How forensic science can fail at solving crimes
We all love a good episode of CSI, where the protagonists gather DNA samples, run them through a machine, and are able to find the bad folks—with a 100 percent accuracy rate and no chance of ever sending an innocent person to prison. Unfortunately, real life is seldom like TV.
According to the Innocence Project, 45 percent of DNA exoneration cases involved misapplications of forensic science that led to wrongful convictions. A collaboration between the Innocence Project and The National Association of Criminal Defense Lawyers found that, of the 268 cases where FBI hair analysis led to a conviction, 96 percent of their findings were the result of flawed forensic testimony.
In 2009, the National Research Council published a report that condemned fields such as bloodstain pattern analysis, firearms examination, fingerprint, bite marks, and impression match analysis as unreliable and lacking scientific basis. While some of these fields have been thoroughly dismissed as pseudoscience, procedures such as DNA and fingerprint analysis are still widely believed to be reliable. This is a level of faith that these practices do not deserve.
DNA analysis has come to play an important role in the justice system as one of the most reliable forensic sciences, but it still has many issues. One of these issues is the fact that all humans have very similar DNA. What makes all of us unique is a 0.1 percent difference in our genetic code. This means that the same DNA profile could potentially match multiple people, especially if the DNA sample is damaged or contains a mixture of DNA from multiple individuals. This issue becomes particularly clear when dealing with close relatives who have more similarities in their DNA. Consider the case of Lynnette White who was murdered in 1988. When the police ran a DNA test on a sample found at the scene, the only match was a boy that was too young to have committed the murder. Later on, it was discovered that his uncle was the murderer.
Similar problems apply to fingerprint analysis. The difference between fingerprints is at times hard to perceive, especially when the fingerprint sample is partial or smudged, which is likely to occur in crime scenes. Take the case of the 2004 Madrid Train Bombing: following the event, the FBI arrested Brandon Mayfield, an Oregon man who had never been in Spain. His arrest was based on the word of three separate examiners who matched his fingerprints to the ones collected from the bag of the detonators. However, it was later found that the same fingerprint also matched someone else’s, who was in Spain at the time. Thus, even the reliability of fingerprint matching is thrown into question. This is without considering the fact that each forensic scientist examines evidence completely differently, as there is a lack of standardardised practices across laboratories. Error rates in fingerprint laboratory matching can vary from 3–20 percent.
The problem is not necessarily that forensic science is always wrong, but that with every DNA analysis or fingerprint match there is a degree of uncertainty and possibility of error that often remains undeclared. The solution is to start treating forensic science as an actual science. Science is fallible and full of uncertainties—and forensic science is no different, no matter how accurate TV shows make it out to be. It is all a matter of statistical probability, p-values, and hypothesis testing—all the stuff that we don’t want to hear about in crime shows, but that should definitely be considered when someone’s life is on the line.