I. Introduction

In Mapp v. Ohio,1 the U.S. Supreme Court extended the due process protections of the exclusionary rule to include all “constitutionally unreasonable searches” that were done without a basis of probable cause.2 In the seven years after Mapp, when homicide rates in the U.S. nearly doubled,3 riots broke out in at least forty-seven U.S. cities.4 During the same era, a heroin epidemic gripped the nation’s urban centers,5 giving rise to street drug markets and associated violence and pressures on law enforcement to curb those markets.6 As violence increased, a turn in the nation’s political culture questioned Mapp’s restraints on police discretion to stop and search criminal suspects.7 Indeed, some writers wondered if the Mapp standard, with its reliance on the exclusionary rule to deter violations of Fourth Amendment rights, had inflicted social costs on the public through over-deterrence of police, leading to elevated crime rates.8

It was no surprise, then, that after those seven years the Supreme Court in Terry v. Ohio9 “uncoupled . . . the two clauses of the Fourth Amendment” that regulated temporary detentions and searches by police.10 Terry dealt with a different “rubric of police conduct”: the beat officer stopping and patting down an individual on the street, more commonly known as an “investigative stop.”11 The Terry test was (and is) thus to balance the scope of the intrusion against the “specific and articulable facts which, taken together with rational inferences from those facts, reasonably warrant [the] intrusion.”12 Justice Douglas, in dissent, labeled this “reasonable suspicion.”13 Although intended to be a narrow departure from Mapp’s standard, it was in fact a big break from Mapp.14 The Court said that the Mapp rule simply did not fit the realities of street policing in an era of rising crime rates.

Under Terry, the police must articulate specific and individualized indicia of suspicion, and those indicia must be salient enough to justify police action. Hunches by police worried the Terry Court.15 The standards then and now do not really tell a police officer doing modern police work how much suspicion is enough to satisfy constitutional standards, or when the quantity of suspicion reaches a threshold of “reasonableness” to justify the intrusion. That question became even more challenging as a series of opinions inflated the scope of “reasonable suspicion” to include pretextual probable cause stops—often minor traffic violations—that open the door to investigations of other crimes,16 or stops where a suspect’s presence in a “high crime area” multiplies less salient factors into actionable suspicion,17 or facially subjective rationales such as “furtive movements” or other “criminal appearances.”

As the fiftieth anniversary of the Terry opinion approaches, it is more than reasonable to ask whether Terry’s move away from probable cause was original sin—whether the dilution and expansion of standards for an investigative stop over time compromised or advanced the very law enforcement interests that animated the Terry opinion. Two criteria of (constitutional) focus are wrapped up in the “law enforcement interest” doctrine: catching offenders and seizing contraband (hit rates), and controlling crime (crime rates). Whether contemporary and expanded Terry standards can achieve or undermine these interests is the primary question for this paper.

Do these sins pay? This is the central question for this article. Sins where officers stop, temporarily detain, question, and possibly frisk a person based on the person’s vague or subjectively perceived actions (appearances, movements) may be less efficient in locating contraband or suppressing crime than stops based on actuarial characteristics (locations). But both may be less efficient than stops based on behavioral indicia of crime. In other words, this article asks empirically whether stops based on indicia that approximate probable cause (based on behavioral indicia that are unambiguously indicative of crime) advance law enforcement interests significantly more than stops based on the more subjective and vague standards that have become commonplace features of contemporary investigative stop programs.

Perhaps these sins pay for only certain types of crime. Terry’s ruling came in the midst of a violent crime spike in the late 1960s through the early 1970s.18 But Terry is now applied broadly for violent and other serious crimes as well as for drug and weapon offenses. And in an era of proactive and “broken windows” policing, minor misdemeanors are theorized as predicates of crime and therefore are indicia of suspicion in and of themselves.19 This leads to the second question for this paper: whether the dilution of standards has differential effects by crime seriousness.

The answers to these questions follow. The first section assesses the doctrinal progression from Mapp to Terry, showing that the original officer safety rationale was eclipsed over time by Terry’s crime control agenda. The second section presents the details of the empirical inquiry on the two research questions. Data on crimes, stops, and arrests from 2004 to 2012 from the Floyd v. City of New York20 litigation are analyzed to address these questions. The data include the bases of suspicion for each stop, and stops are assigned as probable cause or suspicion stops based on the articulated rationale. The third section presents the empirical results. The analyses show significant reductions in crime in neighborhoods (census block groups) with greater numbers of probable cause stops, and ratios of probable cause stops to other stops. The opposite, however, is not evident. Crime neither increases nor decreases in places with higher numbers of non-probable cause stops; those stops simply have no effect on local crime rates. This is an empirical argument about what kinds of police observations of suspicion are indicative of criminal activity, and how acting on those indicia can advance Terry’s public safety agenda.

The final section discusses a set of potential regulatory and doctrinal responses to these results that suggest the application of harm principles to inform the practice of Terry stops that raise privacy and positive liberty interests. This section presents a functional, institutional argument about what kinds of observations of supposedly suspicious activity are susceptible to meaningful review and oversight. It turns out that redemption for Terry’s sins may be close at hand: fleshing out the Terry standard by setting clear rules about what constitutes “reasonable suspicion,” and concretely linking the Terry standard to specific actions indicative of criminal activity will reduce errors in suspicion and better prevent crime.

II. Background

A.    From Mapp to Terry

Bad timing is one factor that led to the shift in standards for investigative stops and searches from Mapp to Terry. Rising violent crime rates through much of the 1960s,21 together with riots in dozens of American cities,22 helped create new social tensions and a legal and policy context in which law enforcement interests eclipsed the restraint on Fourth Amendment violations that was Mapp’s inspiration. As crime rates continued to rise through the next two decades, the focus of Terry’s jurisprudence—and the law enforcement interests that it embodied—shifted from officer safety to public safety. The standards regulating reasonable suspicion, the foundation of Terry, also shifted over time as the public safety interests of Terry hardened and expanded. In this section, the trajectory of this subtle jurisprudential shift is examined, laying the foundation of contemporary Fourth Amendment jurisprudence on the limits of police stop-and-search power.

1. What Mapp did and did not do.

In Mapp v. Ohio, the Supreme Court held that the exclusionary rule applies to state prosecutions.23 The Court had previously held that the exclusionary rule applies to unconstitutionally seized evidence in federal prosecutions in Weeks v. United States.24 But when the Court applied the  Fourth Amendment’s warrant clause to the states in Wolf v. Colorado,25 the Court held that the exclusionary rule was not a necessary component of the Fourth Amendment’s protection.26

Mapp thus stands for the proposition that the probable cause requirement is toothless if not backed by a consequence that “remov[es] the incentive to disregard it.”27 The Weeks Court more fully discussed the details of the probable cause requirement.28 In Weeks, the Court held for the first time that the Constitution requires exclusion from federal criminal prosecutions evidence obtained without a warrant, issued by a judge, supported by probable cause, and describing the object of the search. Implicit in the Weeks decision is the proposition that the “reasonableness” of a given search under the first clause of the Fourth Amendment is defined by the warrant requirement in the second clause of the amendment.29 More specifically, Weeks said the Fourth Amendment protects individuals against “all unreasonable searches and seizures,” which it then defined as those searches done without a “warrant issued as required by the Constitution.”30

Similarly, Mapp described itself as “extending the substantive protections of due process,” that is, the exclusionary rule, “to all constitutionally unreasonable searches”—those done without a warrant issued on a showing of probable cause.31 The meaning of an unreasonable search was a search conducted without a warrant; the two clauses of the Fourth Amendment were linked.32

2. The limits of exclusion.

Earl Dudley notes that the meanings of neither reasonable suspicion to justify a frisk nor Terry stops to justify minor “physical intrusions” are readily apparent from the text of the Terry opinion.33 In Adams v. Williams,34 the first post-Terry decision on stop and frisk, Justice Rehnquist doubled-down on the Warren Court’s reasonableness standard by conflating “stops” with protective “frisks,” and applying the same vague reasonableness standard to both levels of intrusion.35

The phrase “reasonable suspicion” comes from Justice Douglas’s dissenting opinion, when he criticized the Court’s departure from the “certainty” and historical grounding of the probable cause requirement.36 Scott Sundby similarly notes, “Chief Justice Warren’s cautious opinion suggests that the use of the reasonableness balancing test was meant to be viewed as a narrow departure from the norm of probable cause.”37 And Stephen Saltzburg argues that, taking the opinion at face value, “the Court would appear to have decided little.”38 Therefore, for courts in the aftermath of Terry, it was not at all clear how to regulate investigative stops in a novel framework of reasonable suspicion.

However, before the Court reached the merits of Terry’s claim, it discussed the exclusionary rule in a highly suggestive way that highlights the break the Court was making from Mapp.39 In effect, the Court said that the exclusionary rule was powerless and irrelevant to the realities of contemporary beat policing.40 First, the Court noted that, for many interactions, “the police either have no interest in prosecuting or are willing to forgo successful prosecution in the interest of serving some other goal,” rendering the exclusion of evidence useless.41 Second, the Court mentioned the risk of “wholesale harassment” of minority groups by police, but again stated that this “will not be stopped by the exclusion of . . . evidence from any criminal trial.”42

Third and most important, the Court stated that applying the exclusionary rule where it is incapable of stopping police abuse “may exact a high toll in human injury and frustration of efforts to prevent crime.”43 This mention of the crime control objective is muted here, but grows in importance in subsequent rulings, and occupies center stage in the Floyd litigation and in replications of the Terry regime in a swath of U.S. cities and across several countries.44 In effect, the Court took a lesser of evils approach where abuses—both of minorities and the boundaries of investigative stops—are tolerated in return for enhanced crime control, a matter that is challenged in the empirics here.

The focus on the limits of the exclusionary rule at the beginning of the Terry opinion reads as a justification for applying a standard other than probable cause to the types of investigative stops upheld in Terry. The Court says that the Mapp rule does not fit the realities of beat policing because it is too slow to account for public safety concerns and because the remedy—exclusion of evidence—fails to correct police misconduct. In effect, the Court minimized the possibility of deterrent effects on future police misconduct of the suppression of evidence. The Terry standard of reasonable suspicion is thus a vindication of the Court’s public-safety concerns over the trial-focused exclusionary rule of Mapp. Perhaps the Court had a regulatory purpose in mind instead of a deterrence purpose.  The Court was optimistic that police could lean heavily on the internal processes and self-discipline of their institutions to do the work that would have fallen within Mapp’s litigation domain. But the Court also discounted the prospect of an inevitable parade of suppression hearings that would follow the  shift from the more demanding Mapp standard to the subjective, if not inchoate, standard in Terry.

B.    The Terry Standard and the Regulation of Police: Defining—or Failing to Define—Reasonable Suspicion

The new Terry standards did no favor to trial courts by defining in such a subjective way the new standards for street stops that could be challenged. But as the analysis of those standards in this section shows, the Court may have (whether by design or not) mitigated that risk by advancing a standard where subjectivity was subordinated to a highly proceduralized standard.

1. Terry, investigative stops, and a new set of state intrusions.

By requiring reasonable suspicion for stop and frisks, Terry extended the Fourth Amendment to seizures less intrusive than arrests. In effect, the Supreme Court “uncoupled . . . the two clauses of the Fourth Amendment.”45 The Terry Court did not overrule Mapp: the warrant requirement, backed by the exclusionary rule, still applies to the traditional police-at-the-front-door search of an individual’s dwelling.46 But it is hard not to see Terry as a victory for police because it recovers much of the discretion and, to be frank, the power that had been revoked in Mapp,47 and later in Papachristou v. City of Jacksonville.48

Terry dealt with a different “rubric of police conduct” than did Mapp: the beat officer stopping and patting down an individual on the street.49 In footnote sixteen, the Terry majority went further: it refused to consider the question of whether an “investigative ‘seizure’ upon less than probable cause for purposes of ‘detention’ and/or interrogation” violates the Fourth Amendment.50 Then the Court narrowly defined the notion of “seizure” as instances “when the officer, by means of physical force or show of authority, has in some way restrained the liberty of a citizen.”51

Still, even as the Court “emphatically reject[ed]” suggestions that the stop-interrogate-and-frisk interaction is not subject to the Fourth Amendment,52 the Court just as clearly rejected the notion that probable cause was required for the “limited search for weapons” at issue in the case.53 It is not hard to see the standard for such limited searches metastasizing over time into the Floyd regime in New York, with millions of street searches over a decade producing few guns or contraband.54 Instead, the Court adopted a rule from a case involving administrative searches of homes, Camara v. Municipal Court,55 requiring courts to “balanc[e] the need to search (or seize) against the invasion which the search (or seizure) entails.”56

The Terry test, in turn, balanced the scope of the intrusion (e.g., whether the police officer “patted down the outer clothing” or “conduct[ed] a general exploratory search”) against the “specific and articulable facts which, taken together with rational inferences from those facts, reasonably warrant the intrusion.”57 This move in effect distinguishes Terry from Mapp and all cases involving actual arrests. Because “[a]n arrest is the initial stage of criminal prosecution,” it serves very different social interests than a Terry search, which was designed to protect “the police officer, where he has reason to believe he is dealing with an armed and dangerous individual.”58

Although the majority opinion does not address a frisk or pat-down, a Terry frisk also required reasonable suspicion that the individual being frisked presents a danger to the officer or others at the time.59 This specificity requirement is in contrast with the “inchoate and unparticularized suspicion or ‘hunch,’” which is insufficient to justify the intrusion on an individual’s liberty.60 On the other hand, the Court’s disapproval of “hunches” was tempered by its tolerance of “the specific reasonable inferences which [the police officer] is entitled to draw from the facts in light of his experience.”61 Ultimately, the standard is characterized by the Terry Court as “objective,” and must hold up to the “more detached, neutral scrutiny of a judge” who is to ask whether “the facts available to the officer at the moment of the seizure or the search ‘warrant a man of reasonable caution in the belief’ that the action taken was appropriate.” This approach requires balancing the level of suspicion of danger against the invasion of privacy and autonomy, something that an officer in the moment of an encounter with a citizen may be hard pressed to do.62

After nearly four decades, the Terry standard remains rather opaque.63 In describing the government’s burden, William Stuntz analogizes to a statistical determination:

The threshold is not defined mathematically, but one could easily enough think of it that way, and courts and lawyers basically do think of it that way . . . . (Though, I should quickly add, there is no clear agreement on what the right mathematical line is. Probable cause officially means “a fair probability;” in practice, it means, roughly, more-likely-than-not. Reasonable suspicion plainly requires less than probable cause. A good approximation, then, might be something like a one-in-five or one-in-four chance.)64

Thus, the burden of proof is quite low. Given an estimate of a twenty or twenty-five percent chance a crime is about to occur or has occurred, proving a “hunch”—the type of suspicion, with a capacious tolerance for error, Terry condemns—requires a very low probability indeed.

2. Regulating reasonableness.

In the same part of the opinion that explains the shortcomings of the exclusionary rule as a check on street policing, Terry appears to suggest that the rule remains the primary judicial check. For instance, the Court says that, when applying Terry, if courts identify “over-bearing or harassing” conduct, “it must be condemned by the judiciary and its fruits must be excluded from evidence in criminal trials.”65 Still, the Court here is directing lower courts to focus on conduct during the stop, not the indicia of suspicion that motivated the stop.

