Charles Darwin's Darwin Awards 
Darwin Awards
At-Risk Survivors
Urban Legends
Personal Accounts
Slush Pile
Science Essays 
"Brother, Can You Spare a Banana?"
Love Bites
Endogenous Retroviruses and Evolution
Origin of the Novel Species Noodleous doubleous
Aquatic Apes Are People, Too!
Chicken Little Was Right
AIDS, Bubonic Plague, & Evolution
The Skinny On Fat
Forensic Analysis: Achieving Justice
The End of the Universe
Science Essays 
~ Random Story ~
NEW! Gift Shop
Rules  Search
Contact Darwin
Submit a Story
Philosophy Forum


Darwin Awards

Darwin Awards
Science Essays 
Book 4

Forensic Analysis: Achieving Justice 
Maia Smith, Science Writer

Imagine that you are deciding whether to admit a certain type of evidence into court on a criminal case. You know that this type of evidence:

  • Convicts innocent and guilty people with roughly the same frequency.(*)
  • Is extremely susceptible to contamination from outside sources, e.g., through the method of retrieval.
  • Was a key factor in convicting 60 percent of five hundred wrongly sentenced people who were later exonerated by DNA evidence, although it was used in only about 5 percent of cases. (1)
  • Is highly trusted by jurors, who often believe it even if they know the sample is worthless.

What kind of evidence is this? It's eyewitness testimony, and it is a keystone of our justice system. Certainly eyewitness testimony is deeply flawed and has led to many wrongful convictions. For every innocent man in jail, a guilty one runs free. If this were a new high-tech method, like analyzing fiber or DNA, it would have been abandoned as soon as the statistics became known. Yet no human rights activist would dream of advocating for a ban on eyewitness identification in court, despite its low rate of accuracy.

Study after study shows that memory is extremely unreliable and subject to tampering.* Witnesses can be misled by weapon focus, preexisting biases, being shown a lineup of suspects rather than one picture at a time (they tend to pick the person in the lineup most similar to the perpetrator), being questioned by a biased interrogator such as an attorney, and so forth. And bear in mind that studies of witness inaccuracy are done in a controlled setting that tends to decrease the error: None of the study witnesses were offered leniency in exchange for testimony, or had a loaded gun pointed at them while making observations.

The problems with eyewitness testimony are exacerbated in the case of repressed memories. Therapists claim that people subconsciously close off memories that are too painful to live with. This theory is undermined by the vivid accounts of Holocaust survivors and other torture victims. Therapists can sometimes "retrieve" repressed memories, although some techniques generate false memories just as easily.(*) The sketchy reliability of such memories has not stopped the use of this testimony in court, and many innocent people have been jailed after someone claimed horrific abuse (usually sexual) that occurred so long ago that an alibi or defense is unlikely.(2) Jurors, being human, shy away from flatly contradicting a sobbing witness, even if the evidence proves she is deluded.

How does DNA evidence compare to eyewitness testimony?

DNA analysis garnered good press in the 1990s, when it exonerated people who were wrongly sentenced to life in prison. Suddenly, a drinking glass or comb had the potential to identify its user beyond a reasonable doubt. Paternity cases became open-and-shut, as did cases of forcible rape.(4)

DNA analysis is extremely accurate.(3) Errors are so rare (1 in 10,000) that DNA, if available, is nearly always the most accurate method of identification. Problems with DNA analysis are nearly always low-tech, caused by a careless worker. Even a good lab has an error rate of 1 in 200, not because of inaccuracy in the DNA testing itself, but because, for example, samples are mislabeled or contaminated.

This is not a problem if DNA tests are performed after there is reason to suspect a person. However, with the growth of DNA databases, the possibility now exists for someone to be accused based solely on DNA. This creates a measurable risk of a false identification. For example, processing a rape kit often takes years. A scan through a database of DNA profiles could lead to a chance hit on an innocent person who cannot remember what he was doing on the night of September 17, 2002. A false positive rate of 1 in 200 may be acceptable if there are three suspects, two of whom are likely to be eliminated by DNA analysis; it is not acceptable when randomly combing through the linked databases of the FBI and various states, which currently contain close to two million DNA profiles.

DNA's extremely high accuracy rate can also lead to misinterpretation. We may prove beyond a reasonable doubt that the defendant has the murder victim's blood on his shirt, without proving anything at all about whether he participated in the murder. He might have simply bandaged her skinned knee earlier that day. There have already been five cases of identical twins implicated in a crime that only one committed* (although in only one case has there been any difficulty eliminating the innocent twin from suspicion). A 99.5% chance that the test has correctly identified a bloodstain does not translate into a 99.5% chance that the defendant is guilty, but the jury might give the evidence unwarranted weight. Likewise, an attorney might mention the low error rate of a test, without mentioning that the immense database almost guarantees a few chance hits. The problem isn't with DNA evidence per se, but with how it's presented in the courtroom and interpreted by the jury.

So let's examine how the jury trial itself can err.

Juries weigh evidence differently than logic would dictate. Facts, figures, graphs, and experts convey information, but may confuse the jury. Eyewitnesses and vivid descriptions grab attention and sympathy, but often convey no new information. Juries will convict on the basis of eyewitness testimony two-thirds of the time, even if the only witness wasn't wearing his glasses. Without a witness, they convict only one-sixth of the time.(5) A jury is more likely to believe a confident witness than a hesitant one, even though tests prove that confidence does not correlate with witness accuracy. Juries might acquit due to sympathy. And they're more likely to convict if the crime was particularly horrendous, even if it's not clear who committed it.

Forensic analysis is a clash between science and emotion. We have a huge array of tools for analyzing the tiniest shreds of evidence: bloodstains, fibers, bullets, and even repressed memories. At the same time, these are only tools. They can be used for justice or injustice. They can clarify the facts, or simply muddy the waters. Cutting through the gory details and tech-talk still requires a dispassionate, intelligent human mind-the first, last, and only tool we've ever had for achieving justice.


(1) †Huff, C. Ronald. 2003. "Wrongful Conviction: Causes and Public Policy Issues." Criminal Justice Spring 2003, 18 (1).

(2) Sagan, Carl. The Demon-Haunted World: Science as a Candle in the Dark. Ballantine Books, 1997.

(3) "Recovered Memories: Recent Events and Review of Evidence." Interview with Harrison G. Pope Jr., M.D. Currents in Affective Illness, XIII (7), July 1994, 5-12.

(4) DNA analysis is commonly claimed to be 99.99% accurate.

(5) USA Today, posted June 3, 2004, by Richard Willing, describes a few problems with identical twins.

HOME > Science >