Sample Bias

A conundrum which once much puzzled the author is the following: Certain nationalities are, in the author’s experience, an absolutely marvelous lot, intelligent, inquisitive, open-minded, and disinclined to any major political pathologies. Yet, their large home countries are historically and often at-present in a state of repression and poverty which seems entirely incompatible with the nature of this sort of people. These countries ought to be Switzerland or Denmark, not the impoverished despotism they are. How can this possibly be?

TSA vs. My Two-Year Old Girl

This blog post merely recounts one more encounter with the Transportation Security Administration. Like so many other reports, it is merely an anecdote, rather than statistical proof of the thuggish ways of that agency. Yet, it strikes me as at least as cruel as several other incidents that recently have gained national attention, so it might be worth posting and reading.

A Bad Argument Against Terrorist Profiling

In an unfortunately over-elaborate comment on another fine blog, I responded to an argument against using racial, gender, religious, and age profiling against terrorism (such as, for example, in airport screening). Namely, that if such profiling was undertaken terrorist organizations would just switch to using agents which don't fit the profile and we'd all be worse off. I've been repeatedly surprised that this argument arises not only in idle blog comments, but is also made by otherwise more sensible people (e.g., Bruce Schneier) in far more elevated forums. As I'd hate to deprive my devoted readership of several individuals of any pearl of wisdom, let me explain in revised form why that argument makes very little sense.

Before dealing with the argument, let's aside a silly straw man often combined with it: That our only choices are (1) complete disregard for evident surface characteristics or (2) that we rely on stereotypical characteristics of terrorists to the exclusion of all other evidence. Option 2 means performing cavity searches on every young man with a tan and allowing grannies to carry bazookas over their shoulders onto airplanes as long as they also wear a cross, so it follows that Option 1 is the only sensible.

That is—of course—nonsense. Nobody advocates ignoring other indicators of suspiciousness or that any individual, regardless of appearance should be given an automatic pass. The question is whether stereotypical appearance factors should never be given any weight in making any discretionary investigatory decision or that sometimes they should be given some weight—i.e., to use a racial/ethnic/religious/age/gender profile.

To start with what we know: The vast majority of operatives of terrorist organizations seeking to murder Americans (and others) identified from past attempts are young Arab Muslim men. And most of the rest are at least three of the four.

From this it follows that persons of this description must be among the most plentiful, available, and easy to use resources available for such organizations. In other words, this demographic has the highest ratio of reward (in terms of dead or terrorized infidels) to organizational effort (in terms of money, time, and other resources). If another demographic had a better ratio, the terrorist organization would already have switched to them and costlessly increased their effectiveness.

That does not mean that there aren't blond-haired blue-eyed Norwegian grannies who converted to Islam and are now all hot to kill for Allah—surely there must be. What it does mean is that they must be much rarer, harder to use, and generally more expensive means of murdering infidels. If Osama Bin Laden had a granny brigade at his disposal he would already have unleashed it, rather than relying on such a non-diverse and identifiable set of operatives.

So what happens if we institute profiling? Young Arab Muslim men become much more likely to get caught and hence less useful tools. At the same time grannies, given the same amount of total enforcement effort and hassle, become at least a little less likely to get caught and hence more useful tools.

These two effects, while in opposite directions, are not of equal magnitude: The terrorist organization is much more hurt by the declining effectiveness of its chief type of asset than it is helped by the increased effectiveness of a type of recruit it already hardly used because of other reasons. So while there is presumably a shift in the terrorist organization's allocation of efforts, it will on net be worse off (and we better off) after the shift.

How much this hurts them (and helps us) depends on how much more expensive (in the above sense) grannies were for the terrorist organization than young Arab Muslim men before profiling. If currently there was only a slight advantage to using the latter, then the net harm on terrorist organizations of profiling would also be small and so would the net help for us.

While that is possible, this seems unlikely. Terrorist organizations are likely to very much prefer using their current demographic (and hence be hurt a great deal by profiling) for several reasons.

