PURPOSE: To determine whether weight-based similarities among adolescent friends result from social influence processes, after controlling for the role of weight on friendship selection and other confounding influences. METHODS: Four waves of data were collected from a grade 8 cohort of adolescents (N = 156, mean age = 13.6 years) over their initial 2 years of high school. At each wave, participants reported on their friendship relations with grade-mates and had their height and weight measured by researchers to calculate their body mass index (BMI). Newly developed stochastic actor-oriented models for social networks were used to simultaneously assess the role of weight on adolescents' friendship choices, and the effect of friends' BMIs on changes in adolescent BMI. RESULTS: Adolescents' BMIs were not significantly predicted by the BMI of their friends over the 16 months of this study. Similarities in the weights of friends were found to be driven predominantly by friendship selection, whereby adolescents, particularly those who were not overweight, preferred to initiate friendships with peers whose weight status (overweight/nonoverweight) was the same as their own. CONCLUSIONS: Weight-based similarities among friends were largely explained by the marginalization of overweight adolescents by their peers, rather than by the "contagion" of excess weight among friends. These findings highlight the importance of adequately modeling friendship selection processes when estimating social influence effects on adiposity.
PURPOSE: To determine whether weight-based similarities among adolescent friends result from social influence processes, after controlling for the role of weight on friendship selection and other confounding influences. METHODS: Four waves of data were collected from a grade 8 cohort of adolescents (N = 156, mean age = 13.6 years) over their initial 2 years of high school. At each wave, participants reported on their friendship relations with grade-mates and had their height and weight measured by researchers to calculate their body mass index (BMI). Newly developed stochastic actor-oriented models for social networks were used to simultaneously assess the role of weight on adolescents' friendship choices, and the effect of friends' BMIs on changes in adolescent BMI. RESULTS: Adolescents' BMIs were not significantly predicted by the BMI of their friends over the 16 months of this study. Similarities in the weights of friends were found to be driven predominantly by friendship selection, whereby adolescents, particularly those who were not overweight, preferred to initiate friendships with peers whose weight status (overweight/nonoverweight) was the same as their own. CONCLUSIONS: Weight-based similarities among friends were largely explained by the marginalization of overweight adolescents by their peers, rather than by the "contagion" of excess weight among friends. These findings highlight the importance of adequately modeling friendship selection processes when estimating social influence effects on adiposity.