Professional Documents
Culture Documents
Evidence
test for you. Please answer
each of the three questions
below True or False.
1. Scientific tests
by Jonathan J. Koehler conducted over the past
100 years have repeatedly
demonstrated that every-
one has a unique set of
fingerprints.
2. Recent scientific stud-
ies show that the chance
that DNA samples from two
different people will be
identified as a “match” by
a competent, well-trained
DNA examiner is less than
one in a million.
3. Data from scientific
tests conducted over the
past few decades provide a
reliable basis from which
JUDICATURE
of most forensic methods
that have been admitted in
Published by the Bolch Judicial Institute at Duke Law. Reprinted
with permission. © 2018 Duke University School of Law.
U.S. courts.
All rights reserved. judicialstudies.duke.edu/judicature
JUDICATURE 29
The answer to all three of these ques- that have long played an important role to trial judges who are asked to admit
tions is false. How did you do? in our criminal justice system. If they do expert testimony of any sort.
If you read the 2009 report from the so, they will likely find that the (disinter- Like the Daubert opinion itself, the
National Academy of Sciences on foren- ested) scientific community will provide 2000 amendment to FRE 702 empha-
sic science1 or the 2016 report from a very different perspective on the extent sizes the central role that “reliability”
the President’s Council of Advisors on to which forensic science claims have must play in admissibility decisions
Science and Technology (PCAST),2 you stood up to empirical testing than the related to expert testimony. The princi-
probably got all of the questions right. perspective provided by the interested ples that underlie an expert’s testimony
These authoritative reports, investigated examiners who provide forensic science must be reliable, the method an expert
and written by leading scientists in the testimony at trial. uses must be reliable, the application of
U.S., indicate that our forensic sciences those reliable principles and methods to
are badly in need of scientific testing. ROADMAP TO RELIABILITY: DAUBERT & FRE 702 the instant case must be reliable, and the
They also indicate that many of the In its broadest sense, forensic science is testimony must be based on facts and
strong claims made by forensic scientists the application of science to law. Over the data which, presumably, must also be
and their proponents are misleading in past century or so, many different types of reliable. In short, unlike most other forms
light of the lack of scientific data to back forensic science results have been admit- of evidence, expert testimony — includ-
up those claims. ted in U.S. courts, including evidence ing forensic science expert testimony
But who really reads such reports? from fingerprints, palm prints, voice — is inadmissible unless the evidentiary
And who really understands that there prints, DNA, microscopic hair, ballistics, proponent can affirmatively show that it
is not enough science to justify a lot of toolmarks, document examination, shoe is reliable in a variety of ways.
forensic science claims? Certainly not prints, tire tracks, bitemarks, soil, glass, The emphasis FRE 702 places on
the American public. I presented those paint chips, carpet fibers, blood spatter, reliability begs the questions of what it
three statements to a random sample of and more. In the near future, prosecutors means for expert testimony to be reli-
322 jury-eligible Americans and found may seek to introduce biometric tech- able and how a judge can determine
that each statement was judged to be niques including evidence from gaits, whether or not proffered testimony is
true by more than four out of five people, veins, irises, retinas, etc. reliable. The Daubert decision provided
and nearly two out of three people (64 Federal Rule of Evidence (FRE) some useful guidance for assessing the
percent) thought that all three state- 702 (or its state equivalent) governs reliability of proffered scientific expert
ments were true.3 Another recent study the admissibility of expert testimony, testimony in general, and this guidance
asked people to estimate the chance that including testimony pertaining to foren- is applicable to proffered forensic science
a forensic examiner will err in each of five sic analyses. The first part of FRE 702 testimony. Specifically, Daubert and the
different forensic sciences.4 The median essentially requires that an expert witness advisory notes that accompany FRE 702
estimates ranged from one chance be qualified and provide testimony that indicate that a trial judge may consider
in 100,000 (for document examina- will assist the trier of fact. The latter part (1) whether the expert’s theory or method
tion) to one in 10,000,000 (for DNA). of FRE 702, adopted in a 2000 amend- has been tested, (2) whether the theory or
Apparently, then, most people believe ment, provides additional restrictions on method has been subject to peer review
that forensic science results and conclu- the admissibility of expert testimony. and publication, (3) the method’s error
sions are extremely accurate and that FRE 702(b) requires that “the testimony rate, (4) whether the method is a stan-
reliable scientific studies back up those is based on sufficient facts or data.” FRE dard one with controls, and (5) whether
beliefs (see question no. 3 at left). 702(c) requires that “the testimony is the theory or method has been generally
Whether trial judges know differ- the product of reliable principles and accepted in the scientific community.
