Despite their impor tance in assess ing the impact of pol i cies, out come eval u a tions—and in par tic u lar ran dom ized exper i ments—are rel a tively rare. The ratio nal iza tions used to jus tify the absence of out come eval u a tions include such asser tions as “we know our pro grams are work ing,” “they can’t pos si bly harm any one,” and “if they only help one kid they’re worth it.” Us...