Agrees with Seymour Papert that conventional experimentation has shortcomings when used to study effects of educational programs. Cites three research methods that, in combination, can compensate for the shortcomings of both conventional research and the computer criticism suggested by Papert. (PS)