Technology

Jag” and “James Bond” are bloodhounds. They drool a lot but they’re nice dogs. And if you believe their caregiver, Fort Bend County (Texas) Deputy Sheriff Keith Pikett, they’re also CSI specialists, with a sense of smell so keen and an intellect so refined that, far more than just following a scent, they can match suspects to crime scenes and accurately convey their findings.

Michael Buchanek knows these pooches only too well. One day in March 2006 the retired Texas sheriff’s captain answered his door. It was deputies from his old agency, armed with a search warrant. Buchanek’s neighbor Sally had been found strangled in a field five miles away, and Pikett’s dogs had supposedly followed a scent from the rope used by the killer to Buchanek’s home.

Using dogs to track scents is old news. Deputy Pikett and other practitioners of “scent lineups” go it one better. They set up cans in a field. One contains something of the suspect’s, say a shirt, while inside the rest are items belonging to others. Dogs are exposed to a scent from the crime scene and then walked around the cans to see if they alert.

Pikett had been running these tests throughout Texas, where his methods were considered good as gold. He did it this time and reported that, yes, a dog alerted on Buchanek’s can. Convinced that their former colleague was a killer, detectives pressed him to come clean. But Buchanek had come clean. He didn’t kill anyone and wasn’t about to falsely confess.

Buchanek went through hell for five months. Luckily for him, police finally found the real murderer, who pled guilty. Victoria County Sheriff Michael O’Connor was unfazed. “We did the right thing, and the wrong person wasn’t convicted.”

A recent report describes Deputy Pikett’s unusual career. A college graduate with an undergraduate degree in chemistry and a master’s in sports science, Pikett became interested in bloodhounds. By the early 1990’s he was volunteering their services to Texas law enforcement agencies, at first for tracking, then for scent lineups. Although he lacked training in dog handling, followed no protocols and made wild claims of accuracy (his dogs were wrong only once in thousands of trials; they could identify scents many years old) his testimony helped win many convictions. Fort Bend County soon swore him in as a deputy. When a 2002 Texas appeals court opinion declared Pikett a bonafide expert his star rose higher. A Houston citizens’ group named Pikett officer of the year.

That niggling little misfire with Buchanek didn’t slow him down. In 2007 he helped Houston police arrest Ronald Curtis for a series of burglaries, and Cedric Johnson and Curvis Bickham for a triple homicide. Curtis spent eight months in jail before the real perpetrator was caught. Johnson was incarcerated sixteen months before he was cleared; Bickham, eight.

Pikett’s error-plagued sniff-a-thon continued. In early 2009 he gave Yoakum County authorities what they needed to arrest Calvin Miller for rape and robbery. When Miller was quickly cleared by DNA Pikett’s reputation finally began to tumble. In June 2009 a judge in Pikett’s own county ruled that his methods were unreliable. Bad news traveled fast, and everyone he wrongly fingered wound up suing Pikett and the agencies that used him.

Pikett isn’t the only cop charlatan who’s touted canines as ID machines. Pennsylvania trooper John Preston testified in more than 100 cases between 1981 and 1984. In 1981 he used a scent lineup to nail Florida murder suspect William Dillon. One year later his dogs linked another Florida man, Wilton Dedge to a rape. Both were convicted at trial. Decades later DNA proved their innocence; by then Dillon had served 27 years, Dedge, 22.

As scent evidence became more popular technology stepped in. Manufactured in California, the STU-100 “scent transfer unit” purports to suck human scent onto a gauze pad that dogs can sniff. This device was used in the investigation of James Ochoa, arrested in a 2005 carjacking after a bloodhound followed a scent from the vehicle to his home. Threatened with a long prison term, Ochoa pled guilty and got two years. Ten months later DNA proved that another person was the real culprit. Ochoa was released and awarded nearly $600,000. The STU-100 figured in the 1998 arrest of Jeffrey Grant for rape (held four months, he was cleared by DNA and awarded $1.7 million), and the 2003 arrest of Josh Connole for a string of arsons (held briefly, he settled for $120,000 after the real perpetrator was caught.)

Trained canines can track scents and detect vapors emitted by drugs and explosives. When the proof is in the pudding -- one either finds dope or a bomb, or not -- false alerts (and they do happen) can’t lead to a miscarriage of justice. But using a handler’s interpretation of their dog’s behavior as evidence is extremely risky. Lacking a scientific underpinning and validated performance standards, scent comparisons and lineups are nothing more than voodoo. Dogs aren’t calibrated instruments. As living things they are subject to many influencers, yet unlike their handlers they can’t be cross-examined. Could they have been affected by subtle, perhaps unintended cues from their handler? Might they simply have alerted in error?

In 2007, after spending two years locked up because he couldn’t make bail, Riverside County (Calif.) resident Michael Espalin went on trial for setting twenty-one brushfires. The prosecution’s principal witness, junior college biology instructor Lisa Harvey, testified that her bloodhound Dakota tracked a scent from a charred incendiary device to Espalin’s home. Dakota also supposedly matched Espalin’s scent to fire scene vapors collected with a STU-100. According to Harvey the dog could detect scents eight years old. “I don’t know how [scent] stays around for eight years. I just know that it does.”

Jurors didn’t buy her testimony, hanging 9-3 for acquittal. Harvey wasn’t used at the second trial, and Espalin was found not guilty. Taking a cue from Deputy Pikett’s victims, he’s now suing both Harvey and the authorities. One can only imagine how deeply taxpayers will have to dig into their pockets this time.

Orange County (Calif.) District Attorney Tony Rackauckas is a great fan of forensics. So much so, in fact, that he’d like to run a lab. Wouldn’t you know it, there’s one next door!

In 2005 Rackauckas got the Board of Supervisors to part with a cool $500,000 so that he could use DNA for property crimes. But rather than going through the Sheriff’s lab he contracted with a private forensics firm to do the work. Why? Apparently the Sheriff insisted on controlling the process, something that Rackauckas wasn’t willing to give up. Only thing is, CODIS, the FBI’s national DNA databank, only accepts profiles from government labs. No problem: Rackauckas entered into an agreement with the Kern County D.A., who runs his own lab, to upload the data.

Two years later, flush with an additional $875,000 in county funds, Rackauckas set up his very own databank. It’s accumulated the DNA profiles of several thousand misdemeanants and gang members served with injunctions. One of a smattering of “rogue” repositories around the country, the standalone database isn’t bound by State and Federal rules that limit DNA collection to persons arrested or convicted of felonies.

How does Rackauckas get offenders to contribute? Easy -- he “asks.” It’s an offer that many can’t realistically refuse. And now there’s an added inducement: going scot-free! Yes, that’s right. In exchange for $75 and a DNA sample his prosecutors are dismissing non-violent misdemeanors such as petty theft and drug possession. So what if a few cops get “demoralized”? As long as petty violators keep coming, what happens to them down the road seems to be of little public concern.

Just like his counterparts in Kern, Sacramento and Santa Clara counties, Rackauckas wants his own lab, or if not the whole enchilada, at least the sexy part, the DNA. His most recent attempt was in June 2008, while the Sheriff’s Department was reeling from the resignation of disgraced former Sheriff Mike Carona. Proclaiming his office as “the only organization capable of harnessing the vast potential of forensic DNA technology for our community,” he urged Supervisors to place DNA under him.

