Last week, Judge Shira Scheindlin of the U.S. District Court for the Southern District of New York issued an opinion in Floyd v. City of New York, finding the City of New York liable in connection with its stop-and-frisk policy for violating the Fourth Amendment’s guarantee against unreasonable searches and seizures, as well as the Fourteenth Amendment Equal Protection Clause’s guarantee against discrimination. The ruling is important because it validates the perspective of many people—especially minorities—who have observed a pattern of unsupportable and arbitrary police behavior in stopping and frisking people, even as public officials vociferously maintained that police were scrupulously adhering to the dictates of the Constitution.
In reviewing the district court’s comprehensive opinion, I found quite persuasive the district judge’s findings that New York City police, with the blessings of (and often pressure from) their superiors: (1) regularly performed stops in the absence of reasonable suspicion to believe criminal activity was afoot; (2) regularly frisked those who were stopped, in the absence of reasonable suspicion to believe that the targets of the frisks were armed and presently dangerous; and (3) routinely selected people and neighborhoods for stops and frisks at least in part on the basis of race and ethnicity (in particular, on the basis of membership in the African American or Hispanic communities). In short, I found the district court’s assessment of the evidence persuasive and compelling.
In this column, however, I want to focus on a relatively small quibble that I have with one part of the court’s discussion of the Fourteenth Amendment Equal Protection issue. My quibble does not in any way alter my ultimate reaction to the court’s factual and legal conclusions. Nonetheless, I think that my narrow disagreement with a portion of the analysis may help clarify some of the obstacles we face in detecting discriminatory intent, in this and other contexts where the issue arises.
In the course of its analysis of the Fourteenth Amendment Equal Protection Clause question, the district court considered an important question that currently has no definitive answer: What is the baseline against which one ought to determine whether and when police officers (or others) have engaged in discriminatory profiling? That is, what would an officer’s behavior (or the conduct of a group of officers) have looked like if it did not result from racially discriminatory motives? In this column, I will discuss both the district court’s answer to this question and some potentially valid alternative (or additional) answers to the same question.
Racial Profiling
People can mean to refer to any of a number of things when they use the phrase “racial profiling.” In this column, I use the phrase to refer to a decision to stop or frisk a person that is motivated—in whole or in part—by the fact that the person is a member of one or more racial or ethnic groups (in this case, African American or Hispanic). It does not matter, for purposes of this analysis, whether the underlying motivation is a feeling of hatred for people in the targeted group, or whether it is instead the earnest belief that members of that group commit crimes at a disproportionate rate. What matters is the fact that race or ethnicity plays a role in the decision, not the officer’s reason for making decisions on the basis of race or ethnicity.
For an illustrative example, consider the actions of two public school teachers who each call on male students in class more frequently than on female students and who make these decisions on the basis of the students’ gender. One teacher may make this choice because he dislikes girls; he is, in other words, a sexist who calls on boys preferentially out of the desire to express favoritism for his preferred group. The second teacher makes the same choice for a different reason.
The second teacher believes that boys generally prepare for class only when they are called on, while most girls prepare for class no matter what. By contrast to the first teacher, the second teacher feels no animosity toward girls but simply believes his teaching is more effective when he calls mainly on his male students. We would probably be inclined to judge the first teacher more harshly than the second. Nonetheless, both teachers would be engaged in what I have been calling profiling, and both teachers would accordingly violate the Fourteenth Amendment (in the absence of a compelling state interest justifying the discrimination).
Evidencing Racial Profiling
Under the Supreme Court’s 1976 decision in Washington v. Davis, the government does not violate the Constitution by acting in a manner that unintentionally yields racially disparate outcomes. A public official must intend to discriminate on the basis of race in order for the discrimination at issue to count as a violation of equal protection. As I described it in my above definition of racial profiling, then, the sort of profiling of which I am speaking—and the sort which violates equal protection—is discriminatory not merely in effect, but in intent as well.
As with any situation in which the law requires a wrongdoer to have a particular intent, however, the process of proving intent often involves evidence of effect. Some police officers may offer a helpful running commentary while they carry out their jobs, thus proving their racial motivations. But much of the time, it will be necessary to rely not on officers’ declared intent, but on the behavior itself (that is, the simple fact of an officer or group of officers disparately stopping members of different groups) to prove an underlying discriminatory motive. Likewise, in a murder case, the prosecution can prove that the defendant intended the death of her victim even in the absence of direct verbalized evidence of intent (such as a confession from the defendant saying that “I intended to kill that day”). The prosecution can do this by proving that the shooter aimed at the victim’s chest with an automatic weapon and then fired several rounds, after which the victim died of his wounds.
