The role of attribution in weight anxiety and eating disorders in women
This thesis examines weight anxiety and eating disorders amongst women from an attributional perspective. The studies comprise two distinct but interrelated components: investigations of the role of attribution in the development and maintenance of weight anxiety and eating disorders; and analyses of two screening instruments for eating disorders. The study presented in Chapter 2 examines whether, amongst women without an eating disorder, attributions can be extracted in a factor-analytic manner to form relevant stereotypes associated with 'fat' or 'thin' women. The results confirm that clear weight-related factors can be extracted, and these resemble common stereotypes of fatness and thinness. In Chapters 3 and 4 the attributions of both eating disordered and non-eating disordered women were examined. Four groups have been examined: the Anorexic group; the Bulimic group; the Over-Eater group; and the Non-Eating Disordered group. It is proposed that the attributions of eating disordered women and non-eating disordered women will differ; that the various eating disordered groups will also differ in their attributions; and that attributions will differentially contribute to the development and maintenance of different weight anxieties and eating disorders. The results confirm these propositions: this thesis shows that there are attributionaI differences between the women with and without eating disorders; that there are differences in the attributions of the Bulimic, Anorexic and Over-Eater groups; and that there is evidence to suggest that attributions contribute differentially to the development and maintenance of weight anxiety and eating disorders. The internal validity of the subscale structure of the Eating Disorder Inventory (EDI) (Garner, Olmsted and Polivy, 1983) is examined in Chapters 5 and 6, and Chapter 6 analyses the structure of another assessment instrument, the SCANS (Slade and Dewey, 1986). The results confirm the doubts raised about the validity of the subscale structure of the EDI: no clear support for the subscale structure put forward by Garner et al. has been found, and the factor analyses indicate that no single factor structure can be replicated. The factor analysis of the SCANS, presented in Chapter 6, suggests that the subscale structure of this measure is also not well replicated. It is argued that the subscales of both instruments should be used only with caution. Chapter 5 also examines weight anxiety in older women. This study has shown that older women also are anxious about their weight and that their weight anxiety takes a similar form to those of younger women; for this group of women lifestyle might be an influential factor in their weight anxiety. Chapter 7 discusses all the issues raised in the earlier chapters in relationship to the literature. Finally, the findings of this thesis are discussed, models of the role of attributions are outlined and directions for future research are discussed.