Debate  |  Volume 162

The Constitutionality of Stop-and-Frisk in New York City

162 U. Pa. L. Rev. Online 117 (2013)

Last updated: Dec. 9, 2013

Debate - The Constitutionality of Stop-and-Frisk in New York City

Stop-and-frisk, a crime prevention tactic that allows a police officer to stop a person based on "reasonable suspicion” of criminal activity and frisk based on reasonable suspicion that the person is armed and dangerous, has been a contentious police practice since first approved by the Supreme Court in 1968. In Floyd v. City of New York, the U.S. District Court for the Southern District of New York ruled that New York City’s stop-and-frisk practices violate both the Fourth and Fourteenth Amendments. Professors David Rudovsky and Lawrence Rosenthal debate the constitutionality of stop-and-frisk in New York City in light of Floyd and Judge Shira A. Scheindlin’s controversial removal from the case. Professor Rudovsky argues that Floyd shows the important role of data and statistical analysis in assessing the constitutionality of stop-and-frisk procedures. He contends that empirical evidence regarding both the factors for and outcomes of stops and frisks in New York demonstrates that either the legal standard is too permissive or police-stop documentation is not truthful. In response, Professor Rosenthal argues that Judge Scheindlin erred in failing to consider evidence of stop-and-frisk’s efficacy—evidence indicating that the NYPD’s stops are based on reasonable suspicion, a standard considerably less demanding than “preponderance of the evidence.” Additionally, Rosenthal argues that Judge Scheindlin should have considered differential offending by race or other potentially nondiscriminatory explanations for the higher stop rates of minorities.

Continue to full article.

Opening Statement — David Rudovsky

Stop-and-Frisk: The Power of Data and the Decision in Floyd v. City of New York

David Rudovsky is a Senior Fellow at the University of Pennsylvania Law School and a founding partner of Kairys, Rudovsky, Messing & Feinberg LLP. Professor Rudovsky served as counsel in Bailey v. City of Philadelphia, C.A. No. 10-5952 (E.D. Pa. June 21, 2011), discussed infra.

In Terry v. Ohio, the Supreme Court ruled that a person could be seized by the police based on a reasonable suspicion that the suspect was involved in serious criminal conduct, and further, if there were also reasonable suspicion that the person stopped was “armed and dangerous,” the officer could frisk the suspect for the officer’s protection. See 392 U.S. 1, 27, 30 (1968). Over the past forty-five years, the power of police to stop and frisk has greatly expanded and now encompasses all suspected criminal activity, no matter how trivial, and under circumstances where the conduct observed may be fully consistent with innocence. See, e.g., United States v. Arvizu, 534 U.S. 266, 274-75 (2002). Given the extremely broad grounds for stop-and-frisk, the large numbers of persons subjected to these stops, the wide discretion of police departments to decide where to deploy officers and whom to stop and frisk, and the disproportionate number of minorities subjected to these investigative detentions, it is no surprise that the Terry doctrine continues to be highly controversial as a question of constitutional law and as an aspect of the broader issue of effective and fair policing.

In Floyd v. City of New York, Judge Shira A. Scheindlin ruled that New York City’s stop-and-frisk practices violate the Fourth and Fourteenth Amendments. No. 08 Civ. 1034(SAS), 2013 WL 4046209, at *70-75 (S.D.N.Y. Aug. 12, 2013). The court reached these conclusions on a full evidentiary record that included a statistical analysis of over 4.4 million stops in New York City from January 2004 to June 2012. Id. at *1. The U.S. Court of Appeals for the Second Circuit stayed the judgment and remedial order and, in a controversial ruling, removed Judge Scheindlin from the case on grounds that her acceptance of the case as “related” to a previous case over which she presided, as well as her interviews with the press, created an appearance of impropriety. In re Reassignment of Cases: Floyd v. City of New York, No. 13-3088, 2013 WL 5998139, at *1-8 (2d Cir. Nov. 13, 2013). In this Opening Statement, I contend that Judge Scheindlin’s factual findings and legal analysis are correct. I start with a brief overview of the Terry doctrine, follow with a discussion of Floyd and related litigation in Philadelphia, and then provide a critique of both the doctrinal and policing aspects of investigative detentions.

The Supreme Court’s definition of the Terry “reasonable suspicion” standard has been largely a function of the Justices’ subjective assessments as to whether the conduct at issue is predictive of criminal conduct. The Court has repeatedly stated that it would apply “commonsense” judgments and permit officers to make reasonable inferences from a suspect’s behavior in determining whether there were legal grounds for a stop-and-frisk. See, e.g., Illinois v. Wardlow, 528 U.S. 119, 125 (2000). The Court has not required police or prosecutors to demonstrate by empirical data that the characteristics relied upon—for example, that the suspect was acting suspiciously, had fled from police, had bulges in his pockets, or was engaged in “furtive movements”—are actually predictive of criminal conduct. Thus, in Wardlow, the Court permitted a stop-and-frisk where the suspect fled from police in a high crime area. Id. at 124-25. There was no other proof of any criminal conduct and the Court recognized that many persons, particularly in minority neighborhoods, might not want to engage with the police and might flee, though perfectly innocent. Id. at 125. The Court simply assumed that persons who avoided the police in high crime areas could reasonably be suspected to be involved in criminal conduct. But studies have shown that there is no significant correlation between this conduct and criminal activity. See, e.g., Tracey L. Meares & Bernard E. Harcourt, Foreword: Transparent Adjudication and Social Science Research in Constitutional Criminal Procedure, 90 J. Crim. L. & Criminology 733, 790 (2000).

As a related matter, the Supreme Court’s definition of “reasonable suspicion” permits consideration of behavior that is entirely innocent. See Arvizu, 534 U.S. at 274-75. Indeed, stops and frisks have often been justified on vague and subjective grounds. See, e.g., United States v. Erwin, 803 F.2d 1505, 1510 (9th Cir. 1986) (finding that police possessed reasonable suspicion where an arriving airline passenger scanned the passenger area, walked at a fast pace through the terminal, clutched his carry-on luggage, became nervous when police advised him they were investigating drug trafficking, and provided inconsistent statements about needing to visit his sick grandmother).

In addition, the lack of empirical data (a direct result of the failure of most police departments to maintain stop-and-frisk data) has compromised the constitutional assessment of stop-and-frisk practices. We do not know what grounds for stops tend to reveal criminal conduct, including possession of weapons. Nor do we have data from most police departments regarding racial disparities in stops or even the actual number of stops and frisks. As a result, the legal doctrine, police procedures, and community responses to this pervasive practice have been based on conjecture and speculation. Further, case-by-case adjudications are made in a vacuum as courts do not have access to relevant information regarding the universe of stops and frisks in each jurisdiction.

Finally, the stop-and-frisk doctrine has developed primarily as a function of the exclusionary rule, which means that the cases in which courts adjudicate the issue are those where contraband or other evidence of criminal conduct was uncovered in the investigative detention. In this context, significant forces, including police perjury and judicial reluctance to suppress evidence, can distort the fact-finding process and legal analysis.

In New York City, stop-and-frisk practices have generated strong debate on the wisdom and legality of these procedures. This debate came to a head in Floyd v. City of New York. From January 2004 through June 2012, the New York City Police Department made 4.4 million pedestrian stops, of which over 80% were of African Americans or Latinos. Floyd, 2013 WL 4046209, at *1. More than half of those stopped were also subjected to a frisk. The Floyd court found the following facts:

  • The number of stops per year rose sharply from 314,000 in 2004 to a high of 686,000 in 2011. Id. at *3. The 2011 figure represents approximately 8.3% of New York City’s 2011 population, estimated at 8.24 million persons. See Population Div., N.Y.C. Dep’t of City Planning, American Community Survey Demographic and Housing Estimates: New York City and Boroughs (Sept. 2012), available at
  • 52% of all stops were followed by a protective frisk for weapons; a weapon was found in only 1.5% of these frisks. Floyd, 2013 WL 4046209, at *3.
  • 8% of all stops led to a search into the stopped person’s clothing—ostensibly because the officer felt an object during the frisk that he either suspected to be a weapon or immediately perceived to be other contraband. In 9% of these searches, the object was a weapon. 91% of the time, it was not. In 86% of these searches, the object was not contraband. Id. at *4.
  • 6% of all 4.4 million stops resulted in an arrest; 6% resulted in a summons. The remaining 88% resulted in no further law enforcement action. Id.
  • In 83% of the 4.4 million stops, the person stopped was black or Hispanic; in 10%, the person was white. Id.
  • In 2010, New York City’s resident population was 23% black, 29% Hispanic, and 33% white. Id.
  • The officer used force in 23% of the stops of blacks, in 24% of the stops of Hispanics, and in 17% of the stops of whites. Id.
  • Weapons were seized in 1.0% of the stops of blacks, in 1.1% of the stops of Hispanics, and in 1.4% of the stops of whites. Id.
  • Contraband other than weapons was seized in 1.8% of the stops of blacks, in 1.7% of the stops of Hispanics, and in 2.3% of the stops of whites. Id.
  • In 2004, the officer failed to state a specific suspected crime in 1% of stops. This figure rose to 36% by 2009. Id.