And he nearly succeeded. Rackauckas’ move was temporarily short-circuited, first, by acting Sheriff Jack Anderson, who pointed out that he wasn’t consulted, then by the new Sheriff, Sandra Hutchens, who was appalled -- appalled -- at the D.A.’s shameless power grab. A transplant from the far more tightly-wound L.A. County Sheriff’s Department, her recollection of the experience is almost touchingly naive:

But the struggle wasn’t over, not by a long shot. By the time that Hutchens’ outrage hit the papers the Supervisors had thrown Rackauckas a consolation prize, appointing him to a newly created Sheriff’s lab oversight panel. Its two other members are Hutchens and the County Administrative Officer, the latter clearly there as a referee. (Hutchens was so put off by the whole experience that she memorialized it in the official Orange County Sheriff’s Blog.)

Well, why shouldn’t the D.A. run a lab? In 2005 Orange County resident James Ochoa was arrested for carjacking. Ochoa, who had a drug record, was identified by two victims, and a bloodhound also followed a scent from a baseball cap left in the vehicle to his home. But the O.C. Sheriff’s criminalist who processed the cap and other items recovered from the car determined that the DNA wasn’t Ochoa’s. Her report displeased the head of Rackauckas’ DNA program, Deputy D.A. Carmille Hill, who marched into the lab and demanded that Ochoa not be excluded.

Her request was rebuffed. Still, the D.A.’s office wouldn’t drop the charges. Threatened by a judge with a stiff prison term if convicted, Ochoa was unwilling to roll the dice. He pled guilty and got two years. Ten months later the DNA was positively matched to a suspect in another carjacking. Oops! Ochoa got a $550,000 settlement from the cops and $31,700 from the State.

Concerns about such unholy influences prompted a National Academy of Sciences panel to suggest that labs be independent of law enforcement. To their credit, though, accredited labs subscribe to protocols specifically designed to prevent such pressures. But prosecutors who think they’re only there to convict could make enforcing safeguards problematic. Knowing just how unyielding D.A.’s can be when they’re convinced they’re right -- and the Ochoa case is a perfect example -- that’s an uncomfortable prospect.

DNA is also an expensive tool. A recent study of its use in property crimes estimates the average per-case cost of typing and entering profiles as $374 in Orange County, which processes DNA in-house, and $1147 in Los Angeles, which uses an external lab. (Evidence collection costs aren’t included). When there’s a possible hit DNA costs soar, averaging $13,000 per arrest in Los Angeles and nearly $20,000 in Orange County. And that doesn’t include the expense of creating and maintaining a DNA facility, nor of training and certifying investigators and examiners.

Supervisors have dumped more than one and a third million bucks into Rackauckas’ DNA programs. There’s no indication that their generosity was based on a comprehensive review of Orange County’s criminal justice needs. Maybe a study would demonstrate that a back-room DNA operation is a good idea. But giving someone money because of their political juice never is.

As lethal threats to police increase, protection languishes -- but there’s hope

It’s no surprise that Boston cops feel a chill. With criminals wielding powerful semi-automatic weapons whose rounds can sail through walls (and, as in an incident last week, pierce a mattress and strike a 12-year old girl watching T.V.) you’ve got to wonder why anyone would be so foolhardy as to pin on a badge.

Commenting on the tragic event, Boston’s commish bemoaned the proliferation of assault rifles, like the one that wounded the child. They are indeed a significant threat. But there are others. In March a parolee used an AK-type rifle to kill two Oakland SWAT officers who burst into the apartment where he was hiding. Police were there because the man had just shot and killed two patrol officers -- with an ordinary pistol.

And it’s not just “real” criminals who we should worry about. Consider the middle-aged Virginia Beach man who, angry over his eviction, opened up with an AK-47 and a MAC pistol, killing two and wounding three before taking his own life. Or the recent massacre in Alabama where a deeply disturbed 28-year old went on a rampage, slaying ten and wounding six. His weapons? A handgun, a shotgun and two assault rifles.

You’d think that with all the bullets flying around there would be a massive, Federally coordinated effort to improve ballistic protection for police. But you’d be wrong. Compared to the huge bundles of cash that get thrown at DNA, what’s spent on body armor R & D is puny. Firearms lethality has gone through the roof, yet what beat cops wear today -- when they can, if it’s not too hot -- isn’t much different in comfort and protection than what they wore decades ago.

Enough ranting. At the recent NIJ conference your blogger met someone who really knows what he’s talking about. S. Leigh Phoenix (he goes by Leigh) is Professor of Mechanical and Aerospace Engineering at Cornell University. On faculty since 1974, Leigh specializes in composite materials and high performance fabrics. Dr. Phoenix has designed composite overwraps for containers used in the Space Shuttle and space station programs. He’s also been working on ways to measure, predict and enhance the performance of police body armor. If you’re half as interested in keeping cops protected as he is, read on!

When a projectile hits it creates a small pyramid-shaped pocket. Soft armor, which is comprised of many fabric layers, tries to slow down the projectile by pushing back on it at the peak of this pyramid. The best analogy is to a tent, with the central pole representing the projectile. Applying tension to the sides of the tent drives the pole into the ground. As tension on the tent guys increases and the tent’s wall angles become steeper the force on the pole also increases.

When a continuous-weave fabric is struck by a bullet tension waves fan out in all directions along the yarns, traveling at more than ten times the bullet’s speed. Yarn material behind the waves feeds back towards the peak of the pyramid, allowing a relatively deep pocket to form with fairly steep angles (the steeper the better.) Normally the first few fabric layers will be penetrated, which slows down the projectile a bit. It’s the job of the remaining layers to bring it to a full stop.

Yarns used in body armor are more than five times lighter than steel, yet two to three times stronger. They must be very light, stiff and resistant to stretching. These characteristics allow tension waves to travel quickly; they also keep strands from breaking as they’re pulled into the pyramid. Fabrics must also be light, for wearability, and sufficiently flexible to resist crushing and shattering. Some of these factors work against each other, which complicates things.

Fabric works better when the diameter of the impact patch increases. When high velocity bullets with sharp points strike a plate their tips are blunted. Continued contact with the plate causes mushrooming and deposits debris, further reducing the projectile’s velocity. Current ceramic plates are completely sacrificed in the process.

The diagram depicts a hypothetical approach to stopping an armor-piercing rifle round using a combination of ceramic plates and soft armor. Here the “super” ceramic plate (4 mm. layer) has some flexibility and initially blunts the projectile, causing the lead inside the tip (dark area) to splay out. As the bullet continues its copper jacket slides forward and mushrooms and the interior steel core (large pencil-like region behind the lead) tries to push through, but you want to blunt that too, which takes a little more distance. A final fabric panel brings the slowed projectile to a full stop. This concept illustrates a basic tradeoff: you need distance to stop a projectile, but you don’t want to fill the needed space with heavy materials or the vest will be too heavy to wear.

