significantly lower score than the other three groups for factor 1 and 2. On factor 3, however, AN-R had a significant difference from BN only. Probst et al. established 36 points as the cut-off score. They concluded that BAT could differentiate ED patients from normal subjects by the total and subscale scores.6 Furthermore, AN-R had a more positive body attitude than the three other subgroups and BN the least. Regarding the correlation between BAT and other psychometries, factor 2 had a high correlation with depression and anxiety.

In the study by Gila et al. the same three factors as in the original were selected,7 with different item loadings on each as aforementioned. There was no significant difference between anorexic patients and bulimic patients for total BAT score but bulimics had a higher score. Both groups scored higher than the control group. Gila et al. suggested that a cut-off score of 41 would achieve better balance than that of 36 in the original study.7 They considered that differences in the results were also due to mean age of the sample, culture and the prevalence of individual ED.

On factor analysis we could not distinguish between ‘negative appreciation of body size’ and ‘general body dissatisfaction’ used in the original study. Although the original factors try to distinguish one's perception of body size from general body dissatisfaction, in the present study the items of these factors were mixed together and comprised a single factor indicating body dissatisfaction. For example, in the original study, negative appreciation of body size includes items such as item 3 ‘My hips seem too broad to me’, and item 5 ‘I have a strong desire to be thinner’; in contrast, general body dissatisfaction includes points such as item 1 ‘When I compare myself with my peers’ bodies, I’m dissatisfied with my own’ and items 18 ‘I envy others for their physical appearance’. It seems that these two factors differentiate the subject's own attention toward her own body size from the subject's body dissatisfaction when comparing others’ bodies, but in our study all of these items are considered components of one factor.

It is difficult to determine what caused this result. Here we would like to propose the possibility that culture difference has some effect on this result. In Japanese culture we are supposed to be concerned about other peoples’ feelings or evaluations. Our standard is based not on a subjective scale but on a comparative scale. Not only one's own senses but also the imagined opinions of others form the judgment toward one's own body. We consider that this is why our subjects mixed a negative appreciation of body size with general body dissatisfaction. We can also suppose that Japanese subjects cannot distinguish perception and dissatisfaction because they tend to have more of a feeling of disgust toward their own body than the European. Therefore, their perception of body size is influenced by the strong negative feeling toward body image and failure to detect correct body size. We need to reexamine this issue in a further study with a larger sample.

Internal consistency of our version was also similar to that of the original. Correlation with the subscale ‘body dissatisfaction’ of the EDI produced the same result as the original and this supports the convergent validity of the BAT. These results show that the Japanese version of the BAT has reliability and validity.

Our results show that ED patients, especially BN patients, have strong negative feelings toward their own body, which is similar to the findings of the original and Spanish reports. Kobayashi et al. developed the Japanese version of the BSQ.13 This questionnaire was originally developed by Cooper et al.4 and assesses body dissatisfaction and negative attitudes toward the body. It consists of 34 items and factor analysis is not performed. The higher score indicates a more negative attitude toward the body. Kobayashi et al. reported that Japanese patients with BN had the significantly highest score among three groups (BN, AN and control), indicating that BN had the most distorted and most negative body image, and BSQ correlated with many subscales of EDI-2 including body dissatisfaction.13 In contrast, patients with AN had a significantly lower score on BSQ than the control group. At the same time, a significant correlation between BSQ and subscales of EDI-2 such as body dissatisfaction, bulimia, drive for thinness etc. were found among AN. They considered that AN patients use the defense mechanism of denial, causing them to score low points on BSQ. In the present study, although AN-R patients had a higher BAT score than controls on total BAT and factor 2, on factor 1 there was no significant difference between AN-R and controls. At the same time, AN-R had the lowest weight among the three groups. We consider that because of the low BMI, AN-R patients did not feel body dissatisfaction. Instead, they lost familiarity with their own bodies. We can imagine that the reason why AN had a low score on BSQ is because they were satisfied with their body shape rather than using denial. We consider that BSQ is sensitive in detecting negative concern about body shape but not in lack of familiarity with one's own body. This result explains why BAT shows more specific characteristics of body image disturbance that the ED patients have than BSQ. We suggest that after they gain bodyweight, AN patients may have a higher score on body dissatisfaction.