To make an effective showing of racial motive on the basis of disparate outcomes, of course, one must have in mind the sorts of outcomes that would not be disparate. Plaintiffs trying to prove racially discriminatory stops, then, would need to offer a neutral baseline against which to judge those stops.
For a simple illustration, consider our earlier example of the teachers who call preferentially on boys in their classrooms. If we were trying to prove that the teachers are in engaged in deliberate gender profiling, our first choice might be a self-revelation. For example, a teacher might say at the beginning of the semester, “I have a policy of calling on boys rather than on girls, when I have a choice, because I find that it makes for a more interesting class discussion.” Absent such commentary, however, we would look to how frequently the teacher calls on boys versus girls to determine whether he is discriminating.
At first glance, it might seem that if the teacher calls on boys twice as frequently as he calls on girls, it is quite clear that he is discriminating, but we actually need more information to make this determination. First, we might ask how many boys and how many girls are in the class. If there are thirty boys and two girls, then an otherwise suspicious pattern of calling on boys most of the time might turn out to look quite benign. If we judge the frequency of the teacher’s calling on males-versus-females by looking at how many boys versus girls are in the class, then we are treating the male and female proportions of the class as the baseline.
We might select a different baseline, however. What if the teacher generally calls only on students who are raising their hands? And what if the average boy raises his hand five times as frequently as the average girl does? The teacher, in this scenario, might sincerely say that he calls on just about everyone who raises his or her hand at least once during a class but that boys just seem to raise their hands more frequently. In that case, a teacher would not have to be acting on the basis of any gender motive to call on male students much more frequently than he calls on female students. In this example, we judge the teacher’s conduct against a baseline of the proportion of male and female hands in the air rather than (or in addition to) the proportion of male and female students in the class.
One could come up with other baselines for the gender question, but the point is that selecting the right baseline or baselines can be an important part of accurately determining whether we are actually witnessing a disparity that evidences a discriminatory motive.
Baselines for Terry Stops
In Floyd v. City of New York, the plaintiffs and defendants offered very different baselines for determining—based on statistics—whether the police were involved in racial profiling in conducting their stops and frisks. The defendants argued that the appropriate baseline is the proportion of African Americans and Hispanics among the population of criminal suspects. That is, to determine whether police stops and frisks depart from a race-neutral norm, the defendants said the right question is whether the proportion of stops and frisks that are of African Americans and Hispanics deviates from the proportion of criminal suspects in the relevant area who are African American or Hispanic.
The district court judge rejected this proposed baseline, explaining that almost all of the people stopped by the police turned out to be innocent, and we therefore have no reason to expect their demographic characteristics, absent discrimination, to mirror those of the criminal-suspect population. She accordingly adopted the baseline proposed by the plaintiffs, determining that we should be looking at the population demographics in the relevant area and at the crime rates in different neighborhoods, to identify what race-neutral stop-and-frisk numbers would look like.
I think the baselines that the court selected make sense. It turns out, as the judge explained in the body of her opinion, that New York police performed more stops in predominantly African American and Hispanic neighborhoods than they did in predominantly white neighborhoods, even when we hold constant the crime rates in the respective neighborhoods. If two neighborhoods have the same population size and the same crime rate, then it is indeed suspicious for police to be performing more searches in the African American and Hispanic neighborhoods than in the white neighborhoods.
Furthermore, the court explained, the stops and frisks within particular neighborhoods were, as a separate matter, disproportionately visited on members of minority groups, and those stops and frisks tended to be less productive (of actual criminal suspects) than the stops and frisks of white targets.
I agree with the court that all of this is strongly suggestive of racial discrimination. However, I nonetheless see some merit in the defendants’ proposed baseline as well, and I think the court perhaps should not have dismissed it as an additional source of useful information. Its dismissal may betray a conflation of two separate inquiries.
Making a Decision Versus Judging a Decision
From the perspective of a police officer deciding whether or not to stop a potential suspect, it is impermissible for her to take the suspect’s race into account, except in limited circumstances. It would be acceptable to consider race where there is a relatively precise witness identification in connection with a particular crime, and race is one of the identifiers. Say a victim was robbed and told the police, “My assailant was a white male with red hair, green eyes, and very pale skin, between 6 feet and 6 feet 2 inches tall, with a tattoo of the sun on his left arm and missing his right index finger.” With this identification, it would be appropriate for police to rule out all African American suspects and consider whiteness to be one relevant feature of reasonable suspicion.