The Court in fact excluded evidence on these grounds in Sibron v. New York,66  a companion case to Terry. Both Terry and Sibron came to the Supreme Court as appeals from trial court denials of motions to suppress.67 While introduction of the weapon uncovered from the frisk in Terry was admissible, the Sibron evidence was excluded on grounds that the search was not justified by the protective interest that motivated the officer in Terry. Sibron involved a police officer who searched Sibron after observing him “talking to a number of known addicts.”68 The Court noted that the Terry rule only justified limited frisks where particular facts support an inference of danger to the officer, which was not present based on an individual speaking with known addicts.69 The Court went further, distinguishing the Sibron search from the Terry search, which initially “consisted solely of a limited patting of the outer clothing of the suspect for concealed objects,” with the officer “plac[ing] his hands in [Terry’s] pockets” only after “discover[ing] [a concealed] object.”70

The Terry-Sibron comparison illuminates the Court’s concerns about individual dignity. In Terry, the Court justified its application of the Fourth Amendment to frisks (but not stops) because it found a frisk to be “a serious intrusion upon the sanctity of the person.”71 The Court addressed this intrusion by requiring a balancing between the level of intrusion and the level of suspicion. It could be the case that the harm to a person’s dignity from the intrusion of a frisk may be greater than the harm to dignity resulting from a full search incident to arrest. The reason may lie in the difference in the standard for a frisk versus a search: a full search incident to arrest requires probable cause, which suggests specific behavioral and inherently concrete indicia of suspicion. A frisk, in contrast, need only be justified by the more subjective and inchoate standards of reasonable suspicion.

3. How much reasonable suspicion? What indicia?

Both in the run-up to Terry and the decades after, courts never developed a constitutional consensus as to how much suspicion is needed to give rise to reasonable suspicion.72 Nor are there substantive indicia to prioritize or weigh which behaviors or factors matter; the courts have said only that these indicia must be reasonable. Some courts have argued for a test based on the efficacy of stops in detecting crime or locating contraband, but here too, there is no agreement on what constitutes an acceptable “hit rate” that satisfies the reasonableness standard across cases. In Navarette v. California,73 for example, Justice Antonin Scalia suggested that at least five, if not ten percent, of the entire universe of incidents would need to be an accurate “hit” to be indicative of reasonable suspicion. According to Scalia, absent such a showing, the basis of suspicion is not reasonable without further information.74 A similar outcomes test was considered in Floyd to claim that the police were so often wrong in the bases of suspicion for their stops that those bases were categorically faulty.75

But after nearly five decades of Terry, courts have rejected a substantive review of the criteria of “reasonable suspicion.” Instead, courts have consistently decided cases based on some rendering of the reasoning of the officers at the scene (based on a post-hoc account) pursuant to a specific fact, and whether that reasoning was, well, reasonable to an experienced officer.76 But it gets worse. Until recently, a series of cases required that the basis of the information on which reasonable suspicion was determined be reliable.77 But in Navarette, the Court largely abandoned the reliability doctrine by holding that an anonymous 911 call without any corroboration meets a test of reasonable suspicion to justify a stop and seizure.78

Under the real-time demands of police work, and with little oversight to correct misapplication of the perceptual and reasoning processes, the articulation of suspicion often defaults to behavioral scripts that are matched to fill in the empty cognitive spaces in the actual bases of suspicion.79 In three out of four street stops in New York City, for example, police observe a suspect for less than two minutes before proceeding to what New York state law80 defines as an “intrusion.”81 The stop requires officers to perform a quick perceptual and cognitive sorting of complicated and highly contextualized information that shapes the initial evaluation of suspicion. As the interaction unfolds, this sorting is modified and narrowed through interactions and exchanges between the suspect and the officer(s). After all this, the officer then retreats to an unspecified location under uncertain conditions to record the reasons for the encounter, reasons that may have taken place and been cognitively encoded an hour or more before.

It is no wonder that police officers may default to a script. But even with the handy crutch of a script, the cognitive burden to both articulate the reasons for the suspicion, and how those reasons got beyond a (not well articulated) threshold to take action, leaves a wide space for error in perceptions, weighing, and decision-making.

The configuration of Terry and its progeny simply begs the question as to what factors meet the test of articulable and individualized. These cases continue, as did Terry itself, the fiction that there is a threshold of suspicion that renders police action constitutionally permissible. Suspicion in this formulation thus becomes a hurdle model, or a binary category, in which the stop is either constitutional or not.82 Courts worry more than the police about whether there is enough suspicion to get over that hurdle and satisfy the “individualized” suspicion test. And the elasticity of the Terry standards complicates the job of courts to regulate those decisions.83 Officers are left to the extremes of roll call training on the one hand and litigation challenges on the other to define a space in which their actions comport with the shifting territory of the Fourth Amendment.84

C.    Terry’s Crime Control Agenda

Terry’s original sin took two forms. First the majority created the reasonable suspicion standard that allowed subjective assessments of suspects’ behavior to substitute for the more demanding standard of probable cause. This was done, as discussed earlier, in the interest of protecting officers from harm. The Terry Court declined to articulate clear standards of suspicion, defaulting the professional “experience” and judgment of the officer.85 The second sinful act was the doctrinal shift over time from the original officer safety rationale to permitting reasonable suspicion stops in the interest of crime control. This section examines the evolution of this second sin, and describes the rationale for modern Terry practice.

1. Terry’s hidden crime control agenda.

A first reading of the majority opinion in Terry suggests that it had little to do with crime control, and everything to do with the safety of police officers in conducting investigative stops or field interrogations.86 The Court seemed to be well aware that it was making a trade-off: in allowing “something less than a ‘full’ search” at a new and relaxed standard of “reasonable suspicion,” the Court held that the Terry stop “must be limited to that which is necessary for the discovery of weapons which might be used to harm the officer or others nearby.”87

At first glance then, the Terry Court’s concern seemed to be less about public safety generally, but rather the safety of the officer when approaching and questioning individuals like John Terry. The President’s 1965 Commission on Law Enforcement and the Administration of Criminal Justice reported that “[c]ommission observers of police streetwork in high-crime neighborhoods of some large cities report that 10 percent of those frisked were found to be carrying guns, and another 10 percent were carrying knives.”88 But the report did not mention officer injuries or deaths in routine contacts as posing the danger that drove the Terry ruling.

In fact, the only evidence the Terry Court cites about the dangers of policing is a reference to the same 1965 Presidential Commission and is contained in a footnote, worrying that frisks often exacerbate tensions between the police and minority groups.89 In the end, the Court does not tie the crime-control or officer-safety aspects of the opinion to any evidence. In the years between Mapp and Terry, officer deaths rose as overall rates of violent crime rose, but the Terry Court made no note of this.90

But in fact, crime control and public safety were as much on the minds of the justices as was officer safety. Two concurring justices and a third dissenting justice more directly alluded to a general crime-prevention rationale for the Terry holding than does the Court’s majority opinion. First, Justice Harlan, who rejected the Court’s effort to decouple the “frisk” issue from the “stop” issue, concurred to make clear that, to him, the stop was constitutional “only because circumstances warranted [the officer] forcing an encounter with Terry in an effort to prevent or investigate a crime.”91 Thus, for Justice Harlan, the intrusion of the stop and frisk needed the crime-prevention rationale to survive constitutional scrutiny. Although Justice Harlan described his opinion as merely “fill[ing] in a few gaps” in the majority opinion, the reference to preventing crime more broadly than the potential harm to the police officer conducting the frisk nowhere appears in the Court’s opinion.92

Second, Justice White, in the second paragraph of his two-paragraph concurrence, provides “an additional word . . . concerning the matter of interrogation during an investigative stop.”93 Like Justice Harlan, Justice White emphasizes the link between “temporary detention” and the frisk that the majority opinion sought to avoid. But he went further than Justice Harlan to speculate about a possible crime-prevention benefit of frisks that fail to uncover any weapons: “Perhaps the frisk itself, where proper, will have beneficial results whether questions are asked or not. If weapons are found, an arrest will follow. If none are found, the frisk may nevertheless serve preventive ends because of its unmistakable message that suspicion has been aroused.”94 The two concurrences bookend two different theoretical supports for investigative stops: while Justice Harlan appears to have identified the crime-prevention rationale based on police intervention before a crime is committed, Justice White apparently saw a general crime-prevention effect in the failure to uncover weapons as a way to educate citizens about what police officers find suspicious.

Justice Douglas dissented alone in Terry. He was implacable about the necessity of the probable cause requirement for a temporary detention and a frisk or pat down of a suspect.95 In passing, however, he acknowledged—noting the escalating crime rates in the years after Mapp—that “[p]erhaps [the Terry rule] is desirable to cope with modern forms of lawlessness.”96 This seems an oblique reference to a general crime-prevention rationale that goes beyond the narrower interest in officer safety enunciated by the majority.97

In the few states that developed doctrine that differed from Terry, the controlling opinions also incorporated both crime control and officer safety prongs.98 In People v. De Bour,99 officer safety was a less pressing concern than was the broader public safety impetus for the pursuit and frisk of the suspect. The New York State Court of Appeals upheld the introduction into evidence of a gun discovered when police officers asked a man to unzip his jacket.100 The court held the encounter was a “legitimate . . . inquir[y] as to [De Bour’s] identity” because it was without “harassment or intimidation,” “brief,” involved prevention of the “serious crime” of narcotics, “occurred after midnight in an area known for its high incidence of drug activity,” and because “De Bour had conspicuously crossed the street.”101

The crime control function of Terry stops is one of a number of governmental interests that could potentially authorize a stop, but case law after Terry focused increasingly narrowly on violations of criminal law as the primary government interest.102 For example, in United States v. Brignoni-Ponce,103 the Court extended the rationale and basis of Terry stops to a broader government interest: immigration control by roving patrol “to prevent the illegal entry of aliens at the Mexican border” such that warrantless seizures, based on reasonable suspicion, could be used.104 In United States v. Martinez-Fuerte,105 the Court found checkpoints within 100 miles of the border to be reasonable even without an element of suspicion in the stopping of cars.106 These border-control searches could comfortably be described as crime control measures, and the Supreme Court accepts the governmental interest of preventing illegal immigration.107

2. In plain sight: Terry’s explicit crime control strategy.

After Adams v. Williams,108 which affirmed Terry while blurring the line between stops and protective frisks, the Court proceeded to incrementally extend the constitutionality of Terry stops beyond the narrow governmental interests of officer safety or border control, and, in so doing, affirmed its crime control rationale. In Michigan v. Summers,109 the Court extended Terry’s reach to investigative stops incident to a search warrant.110 Here, a division opened within the Court between those who would restrict Terry stops to the already authorized safety (or border control) exceptions and those who sought to extend Terry in favor of police action more closely in tune with a crime control perspective. The dissenters in Summers favored the narrower view. Justice Stewart, joined by Justices Brennan and Marshall, explained that “some governmental interest independent of the ordinary interest in investigating crime and apprehending suspects” must be important enough “to overcome the presumptive constitutional restraints on police conduct.”111 The majority, however, used the border-control-search cases to demonstrate that the exception for limited intrusions that may be justified by special law enforcement interests is not confined to the momentary, on-the-street detention accompanied by a frisk for weapons involved in Terry and Adams. . . . Most obvious is the legitimate law enforcement interest in preventing flight in the event that incriminating evidence is found. Less obvious, but sometimes of greater importance, is the interest in minimizing the risk of harm to the officers.112

Over time, the Court further extended Terry’s subtext authorizing investigative stops as a crime-fighting tool, each time increasing the scope of their permissible contexts. For example, in Michigan v. Long,113 the Supreme Court held that seizure of non-weapon contraband during a weapons search of a vehicle did not violate the Fourth Amendment.114 Minnesota v. Dickerson115 extended Long to contraband found by touch during a pat down for weapons.116 And a further extension of the weapons search rationale came in Maryland v. Buie,117 where the Court authorized a “protective sweep” of an individual’s house where he was arrested pursuant to a warrant. In Hayes v. Florida,118 the Court authorized, in principle, officers in the field to take fingerprints incidental to Terry stops if the officers had a reasonable suspicion that the individual had committed a crime.119

The final stages of Terry’s expansion to crime control are evident in the pretextual stop authorized in Whren v. United States,120 which permits an investigative stop and search once an officer has probable cause to believe that any crime has occurred, no matter how trivial.121 In that case, the Court not only refused to take into account the subjective motivation of the narcotics officers (who had used a traffic infraction as a pretext to stop suspected drug traffickers); it also refused to consider the argument that an objectively reasonable officer, “acting reasonably,” would not have made the stop “for the reason given.”122 Nor did the Whren Court consider the racial lopsidedness of the incorporation of Terry stops in the practices cited in Whren.123 And in Illinois v. Wardlow124 a suspect’s presence in a “high crime area” was validated as a multiplier of less salient factors into actionable suspicion, including facially subjective rationales such as furtive movements or other criminal appearances.125 Yet, neither the Wardlow majority nor any subsequent cases attempted to standardize the parameters of a “high crime area,” completing the subjectivization of what Terry had launched three decades earlier.126 Judge Alex Kozinski, dissenting in United States v. Montero-Camargo127 summed up the Wardlow challenge in the same year: “Just as a man with a hammer sees every problem as a nail, so a man with a badge may see every corner of his beat as a high crime area.”128

3. Modern Terry doctrine.

This trajectory of cases suggests then that from the initially delimited weapons search in Terry, intended to protect police, the Court has thus extended the reasons for which Terry stops may be conducted well beyond its original boundary: law enforcement can use Terry stops to investigate future, ongoing and past crimes; to check identity, even through fingerprinting; to search a car for identification; to search residences for contraband and weapons; and to search luggage, regardless of the presence of the owner.129 By 1986, the Court had stopped discussing the government’s justification for a stop-and-search in terms of broader government interest or of officer safety. By that year, for example, the most detailed articulations of the crime control functions of a Terry stop were general statements such as that in Terry itself: “effective crime prevention and detection.”130 Or, as in United States v. Hensley:131 “solving crimes and bringing offenders to justice.”132 Or, as in Florida v. Royer:133 questioning related to “the suppression of illegal transactions in drugs or of any other serious crime.”134 Finally, in Floyd v. City of New York, the trial court noted that the conduct of Terry stops as part of a crime control “program,”135 suggestive of what Justice Marshall in Florida v. Bostick136 called a “dragnet,”137 violated the original intent of Terry: to conduct investigative stops to identify imminent or ongoing crimes based on articulable bases of suspicion.

The most recent expansion of Terry’s doctrine was actually not about the parameters of suspicion, but addressed the Fourth Amendment regulation of those boundaries, and whether a violation of reasonable suspicion can even trigger Fourth Amendment relief. In Utah v. Strieff,138 the Court held that the exclusionary rule did not apply to evidence discovered after an unlawful stop that turned up an outstanding arrest warrant. The most important feature of the Court’s opinion was its admission of evidence that was obtained by plainly unconstitutional conduct. Officer James Fackrell stopped Edward Strieff as he was leaving a residence that Fackrell believed was a drug selling location.139 “Over the course of about a week, Officer Fackrell conducted intermittent surveillance of the home. He observed visitors who left a few minutes after arriving there. These visits were sufficiently frequent to raise his suspicion that the occupants were dealing drugs.”140 Fackrell’s conclusions about the illegal activity at that spot were based on an anonymous call to a “drug-tip line” and Fackrell’s own personal experience.141 Once stopped, Fackrell discovered that Strieff had a “small” outstanding arrest warrant for a traffic violation.142 Conducting a search incident to arrest, Fackrell discovered drug paraphernalia and amphetamine in Strieff ’s pockets.

The Strieff Court recognized that the stop was unconstitutional.143   So did the Utah Supreme Court, which had nullified the arrest on the drug charges.144 But, because that conduct was (in the eyes of the Court) neither intentional nor flagrant, the evidence was admitted. Applying an attenuation doctrine that severed the police conduct from the causal chain between the stop and the seizure, the evidence was allowed to stand. The decision seems to go in two directions at once. The Court recognized that the discovery of the warrant was unforeseeable: there are no behavioral indicia that someone may have an outstanding warrant, nor was that condition noted in prior cases as a sign, as the Terry Court required, that “crime is afoot.”145 But the Court also wanted to allow the reasonableness of the stop and warrant check, despite the fact that the discovery of an outstanding warrant was unforeseeable. It is rare, except in extraordinary circumstances as in the Ferguson investigation,146 to discover an outstanding warrant during a routine pedestrian or traffic stop.147 In dissent, Justice Sotomayor characterized the warrant check as “part and parcel of the officer’s illegal ‘expedition for evidence in the hope that something might turn up.’”148 Perhaps most important, the attenuation doctrine applied by the Strieff Court essentially scrubs out reasonableness from the Terry formula.

