To evaluate the correctional system while under change and normalcy, researchers studied a large-scale deinstitutionalization experiment in Massachusetts. Study of the experiment had indicated that recidivism was influenced by networks of relationships the youths were involved in. These relationships extended beyond the traditional relationships of school, work, and community programs and included the juvenile's friends, schools, and people met on the street. These relationships were explored through an institutionalization-normalization continuum which allowed researchers to describe the relationships among people in a specific setting as an integrated everage of three constitutive dimensions: social climate, extent of community contact, and quality of community contact. The conception of an institutionalization-normalization continuum can be applied to immediate, short-run and long-run consequences of programs, including social climate within the program and the extent and quality of community linkages (short-run) and new skills or self-images with which the youths leave the program (long-run). While the continuum can describe many aspect of the system, it cannot take into account all the goals of different groups that affect it. Three general types of interest can be distinguished: interest in more therapy and/or linkages, interest in more security and control, interest in making sure the right people's perogatives are respected in the decisionmaking process. These three types can be subdivided into six basic groups, each representing basic interests that must be taken into account if program evaluation is to assess desirability, practicality, and ability to survive. A table can be developed to distinguish relationships among interest groups, and five principles emerge from a mathematical simulation involving variables describing the individual interest groups. These include sequencing, inertia of the people-processing relationship, the formal decisionmaking group as a swing power, responsiveness of the other groups to the people-processing relationship, and short- and long-run effects of extreme tactics. Eighteen references and a tabular representation of the model are included.