Investigating the Scope of Measurement Error in Calls-for-Service as a Measure of Crime

Data analysis was based on the 1977 Police Services Study (PSS), a study that examined numerous facets of policing in 60 neighborhoods served by 24 police departments. Researchers interviewed about 200 randomly selected residents of each neighborhood and collected information on various aspects of respondent experiences, opinions, and sociodemographic characteristics. Systematic observations of police officers at work were also made. Problem codes were used to record information about the nature of approximately 5,000 encounters in the PSS at three points in time: (1) before police-citizen contacts; (2) at the point of initial police-citizen contacts; and (3) at the close of police-citizen encounters. Results of data analysis raised several questions about the validity of citizen calls to police dispatch centers as a crime indicator. Because dispatch records counted only crimes that citizens reported to the police by telephone, they did not include crimes that came to the attention of police officers by other means. Police officers encountered 30 percent more crimes than were called in to police dispatch centers. Variability was observed in the extent of discrepancies between police dispatch crime counts and what police officers encountered across crime types. Implications of variations in errors by crime type are discussed as they pertain to understanding neighborhood-level error counts and how they are associated with other neighborhood characteristics. Further research is recommended to study the scope and extent of measurement errors in CFS crime counts and to examine the relationship between CFS errors and characteristics of spatial aggregates. 11 references, 2 notes, and 8 tables