Justice Sotomayor goes further, claiming that this is hardly an “isolated” event that the Strieff majority claims.149 She describes the same decades of expansion of the Terry logic to justify widespread investigative stops of both pedestrians and vehicles,150 and the risks of humiliating intrusions and abuses during these now routine contacts.151 She goes on to describe the racial skew in the risks of these contacts, describing a “double consciousness” of race and criminality that is instantiated in black and Latino youths.152

D.    Gains and Losses After Terry

The majority of opinions of the courts in the stop-and-frisk cases that followed Terry, as well as recent legal scholarship, argue, “Terry’s regime of stop-and-frisk may well be critical to the fight against violent crime. For that reason, the law enforcement benefits of Terry seem substantial, and the intrusion on liberty that it authorizes seems relatively limited.”153 Yet there has been remarkably little empirical analysis of Terry’s crime control contributions. This crime control agenda, and its claims of efficacy in reducing crime, provides the rationale then to test Terry’s effects on crime.

This essay starts with the notion that, searching for a crime control rationale to justify a broad standard for police intrusions via street stops, Terry’s original sin was forgoing a probable cause standard for investigative stops and substituting an inchoate standard, a standard that is inherently subjective and prone to cognitive distortion, bias and error. Somewhere between that elastic Terry standard in practice today—a practice that often instantiates into policy and program the hunches that so worried the Terry Court—and Mapp’s probable cause standard, lies a threshold of suspicion that can do three things: avoid the petty indignities that have become commonplace in the “new policing,”154 avoid the burdens on the innocents of inefficient stops and intrusions that consume both police resources and citizen trust,155 and contribute substantially to Terry’s crime control agenda. The empirical data in this paper seek out that threshold.

III. Empirical Details

A.    Data and Measures

Data produced in the litigation in Floyd v. City of New York were re-analyzed to address these questions.156 The study period was 2004 though 2012, a lengthy interval to examine trends by month that were sensitive to changes in police stop-and-frisk practices. The data included geocoded records of each stop, which were aggregated to generate counts of stops within police precincts and census block groups for each month.157 The stop data also included police reports of the crime suspected in each stop. These included 133 codes that were reduced to seven categories that reflected the crime categories of interest in the policy debate in New York on the stop regime.158 Crime counts were estimated from crimes reported to the police and geocoded to the nearest street block. These crime reports were then aggregated to generate counts of suspected crimes within precincts and census block groups for each month. The rationales for these units of analysis are discussed infra. The classification categories are shown in Appendix A.

Census data from the 2008 American Community Survey (the midpoint of the time series) were used to generate an empirical description of the social, economic, and demographic conditions for each census block. Although the use of a single time point omits changes in the economic and demographic characteristics of these census blocks during the time-period, only a small portion of the 6475 census block groups were changing dynamically during this interval. I address the effects that temporal trends in areas could have on my estimates by including a linear time trend for each police precinct-month. Precincts are administrative units encompassing census block groups and are substantively important, as this is the spatial unit where uniformed police officers are assigned, and crime control strategies are implemented and managed.

The nine law-defined categories of suspicion that police marked on each stop form were used to state the bases of reasonable suspicion for each stop.159 The boxes included affirmative stop rationales plus an option to check “other” and record the specifics by hand.160 The nine rationales incorporated a set of behavioral categories based on both state and federal case law that would survive a Fourth Amendment test for the individualized stop rationales.161 Officers could check as many boxes as needed to express the basis for the stop. Table 1 lists the categories available for officers to mark the bases of suspicion. In about ninety-five percent of the stops from 2004–2012, officers checked from one to six factors, creating 60,459 possible combinations that express the bases of suspicion for this subset

Table 1. Specific Stop Circumstances and Percent Based on Each Factor



Furtive Movements




Other Stop Circumstance


Evasive Actions


Fits Description


Carrying Crime Objects in Plain View


Drug Transaction


Suspicious Bulge


Actions Indicate Violent Crime


N= 4,575,787

Note: The total exceeds 100 percent due to multiple stop factors indicated per incident.

Source: NYPD Stop-and-Frisk Database, various years.


Three of the nine factors describe observable suspect behaviors that approximate criminal activity: (1) actions indicative of engaging in drug transaction, (2) actions indicative of violent crimes, or (3) “casing” victim or location.162 These factors on their face approximated a probable cause basis for a Terry stop. Each factor is narrow and behaviorally specific, avoiding the vagueness and subjectivity that worried the Terry Court163 and that has translated into recurring constitutional challenges based on Fourth Amendment violations.164 We have only vague ideas about how police discretion is managed in deciding who to stop, and even less information on what exactly they are looking for when they think an action or person looks suspicious.165 While there may be no algorithm to explain how police determinations of suspicious behavior are formed, there are at least observable patterns. The worry in this regime is about unconscious patterns, often racialized, that shape the formation of suspicion based on archetypes such as the “symbolic assailant” and other processes that shape cognition and interpretation of behavioral cues.166 Symbolic cues are clearly problematic, as they have no legal justification.

Judicial opinions make clear that stops based on observations of actions indicative of criminal behavior are constitutional. Actions indicative of a drug transaction that can survive a prima facie claim of probable cause include observed exchange of currency or an object that might contain drugs.167 Some case law suggests that these actions are indicia of criminal drug transactions only if they take place in a “drug-prone” location, although courts have never clarified the meaning of a “drug-prone” location.168 For example, the Strieff Court questioned the formation of suspicion by Officer Fackrell that a particular residence was a “drug location” based on an anonymous tip to a “drug-tip” line about “narcotics activity at a particular residence.”169 Other courts have ruled that a suspected drug transaction with specific behavioral indicia may justify a field interrogation but, absent other factors, cannot justify a frisk.170

Although “casing” can describe a number of different and potentially innocuous behaviors, actions legitimately indicative of casing either a victim or a location can justify a stop and frisk.171 Reasonable suspicion that a person may have been involved in a violent crime can support a stop and frisk, even without other evidence of actual violent or otherwise dangerous behavior.172 So too can threats of violence.173 Still, actions short of behavioral indicia of imminent or ongoing violence run the risk of vague and subjective interpretation by police contemplating a stop, and courts have urged caution in making the leap from “furtive movements” or “evasive actions” to violent crime.174

The behavioral grounding of these three categories provides little room for cognitive error or perceptual distortion, and is consistent with state and federal case law on probable cause.175 In addition, courts have said that observed criminal behavior is sufficient on its own to justify a police stop.176 In contrast, the other six categories of suspicion in these data require subjective judgments and attributions of intent: (1) furtive movements, (2) fits descriptions, (3) carrying objects in plain view, (4) suspicious bulge, (5) evasive actions, or (6) “other.”177 In contrast to observations of specific criminal activity, these factors are vulnerable to cognitive bias and error, as well as racialized attributions of suspicion or criminality.178

By hiving off the three categories of suspicion that are closer in meaning to a probable cause standard, the empirical strategy here is to determine how the use of these three categories of stops influences crime rates, net of other social and crime conditions. I estimate the number of probable and non-probable cause stops in each census block group each month to assess their separate and combined effects on crimes in later months.

B.   Empirical Strategy

All statistical models were estimated as Poisson regressions with standard errors clustered by block groups to control for unmeasured variation and correlation within block groups.179 The regressions include a measure of the total stop, question, and frisk activity (SQF ) per month and a measure of the subset of stops based on “probable cause” justifications. A separate parameter for each month-precinct is included to control for separate trends within the larger precinct units in which block groups are nested.  Precincts are relevant as the management unit to supervise officers and deploy them to locales within the precincts. Assignments of officers change each month within precincts, if not more frequently, based on decisions made by precinct commanders.

The model takes the form:

(1) Yi,b,t = µi + λp,(i),t + β1Di p t + β2Pi + β3Si+ β4D*Pit + β5D*Si + β6X + εi,p,t

where Yi,b,t  is the number of crimes in block group i located in precinct p in month t, λp(i),t is a measure of the crime rate in the block group the month before, D measures the number of stop factors indicated in stops, P measures the percent of probable cause stops, and S measures the total number of stops made in the block group in a month. For this model the parameter (β3) for S is constrained to equal one, so that P and D become rates per overall stop in each block month.180 The regression model also includes a time trend for the month-precinct. An interaction (D*P) between the number of probable cause stops and the average total of stop factors indicated in each stop is also included.

In this model, X represents a set of control variables measuring local social conditions, including racial composition (percent black, percent Hispanic), poverty, age structure, immigrant concentration, average educational attainment, and the housing vacancy rate. These are measured for the midpoint of the time series, 2009, using census data from the American Community Survey’s five-year estimates.181 I also control for block groups in low-crime and low-population business districts, where cues of suspicion leading to stops may be more likely to be formed based on observations of individuals’ behaviors and not priming from the local neighborhood context.182

The initial specifications estimate effects with lags of two months and leads of two months. This empirical strategy allows me to estimate the effects of stops net of the threats of reverse causation. The forward lag, or lead, tests the sensitivity of the results against spuriousness owing to temporal order183 or the possibility of regression to the mean.184 Blocks may receive an increase in overall stops due to a recent crime spike, so that mean reversion would lead to upwardly biased estimates in the regressions of monthly crime rates. To test for residual effects of stops on crime over a longer time-period the models are also estimated with six-month leads and lags.

IV. Results

There are 6,495 census block groups in New York City, as of the 2010 decennial census.185 Table 2 shows descriptive statistics for the city’s block groups. The average residential population is 1,348.9 persons located in land areas averaging 0.047 square miles.186 The racial and ethnic population characteristics suggest the diversity of the city’s population, with percent Non-Hispanic white, Non-Hispanic black, and Hispanic populations nearly equally distributed. Non-

Table 2.  Block Group Descriptive Statistics



Std. Dev.

Total Population



Racial and Ethnic Composition



     % White NH



     % Black NH



     % Hispanic



     % Other NH



Highest Education

     % > High School Grads



     % College Degree +





     Vacancy Rate



Income and Poverty


     % Public Assistance



     % Below Poverty



     Per Capita Income ($)





     Stops per Month



     PC Stops per Month



Crime per Month


     Violent - Felony



     Property - Felony



     Drug Crimes



     Other - Felony














N of Block Groups




Source: American Community Survey, 2006-10 Estimates;
New York City Police Department, Crime Complaints, various years;
New York City Police Department, Stop and Frisk Data, various years.


Hispanic whites are a plurality at 33.3%. Asians comprise 12.6% of the city’s population, the majority of the “Other Race” group of 15.3%. More than one in three adults over the age of twenty-five has a college degree or post-graduate study, and about one in five (22.4%) did not graduate from high school. One in five (21.6%) households live below the federally defined poverty threshold. Median per-capita income is $30,718 per year. The housing vacancy rate, a correlate of crime,187 averages 9.1% of total housing units in the block group.

Indicia of crime and enforcement also show considerable range and skew by census block group. Monthly crime counts appear low at first glance, but when aggregated across census block groups each month, the crime counts add up. The standard deviations again show the skew in these crime counts. Terry stops per month average 4.1 stops, with a standard deviation of 9.3. Of those stops, about half (46.1%) fit the definition of “probable cause” stops (hereinafter PC stops), with a standard deviation of 4.4. Figure 1 shows the distribution of PC and “non-probable cause” stops (hereafter NPC stops) per month over the nine-year period. PC stops were less frequent than NPC stops for nearly every month in the time series until May 2012. In that month, the counts of PC and NPC stops evened out, and the total number of stops declined sharply. The onset of the decline coincided with a class certification ruling in the Floyd litigation that allowed the litigation to proceed to trial.188


[[{"fid":"391","view_mode":"full","fields":{"format":"full","field_file_image_alt_text[und][0][value]":false,"field_file_image_title_text[und][0][value]":"Probable Cause and Non-Probable Cause per Month, New York City","field_caption[und][0][value]":"","field_photo_credit[und][0][value]":""},"type":"media","field_deltas":{"10":{"format":"full","field_file_image_alt_text[und][0][value]":false,"field_file_image_title_text[und][0][value]":"Probable Cause and Non-Probable Cause per Month, New York City","field_caption[und][0][value]":"","field_photo_credit[und][0][value]":""}},"attributes":{"title":"Probable Cause and Non-Probable Cause per Month, New York City","class":"media-element file-full","data-delta":"10"}}]]


The specific stop circumstances that officers mark down for each stop were shown earlier in Table 1. Of the 46.1% that are classified as PC stops, more than half are based on suspicion of “casing,” the same circumstance that animated the 1963 stop of John Terry.189 A judgment that a suspect is casing a person or a location requires a subjective assessment and interpretations of specific behaviors that may be precursors of a crime. Of the three categories of PC stops, this is the most subjective. Suspicion under this category that rises to the level of action by an officer should require a lengthy period of police observation of the suspect or suspects in order to rule out innocent or casual actions and to show that the behavior is sustained over more than just a few minutes. The judgment requires more cognitive work than do judgments based on the other categories, where the behaviors may be more repetitive across events and circumstances, and where the actions and gestures are less ambiguous.190 And that cognitive work also can offset implicit biases in perception that can infect instantaneous or snap judgments of suspect actions, biases based on place, race, or archetypes such as the symbolic assailant.191 In contrast, “furtive movements,” marked as the basis of suspicion in movements, comprise half of the police stops in 2004–12, and represent the most vague and subjective indicia of suspicion.192

Next, Table 3 reports the results of regressions showing the effects of PC stops on crime. Two different model specifications regressions were estimated. The first analyzed the effects of PC stops alone on six different types of crime, with total number of both PC and NPC stops as a control variable. The second version analyzed the effects of PC stops controlling not only for the total number of stops, but also including a measure of the average number of indicia of suspicion marked in the stops conducted in each block group-month observation. In other words, this second estimate represents the effects of PC stops controlling for the totality of suspicion applied by officers in making PC stops in each block-group month. The results are reported as Incident Rate Ratios (IRR). An IRR expresses the change in the dependent variable given a change in the value of the predictor, with a mean of 1.0 indicating no change.193

The upper panel of Table 3 shows the effects of PC stops alone on crime. These estimates examine crimes with a lag and lead of two months. Each model is significant, although the large number of observations (6490 block groups for 108 months) reduces the importance of significance as a measure of model strength. More important are the effect sizes. The IRR estimates range from .924 for violent crimes to .969 for weapons offenses. Interpreting the IRR estimates as rates of change, these models show that for every increase of one PC stop, the various crime types will decline in each block group by anywhere from roughly three percent to seven percent. These are average effects across the block groups of the city over the 108 months of the study interval.

[[{"fid":"396","view_mode":"full","fields":{"format":"full","field_file_image_alt_text[und][0][value]":false,"field_file_image_title_text[und][0][value]":"Regression Estimates for PC Stops","field_caption[und][0][value]":"","field_photo_credit[und][0][value]":""},"type":"media","field_deltas":{"11":{"format":"full","field_file_image_alt_text[und][0][value]":false,"field_file_image_title_text[und][0][value]":"Regression Estimates for PC Stops","field_caption[und][0][value]":"","field_photo_credit[und][0][value]":""}},"attributes":{"title":"Regression Estimates for PC Stops","class":"media-element file-full","data-delta":"11"}}]]

The lower panel of Table 3 shows the effects of PC stops interacted with the total number of stop factors marked by an officer in those stops, or the total quantity of suspicion in each case. The results are again significant in all models, and the IRR estimates this time are larger, ranging from .836 for weapons offenses to .680 for property crimes. Translating this into effect sizes, these estimates show reductions for each increase in PC stops from 16.2% for weapons offenses to approximately 32% for property and violent crimes. Again, these are average monthly effects across the block groups. In an era of declining crime rates in the city,194 these effects based on stop type are quite large. The implication as well is that higher concentrations of NPC stops are unproductive and add nothing to the crime control efforts of law enforcement.

The concentration of PC stops varied in each block-group- month.  The range of PC stop concentration raises the question of threshold effects. At what point do crime rates deflect downward as the concentration of PC stops increases? Figures 2.1 and 2.2 show the marginal effects of PC stops at ten percent intervals in the distribution of PC stops. Marginal effects are the values of a predictor variable on the dependent variable in a regression that is estimated from the specific or fixed values of that predictor with the other predictors held constant at their means or averages.195 In this case, marginal effects are estimated on total crime for each ten percent increment of the percentage of PC stops in a block group-month. Each figure corresponds to the two regression strategies reported in Table 3: PC stops as a predictor (Figure 2.1), and PC stops plus total suspicion as a predictor (Figure 2.2).


