In general, research on the effects of social ties on health behavior is vulnerable to mis-attributed causality (Duncan et al. 2004). A person’s behavior may elicit responses from significant others or change the tone of relationships in subtle or overt ways. Moreover, other factors not directly assessed may contribute to social ties and health behaviors in ways that produce observed links between the two, real or not. Steady progress is being made toward addressing these problems, primarily through the increased availability of longitudinal data that allow for tests of bidirectionality among constructs and the application of more rigorous tests of omitted variable bias (Jaccard et al. 2005, Jeffery & Rick 2002, Sobal et al. 2003). The development of a general class of sophisticated techniques for dealing with unobserved confounds, such as fixed-effects modeling, propensity scores, robustness indices, and instrumental variables, should be used more extensively in the future for dealing with endogeneity of social tie effects (Lundborg 2006, McCaffrey et al. 2008).
