The approach implemented in the four case studies attempts to assess the impact of changes in law enforcement on punishment policy by closely following what happens after particular policy shifts occur. The four studies include the routine preventive patrol experiment conducted in Kansas City, the High-Impact Anti-Crime Program inaugurated by LEAA in eight cities, the Alcohol Safety Action Program, and the 'Rockefeller Drug Law.' Analysis of these intervention experiments reveals specific limitations to this research approach. These limitations include measurability, sensitivity, absence of theory, specificity of findings, and the marginal nature of politically possible shifts in independent variables. Students of deterrence at all levels need to pay more attention to the political science of punishment policy. The macrostatistical studies of deterrence document wide variation in criminal justice policy but reveal little about how the variations happen or why they occur. Concepts of crime and severity of sanctions are too broad and too vague to serve as the foundation for policy research. Two tables, 21 footnotes, 17 references, and an annotated bibliography of deterrence evaluations are provided.