Home About RSS

Evidentiary Cross-Contamination

When it comes to whether expert testimony can be admitted in court, the U.S. is a checkerboard map: each jurisdiction either follows the Daubert rule or the Frye rule. In Daubert jurisdictions, expert testimony may only be admitted if it is (1) relevant; (2) reliable; and (3) rooted in the scientific method. What counts as the “scientific method” varies, but this prong typically focuses on whether hypotheses are subject to empirical falsification and the potential error rate of the test proposed. In Frye jurisdictions, expert testimony may only be admitted if it is generally accepted within its own scientific community. The Frye test has long been criticized as essentially allowing junk science in the courtroom–junk science is, after all, generally accepted in the “junk science community”–but Frye jurisdictions have recently combated this by increasingly aligning the “general acceptance” portion of the standard to Daubert‘s scientific method prong. Currently, the federal system, and most states, are Daubert jurisdictions, but Frye states remain: California, Florida, Illinois, Kansas, Maryland, Minnesota, New Jersey, New York, Pennsylvania, and Washington.

Most people would assume that, Daubert or Frye, expert testimony concerning DNA evidence is admissible. But this paints “DNA evidence” with too broad of a brush. In fact, there are multiple types of DNA evidence, some of which we’ve profiled on this blog. Needless to say, forensic analysis of different types of DNA evidence come with different falsification methods, error rates, and “acceptance” in the forensic community. The admissibility of at least some types of DNA evidence may, therefore, depend on whether the testimony is sought to be admitted in a Daubert or a Frye jurisdiction. And a recent case out of New York–a Frye state–highlights just this distinction.

In People v. Garcia, No. 2650/09, 2013 WL 690600 (N.Y. Sup. Ct., Bx. Cnty. Feb. 13, 2013), the prosecution sought to admit expert testimony of DNA evidence against the defendant. The evidence, however, was based on a low copy number, or LCN, DNA profile. LCN profiles are, as their name suggests, DNA profiles compiled from a low amount of DNA–typically less than 100 picograms. (That, by the way, is one-hundred one-trillionths of a gram, or for those metrically disinclined, three-and-a-half trillionths of an ounce.) To properly analyze such a small amount of DNA, it needs to be amplified, or copied, several times–between 28 to 31 times to be exact. This is contrasted to high copy number, or HCN, DNA profiles, which either do not need to be amplified at all, or only a few number of times. Amplification at this rate presents a serious problem: tiny errors in the beginning of the process lead to enormous errors at the end. To give you an idea of the scope of this problem, a single piece of DNA accidentally introduced into the amplification reaction at the first cycle will, twenty-eight cycles later, produce over 268 million copies.

This presents particular problems in an evidentiary context. Under the Daubert standard, LCN DNA analysis may very well be relevant, but it’s not entirely reliable, at least not as much as HCN DNA analysis. And while it’s rooted in scientific methodology, false positives are difficult to rule out and the general error rate is quite high. To get around this problem, experts have proposed limiting their testimony to simply the percentage of DNA that came from the defendant found in the test tube following the amplification reaction; to avoid, in other words, testifying to whether the DNA “matched” that found at the crime scene. (I am skeptical whether juries, after hearing the term “DNA,” can–or even want to–make out this difference.) LCN DNA analysis may very well be impermissible in some Daubert jurisdictions.

Under the Frye standard, however, LCN is “generally accepted” in the forensic community. Indeed, there are experts who specialize in analyzing LCN DNA profiles. And given that New York is a Frye state, the Garcia court admitted the LCN DNA evidence without a separate hearing. “General acceptance” did not necessitate “unanimous endorsement,” opined the trial court. And the fact that the New York City Office of the Chief Medical Examiner used the technique–even though it’s the only government facility to do so–seemed good enough. Further, it seemed as if other New York courts had already admitted such evidence in their trials.

But to the Garcia court’s credit, it reviewed some of the issues concerning LCN DNA profile techniques, namely, stutter, allelic drop-out, and allelic drop-in. It also discussed the reliability of LCN DNA testing, as well as the statistical reliability of the software used to measure LCN DNA profiles. All issues, of course, for a Daubert jurisdiction rather than a Frye one. The Garcia case’s interpretation of the Frye standard may–like an errant piece of DNA–be just another example of cross-contamination. But a welcome one.

Jacob S. Sherkow, Fellow, Center for Law and the Biosciences, Stanford Law School

2 Responses to “Evidentiary Cross-Contamination”

  1. Hank Greely says:

    But, I take it, you think not really ENOUGH cross-contamination to get a good (better?) result. Do you think Frye states should move to the Daubert standard? Or, at least, construe “general acceptance” in a broader context . . . i.e., not “generally accepted by those people who perform this kind of test for a living”?

    More generally, would you impose a higher standard for admission of scientific evidence in a criminal case than a civil one? We don’t say we’ve got differential standards, but I sometimes suspect we do – with a LOWER standard for criminal cases (at least, for prosecution evidence) than for civil. Thoughts?

    • Jacob S Sherkow says:

      Lots of questions. (1) I certainly do think Frye states should move to the Daubert standard. I think the problems with the Frye standard were pretty well detailed in Galileo’s Revenge, and other follow-on books. And I think it says a lot about the standard that Frye states are moving closer to becoming Daubert states than the other way around.

      (2) Ah, the civil/criminal evidence distinction. Let me first say that I’m generally an evidence law skeptic. What we think of as sequestering information from the jury–objections to the form of questions, limiting instructions, the hearsay exceptions–are truly some of the most egregious fictions we practice in law. But that’s for another day. The purpose of evidentiary exclusions, so I’m told, are to limit the foundations of the ultimate questions of liability to only relevant evidence. So I personally fail to see how a different standards in civil vs. criminal law achieve that goal. Yes, yes, I understand there are constitutional concerns in criminal trials not present in civil ones. But those go towards prejudice, not relevance or reliability. Assuming, of course, there’s some prejudicial concern about certain scientific evidence in criminal vs. civil trials, it could always be excluded on that ground, rather than developing a mish-mash of standards. Or, of course, you could just issue a limiting instruction ;)

Leave a Reply