With research and testing it could be possible to develop considerably lighter ceramics that can better withstand the rigors of the job

There’s another approach. At present all ballistic vest yarn is continuous, allowing material to be sent to the impact point. However, the first few layers are usually penetrated, accomplishing little other than some projectile slowing and blunting. It turns out that a single layer of unwoven yarn can be hit at much higher speed without breaking because it’s not loaded down by the drag of all the other yarns around it, especially as the pyramid deepens. In fact, a two or three inch length of the very strongest yarns can be hit at up to 2500 fps without breaking, even with a pointed-tip projectile.

I’ve given thought to using discontinuous yarns -- small segments, say two inches long -- for the first few layers, which instead of snapping would form a wad around the projectile’s nose as it plows through. That would increase the bullet’s frontal area, slowing it down and helping the fabric underneath do its job. Naturally, it would require a lot of development and experimentation to optimize fiber lengths and combinations. Calculations suggest that it could work with velocities in the 2400 fps range, which covers some rifle threats. Otherwise there will be a need for some ceramic, though maybe a lot thinner than what’s now used.

Helmets have a couple of limitations. First, they must float a distance from the skull so there’s room for deflection. They also lack wide, flat surfaces that can be covered with material to pull into the pyramid. So one can’t just take ballistic vest technology and apply it to helmets. But I think that it’s possible to develop a helmet that’s effective against handguns and light enough to wear on patrol.

What’s happened in the last twenty years to improve ballistic protection for police?

Really, not that much. Kevlar has been tweaked, yielding stiffness/strength combinations that marginally improve its velocity performance. A few new fibers have come out. Zylon, which was used on vests and seemed superior, is now on hold due to degradation concerns. Another fiber, M5, potentially much stronger than Kevlar, hasn’t gone commercial because of manufacturing or other problems. Two ultra high-strength polyethylene fibers, Dyneema and Spectra, are 50 percent lighter than Kevlar and just as strong and stiff. They’ve been used in cockpit doors. They may still be too expensive for wide use in vests but perhaps ideal for the helmets mentioned earlier.

Private industry has a big stake in body armor. Can’t we expect them to lead the way?

Body armor makers sell all they produce, so I don’t see major improvements happening under the present commercially-driven system. I know of an example where extensive manufacturing changes could make yarns stronger, but the company isn’t convinced that the investment would pay off financially. Manufacturers also hold their work very close to the chest. They have their own ideas, needs and priorities and collaborating with them is generally difficult, though I’ve been fortunate in one case.

Funds from government agencies like NSF and the Army are available if you’ve got the right buzzwords, meaning nanotech, biotech, carbon nanotube structures and so forth, but a lot of what gets proposed and funded is unlikely to lead to useful applications in the near term. Funding systematic work on something practical like body armor is difficult because those making the decisions (who never get shot at) consider the topic old-hat and think that the problems have been addressed and solved, which they certainly have not.

Federal law enforcement research dollars are spread very thin, especially when it comes to academic institutions. DOD concentrates on vehicle armor. Their successes are classified, making them unavailable to university researchers.

A lot could probably be done working with present fiber materials, tweaking them with improved processing to increase their strength without changing the basic chemical structure. You could change how fabrics are designed, say, by developing hybrid layered structures. Coming up with an altogether new material could yield big improvements, but we should not underestimate what clever manipulating can do.

To push the frontiers not just as a scientific exercise but with the objective of making significant, practical improvements requires a consortium of knowledgeable, technically-adept researchers who appreciate all the issues, including the need for comfort so that body armor actually gets worn. In other words, one must work on the whole package. We need resources for research and experimentation. We also need an agency or a group of agencies that would host a long-term, comprehensive effort to develop a new materials system that would yield armor that is more protective and comfortable.

Source for figure: S. Phoenix and P. Porwal, “A new membrane model for the ballistic impact response and V50 performance of multi-ply fibrous systems,” International Journal of Solids and Structures (vol. 40, 2003, p. 6724)

Three years in the making, the National Academy of Sciences’ anxiously-awaited “Strengthening Forensic Science in the U.S.: A Path Forward” is finally in, and it doesn’t paint a pretty picture. Although it’s clearly a product of compromise -- the National Institute of Justice reportedly opposed funding the study, then demanded a say in the conclusions -- the report has more than enough meat left on its bones to threaten the interests of labs and self-styled “experts” across the country.

At its most general, the study urges that forensic science live up to its name. Processes used to analyze evidence and make comparisons should be objective, set out in detail, reproducible by others and, as a topper, yield conclusions whose certainty can be accurately estimated, a requirement that places a big question mark next to virtually every identification technique short of DNA. Lamenting the ease with which junk science weasels its way into court, the report’s authors advise establishing a “National Academy of Forensic Science” that would guide research, set standards and certify labs and examiners. To keep unholy influences at bay, they also urge that labs function as independent entities outside the control of both law enforcement and private interests.

It’s a heady agenda that runs head-on into how forensic science is presently organized in the U.S. While many of the more ambitious objectives stand little chance of being implemented in the near term, the report’s disparaging views on some popular forensic matching techniques will surely be welcomed by the defense bar. Here is some of what Chapter Five, “Descriptions of Some Forensic Science Disciplines” has to say:

Fingerprint identification. The Grand-daddy of all identification methods comes under criticism, although not for its validity. (That fingerprints are unique between individuals, an assumption based on decades of observation, has apparently gained support from biological science.) Instead, the issue is reliability: does fingerprint comparison yield identical results across examiners? (For a brief depiction of the process click here.)

Crime scene fingerprints are often of poor quality, leading to subjective judgments that occasionally prove wrong. If the error is a false positive (saying that two prints match when they do not) such as what happened in the Brandon Mayfield case, and more recently at the LAPD crime lab, the consequences can be catastrophic. Meanwhile the identification community resists objectivizing its methods; for example, by using point systems based on minutiae, presumably because setting thresholds would yield fewer matches.

When examiners testify that two prints were deposited by the same person they do so to an absolute certainty. Yet, as the report points out, no judgment can be that “certain.” Indeed, it’s the ability to quantify the probability of error that’s the hallmark of a true science. Whether fingerprinting can be raised to that level remains to be seen.

Shoe prints and tire tracks. Impressions from footwear and tires have “class” characteristics, meaning patterns created during manufacture, and “individual” characteristics, reflecting everyday wear and tear. It’s the latter that are used to match a certain shoe or tire to a certain impression. Like fingerprints, the process is beset by subjectivity and lacks a numerical threshold for calling a match. Unlike fingerprints, it hasn’t been demonstrated that shoe prints and tire tracks are indeed unique, nor that they can be reliably distinguished.
Toolmarks and firearms. Again, class and individual characteristics are applicable. (For an example of firearms identification click here.) As in shoe prints and tire tracks, issues of subjectivity, “lack of a precisely defined process” and the absence of a threshold for calling a match present significant concerns. In 2008 a Michigan State Police audit revealed that Detroit police experts incorrectly matched guns to ammunition in at least three cases, including one that apparently caused a wrongful conviction. (Detroit PD’s entire lab was shut down and its functions were shifted to the State.)
Hairs and fibers. Matching hairs through their physical characteristics has been widely used in sex crimes. What the “experts” haven’t been letting on, though, is the abysmal error rate, with two studies citing false positives of about twelve percent, clearly excessive by any standard. These and other shortcomings led the NAS to declare that, lacking nuclear DNA, there is “no scientific support for the use of hair comparisons.”