ently is an open empirical question. But methods.” FRE 702(d) requires that “the Daubert’s guidance for assessing reli-
regardless of what trial judges know expert has reliably applied the principles ability is sensible, but vague. How
(or think they know) about the forensic and methods to the facts of the case.” exactly is a trial judge to know whether
sciences, they should look to the broader The year 2000 amendment to FRE a forensic theory is sound, an error rate
scientific community for assistance when 702 was offered in response to Daubert is comfortably low, or a laboratory’s
evaluating the reliability of any proffered v. Merrell Dow Pharmaceuticals, Inc.,5 a controls can be trusted? Although these
forensic method, including methods case that conferred a gatekeeper function questions must be answered in a legal 4
30 VOL. 102 NO. 1
arena by a trial judge, they concern tees that examiners will obtain the same As noted earlier, in cases involving
matters that are fundamentally scientific results and draw the same conclusions.10 forensic science evidence (or any other
in nature. As such, judges should look to The report also noted that there is no form of expert evidence), FRE 702 tells
the broader scientific community for guidance standard vocabulary to describe results,11 judges that the principles and methods
when deciding whether proffered scientific which may lead to “imprecise or exag- that were used to create the evidence must
evidence is sufficiently reliable to justify its gerated expert testimony.”12 Notably, be reliable, and that the expert testimony
admissibility at trial.6 That is, trial judges the NAS report took the courts to task related to the evidence must be backed
should lean heavily on the broader scien- for being “utterly ineffective” at pushing up sufficiently by reliable facts or data.
tific community for methodological any of the forensic sciences to test their FRE 104(a) gives judges the authority to
advice about how to determine whether claims and to otherwise conduct them- rule on the admissibility of this evidence,
a scientific technique or claim is suffi- selves in a more scientific manner.13 and judges make this determination
ciently backed up by reliable principles, based on a preponderance of the available
methods, facts, and data. In cases involv- 2016 PRESIDENT’S COUNCIL ON SCIENCE AND evidence. The PCAST report leaves these
ing forensic science methods and claims, TECHNOLOGY (PCAST) REPORT legal standards alone, and instead weighs
such guidance is readily available from a The 2016 PCAST report picks up where in on how a judge (or anyone else) can
2009 report by the National Academy of the 2009 NAS report left off by provid- determine whether a principle, method,
Sciences (NAS) and a 2016 report from ing trial judges with specific guidance or purported fact is scientifically reliable
the President’s Council on Science and for assessing the scientific reliability and valid. In doing so, the PCAST report
Technology (PCAST).7 and validity of proffered forensic science fills a void left by both Daubert and the
evidence. PCAST “is an advisory group 2009 NAS report. It offers clear guidance
2009 NATIONAL ACADEMY OF SCIENCES of the Nation’s leading scientists and to courts from esteemed representatives
(NAS) REPORT engineers, appointed by the President of the scientific community. As part of
The 2009 NAS report on the non-DNA to augment the science and technology this guidance, the report distinguishes
forensic sciences sent shock waves advice available to him from inside the “foundational validity” from “validity as
through the criminal justice system. White House and from cabinet depart- applied” in practice.18 A method is foun-
This report, which was written by a ments and other Federal agencies.”14 In dationally valid if and only if it has been
group of the nation’s elite scientists, 2015, President Obama asked PCAST “shown, based on empirical studies, to
statisticians, judges, and other scholars, to provide advice and recommenda- be repeatable, reproducible, and accurate
concluded that, “[l]ittle rigorous system- tions “that could usefully be taken on . . . under conditions appropriate to its
atic research has been done to validate the the scientific side to strengthen the intended use.”19
basic premises and techniques” in many forensic-science disciplines and ensure These words should not be construed
forensic disciplines. The report detailed the validity of forensic evidence used as highfalutin’, overly cautious, scientific
how many forensic sciences — including in the Nation’s legal system.”15 The mumbo jumbo. The foundational valid-
impression evidence, toolmark and fire- focus of this report was the “forensic ity standard applies to all sciences, and it
arms analysis, microscopic hair evidence, ‘feature-comparison’ methods,”16 which is especially important that it be under-
questioned document examination, include DNA, hair, fingerprints, fire- stood and applied in cases involving
and forensic odontology — “have yet arms, toolmarks, bitemarks, shoe prints, forensic science evidence where match
to establish either the validity of their tire tracks, and handwriting. The elite determinations20 are typically subjective
approach or the accuracy of their conclu- scientists who wrote the report — none judgments made by individual examin-
sions”8 and called for a “major overhaul” of whom served on the 2009 NAS ers. In referring to match determinations
of the U.S. forensic science system.9 The committee described above — indi- as “subjective judgments,” I do not
report repeatedly stated that there is little cated that their focus was on helping mean to imply that there is no basis for
scientific data to indicate the reliabil- judges understand scientific standards those judgments or that the judgments
ity and accuracy of the methods used in for assessing scientific validity, not on are as likely to be right as wrong. I
many forensic sciences. For example, the dictating the legal standards pertain- simply mean that a person, as opposed to
report noted that the standard fingerprint ing to the admissibility of scientific a machine or computer program, makes
method (ACE-V) does not guard against evidence.17 The distinction is subtle but one or more key determinations — such
bias and provides insufficient guaran- important. as which portion of a marking to exam-
JUDICATURE 31
overwhelming evidence of strength and Association for Identification “finds the these words do not seem to be those of a
reliability in forensic science.”35 report lacking in basis and content, and committed forensic-science foe.