[[{"fid":"416","view_mode":"full","fields":{"format":"full","field_file_image_alt_text[und][0][value]":false,"field_file_image_title_text[und][0][value]":"Effects of PC Stops on Crime with 95% CI","field_caption[und][0][value]":"","field_photo_credit[und][0][value]":""},"type":"media","field_deltas":{"12":{"format":"full","field_file_image_alt_text[und][0][value]":false,"field_file_image_title_text[und][0][value]":"Effects of PC Stops on Crime with 95% CI","field_caption[und][0][value]":"","field_photo_credit[und][0][value]":""}},"attributes":{"title":"Effects of PC Stops on Crime with 95% CI","class":"media-element file-full","data-delta":"12"}}]]


[[{"fid":"431","view_mode":"full","fields":{"format":"full","field_file_image_alt_text[und][0][value]":false,"field_file_image_title_text[und][0][value]":"Effects of PC Stops and Total Suspicion on Crime with 95% CI","field_caption[und][0][value]":"","field_photo_credit[und][0][value]":""},"type":"media","field_deltas":{"13":{"format":"full","field_file_image_alt_text[und][0][value]":false,"field_file_image_title_text[und][0][value]":"Effects of PC Stops and Total Suspicion on Crime with 95% CI","field_caption[und][0][value]":"","field_photo_credit[und][0][value]":""}},"attributes":{"title":"Effects of PC Stops and Total Suspicion on Crime with 95% CI","class":"media-element file-full","data-delta":"13"}}]]

Each figure shows that the crime reduction effects of PC stop concentration increase beginning when these stops exceed fifty percent of all stops in a block-group-month.  The effects are stable and modest up to a fifty percent concentration of PC stops, and then increase at successive increments. The sharpest increase in crime reduction is at the highest concentrations of PC stops. Figure 2.1 shows that the marginal effects increase from about 1.75 fewer crimes at fifty percent to over two crimes per block group-month with a sharp increase from eighty percent to ninety percent. These are the effects attributable to PC stops and do not reflect other factors related to crime reductions, which the marginal effect model averages over the full range of PC stops in the marginal effects model. Figure 2.2 shows the same pattern. Estimates of the effects of PC stops together with total suspicion across those stops are stable up to a fifty percent concentration, and increase at each successive increment. The largest increase in marginal effects is at the last increment, from eighty percent to ninety percent.

These analyses show the short-term effects of PC stop concentration at two month projections. This form of residual effect could decay over time as would-be offenders adjust to the increased risk of being stopped by the police, or as police officers rotate into and out of patrol assignments which can interrupt their learning and updating of their practices. An important question, then, is what are the residual effects of PC stop concentrations over longer intervals?196 To test for residual effects, the models in Table 3 were re-estimated in Table 4. Table 4 shows only the IRR for total crime and six specific categories of crime using a six-month lag and lead time parameter. The panel on the left of the table shows the IRR for PC stops only after six months; the panel on the right shows the effects of PC stops plus the totality of suspicion in those stops.

Table 4 and Figures 3.1 and 3.2 show that for both PC stops and PC stops with total suspicion, the crime reduction effects at six months are similar to the effects at two months. There are small and negligible differences in the effect sizes from two to six months for both sets of models. PC stops produce a 6.6% decline in total crime, and crime-specific reductions ranging from 7.7% for violent crime to 3.1% for weapons offenses. The reductions in total crime for PC stops with total suspicion are nearly 30%, with crime-specific reductions ranging from 16.5% for weapons offenses to 32.3% for property crimes. These percentages are based on generally low offense counts, but these monthly reductions aggregate over time to produce important and sizable safety benefits.

[[{"fid":"446","view_mode":"full","fields":{"format":"full","field_file_image_alt_text[und][0][value]":false,"field_file_image_title_text[und][0][value]":"Poisson Regression of Probable Cause Stops on Crime with Six Month Lead and Lag","field_caption[und][0][value]":"","field_photo_credit[und][0][value]":""},"type":"media","field_deltas":{"14":{"format":"full","field_file_image_alt_text[und][0][value]":false,"field_file_image_title_text[und][0][value]":"Poisson Regression of Probable Cause Stops on Crime with Six Month Lead and Lag","field_caption[und][0][value]":"","field_photo_credit[und][0][value]":""}},"attributes":{"title":"Poisson Regression of Probable Cause Stops on Crime with Six Month Lead and Lag","class":"media-element file-full","data-delta":"14"}}]]


[[{"fid":"456","view_mode":"full","fields":{"format":"full","field_file_image_alt_text[und][0][value]":false,"field_file_image_title_text[und][0][value]":"Poisson Regression of Probable Cause Stops on Crime with 6 Month Lead and Lag","field_caption[und][0][value]":"","field_photo_credit[und][0][value]":""},"type":"media","field_deltas":{"15":{"format":"full","field_file_image_alt_text[und][0][value]":false,"field_file_image_title_text[und][0][value]":"Poisson Regression of Probable Cause Stops on Crime with 6 Month Lead and Lag","field_caption[und][0][value]":"","field_photo_credit[und][0][value]":""}},"attributes":{"title":"Poisson Regression of Probable Cause Stops on Crime with 6 Month Lead and Lag","class":"media-element file-full","data-delta":"15"}}]]


[[{"fid":"466","view_mode":"full","fields":{"format":"full","field_file_image_alt_text[und][0][value]":false,"field_file_image_title_text[und][0][value]":"Poisson Regression of Probable Cause Stops with Total Suspicion on Crime with 6 Month Lead and Lag","field_caption[und][0][value]":"","field_photo_credit[und][0][value]":""},"type":"media","field_deltas":{"16":{"format":"full","field_file_image_alt_text[und][0][value]":false,"field_file_image_title_text[und][0][value]":"Poisson Regression of Probable Cause Stops with Total Suspicion on Crime with 6 Month Lead and Lag","field_caption[und][0][value]":"","field_photo_credit[und][0][value]":""}},"attributes":{"title":"Poisson Regression of Probable Cause Stops with Total Suspicion on Crime with 6 Month Lead and Lag","class":"media-element file-full","data-delta":"16"}}]]


The comparative advantage in crime reduction benefits of focusing stops on indicia of suspicion that are more closely aligned with probable cause and behavioral markers are evident in these analyses. It would be important to identify the underlying mechanisms for these effects, but that would require a very different and ethically challenging research enterprise, including strong identification strategies that account for concurrent sources of deterrence, as well as testing the specific underlying mechanisms of deterrence.197 For now, it is not hard to imagine that by narrowing the scope of suspicion to behaviors more closely aligned with criminal activities, the emphasis on accuracy allows the signals of deterrence to be aimed more directly and less speculatively or subjectively at persons who may be deciding about a possible crime. Contemporary theories of deterrence agree that the risks of detection and apprehension are essential to effective deterrence.198

V. Redeeming the Original Sin

There are tradeoffs in crime returns—shown here and in other studies assessing the effects of probable cause stops199 —when stop regimes lean heavily on subjective or inchoate indicia of suspicion over more objective behavioral markers. This tradeoff is one part of Terry’s original sin. Officers can play hunches, but at a price. The original sin then, was less a question of moving away from Mapp’s probable cause standard than it was inviting police to use their authority to conduct temporary detentions and investigations based on the very hunches that worried the Terry Court.200

The Terry Court, and subsequent Fourth Amendment opinions, chose to define neither the substantive criteria of reasonable suspicion, nor how much suspicion is required for police officers to conduct an investigative stop.201 Even when courts exercised caution in expressing what is reasonable suspicion, the standards often were simply an elongated expression of Terry’s binary approach distinguishing reasonable suspicion from probable cause.202 For example, New York State standards are more demanding than the standards in other states for investigative stops203 as well as the Terry standard, yet the De Bour and Holman courts in New York defaulted to a subjective cascade of four increasing levels of intrusion.204 Redemption for Terry’s sins can come in two forms.

A.    Terry at Fifty

Whether in the binary or a more detailed articulation of reasonable suspicion such as De Bour, the Terry component of the new policing seems, as practiced, to have failed at least the crime control prong of Terry’s balancing test.205 Along the way, Terry’s failure to provide substance or quantity to the concept of actionable “suspicion” created a subjective terrain that invited police to use broad assessments of suspicion—at times, hunches—that have raised a steady stream of Fourth Amendment problems.206 While this was not unknown or unanticipated at the outset of the Terry era, the inherent subjectivity of reasonable suspicion became problematic decades later as the “new policing” unfolded and investigative stops became less a practice of individual discretion as a systematic policy-driven program.207 Investigative stops were an essential element of modern proactive policing, and those often were pursued aggressively both in quantity and interaction quality. Yet, these analyses show in New York City that stops based on general categories of suspicion that are not tied to a particular behavior have no crime reduction benefit, even though they were encouraged in an effort to reduce crime.

At the outset of the Terry era, in the midst of spiking crime rates and civil unrest,208 the Terry Court’s comments on reasonable suspicion were framed as making sure that an officer’s actions were reviewable both within police agencies and in the courts.209 Neither the Terry Court, nor later courts reviewing Terry’s standard, ever articulated sufficient detail to allow the police to know whether their actions were constitutional. Substance was almost never part of that discussion. While the reviewability prong of Terry’s doctrine no doubt anticipated a period of sorting out by appellate courts,210 what was a small caseload burden on the courts grew over the years into a long string of contentious decisions, nearly all of which expanded the scope of reasonable suspicion.211 So, one of Terry’s sins was placing a substantial burden of review on federal trial and appellate courts in a succession of suppression motions and constitutional challenges.212 That may well have been the opposite of what the Terry court perhaps sought: to create a procedural rule that would relieve courts of that burden. Asking courts to perform a regulatory function in sorting out constitutional violations in everyday policing is one thing, but asking those same courts to do so in the absence of an articulable standard of conduct is asking too much.

The story becomes more complicated by the fact that some stops, however subjective they may be, will be constitutional under the current case law. Even when a lawful stop proceeds with verbal or physical aggression by the officer, the stop can be lawful under Terry’s expansive view of reasonable suspicion.213 Stops are constitutional so long as officers can articulate facts that are reasonable to other officers given their knowledge and circumstances. But the story is complicated simply by the numbers: when suspicion becomes so inflated as to challenge the boundaries of legality, then it is the practice itself that becomes a contested constitutional matter. The boundary between lawful and unlawful policing is not easy to draw,214 but courts as well as government have done so now on several occasions.215 Still, when courts act as regulators of reasonable suspicion, police officers who are able and willing to spin their behavior in a way that will satisfy judges who reflexively defer to police “expertise,” while officers who are less verbally facile or who are transparent about their subjective assessment and motivations are more likely to be penalized.216

The lesson here is that some bases of suspicion are both constitutional and productive, while others may be constitutional but unproductive, and still others are neither constitutional nor productive. But since all stops have costs that are borne by innocents as well as the guilty, communities and those stopped should not have to pay for those costs that are not worth it. Those stopped shouldn’t have to bear the burdens of police hunches or stops otherwise based on thin suspicion if they in fact are not guilty. This was one part of the “impossibility” that Professor Stuntz wrote about in 1998,217 shortly after the downside of the new policing became a focus of legal, political, and social conflict.

B.    Moving to Regulation

A reset to more concrete indicia of suspicion suggested by the empirical results here hold promise to reverse those sins. Put simply, the Terry standard should be pushed toward a narrower and objective standard in light of this research. The reset involves two domains. One is a constitutional story that also raises regulatory issues. The constitutional story asks whether the procedure and subjective standards in Terry and later cases lead to violations of individual rights, and whether citizens can be asked to sacrifice those rights to social welfare criminal justice interests. The regulatory story is an institutional story: how to design models of oversight and assessment by accountable agents to ensure that the practices remain within the diffuse boundaries of reasonable suspicion, and also emphasizing the use of stops that maximize social welfare goals of crime deterrence. The regulatory challenge is to tether that practice both to the constitutional parameters and to practices that pay.

The prospects for regulation of Terry stops through a more narrowly tailored schema of suspicion are good. At the least, shifting stops toward probable cause or behavioral indicia will shrink the stop circumstances that might otherwise be legally contested, reducing the burdens on trial and appellate courts. A shift in emphasis also creates a vocabulary and logic for internal audit, supervision, and regulation. Officers can be required to answer for what they do, not what they say.218 In contrast, it is not hard to imagine the difficulty of internally auditing the indicia of suspicion for the vague categories of “furtive movements” that were sharply criticized in the Floyd opinion.219

Instead, the process of auditing a claim of “violent crime” or “drug transaction” or even the more subjective marker of “casing a store” can involve a perceptually shared set of behavioral categories that might be more amenable to training on substantive criteria in lieu of procedural ones. Auditing of officers’ expressions of suspicion can promote learning and updating, which should be visible when officers’ actions are viewed across a range of citizen contacts. Auditing internally also creates a context of observational data that can inform collaboration among officers, and for democratic participation by politically accountable agents with police and citizens. The positive returns of collaboration have been observed in studies of police institutional reform in Cincinnati and other smaller departments.220

C.    Harm Reduction

Hewing closer to objective and behaviorally specific markers of suspicion will narrow the circumstances where stops are conducted. A likely result will be a reduction in the scope and magnitude of false positives—low seizure and arrest rates, weak crime control returns—observed in the Floyd and reported by monitors in Bailey v. City of Philadelphia221 litigation. Even though the Terry Court was careful to state that reasonable suspicion was necessary to authorize a frisk, not necessarily the stop itself, its crime control agenda moved reasonable suspicion from the background of the original opinion to the forefront of contemporary case law.222 And in linking Terry stops to a crime control agenda, it was not hard to explain how reasonable suspicion became the basis to justify an investigative stop.223

Beyond the costs of a wrong guess by police that leads to a temporary street detention, the Terry Court worried about a variety of “petty indignities.”224 The indignities of this form of order maintenance in effect piled up from the accumulation of stops, not simply from publicly visible frisks.225 As Terry’s crime control agenda took root, the exposure of citizens, both innocents and those engaged in crime, to a new form street stops grew exponentially. The indignity problem arises not from the indignity of the frisk or the search, but from the context of the stop itself. And the dignity problem also arises not from the sheer prevalence of unproductive stops and the burden on innocents (although that itself is a concern), but from the ways those stops often are conducted.226 Even the most neutral of stops carries emotional freight and the threat of indignities. The concern here is what happens before, during, and after these stops, or how encounters with the police take place and then unfold, rather than on simply the regulatory questions of whether, where, and how often they occur.227

Professor Stuntz identified four types of harm from inchoate and unproductive stops: (1) privacy incursions, or the coercive invasion of one’s property or body; (2) targeting harm, being singled out in public by the police and treated like a criminal suspect;228 (3) the harm of using racial bias to justify these incursions on liberty, or using race as a signal of suspicion if not criminality on black citizens simply by virtue of being black or moving about in a black neighborhood;229 and (4) the risks of verbal and physical force that accompanies stops and searches.230

These psychological and perhaps physical harms risk not only indignities to the person, but also legitimacy costs to the larger community. Intrusive stops that are not based on actual perceived criminal behavior have reduce the perceived legitimacy of the police, and risk the harm of withdrawal of citizens from the co-production (with police) of security and public safety.231 The harms accrue from both direct and vicarious interactions—young persons observing the police are as likely to report lower police legitimacy as are those who have direct experience.232 They are less likely to serve on juries or cooperate with police in investigations of crime reports.233

Assume that legitimacy is produced through the aggregation of interactions within individuals and authorities, and is a language and shared currency of social exchange signals that acknowledge the individual’s dignity, respect, worth and belonging is essential to democratic participation.234 If that were right, then simply reducing the scope of stops and narrowing the bases of suspicion to objective indicia of criminal activity would narrow the scope of indignities and harms. Targeting harms would be reduced by reining in “hunches” or actuarial suspicion based on police officers’ use of collective suspicion or Bayesian attributions of criminal intent.235 And narrowing would also reduce the emotional and psychological toll of these unwarranted intrusions.236 “Why me?” would no longer be a salient question for the residents of many urban neighborhoods who bear much of the burden for the contemporary practice of Terry stops.