More hope is held out for comparing fibers, whose chemical composition can be analyzed with existing tools and protocols. However, since little is known about the effects of manufacturing and wear, reliably matching fibers to specific garments or carpets remains out of reach.

Handwriting. There is some scientific support for the notion that individuals exhibit distinct handwriting characteristics and that these are relatively stable over time. Unfortunately, comparison techniques remain highly subjective, making their validity and reliability difficult to assess.
Causes of fire. Many arson convictions have relied on expert testimony that pour patterns, charring, glass crazing, etc. were caused by accelerants. But the origin of some of these fires, including one that led to an execution, were later shown to have been accidental. (For a brief discussion click here.) According to the NAS, long-accepted indicia of arson are plagued by poor science and subjectivity. Even so, “despite the paucity of research, some arson investigators continue to make determination about whether or not a particular fire was [deliberately] set.”
Bite marks. Bite mark evidence is occasionally used in the investigation of violent crime. Although an odontological dissimilarity might help exclude a suspect, the report concludes that the method’s scientific basis is “insufficient” for matching, and warns that its use has led to wrongful convictions.
Blood spatter. During Phil Spector’s first murder trial a defense expert testified that spatter could reach six feet, potentially placing Spector, whose clothes were flecked with blood, far from the gun (the barrel was in the victim’s mouth when it discharged.) As might be expected, a prosecution witness said that droplets could travel no more than half that distance. (For a brief discussion of the case click here.) Had it been up to the NAS neither witness would have taken the stand. Criticizing the opinions of blood spatter “experts” as overly subjective and driven by advocacy, the report concludes that “the uncertainties associated with bloodstain pattern analysis are enormous.”

What gets admitted into evidence is ultimately up to a judge. Federal practice, on which most State laws are modeled, is set out in Rule 702, Federal Rules of Criminal Procedure, “Testimony by Experts.” Before admitting scientific evidence, judges must determine whether “(1) the testimony is based upon sufficient facts or data, (2) the testimony is the product of reliable principles and methods, and (3) the witness has applied the principles and methods reliably to the facts of the case.”

In the era of C.S.I., with an entrenched forensic establishment that has elevated itself to a near-religion, not even an epidemic of wrongful conviction has managed to slow the choo-choo train of junk science. On the other hand, should defense lawyers take notice of the report, many of today’s quasi-scientific forensic techniques will pass into the realm of voodoo, where they’ve always belonged.

Here’s to their speedy demise.

Did you enjoy this post? Be sure to explore the homepage and topical index!

UPDATES

01/19/10 As the FBI continues reviewing 2,500 cases where analysts used a now-discredited bullet lead comparison technique, judges continue releasing persons convicted on their testimony.

09/10/09: Disturbed by the conclusions of the NAS report, Senators seem to support more funding for forensics research and possibly setting standards for crime labs and analysts.

08/24/09: Financial concerns and opposition by groups such as the National District Attorney’s Association make it unlikely that Congress will create an independent forensics oversight panel

05/17/09: Scientists react to N.A.S. report

Home Top Permalink Share this post

Posted 10/12/08

FORENSICS UNDER THE GUN

Commonly accepted techniques may lack scientific value

On February 17, 2004 Texas inmate Cameron Todd Willingham was strapped to a gurney and given a lethal injection. He had been convicted of arson and murder in a 1991 house fire that killed his three daughters. Evidence against him included the statement of a jailhouse informer who said that Willingham confessed and scientific testimony by the State Fire Marshal’s office that the fire was deliberately set.

Willingham protested to the very end that he was innocent. Now it looks like he might have been right. In August 2008 the Texas Forensic Science Commission agreed to review a 2006 report by five nationally recognized fire experts who refuted the “arson indicators” cited by Texas authorities at Willingham’s trial and said the fire was accidental. One of these indicators, crazed glass, was once thought to be evidence of a superhot fire fed by accelerants. It’s now known to be caused by spraying water on hot glass. According to the experts, another indicator, burn patterns in the floor suggestive of accelerants were meaningless in a fire that burned as hot as the one that destroyed Willingham’s home. And so forth.

In addition the Commission will also be considering the wrongful conviction of Ernest Ray Willis, who spent 17 years on death row for an arson/murder much like the Willingham case. While preparing to retry Willis (his case had been overturned on technical grounds) the prosecutor concluded that the State Fire Marshal’s “scientific” testimony was mistaken and that the fire was accidental. Willis was released.

On March 11, 2004 terrorist bombings in Madrid train stations killed 191 and injured two-thousand more. During their investigation Spanish police recovered fingerprints from inside a bag of unexploded detonators and furnished images of the prints to the FBI.

FBI fingerprint examiners digitized the images and ran them through the national database. They soon identified the prints as belonging to Brandon Mayfield, a Portland attorney who was Muslim and once represented a suspected terrorist in a civil case. Confident in their conclusions, the FBI ignored Spanish investigators who insisted that the prints didn’t match and that the bombers were Moroccan terrorists with no known connection to Al Qaeda or the U.S. On May 6, 2004 the FBI arrested Mayfield as a material witness in the bombings and searched his residence. Within days Spain positively identified the man who left the prints as a known Algerian terrorist. Two weeks after arresting Mayfield the FBI let him go. He got $2 million in taxpayer cash for his troubles.

How could this happen? Found or “latent” fingerprints are nothing like the complete, neatly inked fingerprints taken from job applicants and persons arrested for crimes. Instead, they’re often fragmentary, smudged, distorted and overlapping, which can make it difficult for examiners to identify the “minutiae”, the islands, dots, bifurcations and ridge endings on which comparisons rely. (In this example, a “good quality” latent is on the left, and the same finger inked is on the right.)

Every State and the FBI have large repositories of digitized fingerprint cards. The FBI holds prints for nearly one-hundred million persons, split about evenly between arrestees and applicants. Running recovered prints through these databases yields cards with the closest matches. It’s up to local examiners to order those of interest and microscopically compare them to the latent to see if there’s a fit. Generally at least seven minutiae must match, while only one inconsistency disqualifies. Extrinsic factors such as investigator’s suspicions must never intrude on an examiner’s judgment; if they do, as what apparently happened in Mayfield, the examiner (in the FBI’s case, several examiners and their boss) might mistakenly “find” matching minutiae in the latent that simply aren’t there.

Firing a weapon leaves markings on bullets and cartridge casings that are supposedly unique to that particular gun. If cartridge casings or bullets found at a crime scene or extracted from a body have a sufficient number of identical markings and no inconsistencies examiners will testify that they were also fired by that gun.

That’s the belief. However, a recent report by the National Academies concludes that while “one can find similar marks on bullets and cartridge cases from the same gun,” the assumption that only that gun could have produced those markings “has not yet been fully demonstrated.”

Even if we believe that ballistics evidence is reliable, humans aren’t. In this comparison a recovered bullet in excellent condition is on the left, and a bullet test-fired through the same gun is on the right. We can see that the striations left by the barrel line up perfectly. In the real world, though, bullets are often deformed and fragmented, making comparison difficult. Detroit PD’s lab was recently shut down after State Police auditors found three “false positives”, cases where examiners mistakenly reported a match that didn’t exist.

