Hausman and McPherson provide an evidential defense of welfare economics, arguing that preferences are not constitutive of welfare but nevertheless provide the best evidence for what promotes welfare. Behavioral economics identifies several ways in which some people's preferences exhibit anomalies that are incoherent or inconsistent with rational choice theory. I argue that the existence of these behavioral anomalies calls into question the evidential defense of welfare economics. The evidential defense does not justify preference purification, or eliminating behavioral anomalies before conducting (...) welfare analysis. But without doing so, the evidential defense yields implausible welfare implications. I discuss how the evidential defense could be modified to accommodate behavioral anomalies. (shrink)