The racial component of targeting harms in particular could be addressed through narrowing. Racial tensions are inextricably linked to the drift in reasonable suspicion toward both subjectivity and programmatic overreach.237 Resetting to narrowly defined legal categories may not cure racial disparities, but would likely reduce the disparate exposure to policing by race by narrowing the circumstances for permissible stops. The heaping of indignities from law enforcement and other legal actors on African Americans has special meaning for that community. Glenn Loury explains how the pervasive societal stigmatization of African Americans marginalizes their community from the institutions and norms that “mainstream” society purports to value.238

For African Americans, the sense of gain or loss suffered through the aggregation of social interactions with the state is an important part of collective or shared experiences.239 Conferring respect before the law means conferring social and democratic belonging, a form of social recognition that conveys the shared moral and legal norms between citizens and those who enforce the law. This sense of belonging and social recognition is described in rich detail by Charles Epp and his colleagues in Pulled Over,240 and suggests another, and perhaps more important, potential dignitary benefit of a narrower basis for stops. Epp et al. make an important distinction between being treated respectfully (whether or not lawfully) and being treated legally.241 The respondents in their survey were more concerned with being treated legally than with politeness or other procedural qualities.242 In this form of social recognition in law, we imagine ourselves as how other people see us, and we understand who we are in and through our relationships with others. This form of recognition by legal actors—by applying law equally—affects the ties of groups to legal authority, and to the moral norms that legal actors both express and enforce. In other words, being nice is one thing, perhaps a low cost and generous if not patronizing act on the part of a legal actor holding considerable power, but expressively acknowledging the rights of a person before the law is quite another. Undoing dignitary harms leans strongly toward the latter view.

D.    Regulatory Algebra

Substantive laws that criminalize relatively harmless or benign acts can animate the use of police power to carry out Terry stops to a broad spectrum of behaviors and actions.  But it was never clear that the use of Terry’s stop power was aimed at trivial crimes, or as a pretext for conducting field investigations as a fishing exercise based on hunches.243 Those hunches often are instantiated in current practice,244 which seem to be artifacts of the capacious and inchoate indicia of reasonable suspicion. One remedy is to recalibrate suspicion in the context of Terry stops to move away from subjective criteria and reliance on officers’ experience-based judgments to a regime of objective, behaviorally-defined indicia of suspicion.

The move toward more objective and behavioral bases of suspicion does not mean that the police should abandon the practice of Terry stops. What it does imply is that there is a tradeoff to using this power too broadly, and the regulatory response requires adjusting the thresholds for police contact. As a matter of policy, the broad use of Terry’s stop power is encouraged by the new policing, especially in the context of robust order maintenance regimes.245 Therein lies the trouble. Terry stops should require a higher level of suspicion than an officer’s hunch or subjective appraisal that “crime is afoot”. The Terry Court never said which crimes had to be “afoot” to justify a stop, only that the act was criminal. When the criminal law is so broadly enforced, and when non-criminal violations or local ordinances are integrated with the overall mission of street policing to detect weapons and control violence, the likelihood increases that both benign and serious crimes will be part of the umbrella of suspicion. The burden of proof for administrative violations or low-level misdemeanor offenses is intrinsically lower than for felony offenses and places Terry’s fundamental rules at risk.

Imagine that we have two types of acts—a benign act and a harmful one. Intervening in the relatively benign act, such as a violation of an administrative code, seems to benefit almost no one—there are few public benefits to crime control in that instance, since the range of harm is largely private or de minimus. Even if one accepts that some aspects of disorder may be criminogenic, itself a heavily contested notion, the argument here is that the treatment through criminal enforcement may have iatrogenic effects on legitimacy and cooperation. That is, we may stop someone from smoking in public, or drinking from an open container, playing loud music in a residential area, or jumping turnstiles on public transit. We may signal “order” by enforcing these laws, but their relationship to public safety is path dependent on the questionable relationship between theories of social or physical disorder and crime.246 Worse, such enforcement may engender withdrawal if not resistance to cooperation with the police.247 This may seem like an efficient use of a scarce public good—policing—because “hit rates” may be high, but the yield for public safety is low if these low-level crimes are not gateways to violence or major property crimes.


Figure 4. Probability Distributions for Strength of Evidence

[[{"fid":"476","view_mode":"full","fields":{"format":"full","field_file_image_alt_text[und][0][value]":false,"field_file_image_title_text[und][0][value]":"Probability Distributions for Strength of Evidence","field_caption[und][0][value]":"","field_photo_credit[und][0][value]":""},"type":"media","field_deltas":{"17":{"format":"full","field_file_image_alt_text[und][0][value]":false,"field_file_image_title_text[und][0][value]":"Probability Distributions for Strength of Evidence","field_caption[und][0][value]":"","field_photo_credit[und][0][value]":""}},"attributes":{"title":"Probability Distributions for Strength of Evidence","class":"media-element file-full","data-delta":"17"}}]]

Source: Louis Kaplow, Burden of Proof, 121 Yale L. J. 738, 757 (2011).

More important, the burden of proof in these instances is intrinsically low.248 Policing benign acts may satisfy the demand for metrics of police productivity, but its contribution to crime control is dubious.249 Aggressive enforcement of benign acts also has efficiency costs by distracting police from intervening in the more harmful ones. It is only in the shared space of benign and harmful acts, where it makes sense to intervene in the benign act at a lower standard of proof, and the size of that shared space is part of a contentious debate.250 The social harms from undetected harmful acts will outweigh any private or small-scale benefits from intervening in the benign acts. Figure 4 illustrates how the social good in the form of public safety seems to be greater when we focus our attention on the more serious acts. In other words, do not sweat the little stuff, and focus on more serious acts with more consequential public harms. This is simple regulatory algebra.

VI. Conclusion

Both law enforcement and citizen interests are better served by a recalibration of Terry standards to move them closer to Mapp’s more exacting probable cause standard. A more workable and easily understood standard for regulating police use of the stop power would create a more comfortable space internally for police to monitor, audit, and regulate compliance with constitutional law as well as internal policy. And it also can provide a standard that moves away from the subjective criteria that Terry invited and toward criteria that are less vulnerable to cognitive error, perceptual distortions, and social harms. Secondary benefits for legitimacy may well follow. Penance for Terry’s original sin is within reach.