improper in some of the statements that 2) PCAST offered an overly narrow and
AMERICAN SOCIETY OF CRIME LABORATORY are made.” 44 The Bureau of Alcohol, idiosyncratic definition of scientific valid-
DIRECTORS Tobacco, Firearms and Explosives ity: This effort by critics of the PCAST
The American Society of Crime expressed its “disappointment in the Report to broaden the scope of what
Laboratory Directors (ASCLD) chal- flawed methodology PCAST employed,” constitutes scientific validity must be
lenged PCAST’s definition of a saying that PCAST “did not adequately rejected. Part of what makes the PCAST
scientifically rigorous “black box” consider the numerous research studies report so helpful to courts is that it
validation study as “arbitrary” and that support the validity of firearm and provides clear, sound, and practical guid-
“unhelpful.”36 ASCLD also argued that tool mark forensics.”45 ance about exactly what judges should
forensic science practitioners should look for when considering the scientific
have a hand in the design and conduct CRITIQUING THE PCAST CRITICISMS matter of the foundational validity of a
of the scientific studies to foster “true The sheer volume of professional and method that involves substantial human
advancement . . . of forensic science.”37 government organizations and represen- judgment. In a nutshell, PCAST reminds
tatives that have taken issue with the the world about the wisdom of what
MIDWESTERN ASSOCIATION OF PCAST findings is superficially impres- we learned in our high school science
FORENSIC SCIENTISTS sive. But, of course, it is the logical and classes: the way to know if something works
The Midwestern Association of Forensic scientific merit of those responses that as advertised is to subject it to rigorous and
Scientists (MAFS) characterized must be critiqued, not their volume. repeated empirical testing under conditions
PCAST’s conclusions as “capricious.”38 I have identified six distinct points that are similar to those in the natural envi-
Like ASCLD, MAFS suggested that the raised by the various PCAST critics. I ronment. This has not been done for most
empirical testing methods that PCAST comment on the merits of each of those of the forensic sciences. When PCAST
outlined “are not the only scientific way points below. critics suggest that the daily “experi-
to ensure validity and reliability.”39 Also ence” of forensic examiners vouches for
like ASCLD, MAFS indicated that the 1) The PCAST committee was biased the scientific validity of their work, it
“[e]xperience and daily observation” of against forensic science: It should go with- is important to remind ourselves that
examiners is part of a scientific measure out saying that ad hominem attacks on this is not how science works. Personal
of reliability.40 They wrote, “[t]o not a properly convened scientific commit- experience is no substitute for empirical
include practitioners in the discussion tee are inappropriate and unpersuasive. testing. This doesn’t mean that experi-
would be irresponsible.”41 The PCAST committee, like the NAS ence is worthless. If consumer reviews
committee before it, included some of on Amazon indicate that a weight loss
OTHER FORENSIC ORGANIZATIONS the most accomplished scientists of our pill worked wonders for some people, a
Many of the arguments raised above era. Few of the committee members are potential customer has some justification
were echoed in responses from others primarily focused on forensic science for expecting the pill to help him or her
in the forensic science community. The issues outside of their committee work, lose weight. But these reviews, which
Association of Firearm and Tool Mark and there is nothing in the backgrounds spring from the personal experiences of
Examiners asserted that “[d]ecades of of the committee members as a whole consumers, do not constitute scientific
validation and proficiency studies have that supports a charge of bias. The proof that the pill actually works. The
demonstrated that firearm and toolmark PCAST committee chair, Eric Lander, scientific validity of a claim that a pill
identification is scientifically valid.”42 co-authored a frequently cited paper causes weight loss — or that a forensic
The Organization of Scientific Area in the prestigious journal Nature two method yields true results — can only
Committees Materials Subcommittee decades ago that concluded as follows: be proven using justified, widely agreed
stated that lack of information about an “[T]he DNA fingerprinting controversy upon scientific methods and standards.
