We now attempt to build from the case studies and related experiences to identify and discuss the general factors that can influence the chance of successful implementation of a model-based study. Our discussion starts with the central concern of this book-the mathematical model and its attributesand broadens to include people and their institutions. In considering models for implementation, one can select preexisting models, often available at minimal cost in the public domain; or one can develop new models; orcombining the first two options-one can modify existing models to suit certain agency-specific needs. The individual making this model-selection decision may be the analyst (probably under contract to the user's agency) or the user himself, especially in cases in which an in-house study is being conducted. To address the decision in a systematic manner, the important attributes of each prospective model must be considered. No simple algorithm or formula exists for doing this. Evaluation of alternative models for use in a program is often a highly subjective process. Nevertheless, we attempt here to set out those model attributes that we and other researchers have found to be significant. Undoubtedly, in many programs other model-related factors will surface that we do not address here.