III. Assessing the Stop and Frisk Policy for Racial Bias
The NYPD's stop and frisks did not garner significant attention until they became connected to the death of an innocent man. The shooting of Amadou Diallo in 1999 spurred inquiry into the NYPD's policies, including stop and frisk. Questions were raised about the disparate impact of these policies on minority groups and primarily minority neighborhoods. The answers are conflicting.
A. The Attorney General's 1999 Report
On February 5, 1999, the front page of the New York Times reported that on the previous day, an unarmed man in the Bronx was killed by police officers who had fired at him forty-one times. That first article reported that it was unclear why the police began shooting, but the story quickly became familiar to most New Yorkers. Diallo was an African immigrant, standing unarmed in front of the building where he lived. The four police officers, who were white and in plainclothes, approached him because they believed that he fit the description of a rapist. They began shooting because they believed that he was grabbing a gun, but it turned out to be his wallet.
The shooting drew attention not only from New Yorkers, but also from those in Washington, D.C., forcing the NYPD to reconsider its practices, including stop and frisk. After the four officers were acquitted of all criminal charges, the Bronx District Attorney, Robert Johnson, said: This case raises a lot of issues about police tactics.People in the Bronx have been trying to get the attention of the Police Department for some time . . . and this case will do it.
The NYPD responded by introducing new initiatives and reorganization. Then-Police Commissioner Howard Safir published an editorial in the New York Times, arguing that for most New Yorkers . . . the issue is civility, not brutality, and discussing the NYPD's new Courtesy, Professionalism and Respect initiative to encourage officers to be more polite to those whom they encounter. The Street Crimes Unit was reorganized, making the new, smaller units accountable to regional commanders, and making the unit uniformed rather than plainclothes, a move that was later reversed when gun arrests by the unit dropped by 40%.
In response to public pressure, then-Attorney General Eliot Spitzer opened an investigation into NYPD stop and frisk practices in the spring of 1999, within months of taking office. The resulting report discussed that even before the Amadou Diallo incident in February of 1999, Attorney General Spitzer heard complaints about stop and frisks, primarily from minorities. The report termed the Amadou Diallo incident as the defining moment of police-community relations in New York City. The incident exacerbated the already eroding public confidence in the police, especially among minorities, and prompted the Attorney General to conduct this study.
The study focused on the De Bour level three stop. It used data from approximately 175,000 documented stop and frisk encounters between January 1, 1998 and March 31, 1999, and compared the rates of stop and frisks for people from different racial groups and different locations with crime rate data. The report found a strong statistical correlation between race and likelihood of being stopped, and concluded that crime rate data does not adequately explain these differences, even in predominantly minority precincts.
The report also found that for every stop that led to an arrest, an additional nine stops occurred. The report explained that because an officer need only have reasonable suspicion to stop an individual[,] it is not surprising that, given this lower threshold, many such stops' should fail to result in an actual arrest. The report found further racial disparities in this statistic--for every one arrest, there were 9.5 stops of blacks, 8.8 stops of Hispanics, and 7.9 stops of whites.
In analyzing the stop data in reference to crime rate data, this report chose to rely on arrest statistics. Acknowledging the limitations of arrest statistics, the report nevertheless concluded that they are a more reliable measure than crime complaint data. Using regression analysis, the report found that blacks and Hispanics were significantly more likely than whites to be stopped after controlling for race-specific precinct crime rates and precinct population composition by race.
The report's most compelling statistical finding was that in precincts in which blacks are less than 10% of the population, blacks remained over two times (2.17) more likely to be stopped on suspicion of committing a violent crime than whites . . . [and] these differences are evident after controlling for race- and crime- specific arrest rates (within each precinct). The statistic was the same for stops under suspicion of weapons crimes. In contrast, blacks were half as likely as whites to be stopped under suspicion of property crimes in all precincts. More than anything else, it is this statistic that suggests that officers are specifically targeting blacks for stops.
