ABSTRACT
This chapter shows how to move from archival materials to model-specific data, without in the process losing the critical nuances by which policy recommendations are differentiated from each other. The data construction problem is to extract from archival and other sources as many of those components as possible for each recommendation. The theoretical categorization problem is to "ground" the categories and their interrelations as firmly as possible in the participants' understandings of their statements. The basic issue in data construction is to specify the theoretically relevant components of each recommendation. Similarly, the basic issue in theoretical categorization is to ground the categories and their interrelations in the data. The primary result of the data construction and theoretical categorization will be a set of objects used to generate a computational model of policy recommendations. The generation of a grounded computational model and possible accomplishment of the tasks offer attractive rewards for the imminent completion of the final steps of data construction.
