AbstractEvaluation research assumes the need for evaluation and applies to an array of techniques for satisfying the need. There is another approach though, which rejects the notion of omnibus techniques and the axiom of need. This takes evaluation as a research problem and makes it a by-product of science studies not an objective.A framework is given for evaluating a research program. How the components are defined and measured has consequences for their use, so the design of the evaluation is crucial. Improving evaluation is not only a matter of anticipation; it requires interaction between research performers and the users of the evaluation.