In addition to the race factors, the report found significant problems with the reasons for the stops as marked on the stop and frisk forms. The study concluded that on approximately 15% of the forms, the officers did not provide any reason for the stop. On an additional 23% of the forms, the reasons noted by the officers for the stops were not sufficient for reasonable suspicion. The report hints that these problems should have been noticed by the supervisors reviewing the forms, but stops short of saying that all of these stops were legally insufficient, noting that some of the problems could be explained by clerical errors.
The report did not recommend specific steps that the NYPD take as a result of these findings. Instead, the conclusion of the report called on the NYPD, community leaders, and attorneys to consider the report in their discussions of change. The NYPD responded to the report negatively, criticizing the methodology and conclusions. Then-Commissioner Howard Safir criticized the report for misusing the stop and frisk forms:
What the attorney general fails to recognize is that stop and frisk reports are primarily administrative tools and are used to assist in investigations. . . . His attempts to conclusively evaluate the legality of a stop based solely on what the officer placed on this report is another serious flaw. . . . Clearly, in our look at the analysis . . . we don't see any validity to it.
B. The RAND Report
Eight years later, in 2007, the NYPD hired the RAND Corporation to analyze stop and frisk data in response to continuing controversy over racial bias (RAND Report). First and foremost, the RAND Report concluded that the raw statistics, while easy to compute, often exaggerate racial disparities. This conclusion was based on analysis of the over 500,000 stops that NYPD officers documented in 2006.
The raw numbers of stops that the report referred to show that minorities (blacks and Hispanics) were stopped far more often than whites. When examined based on the reason for the stop, the differences were even more exaggerated--the number of blacks stopped on suspicion of weapons crimes was twice that of Hispanics, and each number overwhelms that of whites. This was true for stops based on suspicion of robbery, trespassing, and drugs, though the rates evened out somewhat for burglary and auto theft.
The RAND Report analyzed stop and frisk data using several benchmarks: 2005 arrests, descriptions of suspects, and the census. The use of such benchmarks, while necessary, are still:
fraught with challenges [because] the method can either detect or hide racial bias due to unobserved or unmeasured factors that affect both the racial distribution that the benchmark establishes and the racial distribution of stops. For example, drug arrests take many forms, including complex buy-bust operations, complaints from residents, and direct observation by officers. Stops documented in UF-250s are generally only from the latter two situations, while the drug-arrest benchmark includes all of these arrests.
The report was most critical, however, of the use of the census as a benchmark, arguing that census benchmarks do now account for differential rates of crime participation by race or for differential exposure to the police. Comparisons to the residential census are not suitable for assessing racial bias.
The RAND Report compared the rates at which whites, blacks, and Hispanics were stopped against each of these types of benchmarks. The report concluded that while Hispanics were stopped at a rate proportional to their representation in the census, blacks were stopped at a rate 50% greater than their representation. On the other hand, blacks were stopped at a rate that is 20 to 30 percent lower than their representation in crime-suspect descriptions, while Hispanics were stopped five to ten percent more. Lastly, using the benchmarks of arrests, the report found that stops of blacks were proportional to their representation among those arrested while the stops of Hispanics were 6% higher.
Lastly, the report considered whether there were race based discrepancies in the manner the stop was conducted and in the outcome, and concluded that after adjustment for stop circumstances, we generally found small racial differences in the rates of frisk, search, use of force, and arrest.
The report's general conclusion was that the raw statistics . . . distort the magnitude, and, at times, the existence of racially biased policing. The report noted that a large scale overhaul of the stop and frisk policy was unnecessary, but nevertheless recommended several improvements. First, officers should clearly explain to pedestrians why they are being stopped. Second, the NYPD should improve training and assessment so that new officers [are] fully conversant with stop, question and frisk documentation policies. Third, the NYPD should improve auditing and oversight of stop and frisks and develop mechanisms to flag officers with out of the ordinary stop patterns.
