Chapter 3: Models of Evaluation

In classic evaluation, three problems were crucial. Is there a match between what actually happens in the real world and what might be expected or desired from some evaluative point of view (value assessment problem)? Is what actually transpires in the world in any way a product of the intervention (effects problem)? And thirdly, are the answers to the first two questions useful for or used by some stakeholder in the intervention (utilization problem)?

When evaluation began to evolve in Sweden in education policy in the 1950s and came up as an innovation in the US around 1965 the three problems were addressed in very distinctive ways. The value assessment issue was conceived as a problem of whether the intervention was attaining its own preset goals. Evaluation meant to answer this goal-attainment question. The effects and utilization issues were approached in similar characteristic fashions. The favored solution to the effects problem was to have academic researchers carry’ out the evaluation to secure the use of best possible scientific methodology. The optimal design, the Cadillac of program evaluation, was the randomized experiment. Academic researchers should randomly create two equivalent groups, an experimental group and a control group. Then, the researchers should administer the intervention to the experimental group but not to the control group. The development in both groups should be measured scrupulously before and after the intervention. Should there occur any differences between the groups in the postintervention measurements, these differences could be attributed to the intervention since all other factors but the intervention were equal due to randomization. All this done, the findings of the evaluation were communicated to relevant decision-makers who were supposed to utilize the findings instrumentally, i.e., in actual decision making and action (utilization issue).