Plaintiffs’ expert, Dr. Jeffrey Fagan, conducted a review of forms prepared by officers after each stop and classified the stops as “apparently justified,” “apparently unjustified,” or “ungeneralizable.” Id. at *17. Significantly, this review accepted as true in every instance the information stated by the police. Id. These indefinite categories reflected a flawed recording system that consisted mainly of checkboxes of highly subjective factors (e.g., “furtive movements,” “evasive response,” “suspicious bulge,” “high crime area”), id. at *17-18, none of which provided a full textual statement of the basis for a stop. The repetitive use of “high crime area” and “furtive movements” as the basis for the stop led Dr. Fagan to conclude that officers were employing a routinized script in asserting reasonable suspicion. Id. In addition to the vague and patterned reasons for stops, the court noted that some stops were not recorded, that 36% of the 2009 stop forms did not identify a suspected crime, that “furtive movements” and “high crime area” (which were each marked in over 40% of the stops) were “negatively correlated” with a summons or arrest, and that only 12% of all stops resulted in an arrest or summons. Id. at *13, *19. On this basis, the court concluded that the number of stops without reasonable suspicion was significantly greater than the 200,000 impermissible stops identified by plaintiffs, Id. at *18-19, and that the City had a policy and custom of Fourth Amendment violations. Id. at *70.

On the race discrimination claim, the court first had to decide which of the competing “benchmarks” proposed by the experts provided the best statistical approach for measuring possible racial profiling. Id. at *19. Dr. Fagan used population and reported crime as benchmarks for understanding the racial distribution of stops. Id. at *20. The City proposed a benchmark consisting of the rates at which various races appear in suspect descriptions from crime victims (“suspect race description data”). Id.

The court credited Dr. Fagan’s analysis and ruled that the City’s benchmark was flawed by the assumption that the racial distribution of stopped pedestrians would resemble the racial distribution of the local criminal population, even if “the people stopped [we]re not criminals.” Id. (emphasis omitted). Given the fact that nearly 90% of all persons stopped were not involved in criminal conduct and that only 13% were stopped pursuant to a specific suspect description, a benchmark of persons actually involved in criminal conduct was not reliable. Id. Crime suspect data may serve as a reliable proxy for the pool of criminals exhibiting suspicious behavior, but not for innocent persons, particularly where the “behavior” descriptions are so vague and often consistent with innocence. It is important to note that even where there are high levels of crime by race, very few persons in the community are responsible. By the City’s logic, all minority residents are properly suspect for the acts of these very few assailants. Cf. United States v. Montero-Camargo, 208 F.3d 1122, 1143 (9th Cir. 2000) (Kozinski, J., concurring) (“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.”).

Dr. Fagan’s standards and regression analysis provided strong support for a finding of race discrimination. More stops were made of blacks and Hispanics, even when other relevant variables were held constant. Floyd, 2013 WL 4046209, at *23-24. Dr. Fagan cited the excess of race-correlated stops beyond that which would be consistent with the local crime rate as evidence of racially disparate treatment. Id. at *24. The analysis also revealed that, regardless of the racial composition of a geographic area, blacks and Hispanics were more likely to be stopped. Id. Once stopped, blacks were 30% more likely than whites to be arrested (controlling for the alleged crime), as opposed to receiving a summons. Id. And minorities were 9-14% more likely to be subjected to the use of force. Id. Most significantly, the hit rate for blacks, as measured by the issuance of a summons or an arrest, was 8% lower than for white suspects. Id. This evidence demonstrates that minorities were targeted for stops based on a lesser degree of suspicion.

Beyond the statistical evidence, the court found evidence of race discrimination through an examination of “institutional” practices, specifically the deliberate indifference of the NYPD to patterns of race discrimination in stop-and-frisk practices. Id. at *24-47, *74-75. These patterns were recognized as early as 1999, when the New York Attorney General issued a report on stop-and-frisk practices that documented unexplained racial disparities in stops. Id. at *25 & n.200. Despite this knowledge, the NYPD put great pressure on commanders and others in the chain of command (down to patrol officers) to increase the number of stops (from 97,000 in 2002 to 686,000 in 2011), Id. at *26-33, but failed to audit the stops for possible racial discrimination. Id. at *38-39. And there was evidence that officers were encouraged to make stops based on racial characteristics or stereotypes—to target the “right people,” young blacks and Hispanics. Id. at *72. The court determined that these factors were evidence of the NYPD’s failure to adequately train, supervise, or discipline officers with respect to racial profiling practices. Id. at *75.

The constitutional flaws in the New York City program are not isolated phenomena. In Philadelphia, from 2008 to 2009, stops increased by almost 20%, reaching an annual rate of over 253,000 for a population less than 1.6 million, thus exceeding the rate of pedestrian stops in New York City. See Research & Planning Unit, Phila. Police Dep’t, Investigation of Person by District (June 7, 2011) (on file with author) (noting a total of 216,832 stops in 2008 and 253,276 stops in 2009); see also Phila. Research Initiative, Pew Charitable Trusts, Philadelphia: The State of the City, A 2010 Update 5 fig. (2010) (estimating the 2009 Philadelphia population at 1.55 million), available at In 2010, a class action lawsuit was filed against the City of Philadelphia, presenting the same Fourth and Fourteenth Amendment claims raised in Floyd. See Settlement Agreement, Class Certification, and Consent Decree at 1, Bailey v. City of Philadelphia, C.A. No. 10-5952 (E.D. Pa. June 21, 2011), available at Soon thereafter, the parties agreed to a consent decree that required the City to conduct stops and frisks within constitutional limits; prohibited stops and frisks solely on the basis of “loitering,” being in a “high crime . . . area,” “acting suspiciously,” or making “furtive movements;” and prohibited the use of race as a basis for a stop except in cases of suspect identifications by race. Id. at 4 (internal quotation marks omitted).

Between 2009 and 2012, the number of stops in Philadelphia dropped from 253,000 to 215,000 per year. See Plaintiffs’ Third Report to Court and Monitor on Stop and Frisk Practices at 4, Bailey, C.A. No. 10-5952, available at However, the high rate of impermissible stops and frisks has persisted. A 2012 audit of a random sample of 1850 stops revealed that 43-47% of all stops and frisks were made without reasonable suspicion. Id. (A police department audit of 2013 stops showed a patrol unit rate of 37% improper stops. Audits & Inspections Div., Phila. Police Dep’t, 75-48A Comprehensive Audit: 2nd Quarter 2013, at 6 (Sept. 6, 2013) (on file with author).) The hit rates in 2012 were even lower than in New York City. Contraband was recovered in only twenty-nine stops (1.57% of all stops) and only three guns were seized (0.16%). Id. at 9. Arrests occurred in only 5.29% of all stops. Id. at 10.

Floyd and Bailey show the important role of data and statistical analysis in assessing Fourth and Fourteenth Amendment stop-and-frisk issues. With respect to Fourth Amendment analysis, police forms provide comprehensive information from which one can determine on a systemic basis whether stops are being conducted with the requisite reasonable suspicion. As we can see from the New York and Philadelphia data, large numbers of these stops facially violate Fourth Amendment standards.

With respect to frisks, the data are even more troubling. In New York City, between 2004 and 2012, there was a 1.5% rate of recovery of a weapon following a frisk, Floyd, 2013 WL 4046209, at *3, while in Philadelphia the rate was 1.1% after the consent decree went into effect. Plaintiffs’ Third Report to Court, supra, at 6-7. It must be remembered that frisks may only be conducted where the officer has reasonable suspicion that the person stopped is armed and dangerous. And while the courts have not quantified the reasonable suspicion standard in terms of expected hit rates for contraband, these rates are well below any reasonable threshold. In other words, if the police are truthfully reporting the basis for their stops, it is apparent that the legal standard is entirely too permissive as it authorizes stops and frisks on the basis of conduct that is not reliably predictive of criminality. The courts’ rulings that actions such as being present in a high crime area, engaging in furtive movements, avoiding police contact, or appearing nervous when stopped reliably predict criminal conduct (or that a suspicious bulge or evasive conduct points to possession of a weapon) are largely unsupported by the data. Alternatively, to the degree that these factors are in fact predictive of criminal conduct, the low hit rates raise serious questions about the truthfulness and accuracy of police-stop documentation.

