Objective: The purpose of this paper is to analyze whether weight gain among adolescents are influenced by their peers weight gain. An extensive array of literature identified the significant effect that peers have on adolescent behavior such as smoking, alcohol consumption, drug use and academic performance, to name a few. Building on the existing literature on peer effects we extend our analysis by empirically quantifying the role of peer social networks to explain the recent escalation of obesity among adolescents and how does this affect their body weight as they make a transition into adulthood.

Data/Methods: We perform our analysis by using all three waves (1994, 1996 and 2002) of the National Longitudinal Study of Adolescent Health (AddHealth). Our models are estimated not only on the adolescents’ obesity status and their BMI, but also on their risk of being overweight and a categorical variable of body weight (under weight, normal weight, risk of overweight and overweight). Since individual behavior may move simultaneously with average group behavior, a standard regression of individual behavior on group means cannot separately identify peer effect. This identification difficulty, or the reflection problem, arises because group behavior is by definition the aggregation of individual behavior. We use several alternative approaches to identify and estimate peer group effect, which include: using individual level and school level fixed effects models to capture unobserved individual-level heterogeneity and common effect of environment; using two-stage method and instrumenting peer weight and peer weight gain with peer family background (parent education) and peer parents’ weight status; utilizing longitudinal  aspect of the data and using lagged peer group behavior as the peer measure; specifying peer effects in a spatial setting and estimating spatial autologistic model using sample information, which should provide  a consistent estimate of the peer effect.

Results: After controlling for the possible endogenity of peer selection we find peer effects to be statistically significant not only during early adolescent years but also during late adolescent years. Specifically we find that having friends who are obese increases your probability of being at risk overweight by 44% and being overweight by 59%. Such effects are consistent among both males and females, with the effect being larger for males. Peers of the same sex have relatively greater influence than those of the opposite sex. These findings have potential important implications for public health interventions and policies. The spread of obesity through peer social networks appears to be a contributing factor in the obesity epidemic and as such this also suggest possible ways to reduce the spread of obesity. The peer network phenomenon can be utilized to promote positive health behaviors and might potentially suggest policy interventions to be more cost effective since health improvement in one person will have a multiplier effect and spill over to others.