German police had been quietly tracking a serial killer for years but in 2007 the case exploded into headline news after the slaying of a young police officer in the city of Heilbronn. DNA evidence identified the suspect as the same individual who had been present at numerous murders, burglaries and break-ins but this latest development seemed to be a worrying escalation in violence. No one had ever seen the suspect’s face but genetic analysis clearly identified her as a woman – an unusual characteristic for a serial killer, but only one of several extraordinary aspects to this progressively troubling crime spree. There was no apparent pattern to the crimes; she apparently struck widely and indiscriminately, even across the border in France and Austria. The only consistent information was the occasional sighting of a man near the scene, leading to speculation that the killer was transsexual. The media nicknamed the killer the Phantom of Heilbronn and the increasingly desperate police offered €300,000 for her capture.
The bounty was never collected. This was not because the serial killer evaded justice, but because she didn’t exist. The DNA was not from a murderer, but from a female factory worker who had been inadvertently contaminating the cotton wool swabs used in the police’s lab tests. The case has become a blot in the history book of forensic science but the lessons extend far beyond the procedures of criminal investigation. If nothing else, the Phantom of Heilbronn demonstrates the amazing psychological power of DNA evidence. The belief that DNA samples mark out individuals like an infallible biological barcode is so powerful that people will begin to hypothesise invincible, transsexual, border-hopping serial killers just to keep the story coherent with the genetic evidence.
While the 1980s marked the start of the laboratory boom in forensic science, recent studies have shown that psychology can have a remarkable effect on the outcome of supposedly objective scientific tests. Fingerprint analysis was the first area where the unsettling influence of subjective bias was discovered. Like DNA, fingerprints are widely considered to identify the individual as clearly as a passport, and in ideal conditions they do, but the process of tying them to an individual from the impressions at a crime scene is a very different matter.
Itiel Dror, a psychologist at University College London, found that forensic fingerprint examiners could come to different conclusions depending on what they knew about the case. In a landmark study, he gave five experts pairs of prints to compare. The examiners were told that the FBI had already mistakenly identified them as coming from the same person. This is a common situation in forensic work – being asked to re-analyse evidence – but being informed of the previous results provided a clear prior suggestion that they didn’t match. In reality, the prints were samples the experts had already judged as matching in the normal course of their work five years previously, but with this new information, only one of the five reached the same conclusion as their earlier selves.
The influence of these biases is likely to have a much wider effect than just on fingerprint analysis. Dror notes that they could affect any area where “the human examiner is the main instrument of analysis”, including fingerprinting, DNA, CCTV images, firearms and document examination. “The contextual influences are many and they come in many forms,” says Dror. “Many… [if not] most of the forensic areas are vulnerable.” These warnings have met with a frosty welcome from many forensic scientists who, despite the evidence, have balked at the suggestion that they could be anything less than objective.
Change has begun to occur, however, because of a combination of these recent studies and some high-profile scandals – not least the detention of Muslim convert Brandon Mayfield after being erroneously connected to the 2004 Madrid train bombings by the FBI. The government report on the incident highlighted how the outcome of the subsequently discredited fingerprint matching – the sole evidence that linked him to the scene – was likely biased by suspicion based on nothing except prejudice about his lifestyle and beliefs.
But in contrast to these skill-based approaches, forensic DNA matching is widely considered to be the result of purely objective lab tests. Nevertheless, recent research by Dror has found that subjective bias could even have an effect here. While a good DNA sample does make for a very reliable comparison, real-world crime scene evidence may have several people’s DNA mixed together. The analysis becomes not a case of “matching barcodes” but of deciding whether any one of the numerous and disjointed “barcode fragments” seem to fit the original. In many cases, a judgment call. In Dror’s study, DNA experts were given results from a mixed sample that was drawn, unknown to them, from a previous real-world case that hinged on whether suspects were present at the scene. After analysing the samples, they not only disagreed with one another but also came to different conclusions depending on whether they had information about the case, or whether they had nothing to go on but the genetic data.
It is worth noting that these findings do not invalidate forensic evidence. Studies also show that despite biases, identification is mostly done reliably, but the fact that outside information can affect decisions remains a worry for the justice system. Justice, it seems, needs to be not just blind, but also psychologically validated.
Vaughan Bell, Guardian