Is there a multi-witness problem with respect to forensic lab tests?

While litigating Briscoe v. Virginia, I conducted – through a research assistant, Andrew Mackie-Mason, now a student at the University of Chicago – a study of transcripts of Michigan drug trials. Michigan adheres to constitutional procedures with respect to forensic lab results; it does not allow prosecutors to prove such results, over the objection of the accused, by introducing the lab report without presenting the testimony, subject to confrontation, of the author. The principal aim of that study was to find out, given this condition, how often a lab witness actually had to appear at trial. The results – reported in my main brief in Briscoe – showed that in the great majority of cases no lab witness appeared. In some cases, the state presented no evidence of lab results; more frequently, the state presented lab results by introducing a written report, the prosecution not wanting to bring the author of the report in as a live witness and the accused consenting to admissibility of the report. The study also demonstrated that when a lab witness does testify at trial, the defense nearly always cross-examines.

In the pending case of Bullcoming v. New Mexico, No. 09-150, the state presented a forensic lab report and also live testimony of a witness from the lab – but the witness did not perform the test and had no personal knowledge of the conduct of it. Like the petitioner in Bullcoming (represented by Jeff Fisher), I believe this procedure was unconstitutional. Anticipating that, as they have done in Melendez-Diaz and Briscoe, many states will contend that adhering to constitutional procedures would be prohibitively expensive, I have, through a group of student research assistants, conducted a more extensive study of Michigan trials. (Let me give due credit right off: The students who did the work are Wencong Fa, Tasha Francis, Justin Gillett, Regan Nunez, Matthew Parelman (who assembled and organized the data), Kimberly Parks, Luke Rachlin, Joe Reiter, Liza Roe, Alex Su, Sean Stiff, and Anna Walker.) This study was primarily intended to determine, in a state that recognizes that forensic laboratory witnesses are not exempt from the accused's right to be confronted with a witness who made a statement admitted against him, whether there was a significant multi-witness problem. That is, is a chain of witnesses from the lab routinely necessary to prove the results of a forensic lab test? The answer is resoundingly negative.

As in the prior study, we used transcripts provided by the State Appellate Defenders’ Office (“SADO”) – my gratitude to Jonathan Sacks, Deputy Director of the Detroit office of SADO, and to other members of the SADO staff, for making this possible. SADO, which is appointed in a random selection of approximately 25% of indigent trial appeals from every county in the state, maintains electronic copies of trial transcripts from its cases. In this study, the students examined all the trial transcripts to which they were able to get access in the Detroit office in three categories: trials involving drug charges from February 14, 2000 to July 15, 2010; trials involving charges of operating under the influence of liquor (OUIL) from February 3, 2003 to April 20, 2010; and trials involving rape charges with penetration (first- and third- degree criminal sexual conduct) spanning from September 15, 1997 to November 25, 2010. I chose drug cases because they generate more presentations of forensic laboratory tests than any other kind. I chose OUIL cases (which we have referred to more informally as DUI) because Bullcoming itself is a case of this type and because forensic lab results are commonly presented in them. And I chose rape cases involving penetration because they are probably the type of case in which DNA tests – a particularly complex form of lab testing – are most frequently presented.

The detailed results of the study are presented on a spreadsheet that you can examine by clicking here. Here is a summary of the results.

The students examined a total of 154 drug cases. In 116 of these, the prosecution presented lab results presented at trial. But in only 59 of these cases did lab analyst actually testify at trial. In 49 of those cases, only one lab witness testified at trial, and never did more than one lab witness testify at trial with respect to a single test. In nine trials, two lab witnesses testified to two separate tests, and in one trial four lab witnesses testified to four separate tests. So all in all, this makes 71 lab witnesses in 154 trials, or about .46 per trial.

Of the 55 DUI trials examined, 41 included the presentation of lab results at trial. In 26 of these cases, one or more lab analysts testified live. In four cases, two witnesses testified live at trial, and there were never more than two. In only one of these four cases did the two witnesses testify with respect to a single test. (In another case, not included among those four, three analysts, in addition to the one who testified live at trial, testified via video-conference. Together, the four witnesses in that case testified to two tests.) In all, 30 witnesses testified live at these trials, for an average of about .55 per trial. I believe there is a ready explanation for why this number is somewhat higher than in drug cases. Very few DUI cases go to trial. Given the relative simplicity of these cases, when one does go to trial there is a rather high probability that it is because the defense believes that for some reason the lab evidence is problematic.

Of the 104 rape cases examined, 25 included DNA lab results presented at trial. These are, of course, particularly serious prosecutions, and DNA testing is more complex than the other types studied. Therefore, it is not surprising that, as compared to the other two classes of cases, analysts testify live at a higher percentage of the trials at which lab results are presented and that it is more common that more than one lab witness will testify at trial with respect to a given test. (Usually when this happens, they are testifying to different phases, one to screening for DNA and the other to profiling; given that the ultimate forensic result is one result (the identification of the DNA) – in contrast to the identification of two separate substances or two different tests of the same condition – it is appropriate for present purposes to consider this one test.) Of the 25 cases, 21 trials included live testimony by lab witnesses – one witness in 12 cases, two witnesses in eight cases, and three witnesses in one case. That is a total of 31 live witnesses in the 104 trials, or about .30 witnesses per trial; if we restrict our attention to the 25 cases in which DNA results were presented, the average is 1.24 witnesses per trial.

The Michigan appellate courts have been very good in insisting that a lab report cannot be introduced, over the objection of the accused, through the testimony of a surrogate witness who had no personal knowledge of the matters reported, and I understand from discussions with SADO lawyers that only rarely do Michigan prosecutors violate this rule. To double-check, I asked the students to look at the appellate opinions in the cases they examined to see whether a violation of the confrontation right was claimed. (Given the way SADO's briefs are filed, it was far easier finding the opinions rather than the briefs themselves.) In some of the newer cases, the court of appeals has not yet ruled (indicated by "Not found" in the last column on the spreadsheet), but in well over 200 of them it has, and in only one of those, People v. Frey, 2009 WL 2244521 (Mich. App.), appeal denied, 775 N.W.2d 788 (Mich. 2009) (marked in red on the spreadsheet), did the court discuss a claimed Confrontation Clause violation. (That court held there was no plain error when the analyst who performed part of a DNA test did not testify at trial – but the analyst who did testify at trial was present while the test was performed.) So it appears that only very rarely has there been what even the excellent, aggressive SADO lawyers believe is plausibly a confrontation violation.

I do not want to over-claim for this study. To make it feasible in a limited amount of time, I have relied on a large group of students, and my instructions developed somewhat as we learned more. I am sure, therefore, there is some variability in the way the cases have been reported. (Some, for example, affirmatively noted when lab results were admitted by stipulation, and some did not.) I suppose some small inaccuracies are inevitable. But I think the basic results are very strong, and would not be materially altered by discovering small glitches: In Michigan, the defense often consents to the presentation of lab results without the need for live testimony. In drug and DUI trials, even when a lab witness testifies at trial, there is virtually never more than one such witness per test, and when DNA results are presented through live testimony there is virtually always either one or two witnesses.

As in the prior study, what we cannot tell from this data is how often live witnesses are presented because that is the preference of the prosecution; for all that appears from the data, it could be that virtually always when a live lab witness testifies it is because the prosecution chooses to bring the witness in.