Non-experimental evaluations of programs compare individuals who choose to participate in a program to individuals who do not. Such comparisons run the risk of conflating non-random selection into the program with its causal effects. By randomly assigning individuals to participate in the...