It would, by contrast, be unacceptable for a police officer to allow the race of a potential suspect to count as a factor in the decision to stop him or her simply because the crime rate reflects a racial disproportion that points statistically to members of the potential suspect’s race. In other words, even if one group is more likely to offend than another group, membership in the first group does not thereby become a legitimate component of reasonable suspicion to stop a particular suspect. To stop the suspect because of this statistical disparity violates equal protection. And this is true even if the racial disproportion in criminal conduct makes it objectively much more likely for a suspect of one racial group to be committing a crime, than for a member of another racial group to be doing so.
Though a stereotype may roughly correspond to statistical reality, official conduct stemming from the stereotype is nonetheless both unconstitutional and grossly unfair to the vast majority of persons in each racial group who are innocent of criminal activity. The same, incidentally, is true for gender and other stereotypes that attach to suspect classifications—their unconstitutionality does not turn on their bearing no relation to statistical reality.
Evaluating the legitimacy and motivation behind prior police conduct, however, is distinct from making a forward-looking decision about whether to stop a suspect. It is impermissible for an officer to decide to stop a person on the basis of the person’s race, no matter what the disproportion in crime rates might be. A disproportion in crime rates, however, could be quite useful, after the fact, in determining whether officers were or were not engaged in stopping people on the basis of race. If the rate at which police stop people of various racial groups roughly corresponds to the rate at which members of the various racial groups engage in criminal activity, then that correspondence could, in the absence of other evidence of discriminatory motive, suggest that police are stopping people in a race-neutral fashion.
To see how this might be true, return to the example of the teacher who calls on male students much more frequently than he calls on female students. One could imagine the following facts: the teacher calls primarily on male students even when the class is evenly divided between boys and girls; the teacher calls on students more frequently in general when he teaches a class that is primarily made up of boys than when he teaches a class that is primarily made up of girls; and even in the classes in which there are more girls than boys in attendance, he still calls on boys a majority of the time. All of this is surely suspicious. However, if it also turns out that boys are, on the whole, much more likely to raise their hands than are girls, then it would be plausible to conclude that the teacher in question might not be engaged in gender profiling.
To the extent that individuals from one group are more likely to engage in criminal activity, by analogy, it would not be surprising to find that members of that group would also be more likely, even without racial profiling, to be stopped by police for suspicious behavior, even though most of the people stopped turn out to be innocent. Non-racial indices of suspicion might correlate with race, just as hand-raising might correlate with gender.
Nothing I say here is meant to dispute the New York Police Department’s responsibility for permitting, encouraging, and even pressuring police to carry out both unreasonable stops and frisks and racial profiling in violation of the Fourteenth Amendment. There is plenty of evidence in the record to support such responsibility. This includes criticism and discipline by superiors of officers who did not stop “enough” people, express suggestions by superiors that police could articulate reasons for stops after the fact, and not-too-subtle suggestions that police should stop the “right people,” with the implication that members of minority groups—and those who live in neighborhoods in which minorities live—should be subject to stops and frisks even when they have done nothing wrong. And the statistically significant emphasis on minority neighborhoods—regardless of crime rates—cannot easily be neutrally explained. It is also significant that little or no attention is paid by the NYPD to reviewing and ensuring that stops occur only based on reasonable suspicion.
I therefore do not quarrel with the district court’s conclusions regarding the unconstitutional behavior of the police and of the police department responsible for ignoring and even encouraging unjustifiable stops frequently aimed at members of minority groups.
Nonetheless, I point out the complexity of baselines, because the problem extends beyond police conduct to, for example, what happens in classrooms. Some personal and demographic characteristics correlate with behavior that is of interest (such as criminal conduct, or a tendency to perform well in school, or an interest in volunteering in class). It is also the case that some personal and demographic characteristics correlate with behavior that is not of direct interest but that may itself correlate with the behavior that is of interest. As a result, one can be operating in a demographically-neutral fashion and nonetheless accurately identify behaviors that correlate with the behavior of interest, resulting in seemingly disproportionate but neutrally intended actions for or against people belonging to a particular group.
My view is that because of these correlations, disparate treatment is generally an inadequate measure of equal protection violations. If police or other official practices disproportionately harm members of one group, then the fact that there is or could be a race-neutral motivation for the practice should not necessarily insulate such practices from constitutional scrutiny. Women and minorities should not have to tolerate practices that are discriminatory in effect, even if they are neutral in intent.
Even if police were acting without any racial bias whatsoever, then, it is still high time that the Supreme Court acknowledge that race-neutrality is something quite different from equal protection, and that the Constitution demands the latter. The “equal protection” of the laws ought to mean that the law and its representatives must treat African Americans and Hispanics as well as it treats white people. Unfortunately, the current Supreme Court is as far from this vision of equal protection as it has been in decades.