Why did the State come in? After a recent murder conviction a retired State firearms examiner conclusively demonstrated that shell casings found at the crime scene came from at least two weapons, not one as the police lab claimed. The judge dismissed the case, which will be retried.

Everyone’s heard of Phil Spector, the celebrity murder defendant whose first trial ended in a hung jury (his retrial will begin any day.) There’s no disputing that the victim, Lana Clarkson, died from a bullet discharged while a gun barrel was in her mouth. Spector claims that he was six feet away when the gun went off. His claim was propped up by blood spatter expert Stuart James, who said that droplets could travel six feet. But Sheriff’s criminalist Dr. Lynne Herold, who admitted she had taken one of James’ courses, said no, that their range was at most three feet. That little duel is likely to replay itself. Meanwhile, what are we to think of blood spatter evidence? Is it meaningful or not?

Maybe CSI isn’t all that it’s cracked up to be. Physical evidence has to be collected, bagged, tagged and interpreted by fallible humans who can slip at any stage of the process, damaging the goods, making them out to be what they’re not, or inferring that they mean something they don’t. It’s happened with arson, fingerprints, ballistics and blood spatter. Last week we mentioned that goofs leading to wrongful convictions have even happened with DNA, which is particularly scary given its aura of infallibility.

According to a recent article in the New York Post, the National Academy of Sciences is expected to shake things up this December with a report that will question the value and accuracy of accepted forensic techniques.

Not to worry, Joe Friday. Looks like shoe leather will be in style a while longer.

Did you enjoy this post? Be sure to explore the homepage and topical index!

Arson investigation issues

UPDATES

01/19/10 As the FBI continues reviewing 2,500 cases where analysts used a now-discredited bullet lead comparison technique, judges continue releasing persons convicted on their testimony.

10/07/09 Texas Innocence Project calls dog scent lineups “junk science” after DNA clears cop of murder

08/31/09 A New Yorker piece concludes that Cameron Willingham was innocent

08/17/09 Israeli scientists manipulate DNA to create fake blood samples

03/16/09 A prosecutor said that 147 cases of Detroit PD lab work so far identified as questionable is only “the tip of the iceberg”

01/15/09 False fingerprint matches that resulted in at least two mistaken arrests have led to a review of nearly 1,000 cases in which two LAPD examiners were involved.

Home Top Permalink Share this post

Posted 10/5/08

LABS UNDER THE GUN

Can police crime laboratories be trusted?

“Of the 33 adjudicated cases from the Wayne County Prosecutor’s Office that were reanalyzed, 3 exhibited Class I inconsistencies. In total, this equates to approximately 10% of the completed firearms cases having significant errors. On average, the DPD firearms unit analyzes 1,800 cases per year. If this 10% error rate holds, the negative impact on the judicial system would be substantial, with a strong likelihood of wrongful convictions and a valid concern about numerous appeals.”

These words aren’t from do-gooders wringing their hands about possible miscarriages of justice. They’re from an official September 2008 report by the Michigan State Police setting out the preliminary findings of an audit of the Detroit crime lab’s firearms unit.

Firearms examiners often test-fire recovered guns hoping to link them to crimes. Firing a weapon leaves markings on bullets and cartridge casings that are supposedly unique to that specific gun. If a sufficient number of identical marks are present in the same locations on cartridge casings or bullets recovered at a crime scene (or extracted from a body) it’s evidence that they were fired by the same gun. Naturally, great care must be taken to insure that there are enough points in common. It’s also critical that there are no dissimilarities; just like in fingerprinting, only one inconsistency rules out a gun as being the source of a particular bullet or cartridge casing.

Michigan State Police auditors reviewed 200 firearms cases. Nineteen had “either Class I or Class II inconsistencies”. A Class I error means that an examiner erroneously declared a match. Such “false positives” can obviously lead to a wrongful conviction. In Class II errors, “false negatives,” a match was overlooked, possibly letting a guilty person go free.

Detroit PD responded by shutting down the entire crime lab -- not just the firearms unit -- and turned over all forensic analysis to the State. It then set out on the unenviable task of reviewing past cases involving testimony by firearms examiners. There was little choice, as defense attorneys immediately announced they would begin questioning everything the lab has ever done. Meanwhile displaced lab employees mounted a protest, claiming that problems in the firearms unit did not affect the good work the lab was doing elsewhere, from fingerprint comparison to DNA. Their view was undercut by the words of their own superior, who in a September 2008 memo reported that the lab’s overworked and underpaid staff was running four-thousand chemistry and biological cases behind. (She retired when the lab closed.)

Alas, Detroit isn’t unique. In November 2002 Houston shut down the DNA section of the police crime lab after an investigation by a TV station revealed a history of shoddy work. A subsequent audit of the lab’s DNA work disclosed “a wide range of serious problems ranging from poor documentation to serious analytical and interpretive errors that resulted in highly questionable results being reported by the Lab.” Issues were also reported in firearms, trace evidence and drug analysis.

What’s worse, at least two cases of wrongful conviction have been attributed to Houston DNA errors. Josiah Sutton served four and one-half years of a 25-year term after the lab incorrectly determined that his DNA was present in a sperm sample. George Rodriguez served seventeen years of a 60-year term; his conviction was due in part to bad witness ID, in part to a mistaken failure to exclude him as a DNA donor, and in part to an incorrect conclusion that a hair found on the victim was likely his.

That’s not all. In January 2008, one and one-half years after the Houston lab’s DNA section reopened, its new supervisor was allowed to resign for helping staff members cheat on proficiency exams. (Amazingly, she was then hired to run the State lab’s DNA section.)

It happens to the best of labs. In May 2005 a grievous analytical error at the Virginia State crime lab, reportedly one of the nation’s finest, prompted Governor Mark Warner to order the re-examination of 150 DNA cases. His move was prompted by a 1985 case, where a prisoner on death row, Earl Washington, came within nine days of being executed before a team of pro-bono defense lawyers finally got him a stay. In 1993, with the threat of execution again looming, a DNA test (which the State partly botched) got Washington’s sentence commuted to life imprisonment. It would take another seven years and a correctly performed DNA procedure to conclusively clear Washington and identify the guilty party. By the time he was finally released in 2000 the innocent man had spent seventeen years behind bars.

Mismanagement and lax quality control have vexed crime laboratories for decades. O.J. might have never been in the position to pull the shenanigan in Vegas except for a lab goof. (His acquittal in the 1994 murder was in large part due to evidence of widespread contamination at the LAPD crime lab.) But trying to keep labs on the straight and narrow with after-the-fact controls such as accreditation visits is a loser’s game. As long as facilities are tidy, paperwork is in order, equipment is in proper repair, manuals are up to date and everyone on staff is certified a “pass” is virtually guaranteed.

Everyone wants to solve crime through science and technology. But as auditors in Houston pointed out, running a good lab is an expensive proposition. When resources are limited -- and when aren’t they? -- it’s easy to wind up with a production-oriented pressure cooker that encourages shortcuts and sloppy work. Throw in a dash of unskilled examiners and a pinch of poor oversight and it’s a recipe for disaster.

Next week we’ll look at issues in forensic techniques, from fingerprinting to ballistics. Stay tuned!

