8 Implementation

Any attempt to implement the models and methods of this book creates in itself a process that requires analysis. We would be poor analysts indeed if the extent of our reasoned thinking about an urban problem stopped with the mathematics. Rarely are we confronted with the simple situation in which all participants in a study share the same priorities and in which the modeling and data collection tasks are inexpensive and straightforward. More likely, things are much more complicated.

Since many readers of this book will eventually attempt to develop and utilize models in an actual urban setting, we wish to close the book by presenting some views on implementation. In the first part of the chapter, to illustrate the variety of forces and factors that can come into play, we provide thumbnail sketches of several "mini case studies" or "war stories" with which we are familiar. We assume that the reader has reviewed the eight steps in an operations research study that were discussed in Chapter 1. The case studies point to the difficulties of defining performance measures, prespecifying all constraints, identifying all decision makers, and retaining sustained commitment of key personnel. Motivated by the cases, in Section 8.2 we attempt to outline in a general way the model-related issues found to be important when attempting implementation. Here, for instance, experience has shown that a model's data-base requirement is often a much more troublesome factor than is model accuracy. Finally, in Section 8.3, we discuss the all-too-critical issues related to people and institutions that affect the success or failure of a model-based analysis. For example, a potentially fatal flaw in this area is for an analyst to work solely through a single individual in any agency-one who speaks "model language" and "agency language." Such an individual may be transferred or promoted by the time the model is ready for use. With the model's major in-house advocate having "vanished" from the scene, the model, even if mathematically sound and sophisticated, may never be used.

Throughout this chapter, we think conceptually of model implementation as taking place within some "intervention program," whose purpose is to analyze and (probably) change one or more operating procedures of an urban agency. While such a program may be rather broad-based, including many types of activities unrelated to models, we naturally focus on its modelrelevant attributes. The intervention program is characterized by a number of inputs [e.g., the model(s), the analyst(s), the user(s), a feasible set of decision options], which give rise to a process [e.g., the use of the model(s), the training of personnel, the collection of data], which transforms the inputs into desirable or undesirable outcomes (e.g., more accessible service, more efficient operation). Our focus in this chapter will be on the input and process features of a model-based intervention program.

Research on the implementation of model-based studies, and policy analytic studies in general, is a new and burgeoning field. Our purpose in this chapter is to discuss critical points directly relevant to implementing methods of this book. Hopefully, those who are interested in additional material on implementation will consult the chapter references or the general references that follow this chapter. There is no simple recipe for guaranteeing implementation success. Our hope here is to establish an awareness of the need to step beyond the mere technical features of a program and to provide a framework for formulating practicable implementation strategies.