Actually, this would seem more like a failure of racial profiling. Statistically serial killers are white and domestic terrorism and threats of domestic terrorism (Liddy’s aim for the head comments) with a military slant prior to 9/11 were overwhelmingly angry white men. So a racial profiler would be making the right decision to go after angry white men rather than going after black men when pursuing a serial killer case. Such a profile you describe would be equivalent to stopping old ladies or frat boys for airline searches.

What was at play here was the same bureaucratic hesitance that kept the FBI from accepting the famous Phoenix memo. There was simply no one in charge then who wanted to take that memo seriously enough to give the order to “watch the Arabs in our midst, especially those in flight school”. In the summer of 2001 that would have ended his or her career. And with the DC Snipers we once again have a case in which political correctness kills.

Are serial killers notably more white than the population at large? There have been some spectacular black serial killings (the Zebra killings, the Atlanta child murders and the Yahweh bin Yahweh killings come to mind). As to terrorism, it seems the Muslim involvement has recently been quite disproportionate to their presence in the population.

I hate to cite J*n*h G*ldb*rg in this venue, but he passes on something a research assistant dug out for him, that out of 514 sniper murders between 1976 and 2000, 55% of the murderers were white—which makes whites underrepresented. The column is at http://www.opinionjournal.com/taste/?id=110002526 .

Still, it isn’t the proportion that matters. It is our absolute expectation in a given incident. If 55% of the time a sniper is white then, given no other information at all, we are justified in having the expectation that any given sniper is 5% more likely to be white than nonwhite. Usually it is not on such slim odds that legitimate attempts at racial profiling is based, though, and there are other factors such as the overt jihad currently underway that sensible people might think more important than the raw statistic that a white sniper is 5% more likely in absolute terms than a nonwhite one.

To illustrate this more concretely: suppose that albinos represent 1% of the population, and yet 10% of snipers on a historical basis were albinos. That means that albinos are disproportionately represented among snipers by quite a significant rate, but still the odds of any given sniper being an albino are small.

My observation of absolute expectation (which is what we saw in the media) stops making sense if we go beyond absolute expectation and into a screening process, just to clarify. If 90% of albinos are snipers then questioning all of them makes sense even if they are a small part of the population. So I guess the answer to the question about racial profiling is that what the media did was not racial profiling at all.

But a profile I thought was a list of characteristics that tell you what to look at more particularly. They are characteristics that inform you as to the relative likelihood that someone you are looking at is the perp. So if 30% of the population is nonwhite but 40% of snipers are nonwhite then if you’re looking at someone the fact he’s white makes it (slightly) less likely he’s the perp. So “whiteness” should not be part of the profile. It’s not a characteristic that makes it more likely the person is the sniper, or that helps you pick out the sniper from the general run of the population.

Right, what I realized after my first comment was that the media expectation wasn’t a racial profile at all, as RC had suggested it might be.

An absolute expectation about a population is fundamentally different from a specific expectation about an individual. It is not contradictory to expect overall that the sniper will be white (which we should have, by a 5% margin, excluding any other info), but at the same time, when looking at a particular white person and a particular nonwhite person, for there to be a (significantly) greater likelihood that the nonwhite is your perp. That seems counterintuitive at least to me but it is a direct consequence of shifting perspective from the overall population to a given actual person.

There is a reason this observation matters, in addition to being interesting for its own sake. The fact that our overall expectation might reasonably be that a perp will be white does not invalidate racial profiling of nonwhites as a prudential matter in order to drastically increase our chances of catching the actual perp. I suppose this may be already obvious to many folks despite its oddity, but it was not to me and may not have been to others.

Put it this way:

“We expect the perp to be White. We are randomly questioning Blacks and only Blacks to increase our chances of catching him” is (or can be, depending on the numbers) a perfectly rational law enforcement approach.

Sure. If 40% of the time the perp is a black, and there are a million whites but only a thousand blacks, then that could be a rational approach especially if you were starting without leads.

There is a very good book called “Mind Hunter” written by John Douglas who, along with a few other people late in the Hoover FBI started the science of criminal profiling. According to Mr. Douglas serial killers are, statistically white males although there are many exceptions to the general rule. Criminal profilers look at a mind-boggling number of variables and of which Race and sex are but two traits. He was involved in the Atlanta child killer case and his profile pointed to a black male in contradiction to many who had the killer pegged as a white male do to the overwhelming statistical facts that serial killers are white. His reasoning was that it would be unlikely for a white person to move thought the large all black neighborhoods of Atlanta without witnesses recalling a white person I the area around the time of the abductions. It really interesting to see how well Douglas’s profile matched up to the killer Wayne Williams.

It is interesting that in the sniper case there are a number of distinct differences between their Modus Operandi and that of a serial killer. Typically a serial killer kills to satisfy some kind of psychological need. Everything from the victim selection to the manor of the kill is carried out in order to satisfy this need. Serial killers never work in teams. This case is more appropriately a mass murder, which yields a very different profile. Typically there is another quality called signature, which is a different concept from MO. MO is what the perp does to carry out the crime; signature is what the criminal must do in order to satisfy a psychological need.

In the book Douglas also describes how authorities will purposely put out disinformation through the media in order to mislead the perp. I’m convinced that the focus of the “white van” was an intentional tactic. It’s a fascinating book.

There is another thing about serial killers. They often have the ability to win the confidence of their victims. The DC shooters did not do this.

If “serial killer” theory applies to people who kill to satisfy a very specific psychological need, and who typically can get the trust of their victims, then maybe it shouldn’t have been applied to killings of a very varied group of people from a distance with no possibility of interaction between murderer and victim. Also, the nature of the killings—the quick unremarked shot followed by the instant disappearance—suggested all along he had an accomplice to stand guard and drive.

Mr. Kalb makes a good point. The lable of “serial killer” in this case seems to be mis-applied. I am going to revist the book I mentioned becase Douglas mentions some of the differances btween a serial killer and a mass murderer.