C. Problems with Assessing the Stop and Frisk Policy
Why did the Attorney General Report and the RAND Report come to opposite conclusions? One explanation is that the studies were conducted almost a decade apart. Another explanation is that the reports were prepared under different political circumstances. Newly-elected Attorney General Spitzer had an incentive to appear tough on an issue important to many voters, while the NYPD hired the RAND Corporation with the hope of showing a lack of racial bias in its tactics.
Another possibility is that the data being assessed is far from perfect because not all stops are being documented. Like the Attorney General's report, the RAND Report focused on level three stop and frisks, which is the lowest level of intrusion at which the NYPD requires officers to document the interaction. Any interactions that, in the officer's view, do not rise to that standard do not require documentation. Although the RAND Report noted that officers have incentives to document all the stops they conduct in order to show their productivity and comply with performance standards, those officers who perform the most stops (and therefore have no reason to worry about their numbers) have no incentive to document at least some encounters. Furthermore, while precinct-level commanders have been documented to use stop and frisk data to assess officer performance, there is no evidence that commanders of specialized units, like anti-crime and narcotics units, do so as well. Officers in these units, who may perform many stops, may not have such incentive to record them. In 2009, the CCRB cited eighty officers for failing to prepare stop, question, and frisk reports. Howard Safir, the police commissioner at the time of the Diallo shooting, admitted that the number of reports filled out may not accurately represent the number of people stopped and frisked.
D. Civil Rights Suits by Groups
The Center for Constitutional Rights (CCR), a non-profit legal advocacy organization tracing its history to the civil rights movement, has filed two suits challenging the constitutionality of the NYPD's actions in implementing their stop and frisk policy. The first suit resulted in a settlement that serves as the basis for the second suit. That settlement mandated the creation of a racial profiling policy and training of officers on this policy. The second suit is ongoing, and represents some possibility of judicial review of the NYPD's practices.
Following the death of Amadou Diallo, the CCR filed a class action lawsuit alleging that the NYPD
implement[s] . . . a policy, practice and custom of unconstitutional stops and frisks of New York City residents by the Street Crimes Unit (SCU) . . . and SCU officers stopped individuals without reasonable suspicion required by the Constitution and often used race and/or national origin as the determinative factors in deciding to stop and frisk individuals, in violation of the Equal Protection Clause of the United States Constitution.
In the Settlement Stipulation, the City of New York continued to deny the allegations against it, but agreed to create and implement a written racial profiling policy that complies with the United States Constitution, to develop protocols to review stop and frisk activity, including whether officers are adequately preparing the necessary documentation and whether the stop and frisk activity is based on reasonable suspicion, and to provide annual in service training regarding the Racial Profiling Policy. The stipulation also generally required that the NYPD continue training officers on stop and frisk procedures and requiring that UF-250s be prepared.
The second suit, filed by the CCR in 2008, alleged that NYPD officers, in violation of the Equal Protection Clause of the Fourteenth Amendment, often have used, and continue to use, race and/or national origin--not reasonable suspicion--as the determinative factors in deciding to stop and frisk individuals. The complaint further alleges that NYPD Commissioner, Raymond Kelly, and Mayor Michael Bloomberg had devised, implemented and enforced policies and practices that lead to these stops, in part by putting pressure on precinct commanders, and through them, individual officers, to conduct more stop and frisks and prepare more UF-250s.
After describing two separate stop and frisk incidents involving the plaintiffs, the complaint recounts the history of alleged stop and frisk abuses by the NYPD, especially the Street Crimes Unit, and the results of the prior lawsuit. The complaint then described a report in the New York Daily News that a precinct captain told officers at a roll call to stop, question, and frisk all black males at a particular subway station in response to a pattern of robberies. The complaint goes on to allege that [d] efendants have applied a facially neutral policy, the anti-racial profiling policy, to Plaintiff, and similarly situated individuals, in an intentionally racially discriminatory manner.