Appendix A           

Aggregate Category

Suspected Offenses



Violent Crime

Aggravated Assault


Aggravated Harassment


Aggravated Sexual Abuse









Minor Violent Crime









Reckless Endangerment


Resisting Arrest




Unlawful Imprisonment


Vehicular Assault

Drug Crime

Criminal Possession of Controlled Substances


Criminal Sale of Controlled Substances


Criminal Possession of Drug Paraphernalia


Other Drug Offenses

Marijuana Possession

Criminal Possession of Marijuana

Marijuana Sale

Criminal Sale of Marijuana

Part I Property Crime





Grand Larceny


Grand Larceny Auto

Minor Property Crime

Auto Stripping


Computer Trespass


Criminal Possession of Stolen Property


Criminal Mischief


Criminal Possession of Computer Materials


Criminal Possession of Forged Instruments


Criminal Tampering


Misapplication of Property


Petit Larceny

Possession of Burglar Tools

Reckless Endangerment of Property

Theft of Services

Unauthorized Use of a Vehicle

Fraud and Related

Falsifying Business Records




Forgery of a VIN




Fraudulent Accosting


Insurance Fraud


Tampering with a Public Record


Unlawful Use of Credit Card, Debit


Criminal Trespass

Prostitution and Related




Quality of Life/Disorder



Fortune Telling






Making Graffiti




Obstructing Firefighting Operations


Obstructing Governmental Administration


Possession of Graffiti Instruments


Trademark Counterfeiting


Unlawfully Dealing with Fireworks


Unauthorized Recording


Unlawful Assembly


Disorderly Conduct


Quality of Life


Riding Bike on the Sidewalk


Alcohol Violation

Sex Crimes and Related







Course of Sexual Conduct




Public Display of Offensive Sexual Material


Public Lewdness


Sexual Abuse


Sexual Misconduct




Forcible Touching

Other Minor Sex Crimes


  • 1367 U.S. 643 (1961).
  • 2Id. at 655.
  • 3Uniform Crime Reporting Statistics, Estimated Murder and Nonnegligent Manslaughter 1961–68, (
  • 4Kerner Comm’n, National Advisory Commission on Civil Disorders 19–60 (1968).
  • 5See, e.g., Michael Agar and Heather Schacht Reisinger, A Heroin Epidemic at the Intersection of Histories: The 1960s Epidemic Among African Americans in Baltimore, 21 Med. Anthropology 115 (2002).
  • 6President’s Comm’n on Law Enforcement and Criminal Justice, The Challenge of Crime in a Free Society 189 (1965) (discussing the links between the rise of heroin addiction and criminal violence in the mid-1960s); see generally 2 Leon Hunt & Carl Chambers, The Heroin Epidemics: A Study of Heroin Use in the United States, 1965–75 (1976).
  • 7See Dallin H. Oaks, Studying the Exclusionary Rule in Search and Seizure, 37 U. Chi. L. Rev. 665, 739 (1970).
  • 8See generally Thomas Y. Davies, A Hard Look at What We Know (and Still Need to Learn) About the “Costs” of the Exclusionary Rule: The NIJ Study and Other Studies of “Lost” Arrests, 1983 Am. B. Found. Res. J. 611 (1983); Peter F. Nardulli, The Societal Cost of the Exclusionary Rule: An Empirical Assessment, in 8 Am. B. Found. Res. J. 585 (Summer 1983); see also Potter Stewart, The Road to Mapp v. Ohio and Beyond: The Origins, Development and Future of the Exclusionary Rule in Search-and-Seizure Cases, 83 Colum. L. Rev. 1365, 1439–40 (1983); William J. Stuntz, The Virtues and Vices of the Exclusionary Rule, 20 Harv. J. L. & Pub. Pol’y 443, 445 (1996).
  • 9392 U.S. 1 (1968).
  • 10See Earl C. Dudley, Jr., Terry v. Ohio, The Warren Court, and the Fourth Amendment: A Law Clerk’s Perspective, 72 St. John’s L. Rev. 891, 900 (1998).
  • 11Terry, 392 U.S. at 21, 35.
  • 12Id. at 21.
  • 13Id. at 38.
  • 14See id. at 27 (“[T]here must be a narrowly drawn authority to permit a reasonable search for weapons for the protection of the police officer . . . .”).
  • 15Id. at 22.
  • 16See generally Whren v. United States, 517 U.S. 806 (1996); see also David A. Sklansky, Traffic Stops, Minority Motorists, and the Future of the Fourth Amendment, 1997 Sup. Ct. Rev. 271, 274–91 (1997).
  • 17See Illinois v. Wardlow, 528 U.S. 119, 125 (2000); see also Andrew Ferguson & Damien Bernache, The “High-Crime Area” Question: Requiring Verifiable and Quantifiable Evidence for Fourth Amendment Reasonable Suspicion Analysis, 57 Am. U. L. Rev. 1587, 1606 (2008).
  • 18See Uniform Crime Reporting Statistics, supra note 3.
  • 19See generally Phillip B. Heymann, The New Policing, 28 Fordham Urb. L.J. 407 (2000); see also Jack Maple with Chris Mitchell, The Crime Fighter: Putting the Bad Guys Out of Business (2000).
  • 20959 F. Supp. 2d 540 (S.D.N.Y. 2013).
  • 21See generally President’s Comm’n on Law Enforcement and Criminal Justice, supra note 6.
  • 22Kerner Comm’n, supra note 4; see also Fred R. Harris & Roger W. Wilkins, Quiet Riots: Race and Poverty in the United States, in The Kerner Report: Twenty Years Later (1988).
  • 23367 U.S. 643, 657 (1961).
  • 24232 U.S. 383, 398 (1914).
  • 25338 U.S. 25 (1949).
  • 26See id. at 33.
  • 27Mapp, 367 U.S. at 656 (quoting Elkins v. United States, 364 U.S. 206, 217 (1960)).
  • 28Weeks, 232 U.S. at 393–94.
  • 29See U.S. Const. amend. IV.
  • 30Weeks, 232 U.S. at 392–93.
  • 31Mapp, 367 U.S. at 655–56.
  • 32Scott E. Sundby, An Ode to Probable Cause: A Brief Response to Professors Amar and Slobogin, 72 St. John’s L. Rev. 1133, 1134 (1998).
  • 33Dudley, supra note 10, at 896.
  • 34407 U.S. 143 (1972).
  • 35See id. at 146 (“A brief stop of a suspicious individual, in order to determine his identity or to maintain the status quo momentarily while obtaining more information, may be most reasonable in light of the facts known to the officer at the time.”); see also Carol Steiker,
    Terry Unbound, 82 Miss. L. J. 329, 338 (2013).
  • 36See Terry v. Ohio, 392 U.S. 1, 37 (1968) (Douglas, J., dissenting); see also Stephen A. Saltzburg, Terry v. Ohio: A Practically Perfect Doctrine, 72 St. John’s L. Rev. 911, 929 (1998). It is almost bitterly ironic that Justice Douglas’s phrase has come to be so closely associated with the Terry standard he disagreed with. It was not until Almeida-Sanchez v. United States, that the Court referred to the Terry standard as “reasonable suspicion.” 413 U.S. 266, 268 (1973); see also Saltzburg, supra, at 945.
  • 37Sundby, supra note 32, at 1135.
  • 38Saltzburg, supra note 36, at 925–26.
  • 39See Terry, 392 U.S. at 12–15.
  • 40Id.; see also Minnesota v. Dickerson, 508 U.S. 366, 381 (1993) (Scalia, J., concurring) (stating “[I doubt that] the fiercely proud men who adopted our Fourth Amendment would have allowed themselves to be subjected, on mere suspicion of being armed and dangerous, to such indignity”). Justice Scalia went on to say, however, that the framers were not concerned with the likelihood that the criminals of that day were carrying concealed firearms. Id.; see also Erich J. Segall, Will the Real Justice Scalia Please Stand Up?, 50 Wake Forest L. Rev. 101, 106–07 (2015) (quoting Justice Scalia’s concurrence in Dickerson and explaining his qualification of his own ambivalence on Terry’s “petty indignities”).
  • 41Terry, 392 U.S. at 14.
  • 42Id. at 14–15; see, e.g., William J. Stuntz, Terry’s Impossibility, 72 St. John’s L. Rev. 1213, 1218 (1998) (discussing a nosology of harms that are inevitable in the proactive policing regimes that emerged in the 1990s); Anthony Thompson, Stopping the Usual Suspects, 74 N.Y.U. L. Rev. 956, 962–73 (1999) (revisiting Terry to show its forgotten racial meaning and implications); Jeffrey Fagan, Ackerman Lecture Series on Equality and Justice: Indignities of Order Maintenance (Nov. 21, 2013) (showing the racial components of the dignity incursions inherent in Terry stops).
  • 43Terry, 392 U.S. at 15.
  • 44See, e.g., Amy Lerman & Vesla Weaver, Arresting Citizenship: The Democratic Consequences of American Crime Control (2014); Ben Bradford et al., Police, Crime and Order: The Case of Stop and Search, in The Sage Handbook of Global Policing (Summer 2016); Tracey L. Meares, Programming Errors: Understanding the Constitutionality of Stop-and-Frisk as a Program, Not an Incident, 82 U. Chi. L. Rev. 159, 163 (2015).
  • 45Dudley, supra note 10, at 900.
  • 46Terry, 392 U.S. at 20 (“We do not retreat from our holdings that the police must, whenever practicable, obtain advance judicial approval of searches and seizures through the warrant procedure . . . .”).
  • 47David A. Harris, Particularized Suspicion, Categorical Judgments: Supreme Court Rhetoric Versus Lower Court Reality Under Terry v. Ohio, 72 St. John’s L. Rev. 975, 985 (1998) (“[T]he conclusion is inescapable that in Terry the police won back a significant part of the power they needed to conduct business according to pre-Mapp standards.”).
  • 48405 U.S. 156 (1972).
  • 49Terry, 392 U.S. at 20.
  • 50Id. at 19 n.16; see also Dudley, supra note 10, at 896–98.
  • 51Terry, 392 U.S at 19 n.16.
  • 52Id. at 16.
  • 53Id. at 25–26.
  • 54Floyd v. City of New York, 959 F. Supp. 2d 540, 559 (S.D.N.Y. 2013); see also Report of Jeffrey Fagan, Ph.D., Floyd v. City of New York, 959 F. Supp. 2d 540 (S.D.N.Y. 2013) (No. 08 Civ. 01034) (hereinafter Fagan Report); Second Supplemental Report of Jeffrey Fagan, Ph.D., Floyd v. City of New York, 959 F. Supp. 2d 540 (S.D.N.Y. 2013) (No. 08 Civ. 01034) (hereinafter Fagan Supplemental Report).
  • 55387 U.S. 523 (1967).
  • 56Terry, 392 U.S. at 21 (quoting Camara, 387 U.S. at 537).
  • 57Id. at 21, 29–30.
  • 58Id. at 26–27.
  • 59Id. at 21.
  • 60Id. at 27.
  • 61Id.; see also Thompson, supra note 42, at 971 (noting that the Court celebrated Officer McFadden’s lengthy tenure and experience in policing the same neighborhood where he stopped John Terry and his colleagues).
  • 62Terry, 392 U.S. at 22 (emphasis added).
  • 63Bernard E. Harcourt & Tracey L. Meares, Randomization and the Fourth Amendment, 78 U. Chi. L. Rev. 809, 850 (2011); see, e.g., Stuntz, supra note 42, at 1215 (“[R]easonable suspicion has never received a solid definition. (Perhaps it can’t).”).
  • 64William J. Stuntz, Terry and Substantive Law, 72 St. John’s L. Rev. 1362, 1362 (2012).
  • 65Terry, 392 U.S. at 15.
  • 66392 U.S. 40 (1968).
  • 67See Terry, 392 U.S. at 5; Sibron, 392 U.S. at 47–48. In Terry, the Court somewhat cryptically noted that its “approval of legitimate and restrained investigative conduct undertaken on the basis of ample factual justification should in no way discourage the employment of other remedies than the exclusionary rule to curtail abuses for which that sanction may prove inappropriate.” 392 U.S. at 15 (emphasis added). However, the Court did not itself suggest any alternate remedies.
  • 68Sibron, 392 U.S. at 62.
  • 69Id.
  • 70Id. at 65.
  • 71Terry, 392 U.S. at 17; see also id. at 24–25 (“Even a limited search of the outer clothing for weapons constitutes a severe, though brief, intrusion upon cherished personal security, and it must surely be an annoying, frightening, and perhaps humiliating experience.”).
  • 72See Camara v. Municipal Court, 387 U.S. 523, 536–37 (1967) (“Unfortunately, there can be no ready test for determining reasonableness other than by balancing the need to search against the invasion which the search entails.”); see also Jeffrey Fagan & Amanda Geller, Following the Script: Narratives of Suspicion in Terry Stops in Street Policing, 82 U. Chi. L. Rev. 51, 58 (2015); Harcourt & Meares, supra note 63.
  • 73134 S. Ct. 1683 (2014) (Scalia, J., dissenting).
  • 74Id. at 1695.
  • 75See Floyd v. City of New York, 959 F. Supp. 2d 540, 559, 575 (S.D.N.Y. 2013) (pointing to the fact that “[t]he rate of arrests arising from stops is low . . . and the yield of seizures of guns or other contraband is even lower,” and noting “that the City’s attempt to account for the low rate of arrests and summonses following stops was not persuasive”).
  • 76Fagan & Geller, supra note 72.
  • 77See Steven Grossman, Whither Reasonable Suspicion: The Supreme Court’s Functional Abandonment of the Reasonableness Requirement for Fourth Amendment Seizures, 53 Am. Crim. L. Rev. 349, 350 (2016) (citing Illinois v. Gates, 462 U.S. 213, 230 (1983); Spinelli v. United States, 393 U.S. 410, 415–16 (1969); Aguilar v. Texas, 378 U.S. 108, 114 (1964)).
  • 78Navarette, 134 S. Ct. at 1686.
  • 79Fagan & Geller, supra note 72.
  • 80See People v. De Bour, 352 N.E.2d 562, 571–72 (N.Y. 1976). In contrast to the two-stage inquiry developed in Terry, De Bour articulates four levels of suspicion correlated with four levels of justified intrusion. See also People v. Hollman, 590 N.E.2d 204, 205 (N.Y. 1992). Most states follow Terry. For example, Massachusetts follows constitutional Terry standards. See Commonwealth v. Torres, 433 Mass. 669, 745 N.E.2d 945, 948 (2001).
  • 81Fagan Report, supra note 54.
  • 82See Harcourt & Meares, supra note 63.
  • 83See Meares, supra note 44, at 172–76.
  • 84See Corey Fleming Hirokawa, Making the “Law of the Land” the Law on the Street: How Police Academies Teach Evolving Fourth Amendment Law, 49 Emory L.J. 295, 319–31 (2000).
  • 85Terry v. Ohio, 392 U.S. 1, 33 (1968) (“We merely hold today that where a police officer observes unusual conduct which leads him reasonably to conclude in light of his experience that criminal activity may be afoot . . . .”).
  • 86The Terry Court states specifically that, “[t]he sole justification of the search in the present situation is the protection of the police officer and others nearby.” Id. at 29. (emphasis added).
  • 87Id. at 26. But see Steiker, supra note 35 (arguing the officer safety standard was extended four years after Terry in Adams v. Williams from the frisk to the stop, via a textual mashup of the facts of the Adams stop with the text of the Terry opinion).
  • 88President’s Comm’n on Law Enforcement and Criminal Justice, Challenge of Crime in a Free Society 94–95 (1967), https: // (
  • 89Terry, 392 U.S. at 14 n.11.
  • 90In the years between Mapp and Terry, total officer deaths (not just those resulting from felony crimes by suspects) increased from 140 in 1961 to 191 in 1968. In the decade after Terry, as violent crime overall increased, officer deaths increased from 194 in 1969 to 215 in 1978, with a peak of 280 in 1974. See Officer Deaths by Year, Nat’l Law Enforcement Officers Memorial Fund, /facts/officer-fatalities-data/year.html ( (including all officer deaths, not just officers killed by criminal suspects). Of the 1466 officers killed from 2005–14, 482 (32.9%) were killed in auto or motorcycle crashes, and 554 (37.6%) were killed by criminal suspects. The remainder died from a variety of natural or accidental causes. Causes of Law Enforcement Deaths, Nat’l Law Enforcement Officers Memorial Fund, /facts/officer-fatalities-data/causes.html ( One writer calculated that felonious deaths of police officers averaged 62.8 per year from 1960–65, and rose to 76.6 in the four years surrounding the Terry opinion, again commensurate with an overall increase in violent crime. See Dan Wang, Data on Police Officers Killed Since 1961, Dan Wang (June 16, 2015), ( (citing evidence from the FBI’s Uniform Crime Reports dating to 1961).
  • 91Terry, 392 U.S. at 34 (Harlan, J., concurring) (emphasis added).
  • 92Id. at 31. Justice Rehnquist continued this conflation of stop-and-frisk rationales in Adams v. Williams, 407 U.S. 143 (1972). See also Steiker, supra note 35.
  • 93Terry, 392 U.S. at 34 (White, J., concurring).
  • 94Id. at 34–35 (emphasis added).
  • 95The gravamen of his dissent is that the Constitution requires probable cause. See id. at 35 (Douglas, J., dissenting).
  • 96Id. at 38.
  • 97But Justice Douglas does not further discuss the issue, saying only that the Terry rule would be justified only by constitutional amendment, not by judicial decision. See id. at 39.
  • 98Most states follow Terry. For example, Massachusetts follows constitutional Terry standards. See Commonwealth v. Torres, 745 N.E.2d 945, 948 (Mass. 2001); see also Commonwealth v. Martin, 4 N.E.3d 1236, 1247 (Mass. 2014) (“[W]e ask whether the stop was based on an officer's reasonable suspicion that the person was committing, had committed, or was about to commit a crime.”); Commonwealth v. Scott, 801 N.E.2d 233, 237 (Mass. 2004) (“[That] suspicion must be grounded in ‘“specific, articulable facts and reasonable inferences [drawn] therefrom” rather than on a “hunch.”’”) (quoting Commonwealth v. Lyons, 564 N.E.2d 390, 392 (Mass. 1990) quoting Commonwealth v. Wren, 464 N.E.2d 344, 345 (Mass. 1984).
  • 99352 N.E.2d 562 (N.Y. 1976).
  • 100Id. at 570. De Bour and two police officers were walking toward each other just after midnight; when the officers and De Bour were within “30 or 40 feet of the uniformed officers,” De Bour crossed the street. Id. at 565. “The two policemen followed suit” and, upon meeting De Bour, one officer asked him what he was doing. “De Bour, clearly but nervously, answered that he had just parked his car and was going to a friend’s house.” The officer asked De Bour for identification; De Bour said he had none. One officer then “noticed a slight waist-high bulge in [De Bour’s] jacket [and] . . . asked De Bour to unzipper his coat. When De Bour complied . . . [the officer] observed a revolver protruding from [De Bour’s] waistband.” At a suppression hearing, De Bour testified that he had been patted down for “two or three minutes” before the gun was found; however, the trial court credited the officer’s testimony and the weapon was admitted into evidence. Id.
  • 101Id. at 570 (emphasis added). The court, in describing the encounter, held that “the encounter did not subject De Bour to a loss of dignity.”
  • 102Terry, 392 U.S. at 20–21 (citing Camara v. Municipal Court, 387 U.S. 523, 534–35, 536–37 (1967). In Brown v. Texas, the Court set out a three-pronged test to determine the constitutionality of Terry stops: “consideration of the constitutionality of such seizures involves a weighing of the gravity of the public concerns served by the seizure, the degree to which the seizure advances the public interest, and the severity of the interference with individual liberty.” 443 U.S. 47, 50–51 (1979). This three-part test is hardly ever accomplished.
  • 103422 U.S. 873 (1975).
  • 104Id. at 879; see also id. at 878 (“As with other categories of police action subject to Fourth Amendment constraints, the reasonableness of such seizures depends on a balance between the public interest and the individual’s rights to personal security free from arbitrary interference by law officers.”).
  • 105428 U.S. 543 (1976).
  • 106Id. at 567. The Court balanced the intrusion created by the checkpoint, evaluated to be considerably less than roving patrols, with the public interest in preventing illegal immigration. The Court also noted that requiring all checkpoint stops to be based on reasonable suspicion would be “impractical because the flow of traffic tends to be too heavy to allow the particularized study of a given car that would enable it to be identified as a possible carrier of illegal aliens.” Id. at 557.
  • 107It may be more accurate to describe these border stops as investigative or detection-based: they seek to discover illegal aliens who have already crossed the border and the smugglers who aid them. The crimes of unauthorized entry and aiding unauthorized entry to the United States have thus already been committed.
  • 108407 U.S. 143 (1972).
  • 109452 U.S. 692 (1981).
  • 110Id. at 705.
  • 111Id. at 707.
  • 112Id. at 700–02 (emphasis added). “In assessing the justification for the detention of an occupant of premises being searched for contraband pursuant to a valid warrant, both the law enforcement interest and the nature of the ‘articulable facts’ supporting the detention are relevant.” Id. at 702–03 (emphasis added) (citation omitted).
  • 113463 U.S. 1032 (1983).
  • 114Id. at 1050.
  • 115508 U.S. 366 (1993).
  • 116Id. at 378–79. The seizure in that case was unlawful because, although the search was authorized, the officer squeezed and manipulated the contraband after concluding that it was not a weapon. It was therefore unrelated to the search for weapons and “amounted to the sort of evidentiary search that Terry expressly refused to authorize.”
  • 117494 U.S. 325 (1990).
  • 118470 U.S. 811 (1985).
  • 119Id. at 817. The Court placed limits on the procedure, ruling that fingerprinting must be “carried out with dispatch” and there must be “a reasonable basis for believing that fingerprinting will establish or negate the suspect’s connection with that crime.” Id.
  • 120517 U.S. 806 (1996).
  • 121For a discussion of the racial implications of Whren, see Kevin R. Johnson, How Racial Profiling in America Became the Law of the Land, 98 GEO. L.J. 1005, 1073 (2010). See also Kevin R. Johnson, The Story of Whren v. United States: The Song Remains the Same, in Race Law Stories (Devon Carbado & Rachel F. Moran eds., 2006).
  • 122Whren, 517 U.S. at 814.
  • 123When the petitioners in Whren introduced evidence showing that a stop for a civil traffic violation—an expression of those non-criminal governmental interests that were foundational to the evolution of Terry doctrine—was in fact a pretextual, race-based stop to look for drugs, the Court thought those concerns to be beyond the scope of the Fourth Amendment. Id. at 813. The Whren Court failed to distinguish what was outside the scope of the Fourth Amendment—the pretextual stop or the racial imbalance—in its application at the time.
  • 124528 U.S. 119 (2000).
  • 125Id. at 124.
  • 126Cf. Ferguson & Bernache, supra note 17 (arguing that courts should use objective, quantifiable measures to determine what qualifies as a “high-crime area”).
  • 127208 F.3d 1122 (9th Cir. 2000).
  • 128Id. at 1143 (Kosinzki, J., dissenting). Judge Kosinzki further explains:

    Does an arrest every four months or so make for a high crime area? . . . [T]o rely on every cop’s repertoire of war stories to determine what is a ‘high crime area’—and on that basis to treat otherwise innocuous behavior as grounds for reasonable suspicion—strikes me as an invitation to trouble. If the testimony of two officers that they made, at most, 32 arrests during the course of a decade is sufficient to turn the road here into a high crime area, then what area under police surveillance wouldn’t qualify as one? . . . I would be most reluctant to give police the power to turn any area into a high crime area based on their unadorned personal experiences.

    Id. (emphasis added).

  • 129See generally Barry Friedman, Unwarranted: Fixing Policing in America (forthcoming 2017) (on file with author).
  • 130Terry v. Ohio, 392 U.S. 1, 22 (1968).
  • 131469 U.S. 221 (1985).
  • 132Id. at 229. The Hensley Court found law enforcement interests outweighed the individual’s Fourth Amendment protection, even for past crimes:

    [W]here police have been unable to locate a person suspected of involvement in a past crime, the ability to briefly stop that person, ask questions, or check identification in the absence of probable cause promotes the strong government interest in solving crimes and bringing offenders to justice. Restraining police action until after probable cause is obtained would not only hinder the investigation, but might also enable the suspect to flee in the interim and to remain at large. Particularly in the context of felonies or crimes involving a threat to public safety, it is in the public interest that the crime be solved and the suspect detained as promptly as possible.