The stop-and-frisk data also provide a reliable basis on which to determine whether racial disparities in stop-and-frisk practices are the result of racial profiling. Claims of racial bias are virtually impossible to prove in a motion to suppress—in fact, the Supreme Court has ruled that racially discriminatory stops are not even the proper subject of Fourth Amendment analysis, see Whren v. United States, 517 U.S. 806, 813 (1996)—or even in a civil rights lawsuit for damages. As the Floyd opinion reflects, benchmarks and sophisticated regression analysis can properly measure discriminatory patterns. For example, statistical findings that whites subjected to a stop-and-frisk are more likely to be carrying a weapon than blacks or Hispanics are probative of the fact that the threshold for stopping minorities is lower than that for stopping whites. Indeed, there is a certain irony in the Police Department’s criticism of the statistical evidence in Floyd, which was relied upon by the court only after a rigorous Daubert challenge. See Floyd v. City of New York, 861 F. Supp. 2d 274 (S.D.N.Y. 2012) (admitting plaintiffs’ expert’s opinions). For many years, police and prosecutors in New York and elsewhere have relied on police “experts” to convict defendants based on untested scientific theories (e.g., hair samples, shoeprints, bite marks, and outdated arson analysis) that have been proven to be unreliable. See Nat’l Research Council of the Nat’l Acads., Strengthening Forensic Science in the United States: A Path Forward 145-150, 155-61, 170-76 (2009), available at

A word on efficacy. Judge Scheindlin precluded the City from presenting evidence to support its position that stop-and-frisk has played a significant role in reducing crime in New York City. See Floyd, 2013 WL 4046209, at *1. The proposition that the stop-and-frisk practices in New York have been successful in reducing crime is both factually questionable and ultimately legally irrelevant. The court made clear that the resolution of the case was not an all-or-nothing proposition. The NYPD can continue to stop and frisk in accordance with constitutional principles. Thus, the issue is not whether stops and frisks generally reduce crime, but whether the unlawful stops and frisks can be justified on the basis of crime control. And, on this question, not only does the evidence show very low hit rates, but it also fails to show any causative effects. Accordingly, in the first quarter of 2013, there was a 50% reduction in the number of stops in New York City (compared with the first quarter of 2012), but violent crime rates have continued to drop at rates similar to those of past years, when the stop rate was far higher. See, e.g., Tamer El-Ghobashy & Michael Howard Saul, New York Police Use of Stop-and-Frisk Drops, Wall St. J. (May 6, 2013), Of course, this evidence does not prove that stops play no role in crime control; rather, it demonstrates that the causal effects of various means of policing and related social and economic factors on crime reduction are not easily determined.

Equally important, as a legal matter, the fact that unrestrained police practices (e.g., dispensing with the probable cause and warrant requirements of the Fourth Amendment) would suppress crime does not mean that in our constitutional system, such “efficacy” can trump restrictions on governmental police powers. Is the City contending that unconstitutional stops are legitimate on the basis of the deterrent effect they have on residents who, knowing that they may be stopped without reason or because of their race, decide not to carry weapons or contraband on the streets?

Finally, those who criticize the Floyd ruling must come to terms with the long-term failure of the NYPD to engage in self-regulation and internal accountability with respect to its stop-and-frisk practices. For years, data that showed that officers had been engaging in large numbers of stops that violated the Fourth Amendment were ignored, and the “scripted” forms supposedly supporting these stops were taken at face value. Claims of racial bias and gratuitous insults and demeaning conduct during stops were also brushed aside, under the mantra that such practices were necessary to effective policing and reduction of crime. Indeed, the expert reports and other evidence in Floyd provided no information not already known to the NYPD. Courts are hesitant to intervene in policing on a systemic level, but in this case, the deliberate indifference by the NYPD made judicial intervention necessary and proper.

Floyd reached the right result for the right reasons. Stops without legal justification are unconstitutional and have a counterproductive effect by reducing the trust necessary for community policing. And worse, if such violations are also racially disproportionate, the injuries cut even deeper. Pro-active, “hot spot,” and community policing can continue to play a role in ensuring public safety. Nothing in Floyd prevents the police from stopping persons based on adequate factual grounds, as long as the practice is not tainted by racial bias. Indeed, reform of these practices can establish the foundation for fairer and more effective community-based policing. These reforms are possible through court-ordered remedies, but only if police departments accept the basic constitutional tenets and hold officers and supervisors accountable for systemic transgressions.

Rebuttal — Lawrence Rosenthal

One-Eyed Floyd

Professor of Law, Chapman University Fowler School of Law.

Perhaps the most curious thing about Judge Scheindlin’s decision condemning the New York City Police Department’s program of stop-and-frisk is how it insists on viewing stop-and-frisk in a relentlessly monocular fashion.

1. There is irony in Judge Scheindlin’s finding that the NYPD’s use of stop-and-frisk amounts to a policy of unreasonable search and seizure in violation of the Fourth Amendment. This tactic, branded as “unreasonable” by Judge Scheindlin, actually seems to work.

Homicide in New York City rose from a rate of 4.7 per 100,000 population in 1960 to a 1991 peak of 31.0, in waves that roughly corresponded to drug epidemics, with the increase concentrated in firearms-related homicides. See Jeffrey Fagan, Deanna L. Wilkinson & Garth Davies, Social Contagion of Violence, in The Cambridge Handbook of Violent Behavior and Aggression 688, 694-701, 694 fig.36.1 (Daniel J. Flannery et al. eds., 2007). In 1991, the NYPD adopted a new community policing model that emphasized foot patrols and combating low-level disorder. See Civil Rights Bureau, N.Y. State Office of the Attorney Gen., The New York City Police Department’s “Stop and Frisk” Practices: A Report to the People of the State of New York from the Office of the Attorney General 47-52 (1999), available at Three years later, after the appointment of a new police commissioner, the department began to place greater emphasis on aggressive stop-and-frisk tactics for minor infractions, adopted a system of statistical analysis that targeted enforcement at “hot spots” of crime, and imposed greater managerial accountability. See id. at 52-56 & 53 n.32.

Between 1991 and 2009, New York experienced the broadest and deepest decline in violent crime of any major American city. See Franklin E. Zimring, The City that Became Safe: New York’s Lessons for Urban Crime and its Control 3-27 (2012). By 2012, New York’s homicide victimization rate dropped to 5.05 per 100,000 population. See FBI, Crime in the United States 2012, at tbl.8 (2013), available at (showing 419 incidents of murder and nonnegligent manslaughter out of a population of 8.3 million). Space does not permit a complete examination of the causes of New York’s crime drop. Elsewhere, I have reviewed the evidence and demonstrated that there is a substantial case to be made that New York’s stop-and-frisk tactics deserve considerable credit. See Lawrence Rosenthal, Pragmatism, Originalism, Race, and the Case Against Terry v. Ohio, 43 Tex Tech. L. Rev. 299, 320-30 (2010). Two additional studies by eminent criminologists have provided additional evidence supporting this conclusion. See Zimring, supra, at 131-44; David Weisburd, Cody W. Telep & Brian A. Lawton, Could Innovations in Policing Have Contributed to the New York City Crime Drop Even in a Period of Declining Police Strength? The Case of Stop, Question and Frisk as a Hot Spots Policing Strategy, Just. Q. 11-18 (Jan. 11, 2013),

The superiority of proactive policing over reactive patrol should be unsurprising. As one criminologist put it, “Police on patrol cannot see enough to intervene very often in the life of the community.” Mark Harrison Moore, Problem-Solving and Community Policing, in 15 Modern Policing 99, 112 (Michael Tonry & Norval Morris eds., 1992). Years ago, a police executive made the same point to me when he told me that he had come to regret that air conditioning had been installed in his department’s patrol cars. In that insight can be found the case for stop-and-frisk. When undertaken with frequency and targeted at hot spots of crime, stop-and-frisk can alter the perceptions of offenders by making apparent the risks of carrying drugs or guns in public. Reactive patrol, by contrast, encourages criminals to intimidate the community so that they do not call police for help. When drugs and guns are driven off the streetscape, the risks of violent confrontation decrease; community decline driven by open and notorious criminality can be reversed. Indeed, aggressive patrol focused at statistical “hot spots” of crime is one of the few crime control policies that has generated fairly consistent evidence of its efficacy. See, e.g., Nat’l Research Council of the Nat’l Acads., Firearms and Violence: A Critical Review 230-35 (Charles F. Wellford et al. eds., 2005); Cody W. Telep & David Weisburd, What Is Known About the Effectiveness of Police Practices in Reducing Crime and Disorder?, 15 Police Q. 331, 333-36, 340-41 (2012).

Like all debates about policing and crime, there are a myriad of factors at work, and rarely can causal statements be made with complete confidence. Yet, if there is any kind of serious case to be made for stop-and-frisk’s ability to drive down crime, surely the judiciary ought to pause before tampering with tactics that may well save lives.

Judge Scheindlin, however, refused to cast her eye to questions of efficacy, believing instead that her “mandate [wa]s solely to judge the constitutionality of police behavior, not its effectiveness as a law enforcement tool.” Floyd v. City of New York, No. 08 Civ. 1034(SAS), 2013 WL 4046209, at *1 (S.D.N.Y. Aug. 12, 2013). Yet, in the seminal case blessing stop-and-frisk as an investigative tool, the Supreme Court concluded that “there is ‘no ready test for determining reasonableness other than by balancing the need to search [or seize] against the invasion which the search [or seizure] entails.’” Terry v. Ohio, 392 U.S. 1, 21 (1968) (alterations in original) (quoting Camara v. Mun. Court, 387 U.S. 523, 536-37 (1967)). To strike this balance, a court must consider the “general interest . . . of effective crime prevention and detection,” an “interest which underlies the recognition that a police officer may in appropriate circumstances and in an appropriate manner approach a person for purposes of investigating possibly criminal behavior even though there is no probable cause to make an arrest.” Id. at 22. Judge Scheindlin, however, refused to consider this interest in “effective crime prevention.” To be sure, none of this means that the Fourth Amendment tolerates unconstrained stop-and-frisk merely because it reduces crime. But evidence of the efficacy of stop-and-frisk offers at least some indication that the NYPD’s stops are based on reasonable suspicion that criminal activity is afoot.

