Statistical monitoring of police force for rapid detection of changes in frequency

Abstract

U.S. Law enforcement agencies are authorized and expected to use the minimum level of force required to maintain law and order. Few civilian law enforcement agencies and rib military law enforcement agencies pro actively monitor the use of force. Furthermore, agencies that do monitor force use methods that produce simplistic data summaries. These data summaries provide late and limited information to decision-makers regarding conditions sufficient to warrant managerial intervention. This study models police force incidents as a Poisson process and monitors the process to detect departures from the model. Police force data is charted using a self-starting control chart scheme. The charts assist the decision-maker in determining if intervention is necessary to correct an out-of-control condition while simultaneously minimizing unnecessary intervention when shifts in the frequency of force are plausibly due to random variation. Force data from military and civilian law enforcement agencies illustrate the methods. Methods are implemented in a Microsoft Excel spreadsheet with Visual Basic macros for ease of use.