Abstract

Prior work has analyzed the extent to which welfare recipients engage in giving money and time to charitable causes (Brooks, 2002, 2004; Author4 & Author1, 2009; Author1 & Author4, 2011), finding that public assistance is negatively associated with donating money with the relationship to volunteering being unclear. Nevertheless, sticky issues of selection bias compel more deliberate thinking about the strength of assertions about cause and effect. In response, we now conduct a multivariate cluster-based subgroup analysis approach to more confidently infer causality about the ways in which welfare receipt affects charitable activity. This approach to dealing with the problem of selection bias capitalizes on the known treatment-associated variance in the X matrix, transforming data to estimate unbiased treatment effects. We contribute to both the substantive and methodological literatures with this work.