First, there is a substantial individual dispersion within each demographic. If the difference in average between recruits from the two demographics were small, the distributions would likely overlap and grannies would be underrepresented in the sample of terrorist, but still a substantial presence. Instead, we see them hardly at all. That means that the difference in average between the two demographics is likely large and shifting from one to the other would impose a large cost on the terrorist organization.

Second, the difference in inherent characteristics between the demographics as to suitability as terrorist agents are likely to be substantial. While there likely are some socially contingent factors, others are so fundamental and significant that I'd estimate them to have a large effect. Lutherans are inherently less likely to consider the majority population of Europe and the U.S. to be infidels worthy of death than Muslims are. Arabs and Muslims today have a far greater cultural acceptance and valorisation of suicide attacks and even terrorism in general. Norwegians are far less likely to feel—rightly or wrongly—profoundly racially alienated from Americans than Arabs are. And throughout history and almost all known cultures, the young and men have always shown far greater propensity—again for good or ill—for the sort of physically dangerous but potentially glorious pursuits like suicide terrorism.

Finally, and most importantly, it does not matter if my predictions—admittedly based in part on broad cultural stereotypes—for the effectiveness of racial/cultural/gender/religious profiling are correct because the practice itself will demonstrate the correct point for stopping:

At the point of maximum harm to terrorist organization efforts to kill us—and for us the optimum amount of profiling—grannies and young Arab Muslim men and everybody else will have become equally (in)effective. At that point, we would expect to observe terrorist organizations to use all demographics equally. By no means should one pursue profiling beyond that point as it would only help the terrorists.

In other words, keep profiling until the demographics of apprehended terrorists look like America. If it is good enough for the U.S. cabinet, it is good enough for Al Qaeda. And just think how pleased their diversity management consultants will be!

More seriously, I've discussed a superior way to compensate the vast majority of innocent young Arab Muslim men (and others) who are inconvenienced by profiling twice before.

Racial Profiling Done Right: An Addendum and Response to Comments

The other day, I suggested a method by which law enforcement could use valid statistical inferences from subjects' surface characteristics (like age, gender, race, and so on) to focus their investigative attention without victimizing those who merely shared those surface characteristics, but were individually quite innocent: offer just compensation for those subjected to additional scrutiny and delay.

If that surface characteristic is apparent racial or ethnic background, the position of two main political tendencies in the U.S. is clear: The Left considers it an intolerable intrusion on the innocent which should be banned regardless of statistical validity. Most of the Right basically says "Though cookies if you happen to be (or look like) a member of any group disproportionately involved in criminal activity."

Neither of these positions seems morally or economically tenable. While race enjoys a different status in U.S. law than other surface characteristics,¹ the non-legal arguments for and against racial profiling are equally applicable to most other surface characteristics.

If your description resembles that of a recently reported bank robber, you are more likely to be questioned by the police, even though you are entirely innocent. Similar consequences await you if you drive a car resembling the getaway vehicle. And if your identical twin makes the FBI Ten Most Wanted list, you can look forward to years of police interference, even if you are the most law-abiding citizen.

The logical consequence of the Left position is that using any less-than-definite surface characteristic to focus investigative attention on specific subjects is wrong. In short, a law enforcement offer must have proof of specific law-breaking before even questioning anybody, lest an innocent be inconvenienced. Carried to this logical conclusion, the Left position would make police investigation of many or most crimes virtually impossible.

The logical consequence of the Right position is hardly better. If you happen to look like a particular criminal, or even look like you are a member of a group disproportionately involved in crime, you'll just have resign yourself to be hassled by the police more than others for the remainder of your days, even if you have never violated any laws yourself. In fact, you should not even be upset that you (and, depending on characteristic, your equally innocent family and many friends) must bear the cost of the enforcement of laws the benefit (assuming it is a sound law) of which is enjoyed by many others who have to make no similar sacrifices.

