Context Is Everything with Racial Profiling

Every day brings new headlines about racial or ethnic profiling. A Secret Service agent who guards the president is not permitted to board an airplane because he is Arab American and cannot satisfy the screeners about his identity. Two white New Jersey state troopers plead guilty to shooting into a van containing four men--three blacks and one Latino--after being instructed to single out such people as drug suspects. These shameful incidents shock our sense of justice. Can anything be said for profiling in a democratic society of equal citizens?

Let’s begin with our values as a society. We should be wary of claims that we must sacrifice our ideals in the name of national security. The ideal most threatened by profiling is the equality of all individuals before the law. Differential treatment must meet a burden of justification--in the case of racial classifications, a very high one. Government may not treat individuals arbitrarily, but must base its actions on information reliable enough to justify its exercise of power over them.

How good must the information be? The law’s answer is that it depends. Criminal punishment requires proof beyond a reasonable doubt, while a tort judgment demands only the preponderance of the evidence. Health agencies must often act with little more than a rational suspicion that a substance might be dangerous. Information good enough for one kind of decision, then, is not nearly good enough for others. The same police officer needs better information for an arrest than for a traffic stop. Context is everything.

So, what should the airport screener do? What would we do in his or her place? Each day, we all face choices that are similar in structure, albeit far less consequential. We must make decisions rapidly about things that matter to us. We know our information is inadequate to the choice, but we also know that we cannot, in the time available, get information that is significantly better. We might make errors of different kinds and know that some would be worse than others. Because we must momentarily integrate all this uncertainty into a concrete choice, we resort to shortcuts. The most important and universal of them is the stereotype, which, by economizing on information, enables us to decide quickly to inadequate information. This problem of limited information is so ubiquitous that we scarcely notice how often we use stereotypes to solve it.

We could not live without stereotypes. They help us predict how others will behave (we assume blacks will vote Democratic, though many do not) and to anticipate what others want (we offer help to disabled people, though some of them find this intrusive). They alert us to possible danger, as when a large, unkempt, angry-looking man approaches us on a dark street (though he may simply be asking directions). Such assumptions are especially important in a mass society where people know less and less about one another.

Stereotypes have an obvious downside: They are sometimes wrong, as my examples show. After all, if they were wrong all the time, no rational person would use them, and if they were never wrong, they would be indisputable facts, not stereotypes. Stereotypes fall somewhere in between these extremes, but it is hard to know precisely where, because we seldom know precisely how accurate they are. Most are probably correct much more often than not; that is why they are useful. But when a stereotype is wrong, those who are exceptions to it naturally feel they have not been treated equally as individuals. They are right, and we share their sense of injustice. When their indignation is compounded by official discourtesy, animus or violence (as in New Jersey), the wound is deeper still.

This is where the law comes in. When we view stereotype-based injustices as sufficiently grave, we prohibit them. Even then, however, we do so only in a qualified way that expresses our ambivalence. Civil rights law, for example, proscribes racial, gender, disability and age stereotypes. Yet it allows some public-interest or business reasons to justify them. Religious groups can hire only co-religionists. Officials drawing legislative districts can, to some extent, treat all members of a minority group as if they all had the same political interests. The military can bar women from certain combat roles. Employers can assume that women are usually less suitable for jobs requiring heavy lifting. Such stereotypes are thought to be reasonable in general, though false as to particular individuals.

Can the same be said of racial or ethnic profiling? Again, context is everything. We would object to a public college that categorically admitted women rather than men on the theory that women tend to be better students--not because the stereotype is false, but because the school can determine academic promise on an individualized basis when reviewing applicants’ files, which it must do anyway. On the other hand, no one would think it unjust for the FBI to screen for Osama bin Laden, who is a very tall man with a beard and turban, by stopping all men meeting that general description. The stakes in apprehending him are immense, and gender, size, physiognomy and dress are valuable clues in making instantaneous decisions about whom to stop. Yet each person stopped is statistically likely to be a false positive and to feel unjustly treated for having been singled out.

Racial profiling in more typical law-enforcement settings can raise difficult moral questions. Suppose that society views drug dealing as a serious vice, and that drug dealers are disproportionately black, although many dealers are white. Would this stereotype justify stopping blacks simply because of their color? Clearly not. The law properly requires more particularized evidence of wrongdoing.

Suppose further, however, that police officers observe a black man engaging in the furtive behavior that characterizes most but not all drug dealers--behavior also engaged in by some innocents. Legally, this stereotype would justify stopping the man. But what if the officer relied on both stereotypes in some impossible-to-parse combination? What if the behavioral stereotype alone had produced a very close call, and the racial one had pushed it over the line?

Although I cannot answer all of these questions, we need to begin asking them. A wise policy will insist that the justice of profiling depends on a number of variables. How serious is the crime risk? How do we assess the relative costs of false positives and false negatives? How accurate is the stereotype? How practicable is it to pursue the facts through an individualized inquiry rather than through stereotypes? If we must use stereotypes, do some rely on less incendiary and objectionable factors?

Stereotypes that seem reasonable when deciding whom to screen for questioning may be unacceptable at the arrest and prosecution stages when more individualized information and procedural safeguards become available. Screeners can recognize the many exceptions to even serviceable stereotypes, and can behave accordingly--so that those being screened understand the reason and do not to take it personally.

Profiling is bound to be part of the post-Sept. 11 dispensation. Clearer thinking and more sensitivity to its potential uses and abuses can help produce both a safer and a more just America.