Cross-sectional studies find a positive relationship between a state's welfare benefits and single motherhood. But is this evidence of a "welfare effect" or rather of cross state differences in social attitudes that influence both policy and behavior? This paper demonstrates that the spatial variation in welfare policy long preceded the spatial correlation of policy and behavior, undermining the social norm hypothesis. But the findings also raise doubts about the role that welfare policy played in the changes in family structure over the century. The correlation between welfare benefits and family structure only appears in 1970, and then only for whites. (Contains 27 footnotes, 6 tables, and 6 figures.)