In short, the only solution which even remotely approaches fairness, efficiency, and logical consistency is something like the one proposed here.

Let me also respond to two anonymous comments criticizing the anti-tickets.

First, it was suggested that making law enforcement pay for the time they take from private citizens would lead to under-enforcement. That seems unlikely. Remember, the anti-ticket is not a social cost, but purely a transfer. The cost is created by stopping and hassling private persons. The question is whether this cost is better borne by whomever the police choose to stop or by the public. As the benefits of good law enforcement investigation accrues to the public (mostly, local) and the decisions whether and how many stops to make are made by public officials (again, mostly local) answerable to the same public, the incentives will only be properly aligned if that same public also bears the cost. If that public decides that the crime reduction caused by certain types of police stops just isn't worth the cost, they should stop. To say that they should continue, but the cost should just be sloughed off by the public onto whomever the police care to stop is perverse.

Second, and this may only be of interest to the legally inclined, it was suggested that the elimination of qualified immunity—a legal doctrine under which individual public employees are not liable for damages caused by their violation of the Constitution, unless it was or should have been extremely clear to them that their course of action was unconstitutional—would be a better solution than the anti-ticket. While elimination of qualified immunity may indeed be a good idea (the author is inclined to think so), it would hardly solve the racial profiling issue for a whole host of reasons:

1. Racial profiling, that is making law enforcement decisions in part on the basis of ethnic appearance, is—contrary to carefully stoked public perception—not unconstitutional (or even generally illegal). See, e.g., U.S. v. Vite-Espinoza, 342 F.3d 462 (6th Cir. 2003) (and cases cited therein).
2. Even if racial profiling was unconstitutional, it shouldn't be. Ethnicity is a bit of available and statistically valid evidence in many circumstances faced by law enforcement officers. To discard it would cause either more innocent people to be hassled, more genuine criminals to go unpunished, or most likely a combination of both. And anybody who argues that it is nevertheless so morally distasteful that it should never be engaged in should also offer an explanation why the same does not hold for the at least equally ubiquitous law enforcement practices of gender and age profiling.
3. Even if racial profiling was unconstitutional, enforcing that right via law suit against individual police officers is almost certainly useless. Very, very few people have time that is so valuable that it would be worth the enormous legal costs of bringing a suit for a lost quarter hour. And in that rare case that the damages would be so large as to justify the legal fees, the individual law enforcement officer (that is, the person against whom a suit would be made more winnable by the elimination of qualified immunity) would be nearly judgment proof.
4. Even if racial profiling was unconstitutional and there were no legal fees, any case would have to rest on proof that the individual law enforcement officer considered race, among other factors, in making the stop or arrest. Without a confession or other extraordinary stupidity, that is going to be nearly impossible to prove in any given case.
5. Finally, even if racial profiling was unconstitutional, there were no legal fees, and the plaintiff has bullet-proof statistical evidence of racial discrimination, any case would still likely be lost. Assume that the plaintiff could prove using a vast statistical study (financed somehow and with the data collected without cooperation of a likely hostile police department) that all else being equal members his or her ethnic group were stopped 50% more often than others behaving in an identical manner. Such a plaintiff would (and should!) still lose. Even if there is a 50% increase in stops due to racial profiling, two-thirds of those stopped would have been stopped regardless. Hence, it is more likely than not that any individual plaintiff would have been stopped anyway and hence suffered no damages due to racial profiling.

Some of the problems of statistical proof and legal financing could be solved by a class action suit. But what would be the remedy in such a case? If you believe that racial profiling is or should be per se unlawful, there could be an injunction. Even in that case, it is hard to see how that injunction could be enforced without a racial quota system for stops and arrests, a measure which even most opponents of racial profiling would blanch at. And if you believe that there should just be money damages for stops, why not skip the whole rigmarole of a law suit and achieve the same end more simply through the anti-ticket?

