The dominant theoretical approaches in the comparative political economy of the welfare state provide alternative accounts for why some governments spend more on social policies than others. In the first, poor voters seek to increase their current income by taxing the rich, and social policy serves to redistribute income from the rich to the poor. In the second account, voters seek social insurance against future job loss, and social policy serves as an insurance mechanism rather than a redistributive one. Both of these accounts share the assumption that voters can clearly distinguish between the redistributive and insurance elements of public policy and, therefore, that individual-level characteristics (income, labor market risks) systematically shape preferences over social policy. Our goal is to examine the soundness of that behavioral assumption. We do so with a laboratory experiment that involves economic production, voting on taxation and fiscal transfers. We treat subjects with social policies that vary in their level of redistribution and insurance to examine how this impacts their preferred tax rate. We complement the experimental evidence with data from original survey questions that assess voters’ knowledge of the distributive characteristics of different social policies in the U.S. Evidence from both settings suggest only marginal support for behavioral underpinnings of the standard insurance model, particularly as the empirical setting more closely approximates the real world.