Evaluation and comparison of different versions of the Body Shape Questionnaire

Abstract

The Body Shape Questionnaire (BSQ) is a widely used scale to assess body dissatisfaction. Several short forms of the BSQ have been introduced. In the present study, the full-length 34-item version of the BSQ and seven derivations were evaluated. Factorial validity was analyzed by confirmatory factor analyses in a representative sample of German females (n = 1080). Treatment sensitivity was evaluated in a sample of 43 women with bulimia nervosa who completed the BSQ before and after cognitive-behavioral therapy. While fit indices of the full-length version of the BSQ were poor, three of the derivations had reasonable fit. If treatment sensitivity is considered, one of the short forms performed best. This derivation is one of the eight-item versions of the BSQ (suggested abbreviation: BSQ-8C). In conclusion, the present study presents the empirical background for choosing between eight different versions of the BSQ. Thus, it is no longer necessary for researchers and clinicians to make their choice based on subjective criteria. Nevertheless, research on the derivations of the BSQ as stand-alone questionnaires is needed.

Introduction

Body image disturbance is a core diagnostic feature of anorexia and bulimia nervosa (American Psychiatric Association [APA], 1994). For instance, patients with anorexia and bulimia nervosa are excessively concerned with body weight and shape, and their self-evaluation is unduly influenced by body weight and shape (APA, 1994). Even for patients suffering from a binge eating disorder (BED), increasing empirical evidence supports the existence of a negative body image, although body image disturbance has not yet been included in the DSM-IV for the diagnosis of BED (APA, 1994). For example, compared with non-eating-disordered individuals with a similarly increased body mass index, patients with a BED are more frequently concerned with body shape and weight, are more dissatisfied with their bodies, and reveal more negative body-related cognitions elicited by a mirror task (De Zwaan et al., 1994, Eldredge and Agras, 1996, Hilbert and Tuschen-Caffier, 2004, Hilbert and Tuschen-Caffier, 2005, Tuschen-Caffier and Schlüssel, 2005). Moreover, it has been shown that body image disturbance is a robust risk factor for the development of eating disorders (e.g. Stice, 2002, Jacobi et al., 2004).

But body image disturbance is a multifaceted construct, encompassing perceptual, affective, cognitive and behavioural aspects of body experience, as well as a broad range of phenomena such as body dissatisfaction, body image investment, and overconcern with body weight and shape (e.g. Cash and Pruzinsky, 1990, Thompson, 1990, Thompson et al., 1999). Many studies investigating body image disturbance in eating disordered patients have been done on perceptual aspects of body image disturbance operationalized as the accuracy of body size estimation (e.g. Lindholm and Wilson, 1988, Probst et al., 1997, Kulbartz-Klatt et al., 1999; for an overview, see Cash and Deagle, 1997). But contradictory results led researchers to question the validity and clinical significance of the accuracy of body size estimation (e.g. Hsu and Sobkiewicz, 1991).

In addition to the research done on body size estimation, a considerable number of studies have been conducted on cognitive-affective aspects of body image disturbance, using a broad range of assessment methods (e.g. self-report questionnaires, interviews, in-vivo cognitive assessment, or thought sampling procedures) and considering a spectrum of cognitive-affective phenomena in body image disturbances (e.g. Freeman et al., 1991, Lovell et al., 1997, Hilbert and Tuschen-Caffier, 2005). One of the cognitive-affective aspects of body image disturbance is body dissatisfaction, which is widespread in eating-disordered and non-eating-disordered females alike, and which seems to be remarkably stable throughout the life-span (e.g. Wood et al., 1996, Goldfein et al., 2000, Tiggemann, 2004). Furthermore, it has been suggested that body dissatisfaction may be a mediator of the relationship between dietary restraint and the development of eating disorders (e.g. Stice, 1994, Ricciardelli et al., 1997). Taking into account that body dissatisfaction seems to be a very relevant aspect of body image disturbance, there is a strong need to use measurements with established reliability and validity for applied and research purposes (e.g. Thompson, 2004).

The Body Shape Questionnaire (BSQ; Cooper et al., 1987) is a self-report scale to assess body dissatisfaction caused by feelings of being fat. It has already been used in many studies on body image disturbance (e.g. Masheb and Grilo, 2003, Tuschen-Caffier et al., 2003). The BSQ is also generally recommended for assessing eating disorder pathology in clinical settings (e.g. Anderson et al., 2004). Psychometric evaluations of the widely used 34-item version have confirmed its unifactorial structure, retest reliability, internal consistency, construct validity, concurrent validity, discriminant validity, and its sensitivity to detect treatment-related changes (Cooper et al., 1987, Rosen et al., 1996, Pook et al., 2002, Pook and Tuschen-Caffier, 2004, Ghaderi and Scott, 2004).

The length of the BSQ, however, has encountered widely varying receptions. Rosen and colleagues (1996) appreciated that the BSQ assesses the concept of body dissatisfaction in a very broad sense. Moreover, the broad conceptualization of body dissatisfaction has been taken to be one of the main reasons for the scale's high treatment sensitivity (Pook and Tuschen-Caffier, 2004). On the other hand, whether 34 items are really needed to assess a single concept has been questioned (Evans and Dolan, 1993). As a consequence, several short versions of the BSQ have been developed.

In one single study, six different forms were introduced (Evans and Dolan, 1993). In designing the forms, two items were excluded due to insufficient factor loadings in a first step. Next, the remaining 32 items were divided into two and four parts, comprising 16 and 8 items, respectively. The item content was not considered relevant when assigning an item to a form. Instead, the authors tried to achieve equal means of forms of similar length. All six newly derived versions showed reasonable psychometric properties in a non-clinical sample. Further analysis even suggested equivalency of the newly derived forms.

A 14-item version of the BSQ was introduced by Dowson and Henderson (2001). There is no information about the criteria used to select an item for this form. Dowson and Henderson (2001) reported sound psychometric properties when examining the 14-item version in a small clinical sample. Ghaderi and Scott (2004) evaluated the same version in a representative sample of Swedish females, a student sample and a clinical sample. In all three samples, the psychometric properties were found to be excellent.

Little is known, however, about whether the version of Dowson and Henderson (2001) performs better than any of the versions introduced by Evans and Dolan (1993). Moreover, it is uncertain whether any of the six versions of Evans and Dolan (1993) is superior to the others when compared in a large sample. Therefore, the present study compared the full-length version of the BSQ and the seven derivations in a representative sample of German females. To gain further information, the different forms were also compared with respect to their treatment sensitivity. For the latter comparison, data from a previously published treatment study (Tuschen-Caffier et al., 2001) were reanalyzed.