2. There is irony as well in Judge Scheindlin’s conclusion that stop-and-frisk amounts to discrimination against minorities, the chief beneficiary of stop-and-frisk. Yet, this reality also remained unseen by Judge Scheindlin. If the NYPD targets minorities, it is not alone; New York’s criminals do the same. At the peak of New York’s crime wave in 1991, the homicide victimization rate in New York was 58 per 100,000 for blacks, 44 for Hispanics, and 8 for whites. Andrew Karmen, New York Murder Mystery: The True Story Behind the Crime Crash of the 1990s 54 fig.2.2 (2000). By 2007, the black homicide victimization rate had declined by more than 42 per 100,000 to 15.9, and the Hispanic rate by more than 39 to 4.9, while the white rate declined by less than 7 per 100,000, reaching 1.55. See Zimring, supra, at 42 fig.2.7. Accordingly, the decline in homicide victimization rates for blacks and Hispanics far exceeded that experienced by whites. Still, the racial skew in homicide victimization continues. The NYPD reported the following victimization statistics for 2012:

Murder and
Shootings New York City
Black 60.1% 74.1% 22.8%
Hispanic 26.7% 22.2% 28.8%
White 8.7% 2.8% 33.1%
Asian/Pac. Isl. 4.2% 0.8% 12.7%

Raymond W. Kelly, N.Y.C. Police Dep’t, Crime and Enforcement Activity in New York City 1, 11, B-1 (2013), available at When a federal judge sets out to remake local policing policy while indifferent to the efficacy of those policies she brands unconstitutional, it is the lives of minorities that are placed at greatest risk.

3. Judge Scheindlin found that Fourth Amendment violations involving stop-and-frisk occurred so frequently in New York that they amounted to both a custom with the force of law and a practice to which policymakers were deliberately indifferent. Floyd, 2013 WL 4046209, at *70-72. Yet, she also acknowledged that about 6% of all stops resulted in an arrest and another 6% resulted in a summons. Id. at *4. The Fourth Amendment reasonable suspicion standard “falls considerably short of satisfying a preponderance of the evidence standard.” United States v. Arvizu, 534 U.S. 266, 274 (2002). It also “accepts the risk that officers may stop innocent people.” Illinois v. Wardlow, 528 U.S. 119, 126 (2000). In this context, a 12% hit rate does not look so bad. It looks even better when one considers that even an apparently unsuccessful stop-and-frisk can create important benefits by deterring potential offenders and cooling off criminogenic hot spots.

Although the Floyd plaintiffs produced evidence relating to nineteen specific stops, 2013 WL 4046209, at *6, such a small sample proves very little about New York City’s policies. With hundreds of thousands of stops at issue, the facts of nineteen discrete incidents cherry-picked by the plaintiffs’ lawyers are statistically meaningless. More significant is the analysis of the plaintiffs’ expert. Relying on data from the UF-250 forms that officers must complete after stops, he estimated that the rate of “apparently unjustified” stops was about 6%. Id. at *16-17. To my eye, this does not look like evidence of a police department that has effectively decreed that noncompliance with the law is its official policy, or that is deliberately indifferent to constitutional requirements. One study of police practices in a mid-sized American city found that 46% of pat-down searches observed by researchers were unconstitutional. See Jon B. Gould & Stephen D. Mastrofski, Suspect Searches: Assessing Police Behavior Under the U.S. Constitution, 3 Criminology & Pub. Pol’y 315, 333 (2004). I have my doubts about the soundness of that conclusion, see Rosenthal, supra, at 330 n.210, but by that standard, the NYPD is doing pretty well.

Judge Scheindlin, however, discounted the significance of the 12% hit rate and the estimated 6% unjustified stop rate with a cascade of speculation. She opined that these calculations are unreliable because officers may not always prepare a UF-250; UF-250s reflect only the officer’s version of a stop and do not contain enough information to ascertain its lawfulness; many of the “stop factors” listed on the UF-250 form are unreliable indicators of reasonable suspicion; over time, officers learned to check off a greater number of factors in order to justify a search; and because the plaintiffs’ expert was overly conservative in his assumption that certain ambiguous factors justified stops. See Floyd, 2013 WL 4046209, at *16-19. Still, it is striking that no expert was willing to opine that the UF-250 database could properly support a finding of a rate of unjustified stops exceeding 6%; it was the nonexpert in black robes who drew that conclusion.

Indeed, Judge Scheindlin’s discussion of the flaws in the UF-250 data, viewed from a different angle, suggests that there was no proper support for her findings. We cannot tell whether the data reflect the rate of unjustified stops or simply the limitations of the UF-250 form. The factors listed on the form may be imperfect, but they are all we have for statistical analysis; and they are hardly irrelevant to the lawfulness of stop-and-frisk. Although Judge Scheindlin deprecated presence in a “high crime area” and “furtive movements” as justifications for a stop, id. at *17-19, the Supreme Court has explained that both “nervous, evasive behavior” and “the fact that the stop occurred in a ‘high crime area’” are relevant to assessing the lawfulness of a stop. Wardlow, 528 U.S. at 124 (citations omitted). Thus, officers’ use of these factors does not mean that a stop is unjustified; indeed, their use of these factors explains why the plaintiffs’ expert was unable to characterize stops as unjustified when these factors were present in combination with others. Similarly, one can only speculate whether evolving trends in officers’ use of the UF-250 form reflect misconduct or, instead, an increasing awareness of the need for care in documenting the justification for stops.

This is not to say that all is well with stop-and-frisk in New York. The number of stops per year increased from 314,000 in 2004 to 686,000 in 2011. Floyd, 2013 WL 4046209, at *3. It is hard to believe that officers’ keen observations alone produced such an astonishing surge. Moreover, there is ample evidence that supervisors pressured subordinates to increase stops. Id. at *26-33. It may well be that as crime continued to decline in tandem with increased stop-and-frisk, command staff pushed the practice to the point of diminishing returns. For example, stop-and-frisk may have spread well beyond the hot spots where it is most likely to be justified—a conclusion supported by evidence of increasing stops in relatively low crime areas where suspects seemed merely “out of place.” See Andrew Gelman, Jeffrey Fagan & Alex Kiss, An Analysis of the New York City Police Department’s “Stop and Frisk” Policy in the Context of Claims of Racial Bias, 102 J. Am. Stat. Ass’n 813, 822 (2007). If this happened, however, it argues only for retaining stop-and-frisk’s original focus on high crime hot spots—certainly not for condemning stop-and-frisk even when it is so targeted.

4. Judge Scheindlin also determined that the NYPD has “a policy of indirect racial profiling based on local criminal suspect data” and that the NYPD and the City “have been deliberately indifferent to the intentionally discriminatory application of stop and frisk.” Floyd, 2013 WL 4046209, at *72. She based these conclusions on evidence that blacks and Hispanics were stopped at elevated rates when compared to a benchmark that considered local population and reported crime rates. Id. at *20, *23-24. But this is again a rather one-dimensional way to view the data. Most obviously, it overlooks the problem of differential offending.

If minorities commit crimes at higher rates than nonminorities, then even absent discrimination, one might expect them to be more frequently subjected to stop-and-frisk tactics. For example, if a stop is properly based on evidence reflecting a 20% probability that a suspect is engaged in unlawful activity, and the underlying offending rate is 2.5% for nonminorities and 5% for minorities, a nondiscriminatory stop-and-frisk regime could nevertheless produce a 12.5% nonminority stop rate and a 25% minority stop rate. Economists argue that elevated search rates for minorities are not troubling as long as minority and nonminority hit rates are comparable; elevated minority search rates may reflect nothing more than an efficient response to differential rates of offending. Cf., e.g., Jeff Dominitz & John Knowles, Crime Minimisation and Racial Bias: What Can We Learn from Police Search Data?, 116 Econ. J. F368, F368-71 (2006) (noting that racial discrimination can be inferred when higher search rates of a high crime–propensity group yield lower hit rates than searches of a low crime–propensity group); John Knowles, Nicola Persico & Petra Todd, Racial Bias in Motor Vehicle Searches: Theory and Evidence, 109 J. Pol. Econ. 203, 205 (2001) (looking at the success rate of police searches across races in order to determine whether those searches are motivated by racial prejudice).

Just as violent victimization rates reflect a racial skew, there is ample reason to believe that differential offending is prevalent in New York. Consider the 2012 NYPD crime data regarding the race of known suspects:

Murder and
Shootings New York City
Black 53.7% 78.2% 22.8%
Hispanic 34.8% 18.9% 28.8%
White 8.7% 2.4% 33.1%
Asian/Pac. Isl. 2.8% 0.5% 12.7%

Kelly, supra, at 1, 11, B-1.

There is also reason to believe that a strategy targeting criminal hot spots would increase this racial skew. Although there is no reliable data on the racial composition of criminal street gangs in New York, the U.S. Department of Justice estimates that nationally, from 1996 through 2011, blacks and Hispanics consistently accounted for more than 80% of total gang membership. Bureau of Justice Assistance, U.S. Dep’t of Justice, National Youth Gang Survey Analysis, Nat’l Gang Center, (last visited Nov. 14, 2013). Similarly, there is evidence that the disproportionate rates at which minorities are incarcerated for drug offenses reflect the fact that they are disproportionately found at open-air drug markets and other locations most easily targeted by police. See Michael Tonry, Malign Neglect—Race, Crime, and Punishment in America 105-07 (1995); cf. R. Richard Banks, Beyond Profiling: Race, Policing, and the Drug War, 56 Stan. L. Rev. 571, 583 (2003) (noting that racial disparities may result from targeting disproportionately minority neighborhoods where it is easier to apprehend drug dealers). Thus, a stop-and-frisk strategy aimed at hot spots of gang and drug crime could well result in higher rates of minority stops, even absent discrimination.