Our control group had a higher average score than the controls in the original study and the BAT score did not correlate with BMI. This suggests that the Japanese women have fewer positive feelings toward their body image regardless of bodyweight. The sensitivity and specificity were not good enough to use as a screening test. This is because the number of ED patients was relatively small and the score of the control group was higher. Thus we need further study in a larger series to calculate the cut-off score.

Probst et al. used the BAT to establish the guidelines for treatment.14 They showed that after 6 months of inpatient treatment with a multidimensional approach, the BAT score of patients improved significantly. Besides its use as a screening test, we may be able to use BAT for assessing how much improvement patients have made.

In the present study we developed the Japanese version of BAT. It is different from the original study in that two factors were found. It has a good reliability and validity. The cut-off point could not be determined because of the small number of ED patients but the ED patients scored significantly higher than the control group. Thus the BAT could be useful in clinical research and practice.


  1. Top of page
  2. Abstract
  • 1
    Cash T, Brown T. Body image in anorexia nervosa and bulimia nervosa: A review of the literature. Behav. Modif. 1987; 11: 487521.
  • 2
    Slade PD, Russell GF. Experimental investigations of bodily perception in anorexia nervosa and obesity. Psychother. Psychosom. 1973; 22: 359363.
  • 3
    Probst M, Van Coppenolle H, Vandereycken W, Goris M. Body image assessment in anorexia nervosa patients and university students by means of video distortion: A reliability study. J. Psychosom. Res. 1992; 36: 8997.
  • 4
    Cooper PJ, Taylor MJ, Cooper A, Fairburn CG. The development and validation of body shape questionnaire. Int. J. Eat. Disord. 1987; 6: 485494.
  • 5
    Ben-Tovim DI, Walker MK. The development of the Ben-Tovim Walker body attitudes questionnaire (BAQ), a new measure of women's attitudes towards their own bodies. Psychol. Med. 1991; 21: 775784.
  • 6
    Probst M, Vandereycken W, Van Coppenolle H, Vanderlinden J. The Body Attitude Test for patients with an eating disorder: Psychometric characteristics of a new questionnaire. Eat. Disord. 1995; 3: 133145.
  • 7
    Gila A, Castro J, Gómez MJ, Toro J, Salamero M. The body attitude test: Validation of the Spanish version. Eat. Weight Disord. 1999; 4: 175178.
  • 8
    Kiriike N, Nagata T. Self-report inventories in eating disorders. Arch. Psychiatr. Diagn. Clin. Eval. 1992; 3: 457464 (in Japanese).
  • 9
    Nagata T, Kiriike N, Matsunaga H et al. Clinical trial of the Eating Disorder Inventory (EDI) in patients with eating disorders. Jpn J. Clin. Psychiatry 1994; 23: 897903 (in Japanese).
  • 10
    Fukuda K, Kobayashi S. A study on a Self-Rating Depression Scale. Psychiatry Neurol. Jpn 1973; 75: 673679 (in Japanese).
  • 11
    Nakazato K, Mizuguchi K. Development of a Japanese version of new anxiety scale State–Trait Anxiety Inventory (STAI). Jpn J. Psychosom. Med. 1982; 22: 108112 (in Japanese).
  • 12
    Surwillo W. Experimental Design in Psychiatry: Research methods for Clinical Practice. Grune and Stratton, New York, 1980.
  • 13
    Kobayashi Y, Tachi T, Murotsu K et al. Clinical trial of Body Shape Questionnaire (BSQ) in eating disorder patients; reliability and validity of the Japanese version of BSQ. Jpn J. Clin. Psychiatry 2001; 30: 15011508 (in Japanese).
  • 14
    Probst M, Vandereycken W, Van Coppenolle H, Pieters G. Body experience in eating disorders before and after treatment: A follow-up study. Eur. Psychiatry 1999; 14: 333340.