It is vitally important to understand the causes of the rise in adolescent overweight and to develop effective public health policies to reverse the current upward trend. Adolescent weight-related behavior (e.g., diet and physical activity) is often influenced by the behavior of peers. If peers do have an influence on adolescent weight, then policies aimed at reducing adolescent overweight could have social multiplier effects in which the impact of the policy is larger in the aggregate than for individual participants. Our study will be the first to estimate the influence of peers on adolescent weight-related behavior while explicitly controlling for potential reverse causation (reciprocal effects) and adolescents'choice of peers (selection). Accounting for these effects is important because if the correlation between adolescent behavior and peers'behavior is not causal, then policies would not lead to social multipliers. The proposed research will address the following specific aims:
Aim 1 : To identify the mechanisms through which peers'weight affects an adolescent's own weight. Using data from the National Longitudinal Study of Adolescent Health (Add Health), we will examine the influence of peers on the following adolescent weight-related behaviors: (a) self-perception of weight, (b) physical (in) activity levels, (c) eating patterns, and (d) weight management activities. Add Health allows us to define peer groups at several levels, including using nominated friendship relationships and grade level. Several empirical strategies will be used to address Aim 1, including a multistage procedure that estimates friend selection based on exogenous variables;an alternative definition of peers at the grade level, which within schools is not affected by selection;lagged value of peer variables using the first two waves of Add Health to control for reverse causation;and the use of indicator variables for each school to control for confounding (unobserved) influences at the school level. We will test for gender differences in the influence of peers. Results for Aim 1 will help policy makers target those behaviors that have the potential for the largest spillovers from targeted adolescents to their peers.
Aim 2 : To estimate the implied social multiplier for policies aimed at adolescent overweight. Using the results from Aim 1, we will calculate the size of social multipliers for each weight-related behavior. Understanding the magnitude of the social multiplier is crucial for assessing the effectiveness and cost-effectiveness of public health policies aimed at improving adolescent weight. Failure to account for social multipliers would undervalue the population-level impact of effective public health policies to reduce adolescent overweight.
If peers have an influence on adolescent weight, then policies aimed at reducing adolescent overweight could have social multiplier effects in which the impact of the policy is larger in the aggregate than for individual participants. The accurate estimates of social multipliers for adolescent weight-related behavior that will be provided by this research are crucial for evaluating the effectiveness of public health policies. Failure to account for social multipliers would undervalue the population-level impact of effective public health policies to reduce adolescent overweight.