Judge Scheindlin made no findings as to differential offending or other potentially nondiscriminatory explanations for differential stop rates, except to acknowledge that in 2011 and 2012, 83% of all known crime suspects and 90% of all violent crime suspects were black or Hispanic. Floyd, 2013 WL 4046209, at *20. She also made no finding of a racial skew in hit rates, as measured by arrests and summonses. The plaintiffs’ expert even acknowledged that arrest and summons data for 2010 to 2012 reflect only “small differences” in hit rates. Second Supplemental Report of Jeffrey Fagan, Ph.D. at 34-35, Floyd, 2013 WL 4046209 (No. 08 Civ. 1034(SAS)), available at This suggests that police utilize equally reliable indicia of suspicion when stopping minorities and nonminorities. Indeed, a RAND Corporation study of 2006 NYPD stop-and-frisk data concluded, after considering arrest rates and the race of criminal suspects, that there was no evidence of discrimination against blacks. See Greg Ridgeway, RAND Corp., Analysis of Racial Disparities in the New York Police Department’s Stop, Question, and Frisk Practices 13-19 (2007).

Equal protection is offended when “the decisionmaker . . . select[s] or reaffirm[s] a particular course of action at least in part ‘because of,’ not merely ‘in spite of,’ its adverse effects upon an identifiable group.” Wayte v. United States, 470 U.S. 598, 610 (1985) (ellipsis in original) (quoting Pers. Adm’r v. Feeney, 442 U.S. 256, 279 (1979)). Accordingly, when the underlying rate of offending differs by race, stop-and-frisk practices could well involve no legally cognizable discrimination despite racially disparate stop rates if the disparity is a consequence of targeting enforcement where offending is greatest or most amenable to stop-and-frisk tactics. A strategy driven by such law enforcement considerations does not run afoul of the Fourteenth Amendment even if it produces racially skewed stop rates, as the Second Circuit suggested in an earlier case involving an investigation focusing on black suspects because they matched a victim’s description of the offender. See Brown v. City of Oneonta, 221 F.3d 329, 333-34 (2d Cir. 2000).

Despite all this, Judge Scheindlin found a pattern of racial discrimination based on differences between stop rates by race and each race’s representation in the local population. See Floyd, 2013 WL 4046209, at *20 (concluding that “[i]f the police are stopping people in a race-neutral way, then the racial composition of innocent people stopped should more or less mirror the racial composition of the areas where they are stopped,” since “nearly 90% of the people stopped are released without the officer finding any basis for a summons or arrest”). This was a serious error. An effective stop-and-frisk strategy should not target the population at large, but rather, should be focused where it can do the most good, such as criminogenic hot spots. In light of the evidence of differential offending in New York, there is little reason to believe that the racial makeup of those who frequent high crime locations will mirror that of the population at large. Even Judge Scheindlin acknowledged that “[c]rime suspect data may serve as a reliable proxy for the pool of criminals exhibiting suspicious behavior.” Id. at *21. The Fourth Amendment, in turn, permits—in fact it effectively compels—a strategy that targets this pool, rather than the population at large. The fact that most of those who are stopped are released therefore does not reflect discrimination, but only the reality that the Fourth Amendment permits the police to act on the basis of evidence of criminality that stops well short of a preponderance.

5. Although Judge Scheindlin invoked “the human toll of unconstitutional stops,” id. at *2, her one-eyed appraisal neglects another human toll. The contours of Judge Scheindlin’s injunction have yet to take shape since she appointed a monitor to “develop, in consultation with the parties, an initial set of reforms to the NYPD’s policies, training, supervision, monitoring, and discipline regarding stop and frisk.” Floyd v. City of New York, No. 08 Civ. 1034(SAS), 2013 WL 4046217, at *5 (S.D.N.Y. Aug. 12, 2013). Indeed, we will likely never know the scope of the remedy that Judge Scheindlin had in mind since the Second Circuit has ordered her removal from the litigation. See In re Reassignment of Cases: Floyd v. City of New York, No. 13-3088, 2013 WL 5998139, at *1 (2d Cir. Nov. 13, 2013). Yet, the history of judicial efforts to overhaul policing is not encouraging. Most who have studied the U.S. Department of Justice’s attempts, in “pattern or practice” cases, see 42 U.S.C. § 14141 (2006), to use judicial decrees to reform policing have found the results disappointing. See, e.g., Rachel A. Harmon, Promoting Civil Rights Through Proactive Policing Reform, 62 Stan. L. Rev. 1, 52-57 (2009); Kami Chavis Simmons, Cooperative Federalism and Police Reform: Using Congressional Spending Power to Promote Police Accountability, 62 Ala. L. Rev. 351, 371-76 (2011). If Judge Scheindlin’s remedy is sustained on appeal, however, and her successor judge and the appointed monitor put sufficient pressure on the NYPD, they probably can drive stop rates down. Police officers internalize few—if any—of the costs or benefits of search and seizure. Stop-and-frisk is a dangerous, unpleasant business. It should not take much to reach a tipping point. If officers face a credible threat of sanctions if someone complains to a monitor, they will likely stop bothering with stop-and-frisk. After all, there is one foolproof way to drive down the rate of unjustified stops—officers can stay in the doughnut shops.

But, once it becomes apparent that a federal judge–run NYPD will no longer intervene forcefully and proactively on the streets, the dynamics for a return to vintage crime rates will be in place. When crime spiked in 1990, New Yorkers at least had the option of demanding more aggressive law enforcement from politically accountable officials. Federal judges, however, enjoy life tenure, and according to Judge Scheindlin, they must resolutely refuse to consider the effectiveness of policing tactics. Perhaps Judge Scheindlin thinks of federal judges as judicial Goldilocks who can design a system that offers just enough—but not too much—deterrence. This is the stuff of which judicial hubris is made. But, given Judge Scheindlin’s announced disinterest in the efficacy of police tactics, one has little reason to hope that she would care if her decrees produced more crime in New York.

Judge Scheindlin wrote, “Fostering trust and confidence between the police and the community would be an improvement for everyone.” Floyd, 2013 WL 4046209, at *2. Yet the available data suggest that what drives the opinions of local police is the local crime rate, not race. See Rosenthal, supra, at 354-55. People who do not feel safe are not likely to repose their trust and confidence in the police. Even worse, as we have seen, the human toll of violent crime in New York falls disproportionately on minorities. Consider a New York in which wealthy white neighborhoods enjoy low crime rates, while, in disadvantaged minority neighborhoods, police keep stop rates low to please a federal judge but fail to intervene effectively to disrupt the dynamics that drive gang- and drug-related crime. That may satisfy Judge Scheindlin, who cannot be bothered with questions of police efficacy. To my eye, however, a New York like that fails to offer, in the most literal sense, equal protection of the law.

Closing Statement — David Rudovsky

Going Beyond “Trust Us” Policing


As John Adams famously stated in the Boston Massacre Trial, “facts are stubborn things.” Professor Rosenthal accuses Judge Scheindlin of ignoring relevant facts and viewing stop-and-frisk in a “relentlessly monocular fashion,” but he fails to acknowledge the substantial array of facts that undermine his arguments.

Professor Rosenthal asserts that Judge Scheindlin erred in refusing to consider evidence of the “efficacy” of the stop-and-frisk program in New York City. He asks: how can a court find a policing program unreasonable if it “seems to work”? Of course, this position requires both proof of the effectiveness of stop-and-frisk practices in controlling crime and support for the legal proposition that conduct otherwise violative of the Constitution can nevertheless be considered legitimate because of its impact on crime. Indeed, the evidentiary bar on efficacy is high, since citizens are being asked to sacrifice their constitutional rights in return for guarantees of safety.

On the factual question, Professor Rosenthal argues that because there was a steep decline in crime in New York City during the period in which stop-and-frisk became a prominent part of policing, we can assume that there must be some causal link. To the contrary, while there is agreement that New York City policing played a significant role in the crime decline, there is agreement as well that there are “myriad” factors at work and that it is extremely difficult to draw causal lines, as Professor Rosenthal concedes. Moreover, he conflates stop-and-frisk policing with a number of overlapping police initiatives, including CompStat analysis of crime patterns, community policing to engage citizens, and “hot spot” policing to identify salient crime areas. The “proof” cited for the proposition that stop-and-frisk was the critical component in New York’s crime drop is surprisingly thin. Professor Rosenthal relies on a study by Franklin E. Zimring, but Zimring is careful to state that there is no specific evidence that stop-and-frisk played a significant role in the decline in crime. See Franklin E. Zimring, The City that Became Safe: New York’s Lessons for Urban Crime and its Control 148-50 (2012). The other cited study also finds the evidence to be inconclusive, as a result of “the lack of adequate data on possible confounding causes.” See David Weisburd, Cody W. Telep & Brian A. Lawton, Could Innovations in Policing Have Contributed to the New York City Crime Drop Even in a Period of Declining Police Strength? The Case of Stop, Question and Frisk as a Hot Spots Policing Strategy, Just. Q. 16 n.11 (Jan. 11, 2013), Further, Weisburd, Telep, and Lawton explicitly state that they cannot determine “whether the increase in [stop, question and frisk] tactics has come at the expense of other innovative policing strategies.” Id. at 18.

