Room: Phillips 403
Time: Mon 13:15 PM-14:45 PM

Adolescent obesity: the importance of peer effects and environmental influences

Chair: Barbara Wolfe

Session Description

The increase in the prevalence of obesity in the United States in recent decades has yielded considerable
attention. Numerous studies have investigated both the antecedents and consequences of being
overweight or obese. The findings that body weight depends not only on biological factors, but also on
behavioral and environmental factors (Cutler, Glaeser, & Shapiro, 2003; Philipson & Posner, 2003;
Christakis & Fowler, 2007) implies that interventions that mitigate behavioral and environmental
influences are important in policies aimed at addressing this growing problem.
Though previous studies suggest that being overweight or obese is in some sense, socially
contagious (Christakis & Fowler, 2007; Trogdon, Nonnemaker, & Pais, 2008), these studies
predominately evaluate the social influences of friends or family members. This is problematic because
individuals clearly do not randomly choose such relationships. Studying peer effects requires addressing
the following potential biases: 1) selection bias as a result of the fact that individuals do not randomly
choose their peers and in fact are likely to choose peers who are “like” them; 2) endogeneity or reflection
bias resulting from the joint influence of the peers on each other and 3) confounding due to shared
contextual or environmental influences. Most of the literature has relied on instrumental variables for
identification. That is, researchers have looked for variables that randomly affect peers’ weight but that
do not directly affect the weight of the individual. Examples include peer’s birth weight and peer’s
parents’ weight (or obesity status). The results, mostly using the National Longitudinal Study of
Adolescent Health (Add Health), find positive effects of peer weight even after controlling for
endogeneity.
This proposed session includes three studies that offer strategies for identifying and measuring
the causal link between social/physical environment and weight gain. Cawley, Han, & Norton’s paper
will use the National Longitudinal Study of Adolescent Health data to take advantage of genetic lotteries
to deal with the reflection bias in estimating peer effects in weight gain among adolescents. Yakusheva,
Kapinos, & Weiss’s paper exploits randomized roommate assignment to study peer effects in weight gain,
using primary data collected specifically for this study. Kapinos & Yakusheva’s paper uses the same
primary data to exploit another natural experiment, randomized dormitory assignment, to study the effect
of physical environment on weight gain of freshman college students.
Taken together, these papers will present findings that could have important practical
implications for public health efforts aimed at reducing obesity, and they will be of great interest to public
health policy makers, medical practitioners, and researchers alike.

Key Terms None Listed

Session Organizer: Kandice Kapinos (University of Michigan)

---

Presentations

1. Identifying the Peer Effect of Obesity Using Genetic Lotteries

   Presenter: John Cawley (Cornell University)

   Discussant: Olga Yakusheva (Marquette University)

2. Peer Effects and the Freshman 15: Evidence from a Natural Experiment

   Presenter: Olga Yakusheva (Marquette University)

   Discussant: Katherine Carman (Tilburg University)

3. Evidence of Environmental Influences on Weight, Eating, and Exercise from a Natural Experiment

   Presenter: Kandice Kapinos (University of Michigan)

   Discussant: Irina Grafova (University of Medicine and Dentistry of New Jersey)