Did you enjoy this post? Be sure to explore the homepage and topical index!

UPDATES

03/04/10 An analyst’s incorrect testimony that blood was found in a vehicle is leading to a reexamination of thousands of old Georgia lab cases to check their accuracy. Related post

01/19/10 As the FBI continues reviewing 2,500 cases where analysts used a now-discredited bullet lead comparison technique, judges continue releasing persons convicted on their testimony.

12/18/09 Deceased NY State analyst falsified reports for fifteen years to cover up his incompetence, throwing hundreds of cases in doubt and creating questions about the rest of the lab. Report

04/08/09 Jarrhod Williams, convicted of two counts of first-degree murder but granted a new trial because the Detroit PD crime lab botched the ballistics was convicted again.

01/14/09 LAPD fingerprint problems lead to review of nearly 1,000 cases

10/18/08 LAPD fingerprint examiners criticized for serious lapses, including mistaken matches

Home Top Permalink Share this post

Posted 9/28/08

MINDBOARDING

Is brain scanning the new polygraph?

Hey, Dick Tracy: don’t knock yourself out pounding the pavement! There’s a far easier way to solve a whodunit. Have a suspect put on a helmet full of electrodes. Then show him a series of photos, including some neutral pictures and some of the crime scene. Looking at the photos will stimulate brain activity, sending electrical signals through the helmet to an EEG machine. You’ll wind up with an electroencephalogram, a chart that identifies the precise regions of the brain that the images stimulated.

Now look closely: if “experiential” areas of the brain “light up” for the crime scene photos, but not for the others, you’ve got your man. Hook him, book him and reward yourself with a trip to Winchell’s! If not, move on to the next chump.

According to an emerging technology known as BEOS, for “Brain Electrical Oscillations Signature,” there are places in the brain that store memories of events that one actively experienced, not just passively observed. Proponents claim that’s what makes it possible to distinguish between a killer and someone who merely discovered a body. Peddled in the U.S. by companies including No Lie MRI and Cephos for use in everything from commercial disputes to intelligence, the technology supposedly far surpasses polygraphy in accuracy. In fact, it was recently used by prosecutors as evidence in a murder case in Mumbai, India. To clear herself, a woman charged of poisoning her husband volunteered for a BEOS test. It wasn’t a wise choice -- the test said she did it. Oopsie!

No Lie and Cephos aren’t alone. A competing technology known as Brain Fingerprinting also gauges the brain’s electrical reaction to visual and aural stimuli, but in a fundamentally different way. Developed by neuroscientist Larry Farwell, it relies on a well-established neurological phenomenon, the so-called “P300 wave,” an involuntary electrical impulse that our brains generate whenever we recognize (have an existing memory of) something, be it an object or a piece of information.

For example, tell a suspect that they’re about to see a picture of the murder weapon, but don’t say what it is. Strap on the helmet (on them, not you) and run a series of slides, say, a gun, a knife, a baseball bat, and what was actually used, Auntie’s embroidery needle. If he emits a P300 wave when the needle comes up, and only when it comes up, have a scrumptious jelly-filled gut buster on us! If not, move on. To his credit, Farwell readily admits that the process has limitations; it won’t work, for example, if word of the needle got out to the public, since everyone would then react to its image. But he claims that when investigators come up with something only the real perp knows, the technology is virtually foolproof.

Alas, neither BEOS nor Brain Fingerprinting have made it into the judicial mainstream. (Brain Fingerprinting claims otherwise, but the episodes cited in its website hardly set a precedent.) According to the landmark Frye decision, before expert scientific testimony can come into court its validity must be widely acknowledged. But the kingdom of the nerds remains highly skeptical. As J. Peter Rosenfeld, a pioneer of using brain waves in lie detection points out, there’s a lack of peer review and replication, the sine qua non of scientific acceptance. Other neuroscientists feel likewise. “Well, the experts all agree,” says Michael Gazzaniga, director of a UCSB mind-research center, referring to BEOS. “This work is shaky at best.”

Unlike the polygraph, which records physiological changes supposedly brought on by the stress of lying, neither BEOS nor Brain Fingerprinting directly measure deception. They’re also far more passive, as no interaction is required between tester and subject. Keeping the two apart prevents contaminating the results, but it also means that EEG technicians won’t get what polygraphers really aim for. It’s the lie detector’s dirty little secret that its real worth isn’t in the squiggles it produces -- the National Academy of Sciences considers those close to worthless -- but on the incriminating statements, admissions and full-blown, tearful confessions that scared, stressed-out subjects occasionally make while in the chair.

But it’s not just about ends -- means are also important. The privacy and liberty implications of brain-wave technology are (pardon the pun) mind-boggling. Just to mention one issue, polygraph subjects are free to clarify and challenge each question before answering. In contrast, EEG screening is purely passive, allowing sneaky administrators to venture into areas far afield of their manifest purpose without the test subject realizing or having a realistic opportunity to refuse.

What’s more, we might not even know that we’re being checked out. Technology now in development allows the remote detection of “anxious” people. FAST, an acronym for “Future Attribute Scanning Technology” (how’s that for an Orwellian nightmare) uses cameras and sensors to screen passers-by for hostile thoughts and intentions, assessing characteristics such as facial expressions and pulse rate. Imagine the false positives that a gaggle of ACLU lawyers would produce!

Well, we’ve got a label for these precious new techniques: Mindboarding. Feel free to use it, but be sure to say that you saw it first on PoliceIssues.com!

Did you enjoy this post? Be sure to explore the homepage and topical index!

Prior posting on polygraphy

UPDATE (3/13/09): Newsweek column by Sharon Begley, “Of Voodoo and the Brain,” challenges the scientific basis of brain imaging

Home Top Permalink Share this post

Posted 5/11/08

BEAT THE ODDS, GO TO JAIL

DNA random match probabilities may be overstated

Looking for a growth industry? Think genetics. With more than one million profiles, California’s DNA databank is the third largest in the world, trailing only those of the FBI and Great Britain. At its 1990 debut the GoldenState’s database only kept track of sex offenders, but it has since expanded to include everyone convicted of a felony. What’s more, starting next year DNA specimens will be collected from every adult arrested for a felony, a move that should increase the databank’s size by 390,000 profiles each year.

When DNA got its start there weren’t databanks, so police had to have someone in mind to make crime scene DNA useful. Now it’s possible to run unknown DNA through massive databanks like California’s hoping for a “cold hit.” A recent example is the case of John Puckett, a previously convicted rapist who is appealing his conviction on a thirty-year old rape/murder. An expert testified that there was only one chance in 1.1 million that the match between Puckett’s DNA and the crime scene sample could have happened at random. With a probability of error that low, prosecutors suggested there was only one explanation: both samples came from the same source. Not unexpectedly, jurors agreed, sending the 70-year old to prison.

Since the human genome is exceedingly large, DNA is only typed at thirteen known places (“loci”) in the strand. Each location has two chemical sequences (“alleles”), one inherited from each parent. Scientists have determined how often specific loci/allele combinations occur in different populations, such as Caucasian males. Single combinations are commonplace and can be present in one out of every three or four persons. Multiple loci/allele combinations occur less frequently. In this example the probability of randomly selecting a DNA profile with four specific loci/allele combinations is 14 in 100,000.

