Both AN and weight restricting behaviours constitute important problems for girls of all ages. While not as severe as AN, weight restricting behaviours are of concern due to their high prevalence and potential adverse long-term effects in the female population. Conversely, while not as prevalent as weight restricting behaviours, the psychosocial and physical costs of AN are disproportionately large. Furthermore, the two constructs are highly related. While it is thankfully true that not all those engaging in weight loss behaviours go on to develop AN, weight restriction methods are the most consistently implicated variables in the pathogenesis of the disorder.
The primary aim of Study 1 was to test of a model of weight restriction cross-sectionally using a path analytic SEM procedure with a sample of Year 8 adolescent girls. The final model provided an excellent fit of the data and was validated by split half procedures. Specifically, it was found that: weight restriction was predicted by weight preoccupation (WP); WP was predicted by upset induced by teasing, body dissatisfaction (BD) and depressive symptoms; BD was predicted by upset induced by teasing, body mass index (BMI) and negative attributional style and; depressive symptoms were predicted by upset induced by teasing and negative attributional style. The original proposed link between BD and depressive symptoms was not supported, but a non-hypothesised link between BMI and negative attributional style was identified.
While cross-sectional investigations allow for some indication of risk factors in the development of a particular disorder or problem, the possibility always exists that results are due to common method variance between the proposed risk factors and dependent variable. The primary aim of Study 2 was therefore to investigate whether the model derived from Study 1 would be upheld longitudinally for a large sample of female adolescents over Years 8, 9 and 10. The results of Study 2 were longitudinally supportive of only some aspects of the proposed model. With respect to the prediction of weight restriction (the major focus of Studies 1 and 2), only BD and WP were found to be important longitudinal predictors. Both BD and WP at earlier time points were found to predict weight restricting behaviour at later time points, and BD was found to have an additional indirect effect on weight restriction through WP as predicted by the model. WP was predicted longitudinally by both weight restricting behaviour and BD, while BD was predicted longitudinally by weight restricting behaviour, upset induced by teasing about weight, BMI and negative attributional style. Depressive symptoms were predicted longitudinally by weight restricting behaviour, upset induced by teasing about weight and negative attributional style. BD was found to have an indirect effect on depressive symptoms through its effect on upset induced by teasing about weight. Finally, upset induced by teasing about weight was longitudinally predicted by BD and BMI, while BMI was predicted by BD over time.
The primary aim of Study 3 was to investigate the continuum hypothesis of eating disorders which asserts that weight loss practices lie on a continuum with non-disordered weight restriction at one extreme and the eating disorders of AN and BN at the other. The continuum hypothesis was tested using risk factors identified as being associated with weight restriction in Studies 1 and 2 (i.e., WP, BD, upset induced by teasing, depressive symptoms and negative attributional style), in addition to factors thought to be more specific to AN (i.e., perfectionism, general self-control, "self-control as self-esteem", and ineffectiveness). A MANOVA was conducted by group with matched samples of anorexic patients (ANs), high weight restrictors (HWRs) and low weight restrictors (LWRs). Support for discontinuity was evident for depressive symptoms, ineffectiveness, perfectionism and general self-control, while clear support for the continuum hypothesis was found for WP and negative attributional style. However, BD, upset induced by teasing and "self-control as self-esteem" did not provide support for either the continuity or discontinuity hypotheses. Those with AN were therefore found to differ from HWRs and LWRs in that they reported less behavioural self-control, and demonstrated higher levels of ineffectiveness, depressive symptoms and perfectionism. The results are discussed in terms of the strengths and weaknesses of each study, the implications of the research, and suggestions for future investigations.