Genetic evidence in the form of DNA profiles has proven to be a powerful tool in the advance of justice by proving the innocence of accused persons. The lack of a match of DNA markers between the crime scene evidence and a suspect can exclude the person from a list of suspects. However, genetic evidence is also used by prosecutors to prove guilt. It is this use of genetic evidence that produces a risk of wrongful conviction of innocent persons.
The great potential of genetic evidence to lead to injustice has to do with statistics. In the case of proving innocence, no statistical assumptions are required. If there is a lack of a match of DNA markers between the suspect and the crime scene evidence, there is no need for statistical analysis. However, when genetic evidence is used to prove guilt, there is a need for a statistical analysis with an accompanying need for statistical assumptions. If the statistical assumptions are incorrect, the results of the statistical analysis are also incorrect and unreliable, and can lead to wrongful convictions.
In order to understand why genetic evidence is a powerful tool for justice when proving innocence and a risky instrument of injustice when used to prove guilt, an understanding of how genetic evidence works is required.
The genetic code present in every cell of our bodies is constructed with DNA (deoxyribonucleic acid). In DNA profiling, areas of genetic material called genetic loci are taken from the suspect and from the crime scene sample. If the genetic loci from the suspect and the crime scene travel the same distance along a gel in a process called electrophoresis, then the crime scene sample and the suspect sample are considered to be a "match" at that particular genetic locus. In DNA profiling, a number of genetic loci are sampled. If any of the genetic loci differ between the suspect and the crime scene sample, then the suspect is excluded as a source of the crime scene sample. In the case of exclusion there is no need for statistical analysis.
However, what does it mean if there is a match between all of the genetic loci taken from the crime scene sample and the suspect? It first must be realized that there are many thousands of genetic loci within the human genome. DNA profiling involves looking at only a small number of genetic loci, with the number of loci tested ranging from ten to a proposed twenty-four loci. A match at all the tested genetic loci means that the suspect cannot be excluded as the source of the crime scene sample. This lack of exclusion based upon all the tested genetic loci does not require any statistical analysis. However, assessment of the probability of a match of all the tested genetic loci between the crime scene sample and the suspect occurring coincidentally requires a statistical analysis. The statistical analysis occurs as follows:
First, a probability value for a coincidental match occurring at each genetic locus must be made. This probability for genetic match at each genetic locus is based upon the frequency of the occurrence of the particular configuration of the genetic locus in the general population of Caucasians, Hispanics and African-Americans. If the suspect is a Caucasian and the frequency of the occurrence of a particular configuration of a genetic locus is one in ten Caucasians, then the probability of a coincidental match between the suspect and the crime scene sample at that particular genetic locus is one in ten (1/10). A probability of match at each individual genetic locus must then be made using data from epidemiological studies that examine probabilities of the occurrence of a certain genetic configuration at each tested genetic locus.
Once the probability of a coincidental match at each of the tested genetic loci is made, then a determination of the probability of having a coincidental match at all the tested genetic loci is made. This is done using the "product rule," whereby all of the individual probabilities are multiplied in order to arrive at the overall probability of a coincidental match at all of the genetic loci. For example, if there is a probability of a coincidental match at each of ten genetic loci and the probability of a match at each genetic locus is independent, then the probability of a coincidental match at all ten genetic loci is (1/10) x (1/10) x (1/10) x (1/10) x (1/10) x (1/10) x (1/10) x (1/10) x (1/10) x (1/10), or one in ten billion.
However, in order to use the product rule all the probabilities of the occurrence of the individual genetic loci must be independent. What this means is that the probability of the occurrence of one particular configuration of genetic locus is not dependent upon the occurrence of another configuration at a different genetic locus. Otherwise a result occurs which is not only wrong but also highly prejudicial because it is clothed in the appearance of a scientifically derived number. But genetic loci are not necessarily independent. For example, blue eyes tend to occur more often in persons with blonde hair than in persons with brown hair.
In order for the expert for the prosecution to arrive at a probability of coincidental match, two statistical assumptions must be made. The first is that the correct probabilities for coincidental match have been made for each tested genetic locus. The next assumption is that the individual probabilities of coincidental match are all independent. The fact that one or both of these assumptions is incorrect is manifested in the results of many probabilities of coincidental matches made by prosecution experts. When probabilities exceed the probability that would occur if the suspect were the only person on earth who could be the source of the genetic sample, then the statistical assumptions and the resultant statistical analysis are clearly incorrect. The facts speak for themselves (res ipsa loquitur).
When the prosecution presents evidence of a coincidental match of one in 100 trillion, that indicates a clear error. There are only seven billion people on earth. It has been estimated that only about 100 billion people have ever lived. The prosecution's evidence for an incidental match is more prejudicial than probative because it is clearly wrong, as it is impossible to get a probability of coincidental match more extreme than one in seven billion. The prosecution's expert may testify that the extreme probability result takes into account possible or "hypothetical" people. This is not logically consistent, since the units for the probability of a match at each individual genetic locus is "people," not "hypothetical" people. After all, a person committed the crime, not a "hypothetical" person.
The Emperor has no clothes.
Ernest P. Chiodo, M.D., J.D., M.P.H., M.S., M.B.A., C.I.H. is a physician and attorney. He received his medical and law degrees from Wayne State University and his Master of Public Health from Harvard University. He is board certified in a number of medical specialties including Public Health and General Preventive Medicine, which is the epidemiological and biostatistical specialty within medicine. He is an Assistant Clinical Professor of Family Medicine and Public Health at Wayne State University School of Medicine in Detroit and an Adjunct Professor of Law at John Marshall and Loyola University law schools in Chicago. Dr. Chiodo has served in a position of high public responsibility, having been the Medical Director of the Detroit Health Department. In that position he was the chief physician in charge of protecting the public health of over one million persons living and working in the City of Detroit. This article is reprinted with permission.
[Ed. note: For more on issues related to the probability of DNA matches in criminal cases, see PLN, Oct. 2010, p.1 and Jan. 2009, p.24].
As a digital subscriber to Prison Legal News, you can access full text and downloads for this and other premium content.
Already a subscriber? Login