Professor Rosenthal fails to mention other studies that find little or no causal relationship between stop-and-frisk and crime reduction in New York City. See, e.g., Richard Rosenfeld & Robert Fornango, The Impact of Police Stops on Precinct Robbery and Burglary Rates in New York City, 2003-2010, Just. Q. 10 (Aug. 21, 2012), Unfortunately, the relationship between crime, stops, other law enforcement practices, and related social and economic factors is difficult, if not impossible, to measure. Cf. David F. Greenberg, Studying New York City’s Crime Decline: Methodological Issues, Just. Q. 2-3 (Jan. 10, 2013), Further, Professor Rosenthal does not address the large decline in crime—not as steep as in New York but still substantial—that was achieved in other cities without massive use of stop-and-frisk. See, e.g., Steven D. Levitt, Understanding Why Crime Fell in the 1990s: Four Factors that Explain the Decline and Six that Do Not, 18 J. Econ. Persp. 163, 168 tbl.4, 172-73 (2004).

Professor Rosenthal also fails to discuss contrary factual evidence from the New York City experience itself. Crime fell by approximately 10% in New York City from 1992 to 1993, two years prior to the onset of NYPD stop-and-frisk policing tactics, and the huge increase in stops from 2002 through 2011 followed several years of a steep decline in New York crime. See Jeffrey Fagan & John MacDonald, Policing, Crime, and Legitimacy in New York and Los Angeles: The Social and Political Contexts of Two Historic Crime Declines, in New York and Los Angeles: The Uncertain Future 219, 240-42, 242 tbl.8.1 (David Halle & Andrew A. Beveridge eds., 2013); Bernard E. Harcourt & Jens Ludwig, Broken Windows: New Evidence from New York City and a Five-City Social Experiment, 73 U. Chi. L. Rev 271, 320 app. fig.1 (2006).

It is remarkable as well that the Rebuttal ignores the most pertinent contemporary data: during the first half of 2013, there was a huge decrease (over 50%) in the number of stops and a continuing sharp decrease (almost 25%) in the number of homicides, compared to the same period last year. Raymond W. Kelly, N.Y.C. Police Dep’t, New York City Police Department Stop Question and Frisk Activity 4 (2012), available at; Joseph Goldstein, City Homicides Drop Sharply, Again; Police Cite New Antigang Strategy, N.Y. Times (June 28, 2013),; Joseph Goldstein, Police Stop-and-Frisk Encounters Plunged in Second Quarter of 2013, Data Show, N.Y. Times (Aug. 27, 2013), And the decline in stops was even greater during the third quarter of 2013—the NYPD made just 21,000 stops, or 80% fewer than during the same time period in 2012. Associated Press, NYC Police Stop-and-Frisk Numbers Down Sharply, N.Y. Times (Nov. 18, 2013), To be clear, I do not argue that these data prove that stop-and-frisk plays no role in the control of violent crime, but it is powerful evidence that the NYPD’s claims on this issue are highly exaggerated.

Professor Rosenthal argues that the hit rate for stops in New York “does not look so bad.” On this normative issue, I strongly disagree. First, the supposed 12% hit rate is itself an overstatement. Recent studies have shown that nearly 50% of all arrests and summons that result from stop-and-frisk activity are dismissed very early in the prosecution phase. See Eric T. Schneiderman, N.Y. State Office of the Attorney Gen., A Report on Arrests Arising from the New York City Police Department’s Stop-and-Frisk Practices 3, 10 fig.7 (2013), available at (noting that, between 2009 and 2012, 15.7% of arrests were not prosecuted, 10.5% were dismissed, and 21.3% were “adjourn[ed] in contemplation of dismissal”). Moreover, a number of these arrests and summonses are the result of post-stop information or police–citizen conflict during the course of a stop, which proves nothing about the legitimacy of the stop in the first place.

Even more important is the data on frisks. New York police frisk suspects they have stopped about 50% of the time (I suspect that frisks are underreported), yet in only 1.5% of these frisks do they uncover a weapon. See Floyd v. City of New York, No. 08 Civ. 1034(SAS), 2013 WL 4046209, at *3 (S.D.N.Y. Aug. 12, 2013). Considering that the Supreme Court has ruled that a frisk requires reasonable suspicion that the suspect is “armed and dangerous,” one can only wonder why the police have been wrong 98.5% of the time. Data from other jurisdictions disclose similar very low hit rates for weapons. In Philadelphia, for example, the hit rate for weapons was 1.1% during the first half of 2012. See Plaintiffs’ Third Report to Court and Monitor on Stop and Frisk Practices at 6-7, Bailey v. City of Philadelphia, C.A. No. 10-5952 (E.D. Pa. June 21, 2011), available at (noting that 265 frisks led to the recovery of only three guns). As I pointed out in my Opening Statement, either the Terry standard has become so open-ended as to allow frisks without cause or the police are asserting grounds for a frisk where none exist.

There is also good reason to believe that the UF-250 checkbox form masks many improper stops. In 1999, the New York Attorney General reviewed instances where police were required to record narratives explaining the reason for their stops. The Attorney General found that based on the information provided in the UF-250 forms, 15.4% of stops were not justified and 23.5% were ambiguous—that is, the forms did not provide sufficient information to allow a reader to determine whether the facts articulated amounted to “reasonable suspicion.” See Civil Rights Bureau, N.Y. State Office of the Attorney Gen., The New York City Police Department’s “Stop and Frisk” Practices: A Report to the People of the State of New York from the Office of the Attorney General 160-64 (1999), available at Similar results have been found in the Bailey litigation and in an academic study cited by Professor Rosenthal. See Plaintiffs’ Third Report to Court, supra, at 4-5; Andrew Gelman, Jeffrey Fagan & Alex Kiss, An Analysis of the New York City Police Department’s “Stop and Frisk” Policy in the Context of Claims of Racial Bias, 102 J. Am. Stat. Ass’n 813, 815-16 (2007).

Finally, it must be remembered that the Floyd ruling on the Fourth Amendment claim was also based on the testimony of individuals who had been subjected to illegal stops. Professor Rosenthal asserts that such testimony was “cherry-picked,” but corroborating evidence demonstrates that the reality of stop-and-frisk in New York City includes a pattern of officer conduct that is needlessly humiliating and intimidating. See, e.g., Wendy Ruderman, For Women in Street Stops, Deeper Humiliation, N.Y. Times (Aug. 6, 2012),

As a legal matter, Professor Rosenthal assures us that nothing permits “unconstrained” stops, but he provides no standards by which to distinguish between proper stops and those that are prohibited by the Fourth Amendment. The argument that efficacy trumps standard Fourth Amendment restraints runs counter to over 200 years of constitutional history. We should not forget the role that the “writs of assistance” played in the adoption of the Fourth Amendment. These writs were used by the British Government to detect and suppress widespread violations by the colonists of laws mandating import duties and taxes. Notwithstanding the fact that the writs were effective, those who drafted the Fourth Amendment’s prohibition on unreasonable searches and seizures intended to prohibit this type of random search. See Boyd v. United States, 116 U.S. 616, 625-27 (1886). The fact that crime can be more effectively suppressed if the police are freed of all restraints on their power to search is quite beside the constitutional point. Indeed, the Fourth Amendment was adopted to ensure that the “liberty of every man [is not placed] in the hands of every petty officer.” Stanford v. Texas, 379 U.S. 476, 481-82 (1965) (quoting Boyd, 116 U.S. at 625). In other cases in which the challenged police practices were similarly effective, the Supreme Court has rejected the “efficacy” argument where the searches were not justified under the Fourth Amendment. See, e.g., City of Indianapolis v. Edmond, 531 U.S. 32, 34-35, 47-48 (2000) (holding a highway drug checkpoint program unconstitutional despite a 4.74% drug-related arrest rate and an 8.96% overall arrest rate).

Professor Rosenthal’s Fourteenth Amendment analysis is equally questionable. Citing the high rates of both victimization and offending by African Americans with respect to violent crimes, he asserts that a program that reduces the level of crime must be legitimate even if it disproportionately affects African Americans. There are a number of flaws in this approach. First, it is simply not permissible to target all members of a racial minority for the acts of very few of that group. Judge Scheindlin was careful to acknowledge that an officer can legitimately consider race when stopping a person who fits a specific suspect description (a category that constituted the basis for a mere 13% of the stops in New York City), Floyd, 2013 WL 4046209, at *20, but that is a far cry from the City’s argument that group actuarial-based suspicion is permissible. To the contrary, stopping individuals based on their race because other members of that race are responsible for higher rates of crime is precisely the kind of discrimination barred by the Equal Protection Clause. See Samuel R. Gross & Debra Livingston, Racial Profiling Under Attack, 102 Colum. L. Rev. 1413, 1415 (2002) (“The essence of racial profiling is a global judgment that the targeted group . . . is more prone to commit crime . . . .”). Further, the Second Circuit has specifically warned that “a description of race and gender alone will rarely provide reasonable suspicion justifying a police search or seizure.” Brown v. City of Oneonta, 221 F.3d 329, 334 (2d Cir. 2000).

