Earlier this month a federal judge held that New York City’s stop-and-frisk policy violates the Fourth and 14th Amendment rights of African-Americans and Hispanics targeted under the policy. Civil libertarians cheered, but the 195-page opinion is poorly reasoned and unpersuasive. Stop-and-frisk might be bad policy; it might unfairly burden minorities. But the plaintiffs—the African-Americans and Hispanics who were stopped—should have lost this case.
Everyone thinks the Fourth Amendment requires police to obtain a warrant before conducting a search. In fact, it only prohibits “unreasonable searches and seizures,” and the Supreme Court held in Terry v. Ohio that a police officer may briefly stop, question, and frisk a person if the officer has a reasonable suspicion, based on articulable facts (rather than vague hunches), that the person is engaged in criminal activity or poses a danger to others. Such stops are known as “Terry stops.” Meanwhile, the 14th Amendment prohibits the government from intentionally discriminating against minorities.
Let’s first look at the facts, and then consider the Fourth and 14th Amendment arguments separately.
According to Judge Shira Scheindlin, from 2004 to 2012, New York police stopped people on 4.4 million occasions. About half of those stops led to frisks, and police found weapons in 1.5 percent of the frisks (indicating that 34,320 weapons were found and presumably confiscated). Twelve percent of the stops resulted in an arrest or summons. The racial breakdowns are as follows.
|Race||City Demographics||Stops by Race of Person Stopped||Use of Force Involved in Stop||Weapon Found After Frisk||Contraband Found After Frisk|
So, police stopped black people more often than they stopped whites even though whites constitute a larger fraction of the city’s population; they used force against blacks more often; and yet they found weapons and contraband less often when they searched blacks than when they searched whites.
After conducting a stop, police fill out a form by checking boxes next to descriptions of the reasons for the stop (for example, “Furtive Movements,” “Suspicious Bulge”). The descriptions are vague; often, the police do not fill out the forms or take them seriously. Some of the box descriptions, like “High Crime Area,” could not by themselves justify a stop since a stop requires individualized suspicion. Judge Scheindlin concluded that at least 200,000 stops thus violated the Fourth Amendment because the officer checked boxes that indicated only generalized grounds for suspicion like “High Crime Area”; that the actual number of stops lacking individualized suspicion was probably far higher given that the police did not always complete the forms, and their form-filling was likely biased; that many more thousands of stops were unconstitutional because descriptions like “Furtive Movements” are too vague and subjective to demonstrate individualized suspicion; and that the police department pressured officers to make as many stops as possible and that many officers were poorly trained. Finally, the fact that only 12 percent of stops resulted in arrests or summonses (and this number probably overstates the true rate because charges were sometimes later dismissed, among other reasons) means that most people who were stopped were innocent of any crime.
Does this behavior violate the Fourth Amendment? Judge Scheindlin does not estimate the number of stops lacking individualized suspicion, and does not explain how many errors justify the striking down of a government policy. But any policy will predictably result in errors. The “reasonable suspicion” standard of Terry v. Ohio is far weaker than “beyond a reasonable doubt” (required for conviction) and “probable cause” (required for a search more intrusive than a frisk)—all of which necessarily result in a large number of false positives. Why doesn’t a 12 percent hit rate (or even lower hit rate) justify the considerably less intrusive tactic of briefly stopping a person and asking him questions? Judge Scheindlin does not identify the error threshold that distinguishes a valid police tactic from an invalid one.
She also ignores an important factor—the baseline criminality of the population subject to stops. If 12 percent of the relevant population engaged in criminal activity, then a hit rate of 12 percent would be no better than random—meaning that police who stop people randomly will be right 12 percent of the time. But if the rate of criminal activity is lower—say, 1 percent of the population—then a hit rate of 12 percent is impressive, and suggests that police do stop people only with reasonable suspicion and score misses only because criminality is so rare. Although Judge Scheindlin does not discuss this point or provide data, it seems likely that baseline criminality is much lower than she implicitly assumes.
Let us turn to the equal-protection claim. Judge Scheindlin held that the stop-and-frisk policy violated the Equal Protection Clause because the police intentionally discriminated against blacks and Hispanics, based on the fact that the police stopped a higher proportion of blacks and Hispanics than whites—beyond what could be justified by the fact that blacks and Hispanics live in higher-crime neighborhoods where more stops would naturally occur.
