OBJECTIVE: The overriding objective of this paper is to outline the steps involved in refining a structural model to explain differences in dieting status. METHODS: Cross-sectional data (representing the responses of 1,644 teenage girls) derive from the preliminary testing in a 3-year longitudinal study. A battery of measures assessed social influence, vulnerability (to conformity) disposition, protective (social coping) skills, and aspects of positive familial context as core components in a model proposed to account for the initiation of dieting. Path analyses were used to establish the predictive ability of those separate components and their interrelationships in accounting for differences in dieting status. RESULTS: Several components of the model were found to be important predictors of dieting status. The model incorporates significant direct, indirect (or mediated), and moderating relationships. Taking all variables into account, the strongest prediction of dieting status was from peer competitiveness, using a new scale developed specifically for this study. CONCLUSION: Systematic analyses are crucial for the refinement of models to be used in large-scale multivariate studies. In the short term, the model investigated in this study has been shown to be useful in accounting for cross-sectional differences in dieting status. The refined model will be most powerfully employed in large-scale time-extended studies of the initiation of dieting to lose weight. Copyright 1999 by John Wiley & Sons, Inc.
OBJECTIVE: The overriding objective of this paper is to outline the steps involved in refining a structural model to explain differences in dieting status. METHODS: Cross-sectional data (representing the responses of 1,644 teenage girls) derive from the preliminary testing in a 3-year longitudinal study. A battery of measures assessed social influence, vulnerability (to conformity) disposition, protective (social coping) skills, and aspects of positive familial context as core components in a model proposed to account for the initiation of dieting. Path analyses were used to establish the predictive ability of those separate components and their interrelationships in accounting for differences in dieting status. RESULTS: Several components of the model were found to be important predictors of dieting status. The model incorporates significant direct, indirect (or mediated), and moderating relationships. Taking all variables into account, the strongest prediction of dieting status was from peer competitiveness, using a new scale developed specifically for this study. CONCLUSION: Systematic analyses are crucial for the refinement of models to be used in large-scale multivariate studies. In the short term, the model investigated in this study has been shown to be useful in accounting for cross-sectional differences in dieting status. The refined model will be most powerfully employed in large-scale time-extended studies of the initiation of dieting to lose weight. Copyright 1999 by John Wiley & Sons, Inc.