Second, Judge Scheindlin credited benchmarks that considered local offending patterns. See Floyd, 2013 WL 4046209, at *5. Plaintiffs’ expert considered the total volume of crimes in small, racially homogeneous areas and employed regression analysis to determine if factors other than race could explain the large racial disparities in New York City stops. See Second Supplemental Report of Jeffrey Fagan, Ph.D. at 2-5, Floyd, 2013 WL 4046209 (No. 08 Civ. 1034(SAS)), available at Professor Rosenthal complains that these benchmarks do not take into consideration “differential offending” patterns by race and asserts that since the stop rates approximate the reports of suspect race data for serious crimes in New York City, the program is free of racial bias.

There are several problems with this argument. There is far from complete data as to the race of those involved in “reported” crimes; the great majority of stops (85%) between 2004 and 2009 were not based on suspicion of violent crime, Floyd, 2013 WL 4046209, at *46 n.424; and only 13% of stops were based on a determination that the suspect fit a specific physical description. at *20. Thus, the notion that stop-and-frisk practices are carefully calibrated to respond to violent crime (or even crimes of possession of weapons) is refuted by the factual record.

Along the same lines, Professor Rosenthal cites studies by economists who suggest that racially disparate stop rates of motorists are justified if they eventually lead to an “equilibrium” in offending rates (on the highways, this usually means drug offenses). See, e.g., John Knowles, Nicola Persico & Petra Todd, Racial Bias in Motor Vehicle Searches: Theory and Evidence, 109 J. Pol. Econ. 203 (2001). Putting aside substantial criticisms of these studies’ methodologies, see e.g., Bernard E. Harcourt, Against Prediction: Profiling, Policing, and Punishing in an Actuarial Age 122-39 (2007); Steven N. Durlauf, Assessing Racial Profiling, 116 Econ. J. F402, F406-07 (2006), and the fact that highway stops involve a universe of drivers, almost all of whom are legitimate “suspects” due to traffic infractions, it is impermissible to treat individuals of one racial group differently on the basis of higher offending levels of that group. Would the advocates of this approach to crime control argue that if one racial group is responsible for twice as many violent crimes as another, the state could enact sentencing laws that double that group’s sentences, on the theory that the increased deterrence will eventually lead to an equilibrium in offending?

Third, Professor Rosenthal’s defense of highly disproportionate rates of prosecution and incarceration of whites and African Americans for drug offenses is highly problematic. The national data demonstrate that whites use, possess, and sell drugs at rates approximately equal to minorities, but the arrest and incarceration rates for drug offenses are overwhelmingly disparate by race. See Marc Mauer, The Sentencing Project, Racial Disparities in the Criminal Justice System: Prepared for the House Judiciary Subcommittee on Crime, Terrorism, and Homeland Security 3-4 (2009), available at (testifying that in 2005, African Americans represented 14% of current drug users, 33.9% of persons arrested for a drug offense, and 53% of persons sentenced to prison for a drug offense); Michael Tonry & Matthew Melewski, The Malign Effects of Drug and Crime Control Policies on Black Americans, in 37 Crime and Justice: A Review of Research 1, 23-25, 25 fig.9, 26 tbl.3 (Michael Tonry ed., 2008). And in New York City, the racial disparities in marijuana arrests are truly astounding. See Harry Levine, The Scandal of Racist Marijuana Arrests—and What to Do About It, Nation (Oct. 30, 2013), African Americans are at great risk in the war on drugs; whites, whether on college campuses, or in middle-class communities—and particularly with respect to marijuana—get a free ride. Is this racial discrimination? Not according to Professor Rosenthal who views this merely as an artifact of policing “easy targets.”

Having failed to come to grips with the facts and the law on both the Fourth and Fourteenth Amendment issues, Professor Rosenthal ends with the usual dire warning: that the Floyd remedial order will lead to “de-policing.” Fortunately, there is no support for the argument that police will retreat to the nearest doughnut shop for fear of complaints or lawsuits, or to avoid what they view as unfair limits on their power. As noted above, before Judge Scheindlin’s ruling, the NYPD substantially decreased its number of stops, yet crime rates continued to fall. Rather than de-policing, the NYPD altered its practices in the summer of 2013, shifting from casting a broad net of low-yield, unproductive stops to a tactical focus on youth gangs and crimes in and around public housing projects. See Joseph Goldstein & J. David Goodman, Frisking Tactic Yields to a Focus on Youth Gangs, N.Y. Times (Sept. 18, 2013),

In other jurisdictions where courts have intervened, there is no evidence of such “de-policing.” See Christopher Stone et al., Policing Los Angeles Under a Consent Decree: The Dynamics of Change at the LAPD 22 (2009), available at The experience in Los Angeles shows that a consent decree and new community policing models can lead to better police practices, higher levels of community cooperation with the police, and lower rates of crime.

A final point. Defenders of New York’s stop-and-frisk practices assert that the decrease in crime has largely benefitted the African American community, which suffers most of the violent crime and, therefore, whatever the indignities suffered, overall the practices provide a distinct advantage to African Americans. Indeed, some suggest that those who do not live in high crime areas should refrain from criticizing any policing methods that drive down crime in those neighborhoods. But it appears quite clear—from the recent mayoral election and from widespread criticism of current policing in New York City—that many African Americans believe the past practices to be unnecessarily harsh, humiliating, and oppressive.

The Floyd decisions on liability and remedial measures were proper and necessary. Court intervention that is properly calibrated to deal with proven constitutional violations can both remediate the violations and leave police with the powers necessary for effective crime control.

Closing Statement — Lawrence Rosenthal

What, Me Worry?


Professor Rudovsky assures us that a hostile judicial takeover of the NYPD is no cause for concern—it will eliminate the supposed evils of stop-and-frisk without unintended consequences. For those of us who can afford to live in safe, stable communities where crime rates remain relatively constant regardless of the policing tactics used, this kind of talk is cheap. For the less fortunate among us, there is more cause for concern.

1. Professor Rudovsky doubts any relationship between stop-and-frisk and crime reduction. His argument, however, encounters a few inconvenient facts, even as he warns us that “‘facts are stubborn things.’”

Professor Rudovsky tells us that crime began declining in New York from 1992 to 1993, before the advent of stop-and-frisk, but he overlooks the fact that, according to the Attorney General’s report that he himself cites, the NYPD began focusing on aggressive patrol and combating low-level disorder as early as 1990, under Police Commissioner Lee Brown. See Civil Rights Bureau, N.Y. State Office of the Attorney Gen., The New York City Police Department’s “Stop and Frisk” Practices: A Report to the People of the State of New York from the Office of the Attorney General 47-52 (1999), available at In fact, an article coauthored by the plaintiffs’ expert in Floyd v. City of New York acknowledged that the crime decline in New York bore a close temporal relationship to the adoption of more aggressive policing tactics. See Jeffrey Fagan, Franklin E. Zimring & June Kim, Declining Homicide in New York City: A Tale of Two Trends, 88 J. Crim. L. & Criminology 1277, 1313-16 (1998). The same expert also coauthored a study demonstrating the efficacy of aggressive patrols undertaken near public housing projects. See Jeffrey Fagan, Garth Davies & Jan Holland, The Paradox of the Drug Elimination Program in New York City Public Housing, 13 Geo. J. on Poverty L. & Pol'y 415, 442-53 (2006). Even more recently, he coauthored an essay, cited by Professor Rudovsky on another point, that concluded that the NYPD’s COMPSTAT program—which uses statistical analysis “to drive the allocation of police resources to crime ‘hot spots’”—was clearly linked to crime reduction: “If we were to focus on only examining the effect of the COMPSTAT program in New York City that started in 1994, we would see a clear influence of this program in reducing robberies and homicides.” Jeffrey Fagan & John MacDonald, Policing, Crime and Legitimacy in New York and Los Angeles: The Social and Political Contexts of Two Historic Crime Declines, in New York and Los Angeles: The Uncertain Future 219, 240 (David Halle & Andrew A. Beveridge eds., 2013).

Professor Rudovsky does not quarrel with the efficacy of COMPSTAT or “hot spot” policing, yet he seems to believe that one can disaggregate the effects of these policing strategies from those of stop-and-frisk. They are, however, inextricably intertwined. Studying statistical data alone does not drive down crime, nor does telling officers to visit hot spots. Without stop-and-frisk, these other tactics cannot alter the calculus of potential offenders on the streetscape. Indeed, as I note in my Rebuttal, a wide variety of studies throughout the nation have demonstrated the efficacy of aggressive patrol targeted at hot spots. And, if stop-and-frisk succeeds in driving down crime, it gets harder to brand it constitutionally unreasonable. To be sure, Professor Rudovsky is correct that evidence of the efficacy of a policing tactic cannot “trump[] standard Fourth Amendment restraints,” but it is equally settled that the Fourth Amendment requires consideration of both the privacy and law-enforcement interests at stake in order to assess that tactic’s reasonableness in the constitutional sense. To ignore efficacy is to ignore at least half of the calculus. Thus, to claim that “citizens are being asked to sacrifice their constitutional rights in return for guarantees of safety” is to assume an unwarranted conclusion about the constitutionality of a practice that may well be predicated on reasonable suspicion of criminal activity, as demonstrated by its ability to drive down crime.

