Where we are in the evaluation of federal social welfare programs

It may seem self-evident that social welfare programs should regularly be assessed and refined in the light of lessons drawn from experience. Nevertheless, systematic program evaluation is a recent development. Modern evaluation practice is generally agreed to have begun in the middle 1960s, when initial attempts were made to evaluate programs enacted or proposed as part of the War on Poverty.' Evaluation has since spread rapidly; today, almost every substantial social program is subjected to some form of e ~ a l u a t i o n . ~ At the same time, evaluation has evolved into both a professional discipline and an i n d ~ s t r y . ~ There now exists a consensus that program evaluation is important and should be an integral part of the policy process. But there is no consensus on the manner in which evaluations should be performed and the way their findings should be interpreted. At the moment, the most heated controversy concerns the relative merits of statistical analysis of controlled social experiments and econometric analysis of actual program outcomes. A less visible, but simmering, dispute questions the logic of the traditional distinction between "process" and "impact" evaluations.