Importantly, social scientists have begun to consider omitted variable bias associated with genetic traits. In some cases, genetically informed designs using large-scale twin data have suggested that long-standing conclusions about the strength of social influences on health behavior may be premature. For example, Harden and associates’ (2008) analysis of Add Health’s sibling subsample (see Harris et al. 2006) revealed no link between peer behavior and adolescent substance use once the genetic heritability of both was controlled. Yet the story is likely more complicated, with social and genetic influences interacting in complex ways. For example, Boardman and associates (2008) also analyzed the Add Health data and found that the genetic heritability of smoking was strongest in schools where smokers were popular. Similarly, Guo and associates (2008) analyzed non-twin data and found that genes related to risky behavior can be turned off in the early life course by positive parenting. Thus, the comparison of social and genetic influences in adolescent behavior is something of a false dichotomy. Gene/environment interactions are apparent throughout life. For example, individuals vary in genetic propensity for nicotine
