Agency-produced statistics reveal agency behavior more than the phenomena they measure

Evidence

Do increases in arrest rates signal improved police performance? Or do they signal deteriorating police performance, indicating an increase in criminal activity and thus in the number of criminals available to catch? Actually, changes in arrest rates may indicate neither, reflecting rather changes in the focus of police patrol. Do decreases in the welfare rolls signal improved bureaucratic performance? They do only if improvement is defined as having fewer people on welfare, the economy that creates the need for welfare remains stable, and welfare agencies do not change their practices affecting clients’ willingness to apply. These are but specific illustrations of the observation that agency-generated statistics are likely to tell us little about the phenomena they purport to reflect, but a great deal about the agency behavior that produced the statistics. — Michael Lipsky, Street-Level Bureaucracy, 401

Categorization

• QUESTION: The welfare-rolls point implies agencies can change intake practices to suppress demand — worth chasing the documented mechanisms for how welfare agencies discouraged applications.
• XREF: connects to: Service demand statistics measure organizational self-presentation more than client need. — Michael Lipsky, Street-Level Bureaucracy: Dilemmas of…, 678

Broader Connections

• Service demand statistics measure organizational self-presentation more than client need. #exemplifies (0.95) — Michael Lipsky, Street-Level Bureaucracy, 678