Comparing "models": the best fit for social predictors of bulimic pathology

The number and variety of theories and research on bulimia nervosa (BN) and its related symptoms indicate that using a multidimensional approach when examining the etiology and risk factors of the disorder is paramount because no single factor seems to be predictive of who will and will not develop BN. More specifically, approaches differentiating among biological, psychological, and social components are often considered when attempting to determine the etiology of BN (Stieger & Seguin, 1999). With respect to social components, different aspects of interactions with the environment (and the people within it) may serve as contributors to bulimic pathology. Although much research has focused on a few specific dimensions of social interactions involved in eating pathology, little research has examined a wide range of social interactions in the prediction of bulimic behaviors and attitudes. Therefore, one purpose of this study was to comprehensively examine the relationships of a wide variety of social interactions with eating pathology, specifically, BN. The social variables included as predictors of bulimic pathology in this study are related to sociocultural encouragement of the thin-ideal, family functioning, and social support. Questionnaires, which also included three measures of bulimic pathology were completed by 289 European American female college students.

The main data analysis was structural equation modeling. This analysis determined if the predictors fit into higher order factor structures differentiating between types of social interactions and sources of social interactions. Two models were tested to establish the best fit in regard to predicting bulimic symptoms. Model A (see Figure 1) investigated the prediction of bulimic symptoms according to types of social interactions. Model B (See Figure 2) investigated the prediction of bulimic symptoms according to sources of social interactions.

In sum, the current study examined whether the social variables included in this study were better predictors of bulimic pathology when grouped according to types of social interactions (such as social reinforcement, family functioning, and social support), or when grouped according to sources of social interactions (the media, family, and peers). Results regarding differences in model fit are discussed, and implications for theory, research, and practice are explored.