Using Social Science Experiments to Improve the Appraisal of Social Programs

There are many social programs that cannot be adequately appraised or evaluated because too little is known about people's behavior to judge the difference made by the program. Political judgements and biases may then unduly prejudice relevant decisions. Health care provides many examples; does charging for treatment have a noteworthy impact upon the nation's health, for instance? The author argues for long-term, large-scale experiments to answer such questions, giving the example of the Rand Health Insurance Study, Long-term observation (without experiment) is also relevant at times; the Panel Study on Income Dynamics in the USA is outlined. Such experiments and observational studies have drawbacks (they are costly in time and money), but their merits are widely underestimated. (Spring 1986)