[Validation of the French version of the Body Shape Questionnaire]

In today's societies, pressures from the idea of thinness are omnipresent and lead to a corporal dissatisfaction with an excessive preoccupation of the body's image. It seems important to have, in France, a device that can evaluate the corporal dissatisfaction degree, for the reason that the troubles of the body's image is a common diagnostical category to the anorexia and bulimia (DSM IV, 1994). Cooper et al. (1987) have developed one-dimensional questionnaire of 34 items in order to measure the worries towards the weight and the shape of the body, called the "Body Shape Questionnaire" (BSQ). Its concurrent validity has been shown with the using of the corporal dissatisfaction under-scale of the Eating Disorders Inventory (EDI; Garner et Olmsted, 1984) and the using of the diagnostical questionnaire: Eating Attitude Test (EAT; Garner et Garfinkel, 1979). The BSQ gives us a way to explore the role of extreme worries towards the body's apparence in the development, the keeping and the treatment of eating disorders. From this point the BSQ is a tool widely used on an international level in researches on the eating disorders. It has been validited in Spain and in Germany whereas it has not been yet subject to a validation in France. For this reason, we proposed to use the BSQ on the french population.

The sample is made of 242 university girl students. The average age is 20.7 years old with an standard deviation of 2.26. The size and the weight helped us to calculate the Body Mass Index (BMI) which is in fact the weight divided by the size squared. The average BMI is 21.06 kg/m2 +/- 2.87. Regarding the evaluating devices, the BSQ is presented as a scale of 34 items marked by 6 different points: 1 never, 2 barely, 3 sometimes, 4 often, 5 very often, 6 always. After getting the authorization form the authors to respect during these kinds of procedures, we have started validation. The corporal dissatisfaction under-scale of Eating Disorder Inventory (EDI, Garner et al., 1991) is made of 9 items and has a scale of 6 points going from "never" to "always". In order to evaluate the accuracy of the Body Shape Questionnaire (BSQ) in France, we made a test/retest. The concurrent validity of the BSQ has been evaluated by the EDI. The 242 subjects have been asked to fill in both questionnaires during the test and the re-test (4 weeks after the test). We have evaluated the accuracy of the BSQ with the calculation of the constancy test/retest (Pearson's r) and the measure of the internal consistency (Cronbach's alpha). Then, we tested the validity regarding an external criteria. The validation procedure is based here on an examination of the correlations between the tests results (BSQ) and another measure taken as criteria (EDI). The solutions proposed by the factorial exploratory analysis have been tested by confirmatory analysis using the 2 index of adjustment: the GFI ("Goodness of Fit Index") and the CFI (Comparative Index of Bentler) which, greater than 0.85, shows a satisfying adjustment and the RMC (redidue) which has to be less than 0.10 for a satisfying adjustment. Finally, inferential statistical analysis have studied the relations between the weight's category and the total result at the BSQ and the results of the factors sprung from factorial analysis.

Correlations between the items of test/re-test BSQ are greater than 0.93. The alpha coefficient is high for the test (0.95) and also for the re-test (0.94). This shows a very good internal consistency between the 34 items of the BSQ. Concerning the EDI, the correlation between the test/re-test are greater than 0.98. The alpha coefficient of Cronbach is high: 0.85 for the test and 0.84 for the re-test. The Pearson's r of the BSQ/EDI test and the BSQ/EDI re-test greater than 0.7 point out the concomitant validity of the BSQ with the EDI. We made a factorial analysis of the BSQ (test) on the 242 subjects. We used the extraction of the main components method with the extraction's rule of the curve for the real values. The transformation method used here is the normalised Varimax. The factorial analysis shows four factors with their own value greater or equal to 1 (15.1; 1.77; 1.48; 1.08). This solution concerning these four factors explains 55.2% of the total variance. The oblical rotations analysis of the four factors seems to point out a high correlation between these latters (from 0.54 to 0.77), this suggests a hierarchical pattern with a single factor which is confirmed with an Cronbach's alpha of 0.95. Then, we chose to gather around each selected factors the items presenting a substantial saturation, greater than or equal to 0.5 and which are saturated by only one factor. The first factor obtained by the factorial analysis of the BSQ has been called "social avoidance and shame of the exposure of the body". The internal coherence of this factor is satisfactory (Cronbach's alpha = 0.87). The second factor obtained has been called "body dissatisfaction compared to the lower parts ot the body (Cronbach's alpha = 0.90). The third factor has been called "using laxatives and vomiting in order to reduce body dissatisfaction" The fourth factor has been called "unsuited cognitions and behaviours in order to control the weight (Cronbach's alpha = 0.76). The solution with 1 and 4 factors has been tested by a confirmatory analysis. The adjustment parameters of the unifactorial pattern were not satisfactory (GFI = 0.76, CFI = 0.83, RMC = 0.5). The 4 factors pattern adjusted better to the data (GFI = 0.86, CFI = 0.90, RMC = 0.5). In order to determinate the differences between the weight/score categories to the BSQ) score in comparison to the 4 factors (coming from the factorial analysis), several Anova have been released as well as post-hoc tests (test of Sheffé). There is a significant effect of the weight category on the corporal dissatisfaction (BSQ result), F (2.230) = 11.34; p < 0.0001. Scheffe's test placed in a prominent position the fact that the subjects having an overweight (99.85 +/- 31.34) have a corporal dissatisfaction significantly greater than the subjects having a normal weight (74.08 +/- 27.94) and presenting a thinness (63.19 +/- 23.61). We pointed out a significant effect of the weight category on factor one "social avoidance and shame of the exposure of the body" [F (2.232) = 20.18; p < 0.001], on factor 2 "body dissatisfaction compared to the lower parts of the body" [F (2.232) = 6.38; p < 0.005] and factor 4 "reduce body dissatisfaction and unsuited cognitions and behaviours in order to control the weight" [F (2.232) = 5.67; p < 0.005]. However, there is no effect in the weight category for factor 3 "using laxatives and cominting in order to reduce the corporal dissatisfaction" [F (2.232) = 5.67; p > 0.05]. Sheffe's test highlighted the fact that the subjects having overweight have a result more important than the ones having a normal weight and presenting a thinness for factor 1.2 and 4.

The BSQ "French version" seems in fact to present the same metrical qualities than its original Anglo-Saxon and Spanish versions. The concluding results invite us to continue our study of the BSQ including in our patients, the bulimic subjects and the ones who are worried about corporal aspect. Its implications on the clinical field could be really interesting especially on people suffering from the eating disorders. Then, the BSQ could provide us a way to explore the role of an extreme worry of the body's appearance in the development, the maintenance and the treatment of the eating disorders.