Studies from previous decades provided detailed information about shooting pat-terns in
(e.g., Fyfe’s 1978 examination of New York City Police firearms discharges and Geller and Karales’ (1982) similar work regarding the Chicago Police Department). Except for work of this sort, which extracts internal data kept by individual law enforcement agencies, we have no sound, empirically grounded, idea of how many people across the nation are shot at by police officers, how many are struck by police bullets, or how many of the individuals who are shot die from the wounds they suffer. While some national data sources on
at the hands of the police do exist, the counts they contain are notoriously inaccurate (Fyfe, 2002; Klinger, 2008; Loftin, Wiersema, McDowall, & Dobrin, 2003; Sherman & Langworthy, 1979; and below) and—by definition—provide absolutely no information about police firearm discharges that do not cause fatal injury.Because we do not have sound national data on police shootings we are not able to answer simple empirical questions such as, “How often do American police officers discharge their firearms at citizens?” “How many citizens are struck by bullets fired by American police officers each year?” And, “How many citizens are killed by police bullets in the United States each year?” It also means that we cannot track how police use of deadly force varies at the national level from year-to-year (or any other tempo-ral unit), comprehensively compare its use between jurisdictions (or any other spatial aggregate) across the nation, or seek to empirically identify the correlates and determi-nants of police shootings across time and space in the United States.Despite the fact that sound national data do not exist, researchers have persisted in using “official” counts of fatalities at the hands of the police as a measure of deadly force by American police officers, including in studies that seek to explain variability in deadly force usage across time and space (e.g., Jacobs & Britt, 1979; Jacobs & O’Brien, 1998; Smith, 2003; 2004; Sorensen, Marquart, & Brock, 1993).
This later tradition of research is especially troubling because it produces findings about the correlates of deadly police violence that are of dubious validity. The remainder of this article is devoted to outlining the liabilities of “official” death counts as indicators of deadly force and other deadly police action across the United States, describing and demonstrating the problems inherent in using
count of citizens killed by the police to measure deadly force usage, and presenting some ideas about how to develop a data collection program that will provide valid and reliable nationwide information about police firearm usage and police-involved deaths of citizens in the United States.
“Official” Data on Deadly Police Action
There exist three sources of data with national reach that have some relevance to the construct “police use of deadly force.” The first data source is the National Center for Health Statistics’ (NCHS) National Vital Statistics System (NVSS), which keeps records on “deaths by legal intervention,” a mortality category that is intended to capture homicidal deaths caused by law enforcement personnel. Academics have long noted the weakness of the NVSS “deaths by legal intervention” data as an