John: Education is a special case. Jobs programs (Remember CETA? It was huge.) pretty much disappeared, in some important degree because the evaluations were so negative. Serious social scientists, even of the Left, slowly but finally have come around to accepting that children of never-married women really do worse than children of married biological parents, even after controlling for all of the usual suspects (income, race, education, etc.). Welfare got a bad name among social scientists, and that helped with welfare reform. Even in education, the evaluations of bilingual education were helpful in undercutting its political support. But education is the last refuge of policy never-never land, and I grant your point. Maybe the increasing examples of vouchers and charter schools owe something to evaluations of education, but otherwise — mushy curriculum, touchy-feely self-esteem programs, the insistence that all kids can succeed on the academic track, refusal to be honest about what needs to be done to rescue inner-city schools — a lot of good policy analysis has been just as ignored as you say.