  • 133460 U.S. 491 (1983).
  • 134Id. at 499.
  • 135See Floyd v. City of New York, 959 F. Supp. 2d 540, 600–01 (S.D.N.Y. 2013); see also Meares, supra note 44. Indeed, the use of Terry stops or investigative stops as a crime control program had a long history that predated Terry. Stop-and-frisk tactics were used programmatically in the 1960s as crime rates increased, and were targeted in poor minority neighborhoods where they became important contributors to urban unrest. See, Kerner Comm’n, supra note 22; see also Robert M. Fogelson, From Resentment to Confrontation: The Police, the Negroes, and the Outbreak of the Nineteen-Sixties Riots, 83 Pol. Sci. Q. 217, 217–18 (1968) (linking the urban riots in the 1960s to a string of “extraordinary and unjustified police actions”).
  • 136501 U.S. 429 (1991).
  • 137Id. at 441 (Marshall, J., dissenting). The broad range of modern searches under Terry’s crime control rationale, including programs like stop and frisk in New York, Los Angeles, Chicago, and Philadelphia, and a wide range of administrative searches including checkpoints and school sweeps, all express the crime control and deterrence rationales that were incubated in Terry and developed over the next five decades. These developments all were done without democratic regulation or political processes. See, e.g., David Sklansky, Democracy and the Police (2008) (arguing that the democratization of the police, not regulation by courts, will produce more effective and accountable policing); Rachel Harmon, The Problem of Policing, 110 Mich. L. Rev. 761, 777 (2012) (emphasizing that constitutional law cannot fully address problems of police regulation); Carl McGowan, Rule-making and the Police, 70 Mich. L. Rev. 659 (1974) (arguing that police departments, not courts, should formulate rules to govern policing); Christopher Slobogin, Panvasive Surveillance, Political Process Theory, and the Non-delegation Doctrine, 102 Geo. L.J. 1721 (2014) (arguing that ordinary democratic processes of legislative authorization and police rulemaking should govern panvasive searches and seizures).
  • 138136 S. Ct. 2056 (2016)
  • 139Id. at 2060.
  • 140Id. at 2059.
  • 141Id.
  • 142State v. Strieff, 357 P.3d 532, 536 (Utah 2015), rev’d, 136 S. Ct. 2056 (2016).
  • 143136 S. Ct. at 2061.
  • 144357 P. 3d at 544.
  • 145Terry v. Ohio, 392 U.S. 1, 5 (1968)
  • 146U.S. Dep’t. of Just., Civil Rights Div., Investigation of the Ferguson Police Department 47, 55 (2015), https: // /files/opa/press-releases/attachments/2015/03/04/ferguson_police_department_report.pdf (
  • 147Brief of Dr. Ian Ayres et al. in Support of Petitioner, Petition for Certiorari to the United States Court of Appeals for the Eighth Circuit at 6–8, United States v. Faulkner, 636 F.3d 1009 (8th Cir. 2011) (No. 11-235), 2011 WL 4479100, at *6–8.
  • 148Utah v. Strieff, 136 S. Ct. 2056, 2066 (2016) (Sotomayor, J., dissenting) (citing Brown v. Illinois, 422 U.S. 590, 603–05 (1975)).
  • 149Id. at 2068–69.
  • 150Id.
  • 151Id. at 2069–71.
  • 152Id.; see also I. Bennett Capers, Policing, Race, and Place, 44 Harv. C.R.-C.L. L. Rev. 43, 44 (2009).
  • 153See Lawrence Rosenthal, Pragmatism, Originalism, Race and the Case against Terry v. Ohio, 43 Tex. Tech L. Rev. 299, 329–30 (2010). However, the intrusions may be anything but “relatively limited.”
  • 154See Josh Bowers, Probable Cause, Constitutional Reasonableness and the Unrecognized Point of a “Pointless Indignity, 66 Stan. L. Rev. 987, 1013–14 (2014); see generally Phillip B. Heymann, The New Policing, 28 Fordham Urb. L.J. 407 (2000).
  • 155See Stuntz, supra note 42.
  • 156Floyd v. City of New York, 959 F. Supp. 2d 540 (S.D.N.Y. 2013). As part of discovery in the litigation, data were provided to plaintiffs on stops, crimes, and arrests for each year in the study period.
  • 157Census block groups are small statistical and spatial divisions of census tracts. They range in size from 600 to 3000 persons. Block groups cover contiguous areas, and are always located within a census tract (never crossing tract boundaries). The boundaries are defined by streets or highways, railroads, streams, and other bodies of water, and/or other visible physical and cultural features. See U.S. Census Bureau, Geographic Terms and Concepts-Block Groups, https: // /gtc_bg.html (
  • 158See Floyd, 959 F. Supp. 2d at 580–81; see also Fagan Report, supra note 54; Fagan Supplemental Report, supra note 54.
  • 159See Fagan Report, supra note 54, at 22. The checkboxes were incorporated into the standard reporting form for stops, the UF-250. They were a set of indicia of suspicion derived from the aggregate experiences of officers who had been conducting stops over many years. See id. at 48–49.
  • 160Id. at 22.
  • 161See id. at Appendix F; see also Davis v. City of New York, 902 F. Supp. 2d 405 (S.D.N.Y. 2012) (No. 10 Civ. 0699) (hereinafter Fagan Davis Report); Fagan & Geller, supra note 72.
  • 162See, e.g., Terry v. Ohio, 392 U.S. 1, 28 (1968) (upholding stop and frisk when officer suspected three men of casing a store in preparation for a daytime robbery); United States v. Padilla, 548 F.3d 179, 187–88 (2d Cir. 2008) (holding that a detective’s observation of two men quietly following another individual into a secluded area in the dark and out of the individual’s peripheral vision “supported the detective’s suspicion that the two men might have been targeting the disheveled man for a robbery” and justified a stop and frisk); People v. Richard, 668 N.Y.S.2d 386, 387 (N.Y. App. Div. 1998) (“Reasonable suspicion supporting the forcible detention of defendant was supplied by lengthy police observations of defendant’s complex, unusual, and suspicious pattern of ‘casing’-type behavior, strongly suggestive of a known series of armed robberies in the neighborhood that targeted movie theaters in particular, coupled with the fact that defendant met a general description of one of the robbers.”).
  • 163See Stuntz, supra note 42. In Brown v. Texas, the U.S. Supreme Court struck down Texas’s stop-and-identify law as violating the Fourth Amendment because it allowed police officers to stop individuals without “reasonable suspicion, based on objective facts, that the individual is involved in criminal activity.” 443 U.S. 47, 51 (1979); see also Fagan & Geller, supra note 72.
  • 164See Floyd v. City of New York, 959 F. Supp. 2d 540, 556 (S.D.N.Y. 2013).
  • 165See William H. Simon, Rethinking Privacy, Boston Rev. (Oct. 20, 2014), /books-ideas/william-simon-rethinking-privacy-surveillance (
  • 166Dorothy E. Roberts, Race, Vagueness, and the Social Meaning of Order-Maintenance Policing, 89 J. Crim. L. & Criminology 775, 802 (1999); see also Fagan & Geller, supra note 72.
  • 167See, e.g., People v. Sierra, 683 N.E.2d 955, 956 (N.Y. 1994) (holding that officers had reasonable suspicion to stop a suspect calling “over here, over here” in an area known for drug trafficking who subsequently fled upon noticing the officers). But see People v. Thompson, No. 2002-1635, 2004 Lexis 873 (N.Y. App. Div. June 9, 2004) (holding that an officer did not have the authority to request that a suspect reveal what was in his hand because the suspect engaged in “some sort of exchange” in a drug prone location).
  • 168See, e.g., People v. Jones, 90 N.Y.2d 835, 837 (N.Y. 1997) (holding that a transaction that (a) involved the exchange of currency, (b) took place in a drug-prone location, and (c) was observed by an experienced officer who was trained in the investigation and detection of narcotics, supported a finding of probable cause).
  • 169Utah v. Strieff, 136 S. Ct. 2056, 2060 (2016) (disputing the reasonableness of Officer Fackrell’s seizure of Edward Strieff based on his presence at a location where Fackrell suspected that drugs were being sold).
  • 170Despite the purported link between guns and drugs that the Court assumed in this case, the fact that a suspect might have participated in a drug transaction does not instantly ensure that an officer has reasonable suspicion to believe a suspect has committed a crime and a reasonable suspicion that the suspect is armed or dangerous. But compare People v. Perolta-Rua, 579 N.Y.S.2d 283, 285 (N.Y. App. Div. 1992) (finding that experienced officer’s knowledge “that drug dealers often carry weapons” was one factor supporting stop and frisk), with United States v. Gonzalez, 362 F. Supp. 415, 424 (S.D.N.Y. 1973) (deciding, under New York law, that stop and frisk was improper because “[n]one of the agents who testified expressed any concern that Torres might be armed and dangerous, and it is evident, even from their own testimony, that they grabbed his paper bag because they hoped to find narcotics, not a weapon”).
  • 171See Ohio v. Terry, 214 N.E.2d 114, 116 (Ohio Ct. App. 1966), aff ’d, 392 U.S. 1 (1968). In the original stop of John Terry, Officer Martin McFadden observed Terry and Richard Chilton, pacing back and forth in front of a jewelry store in Cleveland’s commercial district for twelve minutes before closing time. When Terry and Chilton were joined by a third man, Officer McFadden began a field interrogation. Terry’s mumbled responses further raised McFadden’s suspicion and led to a pat down and then search of Terry’s clothing that produced a loaded automatic gun. The Terry Court celebrated Officer McFadden’s experience as sharpening his ability to distinguish innocuous behavior from his decision that crime was “afoot.” Terry v. Ohio, 392 U.S. 1, 31 (1968).
  • 172See People v. Mack, 258 N.E.2d 703, 707 (N.Y. 1970) (“Where . . . the officer confronts an individual whom he reasonably suspects has committed, is committing or is about to commit such a serious and violent crime as robbery or, as in the instant case, burglary, then it is our opinion that that suspicion not only justifies the detention but also the frisk, thus making it unnecessary to particularize an independent source for the belief of danger.”); see also People v. Schollin, 682 N.Y.S.2d 48, 49 (N.Y. App. Div. 1998) (upholding pat down of suspect when officer believed that victim had been shot in face); People v. Paul, 658 N.Y.S.2d 275, 276 (N.Y. App. Div. 1997) (upholding stop and frisk where officers heard numerous gunshots and saw two persons running from location where shots were fired).
  • 173See People v. Mitchell, 601 N.Y.S.2d 100, 100 (N.Y. App. Div. 1993) (upholding search of suspect’s suitcase when woman told officers that suspect had verbally threatened to shoot her with shotgun).
  • 174See People v. Howard, 542 N.Y.S.2d 536, 538 (N.Y. App. Div. 1989) (finding that police had no reasonable suspicion to frisk suspect who repeatedly looked up and down the street and down subway stairs at 10:00 P.M. in high-crime area and who had reached into his jacket several times); see also People v. Alvarez, 778 N.Y.S.2d 27, 27–28 (N.Y. App. Div. 2004) (finding that pat down was proper where, inter alia, police responded to radio call based on anonymous tip, heard suspicious noises coming from apartment, and witnessed suspect attempting to flee scene by climbing a fence. Given the vagueness of this standard, it is impossible to say whether a “sight” or “sound,” standing alone, would justify a stop and frisk.).
  • 175Fagan Report, supra note 54, at 40–53.
  • 176See Opinion and Order, Floyd v. City of New York, 959 F. Supp. 2d 540 (S.D.N.Y. 2013) (No. 08 Civ. 1034), at 579–80.
  • 177Fagan & Geller, supra note 72. “Other” stop factors were checked off at frequencies that varied by type of suspected crime. The content of the text strings that accompanied this factor was analyzed as part of expert reports in litigation. See Fagan Supplemental Report, supra note 54. The text strings for the “other” factor were a diverse set of observations that were at times specific (e.g., “smell of marijuana smoke”) and at times bizarrely vague (e.g., “looks like a perp”). See Fagan Davis Report, supra note 161. There was no discernable pattern that would sustain meaningful disaggregation or classification. When applied to judgments about the constitutional sufficiency of a stop event, they were far more likely to lead to a conclusion of “indeterminate” or “insufficient.”
  • 178See, e.g., Robert J. Sampson, When Things Aren’t What They Seem: Context and Cognition in Appearance-Based Regulation, 125 Harv. L. Rev. F. 97, 99–102 (2012); see also Geoffrey P. Alpert et al., Police Suspicion and Discretionary Decision Making during Citizen Stops, 43 Criminology 407, 422–23 (2005) (showing that whether a suspect is black influences an officer’s decision to form suspicion based on non-behavioral versus behavioral cues); Adam M. Samaha, Regulation for the Sake of Appearance, 125 Harv. L. Rev. 1563, 1620–34 (2012) (describing the New York stop-and-frisk regime as “appearance-based” regulation based on perceptions of disorderly places or people); Robert J. Sampson & Stephen W. Raudenbush, Seeing Disorder: Neighborhood Stigma and the Social Construction of “Broken Windows”, 67 Soc. Psychol. Q. 319, 330–34 (2004) (showing empirically that perception of disorder in neighborhoods is correlated not only with observation of disorder but also with the racial composition of the neighborhood).
  • 179Richard A. Berk & John M. MacDonald, Overdispersion and Poisson Regression, 24 J. Quantitative Criminology 269, 271–72 (2008).
  • 180John M. MacDonald & Pamela K. Lattimore, Count Models in Criminology, in Handbook of Quantitative Criminology (Alex R. Piquero & David Weisburd eds., 2010).
  • 181See U.S. Census Bureau, American Community Survey Data Releases (2009) https: // (
  • 182See Fagan Report, supra note 54, at Tables 7 and 8 for examples of the sensitivity of estimates of Terry stop patterns when controlling for these non-residential local conditions.
  • 183See, e.g., Christopher Wildemann, Paternal Incarceration and Children’s Physically Aggressive Behaviors: Evidence from the Fragile Families and Child Wellbeing Study, 89 Soc. Forces 285, 293 (2010).
  • 184See Orley Ashenfelter & David Card, Using the Longitudinal Structure of Earnings to Estimate the Effect of Training Programs, 67 Rev. Econ. & Stat. 648 (1985).
  • 185Census 2010, Social Explorer, (
  • 186The standard deviation for the land area is .738 square miles with a median size of .025 square miles, indicating a wide range of sizes and densities of census block groups. Many of the larger block groups are commercial or industrial areas with very low populations.
  • 187See Lauren J. Krivo & Ruth D. Peterson, Extremely Disadvantaged Neighborhoods and Urban Crime, 75 Soc. Forces 619, 634 (1996); see also Lin Cui & Randall Walsh, Foreclosure, Vacancy and Crime, 87 J. Urb. Econ. 72, 75 (2015).
  • 188See Opinion and Order, supra note 176 (ruling that plaintiffs satisfied the standards for class certification under Rule 23(a) of the Federal Rules of Civil Procedure based on numerosity, commonality, typicality, and adequacy) (citing Teamsters Local 445 Freight Div. Pension Fund v. Bombardier Inc., 546 F.3d 196, 201–02 (2d Cir. 2008)).
  • 189See Terry v. Ohio, 392 U.S. 1, 6 (1968).
  • 190As in money changing hands.
  • 191See, e.g., Fagan & Geller, supra note 72; Samaha, supra note 178; Sampson & Raudenbush, supra note 178.
  • 192Floyd v. City of New York, 959 F. Supp. 2d 540, 8, 11, 41, 43–45, n.760 (S.D.N.Y. 2013) (reporting testimony from officers explaining that “furtive movement is a very broad concept,” and could include a person “changing direction,” “walking in a certain way,” “[a]cting a little suspicious,” “making a movement that is not regular,” being “very fidgety,” “going in and out of his pocket,” “going in and out of a location,” “looking back and forth constantly,” “looking over their shoulder,” “adjusting their hip or their belt,” “moving in and out of a car too quickly,” “[t]urning a part of their body away from you,” “[g]rabbing at a certain pocket or something at their waist,” “getting a little nervous, maybe shaking” and “stutter[ing],” “hanging out in front of [a] building, sitting on the benches or something like that” and then making a “quick movement,” such as “bending down and quickly standing back up,” “going inside the lobby . . . and then quickly coming back out,” or “all of a sudden becom[ing] very nervous, very aware”).
