Using Genetic Data to Estimate Causal Influences in the Obesity-SES Relationship

Associational studies can identify relationships between variables, but -- on their own -- cannot untangle causal relationships. The problem of causal inference is especially central in the social sciences, where multiple poorly understood, difficult to characterize, and interacting causal processes may be relevant. The use of genetic information as instrumental variables (IVs) for physical traits -- sometimes referred to as Mendelian Randomization (MR) -- presents a general approach to causal interpretation under appropriate circumstances. To date, MR-based research design has largely been confined to public health and biomedical research. With the increasing availability of genetic information, and the linkage of genetic information to behavioral and social data, MR (when used appropriately) has the potential to provide social science researchers an important new technique to engage in well-founded causal inference on a range of socially important questions.