¹ Among the surface characteristics frequently used by police, gender comes the closest to race in legal status. Under the U.S. Supreme Court's interpretation of the Constitution, gender discrimination is nearly as difficult to justify as racial discrimination (i.e., nearly impossible) and all progressive legal scholars I've ever discussed the subject with support closing whatever gap there may yet exist. That leaves the issue, alluded to before, of why racial profiling is a cause célèbre to the Left, but the at-least equally prevalent gender profiling is a non-issue.

Racial Profiling Done Right

After our most recent excursions into areas of policy where even clear thought fails to readily yield an optimal solution, let's instead wade into a current controversy for which there is a simple solution which ought to be acceptable to all concerned but which never seems to be propounded: racial profiling. That is the practice, often statistically justified, of law enforcement to make discretionary decisions (such as whom to stop, question, or search) in part on the basis of apparent ethnic origin of the subject.

Let's set aside one issue: If racial profiling is used to enforce bad laws, as will arguably be the case under the recent Arizona immigration law, the outcome will be bad. But in this case, that is because the law is bad, and any effective tactic to enforce a bad law should be deemed equally bad. Hence, the argument here is with the law, not the enforcement tactic. So instead let's focus on using racial profiling to enforce otherwise-sound laws, like those against murder, rape, theft, or terrorism.¹

Attitudes towards racial profiling generally align along political lines: Conservatives, arguing rightly that racial profiling will often be a statistically valid and efficient way of focusing police resources while minimizing the total intrusion on private persons, mostly endorse it. Modern liberals, arguing rightly that most of the intrusion will be experienced by innocent members of some ethnic groups and that state action conditioned on ethnic status is contrary to this country's stated principles², are against it.³

But here is a solution which would allow law enforcement to capture the efficiency gains of racial (and other) profiling without victimizing anybody: Just compensate anybody whose time is taken by the government, just as you would if it involved their property. In other words, every time a police officer stops and questions a person or subjects them to additional delay, at the end of the encounter the officer must write a kind of anti-ticket for $X for each minute between the stop and the time the subject is let go (or, upon the development of probable cause, arrested). The subject in turn could turn the ticket to cash at any bank.

The beauty of this approach is that it properly socializes a public good, the search for, arrest, and detention of real criminals. But rather than imposing the cost of this public good disproportionately on minorities (ethic, gender, or age) the vast majority of which, including most who are stopped, are entirely innocent, the cost will also be born by the public at large. As a matter of general principle, policies will always be more successfully planned and carefully executed if their costs and benefits are borne by the same entity who controls them. In the case of public safety and police officers, the public at large is already the beneficiary and at least the best candidate for controlling law enforcement policy. As the benefits and control already lie with the public, so should the costs.

A few objections against this policy do not hold much water:

• The first of these is that it would cost too much. But the policy only costs anything in the most narrow government accounting sense. In the economic sense, it is no cost at all: just a transfer of money from one pocket to another. And in so far as this transfer improves incentives for efficient behavior by law enforcement, there is a general gain in social welfare.

  It is true that groups who ordinarily get stopped by the police less often (whites, women, the very young and the middle-aged and older) would on average end up with less money in their pockets while those who are stopped often would end up with more. But that is just a question of cost distribution. If those who now pay for police stops find the cost excessive, then it always was—it just previously came out of somebody else's pocket. In that case, police procedure should have been adjusted all along.

• The second objection is that it would pay even those arrested and convicted, that is, usually real criminals. That is true, but less significant than it seems. Waiving the anti-ticket if there is an arrest (or allowing any other sort of waiver) would just create too big an incentive for an unethical cop who stopped an innocent to offer an implicit or express bargain: You waive the anti-ticket and I won't fabricate cause for arrest (e.g., break your tail light).

