X. Research on Racial Profiling
In the wake of the 1998 New Jersey Turnpike shooting, a fairly broad scholarly effort ensued to establish whether or not police engaged in racial profiling in other jurisdictions. In major cities, “stop and frisk” questioning of pedestrians also fell under the racial profiling umbrella. Dozens of studies have been conducted, yielding mixed results. In the earliest round of inquiry, most reputable studies indicated that minorities were stopped in numbers disproportionately higher than their representation in population statistics (Engel, Calnon, & Bernard, 2002). Later studies have been mounted after the controversy spurred reform of practices and greater scrutiny; the results of emerging studies also document disparities, but generally are less inclined to conclude that discrimination drives them. However, the later studies also occur in a climate where police and civic officials are well aware of the racial profiling debate. Police agencies are more likely to have taken steps to minimize disparity, and are likely more politically astute in their interpretation of findings.
Whether such disparity represents police prejudice or is the by-product of patterns of civilian behavior has yet to be answered definitively. Because policing is locally based, practices differ by jurisdiction. One comprehensive study indicated wide diversity of practices across the 127 different jurisdictions in Massachusetts (Farrell, McDevitt, Bailey, Andresen, & Pierce, 2004). That study also uncovered evidence of a possible gender bias in some jurisdictions (i.e., the propensity of male officers to stop single, young, female drivers), an artifact not pursued in other studies.
Studying racial profiling has several methodological hurdles. First is establishing an appropriate population baseline against which to compare the proportion of minority stops. Second is the limited number of variables available in official records, the most common source of data for profiling studies. Because of these limitations, the most difficult task is determining whether any disparities are grounded in prejudice, or are a product of real (rather than presumed) criminal conduct.
The original New Jersey and Maryland turnpike cases were relatively easy: The proportions of drivers on the highway could be established by direct observation, as could the proportions of speeders and other motor vehicle code violators. Observable vehicle code violations constituted the sole basis for initiating a stop; the choice to request or initiate a search was less directly observable.
Away from limited-access highways, the question becomes more complex. Police officers working in neighborhoods populated by minorities rightly argue that most of the people they encounter will be of minority status, whether the officers are prejudiced or committed to equality. Furthermore, police investigative stops in cities and towns are often prompted by citizen complaints, providing specific descriptive information that is far more detailed than a profiling “hunch.”
Studies note disparate patterns of highway use (both by time and route) by minority drivers, with differences linked to residential, employment, and recreational patterns (Meehan & Ponder, 2002a). Certain highways are transportation corridors from exurbs and suburbs into the core cities, and the commuting drivers may represent different ethnic proportions from the towns’ resident populations. One post-shooting study in New Jersey suggested that young minority males comprised a much greater proportion of drivers using excessive speed (Lange et al., 2005). It retroactively provided nominal support for the original profile, but remains untested against actual vehicle stop practices.
Another study found that even within a single jurisdiction, racial disparity in stops increased with distance from the city border. Minority drivers were far more likely to be stopped in homogeneous white suburban neighborhoods farthest from the city line, where black or brown faces stood out. Police activity was relatively race-neutral in areas abutting a core city, where a heterogeneous population mix was the norm (Meehan & Ponder, 2002b).
More critical to the interpretation of police stop data is the nature of formalized record keeping. Most police records systems were designed to facilitate case tracking and offender identification. Even in those jurisdictions that have adopted more rigorous tracking by race and ethnicity, either prospectively or in response to consent decrees or public criticism, few forms reveal the actual motivation of the officer. The Supreme Court decision of Whren et al. v. United States (1996) has legitimized the police use of the pretext stop, initiating action on the basis of a violation unrelated to the officers’ real intent to conduct an investigation. Almost all vehicle stops meet the low threshold for probable cause (a moving violation or equipment defect), and are therefore legal.
Police records systems generally record only the fact of a stop, not its context. Computer analysis of aggregate stop data is insufficient to establish motivation because such information is not recorded. It remains difficult to determine whether disparate patterns represent overt bias, subtle bias, or a statistical artifact of police deployment based on crime and call patterns, or other localized phenomena.
Analyzing stop data at the aggregate level avoids any scrutiny of individual officers’ work habits. Such techniques have political benefits, but if officers in small groups engage in prejudicial behaviors unrepresentative of the agency practices, the bias may be masked statistically in the larger agency patterns. The impact of their actions upon the minority citizens will remain vivid, however, and the gap between what is experienced by citizens and what can be demonstrated statistically will remain a cause of friction.