Robert Todd Carroll
about the newsletter
Skeptic's Dictionary Newsletter 15
November 9, 2002
(Past issues posted at http://skepdic.com/news/)
1) New or revised entries in The Skeptic's Dictionary & Skeptic's Refuge
Since the last newsletter, I have
I didn't get a mandate from readers of the last newsletter regarding putting warning stickers on Bibles (contains myths, not scientific theories, etc.), but I did get an offer from Bruce who said he'd put them in Gideon Bibles when traveling.
Mark suggested that I put an entry on myself in the Skeptic's Dictionary, to remind readers "to look sideways at the skeptic as well as at everything else." I'm not sure what it means to look at something "sideways," but if he means readers need to be reminded to be skeptical of what I write, I don't think they need to be reminded. They remind me of my faults to a fault. I've learned to handle most of the abuse (that little trash can icon is a blessing), but I still get hot and bothered when it is pointed out to me that I've made another major error. (I get over it, though, as soon as I correct it. Then I feel good and confident for another ten minutes, until the next blunder is revealed.)
I've received several inquiries like the following from Gina Martin.
I'd begin with Profiles in Injustice: Why Racial Profiling Cannot Work by professor David A. Harris. His book thoroughly debunks racial profiling--using race or ethnic appearance to predict criminal behavior. However, there will be serious problems with any kind of profiling.
The data itself is inherently unreliable. Many, perhaps most, crimes are not reported. Data can only be gathered from reported crimes. The profiler must assume that those who commit crimes that are not reported are significantly similar to those who commit crimes that are reported. The profiler's assumption could be right, or it could be wrong, but there is no way to ever find out. But this should not be too serious a problem for most of the kinds of crimes profilers would be interested in, e.g., the serial rapist, the serial killer, the burglar operating within a fixed geographical area.
Secondly, not all reported crimes end in arrests. The profiler must make another assumption, namely, that those arrested for reported crimes are significantly similar to those who are not arrested. Again, there is no way to ever find out if this is true.
Thirdly, not all those arrested for reported crimes are convicted. They may be innocent or there may not have been enough evidence to convict them even though they were guilty.
Fourth, not all those convicted are really guilty.
Fifth, some crimes are so rare that the data will always be sparse, yet this may not prevent the profiler from speculating with confidence.
Finally, applying a profile has to be done without knowing whether the data you have collected regarding crimes committed by the one you are profiling is complete. For example, it was only after the Beltway killers were caught that investigators found out that they probably committed several other murders in several other states.
Thus, for every profile there is data in the database that shouldn't be there and data that should be there is absent.
Profiling is more guesswork and art, than science. It seems that there is likely to be a great deal of wishful thinking, self-deception, selective thinking, and confirmation bias among the profilers and their advocates. Thus, one should not make too much of highly touted success stories. All the evidence has to be considered when evaluating the effectiveness of profiling in lessening the time to apprehend and convict lawbreakers. When all the evidence is considered it will appear beyond a reasonable doubt that profiling is another law-enforcement pipe dream on par with the polygraph, voice stress analyzers, brain scanners, the Quadro Tracker, the DKL Lifeguard, and psychic detectives.
A few examples might help cast some reasonable doubt on the efficacy of profiling. What profile did the FBI use when they interrogated Cary Stayner about the Yosemite murders? Whatever it was, they let him go because he didn't fit the profile. He then went on to kill Joey Armstrong. (Gina has already reminded us of the Richard Jewell fiasco.) Some think the investigation of Steven Hatfill is another example of profiling gone awry. Another example is that of Edward Humphrey, who was all but tried and convicted by the police and the press for a series of horrible murders of young women in Gainesville, Florida. Eventually, Danny Rollings, the real killer, was apprehended. The police were using Ted Bundy as their model to profile this killer.
In the October 21, 2002 issue of Newsweek there is a graphic on page 27 regarding geographic profiling and "hunting" the "Tarot card killer." Eleven attack sites are plotted and we are told that "a map of the killer's probable home base is created with a program that takes into account known patterns of serial killers." But we now know that these killers lived in their car and had a mobile home base. Furthermore, we now suspect that these 11 crimes provide an incomplete list of the total number of shootings perpetrated by Muhammed and Malvo. This illustrates one of the problems with geographic profiling: you don't know until after the fact whether you have all the data. (When the two were caught while sleeping in their car, they were nowhere near the area the profilers had designated as being the "most likely" place for their home base. They had no home base.)
On the other hand, behavioral profiling would have eliminated half the human race as suspects in the search for these snipers. Women serial killers don't use sniper tactics on strangers. It was a pretty safe bet that the sex of this killer or killers would be male.
In another case, the profilers were way off the mark in their predictions. The FBI profile of the Unabomber (Ted Kaczynski) was
Outside of being right about his being a white male and a loner with no girl friend, this profile does not strike me as being "strikingly close to Kaczynski." Yet, that is how the Sacramento Bee described the FBI profile in its headline on April 7, 1996. There were also media reports that claimed that the composite drawing of the Unabomber released by the FBI was a good likeness of Kaczynski. Judge for yourself.
Next, consider something like trying to profile a serial killer by gathering facts about known serial killers. John Wayne Gacy, Ted Bundy, David Berkowitz, Angelo Buono, Ken Bianchi, Richard Speck, Edward Gein, Tom Ridgway, Richard Ramirez, Ted Kaczynski, Jeffry Dahmer, Aileen Wuornos, Dorothy Puente, Juan Corona, Carlton Gary, Wayne Williams, and Andrew Cunanan were all serial killers. Most of them were white men. Buono and Bianchi worked together, but most serial killers don't have partners. After collecting as much data as is available about serial killers, the profiler concludes that most serial killers are white males who act alone and are in their late 20s to early 30s. But, there won't be just one type of serial killer, so there won't be just one profile for the serial killer; there will have to be several. The danger is that a profiler will apply one of the profiles too early in an investigation and not only prevent the real killer from being caught sooner rather than later, but may well be driven by confirmation bias to track down and find the evidence necessary to arrest and convict an innocent person who happens to fit a good part of the profile the investigator is working from. It may even ultimately be the case that the profile serves more to convince the investigator that he or she is on the right track, than to actually aid in the arrest and conviction of the guilty. It is only after the fact that we can ever be confident that the profile and criminal fit like a hand in a glove, was partially correct and partially incorrect, or was totally off the mark. To assume you know what kind of person committed a crime before there is adequate evidence to justify that conclusion can have serious consequences. Because of this attitude, innocent people have been treated as if they were guilty, guilty people have been allowed to go free and commit more crimes, and some poor people have ended up dead.