Duke 2008 | ASHE | Equity and Efficiency in Health and Healthcare

This study is the first to estimate peer effects for adolescent weight. Peers can influence a variety of weight-related choices including healthy eating patterns (e.g., cafeteria vs. snack shop), dieting, physical activity (e.g., extra-curricular sports), and perceptions of acceptable weight. If peers do have an influence in determining adolescent weight, then policies aimed at reducing adolescent overweight could have social multiplier effects where the impact of the policy in the aggregate is larger than for individual participants. We use data from the National Longitudinal Survey of Adolescent Youth (Add Health), a nationally-representative, school-based survey of youth in grades 7 through 12. One of the unique features of this data is that students were asked to list up to 10 of their closest friends (five male and five female), which allows us to define peer groups using nominated friends within schools. We identify 2,800 students from the saturation sample who for which we can match at least one of their friends' weights. The dependent variables of interest are body mass index (BMI; kg/m2) and an indicator variable for being at-risk or overweight defined using standard definitions of adolescent overweight based on historical growth charts. The main variables of interest are the mean weight among friends and the proportion of friends that are at-risk or overweight. Endogenous peer groups are accounted for using a combination of school fixed effects, instrumental variables, and alternative definitions of peers (i.e., grade level peers). Add Health provides a number of variables for peers that are supported by the literature as being predictive of peers' weight and that are plausibly exogenous to the adolescent's own weight: birth weight, parents' obesity, and parents' health. Mean peer weight is correlated with adolescent weight, even after controlling for endogenous peer groups. Statistical tests indicate that our instruments are strong in the first stage and pass over-identification tests. The impact of peer weight is larger among females and adolescents with high body mass index. The results do not rule out social multipliers for adolescent overweight policies.