The remedies sought by the complaint are as follows: 1) a declaration that the NYPD's policy, practice and/or custom of suspicionless stops and frisks challenged herein is unconstitutional; 2) injunctive relief enjoining the NYPD from continuing suspicionless and racially based stops and use of quotas; 3) improved training and supervision; 4) compensatory damages; and 5) attorney's fees and costs.
Shortly after filing suit, the plaintiffs moved for production of all the UF-250s filed by police officers since 1998. Defendants argued that portions of the data, including information about the location of the stop; the identity of the person stopped; identity of the reporting officer; and information about any subsequent arrest or summons, either were not relevant or were protected by the law enforcement privilege. If forced to produce it, the defendants insisted that the information should be under protective order to prevent public disclosure. The court agreed in part with the defendants and ordered that the names of the persons stopped and of the reporting officers be placed under a protective order, but granted the plaintiffs' motion for the remainder of the data. The court ordered that the same categories of information previously released to other entities, including the RAND Corporation, be released, and further held that the exact stop location (rather than merely the precinct) should be released because stops are routinely justifie[d] by noting that the suspect is in a high-crime area . . . and more specific information is needed to determine whether, as plaintiffs claim, high-crime area is used as a proxy for the race and ethnicity of the area's residents. The court also agreed that the plaintiffs need some way of identifying the officers conducting the stops in order to assess the defendants' theory that only certain specific officers are engaging in racially motivated stops, and to identify those officers they wish to depose. The court declined, however, to make available the names, addresses, and phone numbers of those stopped, stating that as no class has yet been certified, those individuals stopped are not yet parties to this litigation, and their privacy interests must be respected.
Two years later, plaintiffs learned that the IAB was conducting investigations into allegations by officers from three different commands that they were subjected to quotas, and moved for production by defendants of all the documents from the IAB investigations into these allegations. Defendants argued that these files are confidential, contain irrelevant information, and are protected by the law enforcement privilege. The court conducted an in camera inspection of the documents and concluded that the vast majority of documents in the IAB files are irrelevant to plaintiffs' claims. The court further found that although the defendants met their burden that the law enforcement privilege is applicable to these items, the plaintiffs have successfully rebutted the strong presumption against lifting the privilege. The court discussed the difficulty that plaintiffs had in proving their allegations of the existence of quotas and that the allegations made by officers of the existence of such quotas provides at least unexplored avenues of discovery. The court then weighed the interests involved and concluded that the information in the files that the court deemed relevant was to be disclosed, subject to a protective order.
In October 2010, the plaintiffs submitted their expert report prepared by Jeffrey Fagan, a law professor at Columbia University and a noted constitutional law scholar. Fagan criticized the RAND Report's methodology, especially its use of the race of suspects of violent crimes as a benchmark. He pointed out that violent crimes comprised fewer than ten percent of the crimes reported in the 2005-2006 period used in the study and that almost half the violent crime complaints did not report the race of the suspect.
Fagan assessed the UF-250 reports for stops between 2004 and 2009 provided to the plaintiffs as a result of the discovery order. He considered the plaintiffs' Fourteenth Amendment claim of disparate treatment and their Fourth Amendment claim that many stops lack legal justification. With regard to the first claim that NYPD officers use race as a factor in determining whether to stop and frisk an individual, Fagan found that:
The results show consistently, across the most policy-relevant and frequent crime categories, that racial composition predicts stop patterns over and above any predictions made by crime or other factors. In effect, overall stop patterns in the precincts were predicted more by the Percent Black [a variable indicating the percentage of the precinct population that is black] and Percent Hispanic (compared to Percent White) than by observed crime.