Just like with fingerprints, a single dissimilarity between DNA profiles means that they’re not from the same person. If no differences are observed a sufficient number of identical loci/allele combinations must be present to suggest that they have a common origin. How many is enough? There’s no set answer. Five and six loci/allele combinations can yield probabilities of a random match in the one-in-a-million range; while seven or more can generate probabilities in the hundreds-of-millions, billions, trillions or even quadrillions. (For an online tool that lets users run a sample profile, click here.)

When a suspect is independently developed a subsequent DNA match obviously carries enormous weight. Still, the DNA match alone is not a probability of guilt -- it’s an estimate of the likelihood that DNA drawn at random will match the profile of crime scene DNA. (Probability of guilt requires that all other pertinent factors be considered. This requires use of Bayes’ theorem.) Random match probabilities also assume that we only draw once from a population. But that’s not what happens in cold hits. No one knows whether the match that cooked Puckett’s goose came on the computer’s first draw or last (at the time California had 338,000 DNA profiles online.) Had the expert witness followed the recommendation of the National Academy of Sciences he would have multiplied the random match probability of one in 1.1 million by the number of draws (338,000), yielding a true random match probability -- in effect, the chance of mistakenly identifying an innocent person -- of one in three.

Interestingly, the expert told a reporter that he didn’t mention the adjustment, which he agreed was a superior approach, because the judge wouldn’t allow it. After the trial jurors said that the probability of one in 1.1 million was a key factor in deciding to convict. Asked if correcting it might have affected the verdict a juror said, “of course it would have changed things. It would have changed a lot of things.”

Bigger DNA databases will yield more matches. While that seems beneficial, more profiles mean more draws, so the probability that matches may be caused by chance will increase. Of course, random match probabilities with denominators that approach or exceed the population of the U.S. or the planet will remain noteworthy. In any event, understating the probability that a match might point to the wrong person is no solution. At least one expert has already warned that an invaluable tool for freeing the innocent -- DNA -- could inadvertently become an instrument of wrongful conviction.

Only days ago Puckett’s appeal was argued before the California Supreme Court. Its decision is expected soon. In the meantime keep away from the lottery. The probability of hitting it is so low that if you do, it could be evidence that you fixed it!

Did you enjoy this post? Be sure to explore the homepage and topical index!

UPDATES

02/24/10 Puckett case featured in the Washington Monthly

06/16/08 Ruling in People v. Nelson, where the random match probability in the general population was in the quadrillions, the California Supreme Court decided it’s permissible to introduce it even though the match was against a database. The “database probability” can also be introduced.

07/20/08 Arizona DNA data challenges rarity of DNA profiles, especially when fewer than 13 loci are available for comparison

05/05/08 Ninth Circuit rules on DNA random match probability, calls its misuse a due process violation.

Home Top Permalink Share this post

Posted 12/7/07

C.S.I. THEY’RE NOT

Lab goofs and dueling “experts” give forensics a black eye

New YorkState’s Inspector General recently recommended that criminal charges be considered against the retired director of the New York Police Department’s crime lab and three former analysts for botching thousands of drug tests in 2002. Investigators claim that analysts took shortcuts when analyzing large seizures, falsely certifying that every container of suspected drugs was tested, and that managers who suspected something was amiss turned a blind eye. The lapse caused NYPD to start re-examining 3,000 individual drug tests last March. However, by that time more than 700 had been destroyed, bringing every conviction based on those tests into question.

Problems at crime labs are nothing new. In June 2007 an investigative panel cast doubt on thousands of convictions in Houston, calling its police lab deficient “across the board,” with serious errors in ballistics, drugs, DNA and serology. The damage was not merely hypothetical, with mistakes responsible for at least three wrongful convictions: Ronald Taylor, who served 14 years because the lab missed finding the real perpetrator’s DNA on a bedsheet, and George Rodriguez and Josiah Sutton, who served 17 and 4 ½ years respectively due to faulty serology. Nearly two-hundred other cases are on review.

In May 2005 Virginia’s Governor ordered a review of 150 cases processed through the State’s crime lab after two botched DNA tests nearly led to the execution of Earl Washington, Jr., who served 18 years after being wrongfully convicted of rape. Washington was only nine days away from lethal injection when discrepancies in the case prompted the prior Governor to commute his sentence to life imprisonment. A properly conducted DNA test later proved that the perpetrator was an already-convicted serial rapist. Auditors attributed the Virginia lab’s sloppy work to pressures to increase productivity. A Federal civil jury awarded Washington $2.25 million in compensation.

Two months after terrorists bombed a Spanish train, leaving 200 dead and 1,400 injured, FBI agents arrested Portland attorney Brandon Mayfield as a material witness. FBI fingerprint examiners said they matched Mayfield’s fingerprints to latent prints found by Spanish police on a bag of unexploded detonators. Confident that they had the right man (Mayfield is Muslim and represented a suspected terrorist in a civil action), the Feds refused to believe Spanish experts who insisted that the prints were not Mayfield’s. A chastened FBI eventually apologized when Spanish investigators positively identified the fingerprints as belonging to an Algerian suspect.

It’s not just lab goofs that give forensics a black eye. In the recent Phil Spector trial renowned experts argued about, well, everything -- from the cause of the injury to the victim’s tongue, to how far blood spatter can travel, to whether the victim could have coughed after being shot. Spector’s trial is remarkably similar to the 2004 murder trial of Idaho resident Craig Perry, who insisted that the uncle he was accused of shooting committed suicide. Thanks to blood spatter expert Stuart James, the same witness who raised enough doubt to hang Spector’s jury, Perry won an acquittal. (Demonstrating the whimsical, musical-chairs aspect of forensic “science,” another of Spector’s experts, Dr. Vincent Di Maio, testified against Perry. Back then Di Maio was still Chief Medical Examiner for San Antonio and working for prosecutors.)

A litany of lab disasters, dueling experts, wrongful convictions and bizarre acquittals (O.J. and Robert Blake come to mind) have done little to reassure a skeptical public about the merits of physical evidence. Police, prosecutors, courts and juries must be confident in the accuracy of laboratories and the trustworthiness of government witnesses. That’s hard to do when labs and experts are captive parts of the law enforcement establishment. Regaining confidence in forensics calls for a national system of independent, government-funded laboratories, much like the National Institutes, that are operated and controlled by top-notch scientists. Anything less is not good enough.

Did you enjoy this post? Be sure to explore the homepage and topical index!

Junk science links US DOJ/OIG report on Brandon Mayfield

Home Top Permalink Share this post

Posted 11/27/06

THE MYTH OF PROFILING

Pop psychology can lead investigators astray

“That really ordinary guy living next door could be a serial killer.” Dave Shiflett of the Bloomberg News says that’s the lesson we can draw from “Inside the Mind of BTK,” John Douglas’s new book about the infamous serial killer Dennis Rader, who tortured and murdered ten Wichita women between 1973 and 1991.

But, wait! John Douglas is the most famous FBI profiler ever, an author of several true-crime best sellers and the model for Jodie Foster’s superior in “Silence of the Lambs”. If a sick puppy like BTK can seem so “ordinary”, how could he be identified through profiling?

