As new and innovative social welfare programs are being attempted, there has been an increased concern with evaluating the effectiveness of such programs. To what degree is a new program effective? For which kinds of clients is each type of program effective? What elements are crucial in a program which has been judged to be effective? These are just a few of the questions that evaluators would like to answer.

There is a large literature on evaluation research--some of it reporting or reviewing the results of specific evaluations (6, 8, 9, 10, 11) and some of it presenting general discussions, essays or models (1, 2, 3, 4, 5, 6, 7, 8). This literature outlines many important problems and suggests imaginative solutions to a number of those problems. However, in our judgment, some of the literature has viewed all evaluations as alike; insufficient attention has been given to the basic differences between different types of evaluation. The objective of this paper is to clarify this issue.

We argue that there are three distinct types of evaluation that can be articulated on the basis of the intent of the evaluation: (1) The evaluation is a test of some basic but general scientific hypothesis that underlies or motivates a proposed intervention. 2) The general scientific hypothesis is found or assumed to be valid, and the evaluation is an assessment of the effectiveness of a particular program that supposedly embodies the hypothesis. (3) After a particular program has been shown to be effective at some point in time, the evaluation is a continuing assessment of the effectiveness of the program. An awareness of these three types is important because the methods of evaluation are partly dictated by the intent of the evaluation.

These three types will be referred to, for the nonce, as evaluation as- experiment (testing a scientific hypothesis), evaluation-as-assessment (judging the effectiveness of a particular program), and evaluation-as monitoring (continuously examining the effectiveness of the program). Each of these will be discussed in turn.