The logic of the evaluation of educational and other action programs is discussed from a methodological viewpoint. However, no attempt is made to develop methods of evaluating programs. In Part I, the structure of an educational program is viewed as a system with three components--inputs, transformation of inputs into outputs, and outputs. Part II discusses the necessary condition for a program to be a system (the presence of feedback loops) citing as one example the school system with an evaluation unit. In Part III, the possibility of mapping experimental designs into social space characterized by feedback loops is confirmed while refuting statements by Stufflebeam to the contrary. Part IV examines the historical precedence for the findings and concludes that it is possible, from a methodological viewpoint, to implement a rigorous experimental design and also to provide feedback for managerial decisionmaking in the context of action research. (HW)