Racial Profiling

Lessons from the Drug War

By K. Jack Riley

Jack Riley is director of RAND Public Safety and Justice.

Police allegedly stop drivers because of their race or ethnicity, rather than because of suspected law or code violations. Colloquially referred to as racial profiling, these alleged actions have become a major source of antagonism between police and minorities in many communities. The contemporary debate about police perpetration of racial profiling has its roots in counterdrug profiles that were developed in the 1980s. The use of profiling in the drug war offers insight as to how profiling might—and should—evolve in the war on terrorism.

How Drug Profiling Did Not Work


A state trooper at Boston's Logan International Airport displays a portable digital device that uses a driver's license number to check passenger information, vehicle registration, and outstanding warrants. Troopers will use the device during random passenger inspections.

In the 1980s, the U.S. Drug Enforcement Agency (DEA) and customs agents developed profiles of U.S. land-based distribution networks by observing how drugs moved out of source countries and through border checkpoints. The DEA's Operation Pipeline taught state troopers to look for a variety of characteristics in drivers, including nervousness, an abundance of cash, lack of luggage for long trips, and inconsistent passenger and driver stories about such things as the destination, purpose of the trip, and the names of fellow passengers. Race indirectly entered the equation in that DEA characterized certain retail and wholesale markets as controlled by racial and ethnic groups, such as Jamaicans, Haitians, Colombians, Nigerians, and Puerto Ricans.

Problems soon emerged. First, the drug couriers quickly learned what the profile flags were and adjusted their methods accordingly. Second, profiling evolved without much thought given as to how to document its utility. Consequently, when questions were first raised about racial disparities in enforcement, officials had a weak empirical basis from which to defend their activities. Civil rights activists, however, were also caught in a bind: They had compelling anecdotes about profiling but no information on the racial distribution of driving and of driving violations. Therefore, the activists had no denominator to which to compare the numerator of law enforcement stops.

The result has been a mess that law enforcement agencies and the minority communities are slowly addressing. In many communities, citizen support for law enforcement has eroded seriously, and community members have pressed for data collection about police stops. Many officers oppose data collection using individual identifiers out of fear that their jobs are on the line if their numbers are "out of line." Dispirited officers have slowed their behavior—documented through reduced ticketing and arrest rates—particularly in situations where their exercise of situational discretion could be challenged. Law enforcement leaders are grappling with structuring data collection systems that satisfy the public but provide officers with the incentive to engage in effective policing.

How Terrorism Profiling Could Work

We must ensure that terrorism profiling does not develop on a similar trajectory. Terrorism, more clearly than drugs, illustrates the need to distinguish between strategic and tactical profiling. Strategic profiling helps build intelligence about the shape of a problem, but it has no specific and immediate use in security. The events of Sept. 11 revealed that any strategic profile must focus on identifying radical Islamic males, particularly those sympathetic to, or receiving training from, Al Qaeda.

In contrast, tactical profiling is used to provide security at specific potential targets (such as airports), and it is the kind of terrorism profiling that people are likely to encounter most frequently. Tactical profiling cannot proceed from as narrow a base as strategic profiling for the simple reason that terrorists are dedicated to ensuring that they can complete their tactical mission, and they will use deception to do so. The Al Qaeda training manual instructs its adherents to blend in through disguise and the avoidance of practices (such as prayer) that draw attention. Suicide bombers in Israel have disguised themselves as blond European tourists and Israeli soldiers. What, then, should we seek to accomplish with profiling in the tactical environment? Three objectives stand out:

  • Profile "out" where possible.

  • Increase the randomness of inspections.

  • Make the system defensible against claims of racial profiling.

Profile out, not in. We can conserve scarce enforcement resources if we can profile people out of, rather than into, security procedures. Although the specifics vary substantially, the basic approach would be to allow certain individuals whose credentials have been verified (those with security clearances, for example) to avoid targeted security checks (but not necessarily the random checks described below) at airports and other places where such checks are implemented. Note that such a system could be racially neutral, although it may not be if those profiled out are concentrated in certain ethnic groups.

Increase the randomness of inspections. Profiling out reduces the pool of people who need to be subjected to time-consuming and potentially objectionable security procedures. This leads to the next opportunity: If enough people are profiled out, we could randomly submit a substantial portion of the residual population to inspection. True random selection is an excellent deterrent, because it is impervious to the disguise and deception that terrorists might employ. A side bonus of random selection methods is that they are inherently racially neutral.

Create defensible profiling. What makes tactical profiling defensible? Three things:

  • good baseline information about the numerator (who is stopped) and denominator (the size and characteristics of the population from which those stopped were selected)

  • clear analysis of the "yield" from profiling

  • implementation of effective oversight.

Oversight must include a willingness to listen to complaints—and act on them. Law enforcement agencies are finding that community involvement in designing and reviewing the monitoring system is an important component. The system should be rigorously tested with both dummy contraband and undercover people fitting various profiles. The testing will both teach us about structural problems that the terrorist might exploit and bring to light any biases that might exist. Development of a data baseline will also ensure that we understand the racial implications of the system.

Unquestionably, the war on terrorism raises uncomfortable questions about balancing civil liberties with civil defense. Narrowly constructed profiles used in tactical protection situations—particularly those profiles based on race as a primary predictor—seem doomed to failure and risk generating ill will against counterterrorist efforts. Our collective interests are served by designing a system that is as efficient, effective, and defensible as possible. Such a system need not place an undue burden on any racial or ethnic group.

The author thanks RAND colleagues John Godges, Sarah Hunter, Andrew Morral, and John Woodward, and former RAND colleague Gerald Kauvar, for their thoughtful comments on early drafts. Captain Ron Davis, of the Oakland Police Department, and Heather Mac Donald, of the Manhattan Institute, also provided important insights on the issue of racial profiling. None is in any way responsible for the content.