New York had argued that the police merely focused resources on people most likely to commit crimes according to police data. It argued that if, in a particular area, most crimes are committed by blacks, then the police are justified in targeting black people, even if most residents are white. Judge Scheindlin rejected this argument because “there is no basis for assuming that the racial distribution of stopped pedestrians will resemble the racial distribution of the local criminal population if the people stopped are not criminals” (her emphasis). Maybe. But if a police officer accurately perceives that someone is about to snatch a purse, but stops the person before he does so, and so must release him because he does not carry contraband or a weapon, the officer has acted properly, not wrongly. It is unlikely that all of the people stopped were not criminals.
Judge Scheindlin makes a stronger point that the hit rate for whites is higher than the hit rate for blacks—indicating that police are inefficiently directing resources against blacks when they could get more bang for the buck if they stopped more whites, who are more numerous and possibly more responsive to criminal sanctions than blacks because they have more legal opportunities (a point made in this paper). Nor is there evidence for Mayor Michael Bloomberg’s claim that the stop-and-frisk policy caused the decline in crime in New York City, which began long before this policy did.
So the New York system of racial profiling may be ill-designed or poorly executed—or not; the evidence doesn’t tell us one way or the other. But the plaintiffs carry the burden of proof; they must show the policy is motivated by discrimination and cannot prevail merely by pointing out the weakness of the evidence that supports it. (As criminologists will tell you, evidence supports the effectiveness of hardly any police practice.)
To see the problem with Judge Scheindlin’s view, consider a police officer located in an area where a great many muggings take place, and (according to victim reports) most of the muggers are black. He observes someone who appears to be tailing someone else. The behavior may be innocent, or it may signal the start of a crime. Is the officer entitled to take into account the target’s race in determining whether to stop the target and ask him some questions?
Surely, the officer could take into account the age of the target (elderly people don’t mug people) and the target’s sex (women don’t mug people very often). Why not race—if crime statistics indicate that in this area muggings are usually committed by blacks? In fact, it would be hard to understand how the race of the person would not play a role in determining whether reasonable suspicion existed—how a conscientious police officer could ignore the target’s race. On the court’s view, this officer would have to somehow scrub the racial identity of the target from his mind before deciding whether the person was acting suspiciously. This is a sure way not to eliminate racism from policing, but to increase the level of perjury among police officers, who are likely to disregard rules that make no sense to them.
The contrary argument is that if people know that police concentrate on blacks, then white people will flood into the mugging business, secure in their immunity. Yet if that is so, the police will adjust as mugging victims begin to report that they were victimized by white people.
Consider another example. A victim is mugged, and tells the police that her attacker was a tall black man wearing a red sweatshirt. The police put out a bulletin describing the suspect. May the police identify the race of the target in the bulletin?
Most people would say yes (and the courts have approved this practice). But it too puts a burden on African-Americans. Innocent tall black men wearing red sweatshirts will be stopped while innocent tall white men wearing red sweatshirts will not be stopped. And while the failure to identify the race of the suspect would likely make it harder to catch him, the magnitude of the effect is unknown just as it is for stop-and-frisk. Judge Scheindlin does not believe that speculative claims about effectiveness justify racially burdensome policies.
Judge Scheindlin argues that stop-and-frisk is different because in my example race is used for the identification of a specific perpetrator, and the policy—to identify people in part on the basis of racial characteristics—is “race-neutral.” But these distinctions break down quickly. If police know that the muggings in an area are caused by blacks, or that a Hispanic street gang controls an area in which drugs are sold, then the stop-and-frisk policy is oriented toward specific perpetrators, and it is race-neutral in the sense that it draws on demographic patterns regardless of what they are.
This was not an easy case. Stop-and-frisk humiliates and inconveniences young black and Hispanic men and may reinforce unfair perceptions about race and crime, while creating uncertain benefits for the public. But all laws and police practices create uncertain benefits—social science methods are too weak to prove the effectiveness of any given policy—and all of them will impose unequal burdens in racially diverse cities where different groups commit crimes at different rates. These tradeoffs are best left to the political process. We should expect better reasoning from judges when they strike down entire policing programs that the mayor and police commissioner believe keep a city safe.