When we hear the phrase “DNA test” in the criminal context, we likely think of a single, uniform test that can almost perfectly match the perpetrator’s DNA to a DNA sample found at the crime scene. The reality, however, is far more nuanced. A multitude of tests, with varying degrees of precision, are routinely used to compare two DNA samples in criminal cases, e.g., RFLP analysis, STR analysis, AmpFLP, and VNTR. STR analysis, standing for “short tandem repeats,” is the most common DNA “profiling” method used today. All of these methods, though, rely on the same principle: that different people will generally have different sequences at varying locations in their DNA. Because the incidence of two different people having the same sequence at a particular location is the result of chance, a forensic scientist can calculate the probability that two samples match by chance alone. The lower the probability, the more likely it isn’t luck; the two samples probably came from the same person.
On September 18, 2012, the Third District Court of Appeal in California ruled that one type of DNA profiling, Y-STR analysis, could be admitted as evidence against a criminal defendant without the need for a separate hearing, a Kelly hearing, to determine whether the method was scientifically valid. The case is interesting because–apparently from the court’s opinion–it’s the first time such a test had been used and challenged in California even though Y-STR testing has been routinely used for over a decade in other states. The court noted that the prosecution, in relying on Y-STR profiling, “offered no new science, no breakthrough technology, and no untested kits or tests” that necessitated a separate evidentiary hearing. (Op. at 2.)
The opinion is commendable for its thorough discussion of the science behind Y-STR testing, including important differences between the statistical methodology required to analyze Y-STR profiles from “regular” STR tests. Specifically, in regular STR tests, the different STR locations are from different chromosomes and are, therefore, “independent.” (The presence of one doesn’t affect the presence of another.) You may recall from statistics that the total likelihood of independent events can be calculated using the “product rule.” Thus, if the percent chance of a particular sequence on chromosome 1 is ½, and the percent chance of a particular sequence of chromosome 2 is ¼, then the chance of the two together is 1/8.
Y-STR testing is different. All of the sequences are from the Y-chromosome (hence, Y-STR). Thus, the sequences are not “independent.” As such, the percent chance of a particular sequence at each location can’t simply be multiplied together. A more complicated analysis–confidence interval modeling–must be used. While this presents challenges of its own, its biggest drawback is that it’s much less powerful.
Presiding Justice Vance W. Raye, the author of the opinion, nailed all of these nuances. It’s unfortunately rare that state appellate courts spend the time to discuss what are often critical differences in various forms of scientific testing. Kudos to him and the Third District Court of Appeal.