Professor Rudovsky argues that cities other than New York also experienced crime declines, but he pays little attention to Franklin Zimring’s exhaustive demonstration that no major city experienced a crime decline as steep and long lasting as that of New York. See Franklin E. Zimring, The City that Became Safe: New York’s Lessons for Urban Crime and Its Control 3-27 (2012). He also cites a study showing little correlation between stop-and-frisk and crime decline in New York from 2003 to 2010, see Richard Rosenfeld & Robert Fornango, The Impact of Police Stops on Precinct Robbery and Burglary Rates in New York City, 2003-2010, Just. Q. 10 (Aug. 21, 2012),, but ignores the point made in my Rebuttal—in the last decade, it may well be that stop-and-frisk passed the point of diminishing returns, which would explain the lack of correlation between stop-and-frisk and crime decline during that period. Indeed, the authors of the study that Professor Rudovsky cites acknowledge both that their conclusion is “[i]n contrast with prior research,” and that the total rate of documented stops tripled between 2003 and 2010. See id. at 2-3. Having observed the success of stop-and-frisk, the NYPD may have expanded its use beyond the criminogenic hot spots where it is most likely to pay dividends. Overuse of stop-and-frisk during the past decade likely explains as well the NYPD’s ability to keep crime rates low over the past eighteen months even as it has reduced its reliance on stop-and-frisk. An unnecessary increase in stop rates since 2003, however, supplies an argument for returning stop-and-frisk to its original focus on hot spots—not for the complete overhaul ordered by Judge Scheindlin, with stated indifference to questions of efficacy.

When it comes to the efficacy of stop-and-frisk, it is striking that Professor Rudovsky offers no alternate explanation for New York’s crime decline. As I note in my Rebuttal, no other explanation presents itself. It is true, as Professor Rudovsky notes, that “the relationship between crime, stops, other law enforcement practices, and related social and economic factors is difficult, if not impossible, to measure.” But, if there is even a serious case to be made for the efficacy of stop-and-frisk, surely we should hesitate before abandoning a tactic that may well save lives, mostly those from impoverished and disproportionately minority communities. Abandoning a tactic that may well work, after all, amounts to a gamble with people’s lives.

2. Like Judge Scheindlin, Professor Rudovsky speculates that the 12% hit rate reflected in the New York stop-and-frisk data may be overstated. Yet, the evidence he offers makes a thin gruel. That many charges arising from stops and frisks are dismissed at some point prior to trial, for example, is not evidence that these charges were unsupported by probable cause; the standard for conviction, of course, is higher than the standard required to justify a stop or an ensuing arrest. Beyond that, prosecutors may abandon a case for a myriad of reasons unrelated to the propriety of the arrest or stop that may have preceded it. Professor Rudovsky also disregards the rather low rate (6%) of unjustified stops calculated by the plaintiffs’ expert, see Floyd v. City of New York, No. 08 Civ. 1034(SAS), 2013 WL 4046209, at *16 (S.D.N.Y. Aug. 12, 2013), and instead cites the higher rates calculated by the New York Attorney General in 1999 based on the UF-250 forms then in use. See Civil Rights Bureau, supra, at 160-64. That iteration of the form, however, was intended solely as an investigative tool; it was neither intended nor used to document the justification for stops. See James J. Fyfe, Reaction Essay, Stops, Frisks, Searches, and the Constitution, 3 Criminology & Pub. Pol’y 379, 392-94 (2004). The 1999 Attorney General’s report, in short, proves little. And, as I note in my Rebuttal, even the current form has dramatic limitations as a tool for identifying unjustified stops. About this, Professor Rudovsky is silent.

Professor Rudovsky highlights Judge Scheindlin’s finding that only about 1.5% of frisks recover weapons, though this amounts to a recovery of some 33,882 weapons. See Floyd, 2013 WL 4046209, at *13 & n.110. One wonders exactly how high the probability that a suspect is armed and dangerous needs to be before Professor Rudovsky would permit a frisk. It is hard to brand an officer unreasonable if, when stopping a suspect reasonably suspected of criminal activity while in a statistical hot spot of violent crime, that officer does not require much in the way of additional predication before performing a protective frisk. After all, the officer’s life is at risk. For just this reason, if judicial intervention places too much pressure on officers to refrain from frisks, especially in hot spots of violent crime, the likely result is that officers will simply abstain from stopping suspects altogether. As Chief Justice Warren cautioned, excessive judicial efforts to restrict officers’ ability to protect themselves “may exact a high toll in human injury and frustration of efforts to prevent crime.” Terry v. Ohio, 392 U.S. 1, 15 (1968). If we are to take seriously the notion that the Fourth Amendment’s command of reasonableness requires that we balance privacy and law-enforcement interests, surely we should regard an officer’s ability to err on the side of caution when safety is at risk as an interest of the highest order.

3. Professor Rudovsky and I can agree that “stopping individuals based on their race because other members of that race are responsible for higher rates of crime is precisely the kind of discrimination barred by the Equal Protection Clause.” But, the small differences in hit rates in New York for stops and frisks of members of different racial groups suggest that persons are not being stopped because of their race, but based on equally valid indicia of reasonable suspicion. Beyond that, Professor Rudovsky never explains why, in light of the ample evidence of differential offending by race in New York—evidence with which neither the plaintiffs nor Judge Scheindlin quarreled—we should expect stop rates to be representative of the racial composition of the population at large, rather than that of the pool of offenders.

Especially for a stop-and-frisk strategy focused at hot spots of violent crime, we should expect stop rates to reflect the racial composition of the subset of those likely to frequent such hot spots, not that of the general population. Similarly, even if whites and blacks use drugs at roughly equal rates, a stop-and-frisk strategy focusing on the kind of street-level distribution most vulnerable to that tactic should be expected to reflect the racial composition of the pool of street-level traffickers, not the population of drug users at large. If we focus on the subset of the population most likely to frequent the statistical hot spots of violent crime on which stop-and-frisk focuses, then the racial composition of the pool of offenders in New York would seem to be a far better proxy than the general population. Judge Scheindlin even acknowledged that “[c]rime suspect data may serve as a reliable proxy for the pool of criminals exhibiting suspicious behavior.” Floyd, 2013 WL 4046209, at *21. This is precisely why crime suspect data offers a far better proxy for those most likely to be subjected to stop-and-frisk based on reasonable suspicion of criminal activity in a nondiscriminatory regime than a benchmark based on the racial composition of the population at large. Cf. United States v. Bass, 536 U.S. 862, 862-64 (2002) (per curiam) (rejecting an equal protection challenge to a capital prosecution based on statistics showing that the United States charges blacks with death-eligible offenses at higher rates than whites and enters into plea bargains more frequently with whites “because respondent failed to submit relevant evidence that similarly situated persons were treated differently”). Officers on patrol are looking for offenders, not the innocent.

It seems that Judge Scheindlin and Professor Rudovsky are distressed that the NYPD is not stopping enough innocent whites. Yet, if the NYPD stopped more innocent whites, the hit rate for stops of whites would fall below that for minority stops, providing potential white plaintiffs powerful support for their own equal protection claim that the NYPD discriminated against them on the basis of race. Thus, given the population benchmark employed by Judge Scheindlin and endorsed by Professor Rudovsky, the NYPD would violate the Equal Protection Clause no matter what it did. These are the inescapable conundrums born of a failure to take account of the undisputed evidence of differential rates of offending by race in New York City.

4. Professor Rudovsky dismisses fears that placing judicial pressure on stop-and-frisk will overdeter police. Yet, as even critics of urban policing acknowledge, there is substantial evidence that police respond to criticism of perceived racial profiling in just this fashion. See, e.g., Frank Rudy Cooper, Understanding “Depolicing”: Symbiosis Theory and Critical Culture Theory, 71 UMKC L. Rev. 355, 357-64 (2002). Economists have modeled and studied the phenomenon, and caution that proper training and supervision of officers are critical to ensuring that police reforms aimed at reducing perceived misconduct do not produce higher rates of unlawful behavior. See, e.g., Paul Heaton, Understanding the Effects of Antiprofiling Policies, 53 J.L. & Econ. 29, 57-58 (2010).

Under other circumstances, the risk of overdeterrence might be regarded as acceptable since politically accountable officials have potent incentives to ensure that police fight crime effectively. Now that crime in New York is low, the focus seems to have turned to police reform, but if crime were to rise, the ordinary processes of political accountability would cause the pendulum to swing. But, when an unaccountable, life-tenured federal judge seizes control of a police department, declares her disinterest in the efficacy of police tactics, and then appoints a monitor with a mission to reduce the prevalence of stop-and-frisk, one has to worry how the police will react, and whether the ordinary political checks and balances will remain effective if crime begins to spiral. At least, I worry. Professor Rudovsky, not so much.

 Previous Debate

The Contraception Mandate and Religious Freedom

Steven D. Smith and Caroline Mala Corbin

Next Debate 

Off-Label Drug Promotion and the First Amendment

Stephanie M. Greene and Lars Noah