Abstract
This article describes how microsimulation analysis was used to help design a social experiment currently being conducted in two provinces in Canada. To the authors' knowledge, microsimu lation has never been used before for this purpose, although the technique has been used to assist development of a couple of nonexperimental demonstration programs. For the Canadian experiment, the microsimulation analysis was used primarily for choosing among alternative program models and for refining the selected model, but it had other important uses, such as helping to project the potential financial liability to the Canadian government. The authors conclude that microsimulation should be given serious consideration in the design of future experiments, whenever an appropriate simulation model is available.