error rate for microscopic hair compari- has been resolved. There is no scientific 3) PCAST ignored strong evidence that
son evidence “should not be interpreted reason to doubt the accuracy of forensic proves the scientific validity of various foren-
to suggest that the discipline is any DNA typing results . . . .” Although this sic sciences: In response to this potentially
less scientific.”43 The International conclusion may have been premature, devastating charge, PCAST invited the 4
34 VOL. 102 NO. 1
FBI and other agencies who made this (1999),48 the gatekeeping function for suggest otherwise blatantly distorts the
claim “to identify any ‘published . . . trial courts identified in Daubert extends shared understanding among scientists
appropriately designed studies’ that had to all expert testimony offered under FRE of what it means for a method to have
not been considered by PCAST and that 702. This means that expert testimony, been scientifically validated.
established the validity and reliability of whether scientific or not, must still be In sum, a critique of the criticisms
any of the forensic feature-comparison reliable to be admissible. Although trial leveled against the PCAST report
methods that the PCAST report found to judges have latitude when assessing the supports the view that PCAST and the
lack such support.” No such studies were reliability of expert testimony, a lower NAS have it right: An assessment of
provided. Indeed, the FBI ultimately reliability standard is not automatically the reliability and validity of the foren-
conceded that there were no such studies in play once the evidentiary proponent sic sciences requires testing, and many
after all.46 Nevertheless, forensic scien- declares a willingness to have its evidence of those tests have yet to be performed.
tists commonly offer sworn testimony reviewed as non-science for admissibility As a result, we know surprisingly little
that relevant validation studies exist purposes. Regardless of whether forensic about how accurate forensic science
and that they personally believe that the science is characterized as 100 percent testimony is.
method in question is reliable. Needless science, part science and part technical
to say, such testimony does not suffice as knowledge, or 100 percent technical ERROR RATES: WHAT DO WE KNOW?
proof of scientific validity under the stan- knowledge, the reliability and validity If the legal standard for admitting foren-
dards imposed by Daubert and FRE 702. of the methods used by examiners to sic science evidence is followed, then the
4) PCAST usurped the role of judges reach their subjective conclusions must evidentiary proponent must show that
and juries by inserting its own opinions be demonstrated affirmatively. the methods that produced the forensic
about forensic science: As noted previ- 6) Practitioners’ personal experiences and result are themselves reliable. The most
ously, the PCAST report was quite clear observations should be given weight when important indicator of the reliability of
about differentiating between scientific assessing the scientific validity of foren- a forensic method is the rate at which
matters pertaining to forensic science sic science: This claim, which also lies trained examiners who use that method
that were clearly within its charge, and behind the critics’ claim in point 2 err: the lower the error rate, the greater
legal matters that did not concern either above, reveals how “motivated reason- the reliability of the method. Of course,
PCAST or the broader scientific commu- ing”49 can distort the judgments of in an actual case in which an unknown
nity.47 Whereas legal policymakers, professionals. The experiences and print or marking is compared to one or
judges, and juries must decide matters casual observations of forensic scientists more knowns, ground truth is absent. In
such as general admissibility standards may aid future scientific study by, for such cases, we cannot be sure whether a
for scientific evidence and whether a example, identifying hypotheses, ideas, correct result is achieved because there is
proffered method has met those legal patterns, correlations, etc. But experi- no independent way to verify the accu-
standards, scientists are best positioned ences and unsystematic observations must not racy of the examiner’s conclusion. But in
to advise on the scientific standards for be confused with systematic scientific study. a properly designed test in which prints
scientific validity. Judges must firmly reject the notion or markings are produced from recorded
5) Forensic science evidence should not be that experience — even a great deal of knowns, ground truth is available, and
held to scientific standards of validity because it — contributes to the scientific vali- an examiner’s error rate (or a laboratory’s
the evidence includes technical or specialized dation of a method. People experience error rate or the error rate of a method in
knowledge: Ordinarily, forensic science and observe many things that system- general) may be computed.
