The traditional approach to police administration focuses on strict hierarchial subordination. This approach is based upon three major assumptions of the police environment and personality: (1) that there is certainty and regularity in the police work environment; (2) that police officials have a bureau-type personality receptive to following rules; and (3) that there is nearly absolute dependence of subordinates on their superiors for receiving rewards and avoiding penalties. Police textbook theory assumes that supervision has a positive impact on desirable police officer activities and worker output, especially when the supervision is physically present. Using data gathered during phase two of the Police Services Study, which examined 24 police agencies and 60 neighborhoods in the Rochester, N.Y.; St. Louis; and Tampa-St. Petersburg, Fla. areas, the dissertation constructs a police production process model and conducts three separate levels of analysis (the encounter, the shift, and the neighborhood) to identify and link the components of production, delivery, and outcomes of police services. These analyses suggest that assumptions of the impact of supervision on police officers have been overly strong and in some cases not plausible. It was found that supervision had no effect on neighborhood outcomes and that supervisors seldom gave orders when present at encounters. Suggestions for future policies concentrate on increasing the emphasis on a collegial attitude and approach to training, recasting the negative structure of incentives, and increasing the citizen component of officer performance evaluation. Tables and a bibliography of over 200 entries are included. Footnotes follow each chapter.