Date of Award


Document Type



Edith Cowan University

Degree Name

Doctor of Philosophy


School of Psychology


Faculty of Community Services, Education and Social Sciences

First Supervisor

Dr Moira O'Connor

Second Supervisor

Dr Loraine Corrie


The purpose of this research was to study the association between adult mental representations of curly attachment relationships and eating disorders, and to compare attachment classifications between anorexia nervosa and bulimia nervosa. The sample consisted of 62 women with eating disorders, divided between anorexia nervosa [restricting type (n=20) and binge-eating/purging type (n=l6)], and bulimia nervosa (n=26). The measures were (a) a structured interview based on the ''Diagnostic Survey for Eating Disorders" (Johnson and Connors, 1987) and the DSM IV criteria for the classification of eating disorders, and (b) the Adult Attachment Interview (AAI) (George, Kaplan and Main, 1985). The AAI was audio-taped, transcribed verbatim, and then scored by two independent scorers, following Crittenden (1998, 1999)'s dynamic-maturational model of attachment. The hypotheses were: (a) women with eating disorders show a high proportion of insecure attachment classifications, (b) the insecure attachment patterns are different (dismissing versus preoccupied) between women with anorexia nervosa and women with bulimia nervosa, and (c) the mental representation of early attachment of women with eating disorders with their fathers is characterised by negativity. The results showed that: (a) 60 out of 62 participants had an insecure attachment classification; (b) 65.4% of women with bulimia were classified preoccupied; of those with restricting anorexia, 42.1% were classified dismissing and 42.1% were classified combined dismissing/preoccupied; and for women with bulimic anorexia, 33,3% were dismissing, 46,7% preoccupied, and 20% combined dismissing/preoccupied [X² (4, N=60) = 11.337, p = .02]; and (c) concerning the "degree of negativity" in the mental representation of early attachment to the father, bulimic women were slightly more negative (57.7%) than positive (42.3%), restricting anorexics were more positive (70.0%) and bulimic anorexics were more negative (75.0%). These results were significant [X² (2,N = 62) = 7.589, p = .02]. The implications are: (a) For research, the application of the Adult Attachment lnterview to a clinical population; (b) for treatment, to differentiate appropriate treatments between types of eating disorders; and (c) for prevention, to underline the importance of early support and intervention in families "at risk".