The purpose of this study was to develop, implement, and evaluate a primary and secondary prevention program for eating disorders. The intervention program in this study was adapted from a peer leader intervention similar to Kelly et al.'s (1992) AIDS prevention program. Four sorority houses comprised the sites used to test the effectiveness of the prevention program using a quasi-experimental wait-list control design. Participants in each house completed assessments of dieting behavior, dietary restraint, body image esteem, and health behavior self-efficacy on two occasions. Two sororities received the peer leader preventive intervention between these testing times while the other two sororities served as controls.

The experimenter hypothesized the intervention would decrease dieting behavior and dietary restraint while increasing body image esteem and self-efficacy for health behavior. To improve upon a previous pilot study, sororities were selected over female dormitories. Several factors led to the selection of the sorority houses rather than the dormitories. First, the diffusion of innovations model from which the peer leader program was developed assumes a closed cohesive group of individuals so the intervention can diffuse throughout the community. Second, the experimenter hypothesized that members of sororities would report higher rates of dieting relative to females who live in dormitories. Thus, sororities were selected as intervention sites since their members are self-selected into a cohesive group and they are potentially at a higher risk for developing eating disorders.

Peer leaders were selected from experimental sorority houses and trained to advocate no-dieting, healthy eating, and exercise for sorority members living in their house (i.e., primary prevention). These peer leaders were also trained to make effective treatment referrals for students with existing eating problems (i.e., secondary prevention). Manipulation checks suggested the training program for selected peer leaders was successful and that peer leaders effectively implemented the program via conversations in the experimental sorority houses.

Community posttest analyses were conducted to determine intervention effectiveness after the pretest analyses showed the groups did not differ. Community analyses at posttest suggested peer leaders benefited slightly from their training program. However, no significant differences were found between experimental participants and control participants. However, there was a minimal dose-response relationship for the intervention. Participants who knew the meaning of the Don't Diet symbol used in the intervention at posttest (i.e., implying they received a stronger dose of the intervention) were dieting less according to some variables than participants who did not know the meaning of the symbol. Participants who knew the meaning of the symbol also reported significantly greater body image esteem and self-efficacy for exercise.

In conclusion, although the effects of the intervention were not significant at the community level between experimental and control sororities for the hypothesized variables using the planned analyses, exploratory post-hoc analyses showed some positive effects for a subset of participants who were more knowledgeable about the intervention program.