With regard to the second claim, that officers lack reasonable suspicion to effect many of the stops, Fagan examined both the prevalence of stops lacking reasonable suspicion and whether blacks and Hispanics were more likely to be stopped with inadequate reasonable suspicion. Fagan examined the forms to determine whether the stops were justified, unjustified, or of indeterminate legality. First, Fagan found:
furtive movements and high crime area are the two most common items checked off on the UF-250. . . . [T] hese two categories are notable in two ways: they both are subject to subjective and highly contextualized interpretation, and they both--either separately or in conjunction with one another--are legally insufficient to justify a stop.
Fagan also found that when furtive movements and high crime area were indicated on the form, the arrest rates were lower than when they were not indicated. Fagan concluded that these indicia seem[ ] to be invoked so often and in such disparate circumstances to suggest that [they are] almost meaningless.
Using the above criteria, Fagan found that 68.9% of all stops were justified, 24.4% were indeterminate, and 6.7% were unjustified. Fagan noted troubling patterns in stops for trespassing specifically, where 55.38% of stops (in non-radio run instances) were indeterminate, adding that this may result from the design of the stop factors, since other was checked off in 60% of the stops for this suspect crime. Additionally, Fagan found that:
The highest rate of unjustified stops was for weapons offenses. Nine percent of the radio runs and 12.3 percent of the non-radio runs were classified as unjustified. This results in large part from the extensive use of furtive movements as a stop justification for weapons offenses: 60 percent of stops where a weapons offense was suspected were justified in whole or part by furtive movement. Another legally indeterminate stop factor, suspicious bulge, was cited in 34.4 percent of weapons stops, and high crime area in nearly half the weapons stops.
Fagan then examined the hit rates of the stops, defined as whether a stop leads to an arrest of a suspect or a summons, and whether contraband is seized. Fagan found that 5.37% of all stops resulted in an arrest, while 6.26% resulted in the issuance of a summons. Guns were seized in 0.15% of all stops, and contraband of all types (including weapons, drugs, and stolen property) were seized in 1.75% of all cases. Fagan compared this to the rates in random checkpoint cases--an 8.96% arrest rate where officers were searching vehicles for drugs and a 1.6% rate when officers were searching for drunk drivers--and concluded that the NYPD stop and frisk tactics produce rates of seizures of guns or other contraband that are no greater than would be produced simply by chance.
Lastly, Fagan looked at racial disparities in stop outcomes. He found that blacks were significantly less likely to receive any sanction, but there was no statistically significant difference between Hispanics and whites. Furthermore, Fagan found that [a] mong those receiving any sanction, Blacks were 31.4% more likely than White suspects to be arrested versus summonsed, but [t] here was no significant difference for Hispanics or Other Race suspects compared to Whites.
Even with these findings, the defendants still moved for summary judgment on the overarching claims against the City of New York and the NYPD as well as on the claims of the named plaintiffs. The court found multiple triable issues of fact, allowing most of the case to proceed to the trial stage. First, the court found that there was significant conflicting evidence on the questions of whether the NYPD is providing adequate training to officers on the law of reasonable suspicion, the department's required documentation procedures, and the racial profiling policy. The court also denied summary judgment on the issue of whether or not the City has acquiesced in a widespread custom or practice of unconstitutional stops and frisks. Next, the court denied summary judgment on the issue of illegal quotas for stop and frisks imposed by the NYPD, finding that that the testimony of officers, the recordings of roll calls, and the findings of a labor arbitrator as to the existence of such quotas presented a stark contrast to the City's denial of their existence. Finally, the court found issues of material fact on whether race is the determinative factor in the decision to effect a stop and whether the NYPD's stop and frisk policies have had a disparate impact in the form of a widespread pattern of race-based stops, reflected in the differences between the RAND Report and Fagan's report.
The survival of this case past the summary judgment stage represents perhaps the best chance for substantial judicial review of the practices of the NYPD. In particular, the plaintiffs' focus on the issues of the training and supervision of officers as well as the use of quotas is most likely to lead to real change, as these represent the clearest avenues of reform within the NYPD.