  • 193See, e.g., Joseph Hilbe, Negative Binomial Regression 111–12 (2011); Berk & MacDonald, supra note 179; Sander Greenland, Dose-Response and Trend Analysis in Epidemiology: Alternatives to Categorical Analysis, 6 Epidemiology 356, 359 (1996).
  • 194See New York City Police Dep’t, Historical New York City Crime Data, /historical_nyc_crime_data.shtml (
  • 195Richard Williams, Using the Margins Command to Estimate and Interpret Adjusted Predictions and Marginal Effects, 12 Stata J. 308, 323–24 (2012); see also Michael J. Hanmer & Kerem O. Kalkan, Behind the Curve: Clarifying the Best Approach to Calculating Predicted Probabilities and Marginal Effects From Limited Dependent Variable Models, 57 Am. J. Pol. Sci. 263, 275 (2013).
  • 196See Lawrence W. Sherman, Police Crackdowns: Initial and Residual Deterrence, 12 Crime & Just. 1, 11–12 (1990) (showing that the majority of empirical studies on police “crackdowns” or concentrated patrol strategies report initial but not long-term deterrent effects on crime after two months, despite continued patrol pressures and allocations of officers over a longer interval).
  • 197See generally Robert Apel, On the Deterrent Effects of Stop, Question, and Frisk, 15 Criminology & Pub. Pol’y 27 (2016).
  • 198See Daniel Nagin, Deterrence in the Twenty-First Century, 42 Crime & Just. 199, 201 (2013) (reviewing empirical and theoretical scholarship on deterrence over the past three decades); see also Robert Apel & Daniel S. Nagin, General Deterrence: A Review of the Evidence, in Crime and Public Policy (James Q. Wilson & Joan Petersilia eds., 4th ed., Oxford Univ. Press 2011); Paul H. Robinson & John M. Darley, Does Criminal Law Deter? A Behavioral Science Investigation, 24 O.J.L.S. 173 (2004) (articulating the behavioral mechanisms of deterrence).
  • 199See, e.g., John MacDonald et al., The Effects of Local Police Surges on Crime and Arrests in New York City, 6 PLoS ONE, e0157223 11 (2016), ( /warc /W7KN-GEVN).
  • 200Terry v. Ohio, 392 U.S. 1, 27 (1968) (“The officer must be able to articulate more than an ‘inchoate and unparticularized suspicion or hunch’ of criminal activity.” The officer’s intuitions, gut feelings, and “sixth sense” about a situation are disallowed.); see also Craig S. Lerner, Reasonable Suspicion and Mere Hunches, 59 Vand. L. Rev. 407, 419 (2006).
  • 201See, e.g., Navarette v. California, 134 S. Ct. 1683, 1695 (2014) (Scalia, J., dissenting) (discussing the vague empirical standards to determine whether the criteria of reasonable suspicion in vehicle or pedestrian stops as applied is “reasonable”); see also City of Indianapolis v. Edmund, 531 U.S. 32, 44 (2000) (stating that a “hit rate” of approximately nine percent at a random checkpoint does not justify a general purpose of stopping vehicles in a search for criminal activity).
  • 202See, e.g., Ornelas v. United States, 517 U.S. 690, 696 (1996) (“The principal components [of reasonable suspicion] . . .” are “the events which occurred leading up to the stop or search” and “the decision whether these historical facts, viewed from the standpoint of an objectively reasonable police officer, amount to reasonable suspicion[.]”).
  • 203For example, Massachusetts follows constitutional standards in Terry v. Ohio. See Commonwealth v. Torres, 745 N.E.2d 945, 948 (Mass. 2001); see also supra note 98.
  • 204Tilem & Associates, New York Search and Seizure Law: Street Encounters with Police in New York – The Four Levels of Intrusion, N.Y. Crim. Att’y Blog, ( People v. De Bour specifies four distinct levels of intrusion by a police officer, each linked to a permissible response. 40 N.Y.2d 210, 213 (N.Y. 1976). (1) Approaching a suspect to request information through non-threatening questions requires an objective or credible reason to interact with the suspect. See People v. Hollman & Saunders, 79 N.Y.2d 181, 184 (N.Y. 1992). (2) Asking pointed questions that might lead an officer to believe that the person is suspected of a crime is based on the common law right of inquiry, and requires founded suspicion. See id. at 185. (3) A stop, or temporary but forcible detention, requires reasonable suspicion. Reasonable suspicion can also justify a frisk, if the officer fears that the suspect has a weapon.
    See N.Y. Crim. Proc. Law § 140.50 (McKinney 2010); see also, Terry v. Ohio, 392 U.S. 1 (1968). (4) Arrest and full search of a suspect requires probable cause. See United States v. Watson, 423 U.S. 411, 432 (1976); Chimel v. California, 395 U.S. 752 (1969).
  • 205Terry’s balancing test was cast in 1968 as weighing the level of intrusion on the citizen versus the officer safety concerns. Terry v. Ohio, 392 U.S. 1, 29 (1968); see Rosenthal, supra note 153. But over time, the balancing test devolved into crime control payoffs—whether the intrusions were necessary to address public safety concerns. See, e.g., Heymann, supra note 154; Meares, supra note 44.
  • 206See Lewis R. Katz, Terry v. Ohio at Thirty-Five: A Revisionist View, 74 Miss. L.J. 423, 427 (2004) (claiming that Terry failed to achieve its stated purpose of tying the practice [or investigative stops] to the Fourth Amendment reasonableness standard); see also Harcourt & Meares, supra note 63 (claiming that intractable problems in regulating police discretion in Terry stops suggest that discretionary stops based on reasonable suspicion should be replaced by randomized stops that are tailored in probability to local crime rates).
  • 207See, e.g., Heymann, supra note 154; Meares, supra note 44.
  • 208See Report of the Nat’l Advisory Comm’n on Civil Disorders (1967); see also William Stuntz, The Collapse of American Criminal Justice (2011).
  • 209See Terry v. Ohio, 392 U.S. 1, 13 (1968) (“Thus, in our system, evidentiary rulings provide the context in which the judicial process of inclusion and exclusion approves some conduct as comporting with constitutional guarantees and disapproves other actions by state agents.”).
  • 210See, e.g., United States v. Martinez-Fuerte, 428 U.S. 543 (1976); United States v. Brigoni-Ponce, 422 U.S. 873 (1975); People v. Holman, 590 N.E.2d 204 (N.Y. 1992); People v. De Bour, 352 N.E.2d 562 (N.Y. 1976).
  • 211Barry Friedman & Maria Ponomarenko, Democratic Policing, 90 N.Y.U. L. Rev. 1827, 1899–1900 (2015).
  • 212A quick glance at the docket of the Special Litigation Section of the Civil Rights Division in the U.S. Department of Justice shows a roster of active consent decrees in twelve cities, nearly half of which can be traced either to a pattern of Fourth Amendment violations in stops, or unwarranted racial disparities in the practice of Terry stops. See Civil Rights Div., Special Litigation Section Cases and Matters, U.S. Dep’t of Justice, /special-litigation-section-cases-and-matters0#police (]; see also Rachel A. Harmon, Promoting Civil Rights Through Proactive Policing Reform, 62 Stan. L. Rev. 1, 15 n.50 (2009).
  • 213See Rebecca Kaplan, NYPD Chief: Policing Involving Use of Force, It Always Looks Awful, CBS News (Dec. 7, 2014), (
  • 214See, e.g., City of Chicago v. Morales, 523 U.S. 41, 64 (1998) (showing the difficulty of designing a constitutionally valid basis for exerting police authority in the context of activities that might or might not be markers of criminal activity).
  • 215See, e.g., Harmon, supra note 137; Sonja B. Starr, Explaining Race Gaps in Policing: Normative and Empirical Challenges (Univ. of Mich. Law & Econ. Research Paper No. 15-003, 2015).
  • 216See Lerner, supra note 200; see also Anna Lvovsky, The Judicial Presumption of Police Expertise, 130 Harv. L. Rev.  (forthcoming 2017) (analyzing and critiquing the widespread judicial deference to police testimony and expertise in resolving Fourth Amendment claims).
  • 217See Stuntz, supra note 42; see also Meares, supra note 44.
  • 218Fagan & Geller, supra note 72; see also Floyd v. City of New York, 959 F. Supp. 2d 540, 578 (S.D.N.Y. 2013) (criticizing plaintiffs’ expert for being too conservative in estimating the extent of constitutionally flawed bases of suspicion for stops).
  • 219See supra note 192 and accompanying text.
  • 220See Charles F. Sabel & William H. Simon, Due Process of Administration: The Problem of Police Accountability (Columbia Pub. Law Research Paper 14-420, 2014); see also Eliot H. Schatmeier, Reforming Police Use-of-Force Practices: A Case Study of the Cincinnati Police Department, 46 Colum. J. L. & Soc. Probs. 539, 557–59 (2012).
  • 221See Pl’s Fifth Report to Court and Monitor on Stop and Frisk Practices, Bailey v. City of Philadelphia, C.A. 10-5952 (E.D. Pa. 2015).
  • 222See Meares, supra note 44; see, e.g., Wardlow v. Illinois, 528 U.S. 119 (2000) (suggesting that ambiguous behaviors in a high crime area could heighten suspicion and justify a stop). But see United States v. Montero-Camargo, 208 F.3d 1122, 1141–44 (9th Cir. 2000) (Kozinski, J., dissenting) (criticizing the overbreadth of “high crime area” as a component of reasonable suspicion).
  • 223See Harcourt & Meares, supra note 63.
  • 224Terry v. Ohio, 392 U.S. 1, 16–17 (1968) (“Moreover, it is simply fantastic to urge that such a procedure performed in public by a policeman while the citizen stands helpless, perhaps facing a wall with his hands raised, is a ‘petty indignity.’ It is a serious intrusion upon the sanctity of the person, which may inflict great indignity and arouse strong resentment, and it is not to be undertaken lightly.”); see also Akil Amar, Terry and Fourth Amendment First Principles, 72 St. John’s L. R. 1100, 1101 (1998); Steiker, supra note 35.
  • 225See Stuntz, supra note 42; see also, Josh Bowers, Probable Cause, Constitutional Reasonableness, and the Unrecognized Point of a Pointless Indignity, 66 Stan. L. Rev. 987, 989 (2014)
  • 226See Apel, supra note 197 (noting the potential chilling effect on innocents who may be reluctant to move freely in public spheres fearing unwarranted Terry stops by police).
  • 227See Stuntz, supra note 42 (citing two dimensions of Fourth Amendment regulation that ignore interaction content and demeanor).
  • 228See Sherry F. Colb, Innocence, Privacy, and Targeting in Fourth Amendment Jurisprudence, 96 Colum. L. Rev. 1456, 1501 (1996); see also Charles R. Epp et al., Pulled over: How Police Stops Define Race and Citizenship 134–51 (John M. Conley & Lynn Mather eds., 2014). The first two types of harm are often joined. An innocent person could reasonably ask why me? Why would a police officer use her discretion to single out me of all people absent some concrete evidence or signs that I was up to no good? Why would s/he have a “hunch” that I am a criminal? Targeting harm then, encompasses both an innocence harm plus the basic harm to autonomy of targeting.
  • 229Epp et al., supra note 228. To a similar extent, the same harms accrue to Latino young men, as well as Latino and black young adults. See Victor M. Rios, Punished: Policing the Lives of Black and Latino Boys (2011). Bennet Capers refers to this as a form of “public shaming” that is skewed toward black men. See Bennet Capers, Policing, Race and Place, 44 Harv. C.R.-C.L L. Rev. 43, 68 (2009) (stating a similar claims in terms of “public shaming”).
  • 230See Rod K. Brunson & Ronald Weitzer, Police Relations with Black and White Youths in Different Urban Neighborhoods, 44 Urb. Affairs Rev. 858, 869–73 (2009); see also Stuntz, supra note 42; Rios, supra note 229; Michael Powell, Police Polish Image, but Concerns Persist, N.Y. Times, Jan. 4, 2009, at B1 (reporting on interviews with young Latino and black males on physical violence and harassment by police in everyday stop-and-frisk encounters in New York City during the latter half of 2008).
  • 231See, e.g., Jonathan Jackson et al., Why Do People Comply with the Law? Legitimacy and the Influence of Legal Institutions, 52 Brit. J. Criminology 1051, 1063 (2012); Tom R. Tyler & Jeffrey Fagan, Legitimacy and Cooperation: Why Do People Help the Police Fight Crime in Their Communities, 6 Ohio St. J. Crim. L. 231, 236–37 (2008); Tom R. Tyler et al., Street Stops and Police Legitimacy: Teachable Moments in Young Urban Men’s Legal Socialization, 11 J. Empirical Legal Stud. 751, 758 (2014).
  • 232Tyler et al., supra note 231.
  • 233Id.; see also Tyler & Fagan, supra note 231.
  • 234See David Beetham, The Legitimation of Power 15–16 (1991) (defining legitimacy along three dimensions, including rules that are justified “by reference to beliefs shared by both dominant and subordinate”); see also Tom R. Tyler, Psychological Perspectives on Legitimacy and Legitimation, 57 Ann. Rev. Psychol. 375 (2006) (discussing ways in which legitimacy facilitates state exercise of power because individuals view authorities as morally or normatively appropriate).
  • 235See Fagan & Geller, supra note 72.
  • 236See Amanda Geller et al., Aggressive Policing and the Mental Health of Young Men, 104 Am. J. Pub. Health 2321, 2323–24 (2014) (showing elevated rates of anxiety, post-traumatic stress symptoms, and perceived stigma as a function of increasing number of stops and increasing intrusiveness of those stops).
  • 237See Floyd v. City of New York, 959 F. Supp. 2d 540, 575–76 (S.D.N.Y. 2013) (noting that the indicia of suspicion were different for non-white compared to white suspects, and that disparities in “hit rates” could reflect those differences—and weaknesses—in the suspicion that was used to justify stops of black and Latino citizens).
  • 238See Glenn Loury, Anatomy of Racial Inequality (2002).
  • 239See Michael C. Dawson, Behind the Mule: Race and Class in American Politics 77 (1994) (explaining the linked fate concept as a means of explaining the way that African Americans perceive what is in their individual self-interest.); see also Jeffrey Fagan & Tracey L. Meares, Punishment, Deterrence and Social Control: The Paradox of Punishment in Minority Communities, 6 Ohio St. J. Crim. L. 173, 179–80 (2008).
  • 240Epp et al., supra note 228.
  • 241See id. at 115–20.
  • 242Id.; see also Jon B. Gould & Stephen D. Mastrofski, Suspect Searches: Assessing Police Behavior Under the U.S. Constitution, 3 Criminology & Pub. Pol’y 315, 345 (2004) (showing that officers engaged in unconstitutional searches were also more likely than officers who engaged in constitutional searches to be friendly and courteous to suspects).
  • 243The Terry Court abhorred hunches, terming them “inarticulate.” 392 U.S. 1, 22 (1968) (holding that intrusions upon constitutionally guaranteed rights must be based on more than inarticulate hunches).
  • 244Fifth Pl’s Progress Report, Bailey v. City of Philadelphia, supra note 221; Opinion and Order, supra note 176 (linking the low seizure rates to Fourth Amendment violations in carrying out Terry stops).
  • 245Heymann, supra note 154; see, e.g., Debra Livingston, Police Discretion and the Quality of Life in Public Spaces: Courts, Communities and the New Policing, 97 Colum. L. R. 551, 558 (1997); Lawrence Rosenthal, Policing and Equal Protection, 21 Yale L. & Pol’y Rev. 53, 58–59 (2003).
  • 246See, e.g., Bernard Harcourt, Illusion of Order (2001); Jeffrey Fagan & Garth Davies, Street Stops and Broken Windows: Terry, Race and Disorder in New York City, 28 Fordham Urb. L. J. 457, 467 (2000); Sampson & Raudenbush, supra note 178; Sampson & Raudenbush, Systematic Social Observation of Public Spaces: A New Look at Disorder in Urban Neighborhoods, 105 Am. J. Soc. 603, 611 (1999).
  • 247Tyler & Fagan, Legitimacy and Cooperation, supra note 231.
  • 248See Louis Kaplow, Burden of Proof, 121 Yale L. J. 738, 748 (2011) (arguing that strong evidence is necessary to assign liability or culpability since the proof burden can affect the design accuracy of enforcement).
  • 249Anthony A. Braga, Brandon C. Welsh & Cory Schnell, Can Policing Disorder Reduce Crime? A Systematic Review and Meta-Analysis, 52 J. Res. Crime & Delinq. 567, 578 (2015) (reporting non-significant effects of aggressive order-maintenance policing strategies on crime reduction).
  • 250Sampson & Raudenbush, supra note 178.