That, according to a lengthy exposé in The New Yorker (“Dangerous Minds,” 11/12/07), is the problem. John Douglas and his FBI colleagues told Wichita police that BTK was an American male with a decent IQ, that he drove a decent car, liked to masturbate, was selfish in bed, a loner (but could get along socially), uncomfortable with women (but could have women as friends,) maybe married, maybe not (but if married his wife was younger or older,) and so forth. Thankfully, officers managed to eventually solve the case sans profile. Rader was nothing like the FBI suggested. He was married, with children, active in his church and a pillar of the community.

Profiling is one of several psychological techniques, along with investigative hypnosis and the recovery of repressed memories, that gained popularity during the free-wheeling 80’s. Although the latter methods have been thrashed for over-promising, under-performing and generally leading investigators astray, profiling lives on, its findings so elastic that they can seldom be disproven.

It’s when profilers get specific that the nonsense becomes obvious. On the morning of January 21, 1998, Stephanie Crowe, 12, was stabbed to death in her Escondido (Calif.) home while the family slept. Detectives soon zeroed in on her reticent 14-year old brother, Michael. After relentless interrogation, he confessed and implicated two friends. Both got raked over the coals; one confessed while the other didn’t. Police arrested all three. They and prosecutors remained confident in the case even after the coerced statements were suppressed. After all, didn’t the FBI profile conclude that the murder was planned? Didn’t profilers say that the killer had “familiarity, comfort and knowledge” of the residence and the victim’s bedroom?

Months later, while the boys awaited trial, a violent, mentally ill transient whom detectives originally discounted as a suspect (in part because of the FBI profile) was arrested when DNA testing revealed that those spots on his clothes were the victim’s blood. Charges against the boys were dropped and the man was convicted and imprisoned.

On November 5, 2003 Gary Ridgway, the “Green River Killer,” pled guilty to murdering 48 women in KingCounty between 1982 and 1998. The investigation dragged on for twenty years and several FBI profiles, the first prepared by -- you guessed it -- the celebrated John Douglas. Their conclusions: the killer was likely an unemployed transient who had left the area and was either dead or in prison.

Fortunately, the cops had Ridgway in mind all along. Deputies knew that the married truck driver, a local resident, had a reputation for picking up prostitutes and playing rough. In 2001 new DNA techniques matched Ridgway to four of the victims. He got life without parole.

During the 1996 Atlanta Olympics a bomb exploded in a city park, leaving two dead and more than one-hundred injured. FBI agents immediately focused on Richard Jewell, the security guard who found the device before it detonated and sounded the alarm, undoubtedly saving many. But the FBI didn’t see him as a hero. Convinced that the chubby bachelor who lived with his mother fit the profile of a lone bomber, the Feds searched his home and conducted an exhaustive, highly public investigation. Jewell was cleared after two months. But the stain on his reputation never disappeared.

In 2003 police finally caught up with the man responsible. Eric Rudolph had used identical devices to bomb the park and a string of abortion clinics. He confessed and got life without parole.

It’s the patina of science that makes profiling so disturbing, lending confidence in conclusions with no more factual basis than the prognostications of a horoscope. Although recent studies seriously challenge the technique’s reliability, the FBI’s thirty-odd profilers remain on the job, reportedly fielding more than one-thousand requests from local police each year.

More than twenty years after its inception profiling chugs on, the embarrassing detritus of a decade when overburdened police and prosecutors were seduced by the promises of pop psychology. Let’s hope it doesn’t take us another twenty to rediscover that it’s shoe leather, not magic, that solves crime.

Did you enjoy this post? Be sure to explore the homepage and topical index!

UPDATE: Columbine author says that the killers weren’t loners or “socially inept” and that there is no profile for a school shooter

UPDATE: FBI admits there’s no such thing as a serial killer “profile”: News release Study

Home Top Permalink Share this post

Posted 11/20/07

POLYGRAPH: SCIENCE OR SORCERY?

Its usefulness is mostly as a prop

Exposing a stunning breach of national security, Nada Prouty, 37, a former FBI and CIA agent, pled guilty this month in D.C. Federal court to nationalization fraud, illegal computer access and conspiracy. Admitted in 1989 on a student visa, the Lebanese immigrant staged a sham marriage and gained permanent residency. In 1997, now-citizen Prouty was hired by the FBI and allegedly started passing top-secret information about Hezbollah to accomplices. A few years later the super-achiever landed in the CIA, an even better place from where to compromise American secrets.

So what’s the rub? Prouty sailed through FBI and CIA pre-employment polygraph exams, supposedly the toughest in the universe. In all likelihood she would still be a mole except that her name came up during an investigation of her brother-in-law, Talal Chahine, who allegedly channeled millions of dollars to Lebanese militants.

The history of lie detection is replete with disasters. None seems worse than the case of Aldrich Ames, a CIA agent who got rich by exposing his colleagues to the USSR (Ames’ treachery led to the execution of several Soviet citizens who were spying for the U.S.) While pocketing bundles of cash Ames passed two routine CIA polygraphs, and when caught bragged that he had never employed countermeasures.

Ames wasn’t lying. In an exhaustive 2001 report, the National Academy of Sciences concluded that the polygraph is worthless for screening job applicants and employees. It held out a bit more hope when polygraphs are used for investigating specific, known events (i.e., crimes), but cautioned that research that supports this more limited application lacks scientific validity and probably overstates the technique’s accuracy.

That’s a warning to take to heart. Between 1982 and 1998 forty-two women, mostly prostitutes, were murdered in King County, Washington. Most of their bodies were found in or near the Green River. Suspicion soon fell on Gary Ridgway, a truck painter whom prostitutes accused of rough treatment. Ridgway took and passed a police polygraph. In 2001, improved DNA techniques proved that he was indeed the killer. Ridgway was arrested and plea-bargained to life without parole.

Polygraphs are frequently used to narrow the field of suspects. They are routinely administered to the parents and caregivers of missing and abducted children. Results are not reassuring. In the 1997 disappearance of Sabrina Aisenberg, local police, who suspected the parents, called polygraph results “inconclusive,” while an ex-FBI polygrapher hired by the defense insisted that it cleared them. A like controversy dogged the investigation of the 1996 murder of JonBenet Ramsey, where police rejected the findings of a renowned polygrapher who insisted that the victim’s mother and father were being truthful. (The Ramseys refused to be tested by the FBI because its profilers told police that the murder was probably an inside job.)

Leery of being led down the wrong path, many savvy investigators shun the polygraph as a “truth machine” but use it as a prop when physical evidence or witnesses are lacking. Refusing to take a polygraph can land one on the short list of suspects. Even better, a few guilty persons get so intimidated by the black box that they shrivel up and confess even before the test begins. It’s a form of legalized coercion that leaves no bruises and may be impossible to challenge in court.

It’s no surprise that shortcuts to finding the truth are hugely popular. As long as we’re willing to dig in our pockets there will always be someone happy to supply all the elixirs we want. We will soon be reporting on other questionable techniques, including cognitive interviews, profiling, investigative hypnosis and the recovery of repressed memories. Stay tuned!

Did you enjoy this post? Be sure to explore the homepage and topical index!

Home Top Permalink Share this post