Re-assessing measurement error in police calls for service: Classifications of events by dispatchers and officers

Abstract

Police calls for service are an important conduit by which officers and researchers can obtain insight into public requests for police service. Questions remain, however, about the quality of these data, and, particularly, the prevalence of measurement error in the classifications of events. As part of the present research, we assess the accuracy of call-types used by police dispatchers to describe events that are responded to by police officers. Drawing upon a sample of 515,155 calls for police service, we explore the differences among initial call-types, cleared call-types, and crime-types as documented in crime reports. Our analyses reveal that although the majority of calls for service exhibit overlap in their classifications, many still exhibit evidence of misclassification. Our analyses also reveal that such patterns vary as a function of call- and crime-type categories. We discuss our findings in light of the challenges of the classification process and the associated implications.

Unpacking the police patrol shift: observations and complications of “electronically” riding along with police

“…Within the crime context, calls related to violent crimes averaged longer durations than those related to property crimes. Although we recognise that measurement error exists in call-type classifications (as discussed earlier; Simpson & Orosco, 2021), and therefore they must be interpreted with some caution, differences in durations indicate that different calls for service may require different levels of investment by officers. See Table 3 for a summary of call-type demands by prevalence and duration for the top 10 most frequently occurring call-types attended by officers during our sample of shifts.…”

“…And, different styles of dispatching impacted the quality and accuracy of shift logs. Police data can present many analytical complications, and so from this perspective, we contribute to a growing body of literature that has explored the empirical use of such data (e.g., Haberman et al, 2022;Klinger & Bridges, 1997;Nesbary, 1998;Pearce & Simpson, 2022;Ratcliffe, 2021;Simpson & Orosco, 2021;Varano et al, 2009).…”

“…Research has long argued that much police work does not regard crime, but rather social order and peacekeeping functions (e.g., Bittner, 1967;Wilson, 1968). Consistent with this argument, recent research that explored police calls for service from across the United States found that "traffic-related problems, everyday disputes, worries of suspicious behaviors, disorders, disturbances, and general requests for help and assurance" comprised the majority of calls for which officers are dispatched (Lum et al, 2021, p. 17; also see Simpson & Orosco, 2021). If patrol officers are handling these calls, then it would suggest that their work is diverse as a function of these diverse requests.…”

“…We recognise, however, that call-type classifications are not without their limitations. As argued by Simpson and Orosco (2021), calls for service data can exhibit at least some measurement error. Not only is the information initially provided by callers to dispatchers often ambiguous and nebulous, but the public's reasons for requesting police can be complex and dynamic, making the process of classifying events into specific call-types difficult (Simpson & Orosco, 2021).…”

“…As argued by Simpson and Orosco (2021), calls for service data can exhibit at least some measurement error. Not only is the information initially provided by callers to dispatchers often ambiguous and nebulous, but the public's reasons for requesting police can be complex and dynamic, making the process of classifying events into specific call-types difficult (Simpson & Orosco, 2021). Further complicating such process is the fact that calls can change over their lifespan as new information is brought to the attention of police, which can lead to changes in classifications over time (Simpson & Orosco, 2021).…”