Experimentation and Innovations

Abstract

Society cannot afford to institute poorly tested reforms because a substantial fraction of innovations, when carefully tested, prove not to be beneficial.

Poorly controlled investigations are likely to make an innovation appear to perform better than it actually does.

Where applicable, well controlled field trials of innovations can offer valuable information. The better opportunities occur when dealing with effects on small, independent, individual units such as individuals, households, classrooms, or small geographic units. The difficult problems of changing organizations or markets where the chain of causation leading to the output is long seem less likely to yield to this method, though demonstrations may still be useful.

Although individuals must be safeguarded in setting up experiments, the usual arguments against social experimentation such as concern for putting people as risk, opposition to withholding a good, the time-consuming nature of experimentation and its costliness, do not stand up well when compared with the alternatives of haphazard installation of innovations. Such methods give us little chance to appraise the benefits of innovations.

Alternative methods of investigation, although often valuable, have the weakness that they compare different situations as they stand but do not actually make changes in treatment in the field and observe their effect.

Society needs to continue to extend and improve its methods of controlled experimentation in the appraisal of social, economic, and medical innovations.