Data analysis
Data were analyzed using the Statistical Package for Social Sciences for Windows (SSPS, v. 11, Chicago, IL, 1999). Descriptive statistics were used to determine means and standard deviations of all constructs as well as age, gender, and BMI. Mean values for the constructs for boys and girls were compared using independent t-tests to identify any gender or age differences. No differences were found based on age.
Because gender differences were observed all analyses were conducted by gender. Pearson correlations were conducted to examine associations between the psychosocial variables (constructs) for both the original TPB model, then for the expanded TPB model. Because the PBC construct was only weakly associated to girl's behavior and not associated to girl's intention or boy's behavior or intention, the expanded version of TPB model was used for further analyses. In order to examine constructs (attitude, subjective norm, PBC, self-efficacy, and barriers) most predictive of intention and behavior, stepwise regression analyses were performed respectively. Additionally, further stepwise regression analyses were conducted with the construct that was most predictive of behavioral change for boys (subjective norms) and girls (barriers). Both of these issues will need to be addressed in future interventions if behavioral change is to be achieved. The level of significance was set at p < 0.05 for all statistical tests.