Understanding Variation in Health-Related Behaviors: Evidence from American Twins

Nikkil Sudharsanan, University of Pennsylvania
Hans-Peter Kohler, University of Pennsylvania
Jere Behrman, University of Pennsylvania

Health-related behaviors contribute significantly to U.S. morbidity and mortality, yet empirical evidence on causes of within-population variation in health-related behaviors is mixed. For example, schooling is frequently assumed to affect different health-related behaviors, but several studies find that schooling does not explain much variation in health behaviors. This paper advances knowledge on the relative importance of schooling, genetic endowments, and environments using a novel ACE-ß twin model that integrates behavioral-genetic approaches focused on variance decomposition and economics within-MZ twin pair design focused on causal inferences. We find that schooling does not have significant effects on health-related behaviors. A large proportion of variation in BMI and health seeking behavior is due to genetic endowments with a negligible contribution from common environments. This pattern is reversed for cigarette smoking with common environments contributing a large portion to the overall variation. We conclude by discussing the policy implications of our findings.