  As a practical matter, when there is a bona fide arrest and prosecution, the anti-ticket would just be offset against court fees and fines which otherwise would be waived for the typical indigent or judgment-proof defendant. So unless the detention was so long that the anti-ticket would be larger than these fees and fines, there would be no net payment to criminals. And in the rare case of an excessively long detention for a trivial fine smaller than the anti-ticket, a net payment to the "criminal" would not be wrong.

• Next, some may complain that this would just be a license for racist police officers to indulge their preference by hassling innocent minorities. Maybe so, but that seems unlikely. The innocently stopped would all be compensated, so this really amounts to no more than a license for racist police officers to hand out free money to ethic minorities—one that is unlikely to be exercised once the cop realizes that is what it amounts to and, even if it did occur, would be less than tragic for its victims.

  Moreover, the anti-ticket creates its own audit trail. The perhaps mythical, perhaps real racist cop who just likes to inconvenience innocent members of disliked groups to gratify his or her own animus, would quickly run up a large tab of anti-tickets without any record of arrests and convictions to show for it. That should serve as at least some deterrent to such behavior.

• Finally, it may be argued that while the anti-ticket could compensate for lost time, there still remains an element of uncompensated emotional or dignitary damage. I am not entirely unsympathetic to this concern. Having been stopped by the police a few times in my life, introspection suggests that given a choice between any plausible amount received for a hypothetical anti-ticket and not having been stopped in the first place, I'd still choose the latter. But then this author is reasonably well off and may be more sensitive to being subject to involuntary questioning than the average person.

  That said, it seems hard to see how the anti-ticket would not still be a, perhaps incomplete, improvement. There are few ways in which strangers can convey their apologies for having bothered you more convincingly than cold hard cash, going at least some way towards easing the dignitary insult. In addition, any other remedy for abusive police conduct, like unjustified physical assault, would remain in place. The anti-ticket would be on top of such other remedies and purely for the currently uncompensated injury of lost time.

The most serious, but hardly insurmountable problem with the anti-ticket proposal is getting the amount right: Too small and there will be no effect; too large and you encourage an industry of people deliberately acting suspiciously while not actually committing any crime in order to profit off the over-generous anti-tickets. What's more the amount cannot be tailored carefully to the situation if it is to remain administrable and enforceable. Anything more variable than a fixed amount per minute of delay would probably be too complicated.

That said, an amount on the order of $1/minute of delay seems about right. It would compensate most individuals at a rate higher than their marginal cost of time (that is, at the equivalent of a wage of $60/hour after tax), while still likely preventing most schemes to be stopped for profit from being lucrative. Perhaps, legal suits should be permitted (by either police or subject) where the plaintiff would bear the burden of proof that the compensation would either be grossly inadequate or excessive in a particular case; given the transaction costs of the legal system, these would presumably be rare.

So what do my gentle readers think?

¹ There is the argument abroad that racial profiling cannot work against terrorists. Even though some persons worthy of respect make it, I do not believe that it is a very strong position under real-world conditions. But that is an argument for another post.

² It is unclear on what principle the at-least equally common and equally justified law enforcement practices of gender profiling (i.e., focusing their attention on men, rather than women, on the basis than men disproportionately commit violent and, to a lesser degree, non-violent crime) and age profiling (i.e., focusing on subjects in the prime crime ages of 15 to 35 years or so) are not equally controversial. Discarding the requirement of principle, one could speculate.

³ It is hardly original, but worth noting that this is an 180 degree reversal from the position generally taken by these two camps when it comes to the distribution of benefits, such as university admissions or hiring. Some modern liberals will quite frankly admit that they feel that the history of this country justifies a thumb on scale for certain previously disfavored, but now favored groups. Hence color-blindness is not a principle for them but merely a convenient stick to beat up policies harming the groups they favor equally conveniently discarded when it comes to policies where it points the other way. This at least has the benefit of honesty. Most conservatives fail to offer even this much of an explanation for their reversal and one cannot help but wonder if their commitment to color-blindness is confined to cases where its absence inconveniences them and theirs.