supporters are keen on promoting their atic study later proves to be spurious Unfortunately, and perhaps surpris-
disciplines as scientific. It is therefore or false.50 When assessing the scientific ingly, such tests are virtually nonexistent
puzzling to see the NDAA argue that validity of a method involving human in the world of forensic science. This
their evidence should be assessed using judgment, systematic, rigorous empiri- means that there is little basis for
standards that are more lenient than cal testing — scientific testing — is not estimating error rates for any foren-
those used for other sciences. Whether an option: It is a requirement. There are sic science method. As a result, courts
this maneuver is regarded as clever or no shortcuts, and the day-to-day work cannot make a properly informed judg-
desperate, it should fail. As the Supreme experiences of examiners are not a legiti- ment about the reliability of a proffered
Court noted in Kumho Tire v. Carmichael mate substitute for empirical testing. To forensic method.
JUDICATURE 35
39 Amer. J. Pub. Health S107, S110 (2005). Opinion: Demonstration of Cognitive Bias on a
Id. at 1.
40 53
Collaborative Testing Services, CTS Statement Fingerprint Matching Task Through Knowledge of
Id. at 2.
on the Use of Proficiency Testing Data for Error DNA Test Results, 276 Forensic Sci. Int’l 93
41
Id. at 2. (2017).
Rate Determinations at 2 (Mar. 30, 2010).
42
Ass’n of Firearm and Tool Mark Exam’rs, https://www.ctsforensics.com/assets/news/ 61
Dahlia Lithwick, Crime Lab Scandals Just Keep
Response to PCAST Report on Forensic CTSErrorRateStatement.pdf. Collaborative Getting Worse, SLATE (October 20, 2015).
Science 1 (2016). Testing Services provides testing materials 62
The issue of how and how well fact-finders will
43
Org. of Sci. Area Comms. Materials to laboratories across the forensic sciences. use error rate data if and when it is provided
Subcomm., Response to PCAST Call for Occasionally, a study is designed to identify to them may be complicated. See Nicholas
Additional References from OSAC the rate at which examiners err. For example, a Scurich, The Differential Effect of Numeracy and
Materials Subcommittee 2 (n.d.). study by Ulery and his colleagues was designed Anecdotes on the Perceived Fallibility of Forensic
44 to assess the rate at which latent fingerprint Science, 22 Psychiatry, Psychology &
Int’l Ass’n for Identification, IAI
examiners commit different types of errors. See Law 616 (2015), for a review of the empiri-
Response to the Report to the
Bradford T. Ulery, R. Austin Hicklin, JoAnn cal literature, and a finding that proficiency
President ‘Forensic Science in Criminal
Buscaglia, & Maria Antonia Roberts, Accuracy with numerical information (i.e., numeracy)
Courts Ensuring Scientific Validity of
and Reliability of Forensic Latent Fingerprint impacts fact-finders’ use of error rates.
Feature-Comparison Methods’ Issued
Decisions, 108(19) Proc. Natl. Acad. Sci.
by the President’s Council of Advisors 63
Nancy Gertner, Commentary on the Need for
7733 (2011). However, given that the partic-
on Science and Technology (PCAST) in a Research Culture in the Forensic Sciences, 58
ipants in the study were volunteers who knew
September 2016 1 (n.d.). UCLA L. Rev. 789, 790 (2011).
they were being tested, that the study was
45
Bureau of Alcohol, Tobacco, Firearms paid for by an interested party (the FBI), and
& Explosives, ATF Response to the
President’s Council of Advisors on
that some of the authors work for the FBI, the
results of the study should be viewed with
PUBLICATIONS FROM THE
Science and Technology Report 1 (2016). caution. See Jonathan J. Koehler, Forensics or BOLCH JUDICIAL INSTITUTE
46
2017 PCAST Addendum, supra note 19, at 5. Fauxrensics? Ascertaining Accuracy in the Forensic
47
Sciences 49 Ariz St. L. J 36-38 (2018). • Revised Guidelines and Practices for
See supra n. 17. See also 2016 PCAST Report, 54
Collaborative Testing Services, supra note 53. Implementing the 2015 Discovery
supra note 2, at 21 n.11 (“In this report,
at 3. Amendments to Achieve Proportionality:
PCAST addresses solely the scientific standards
for scientific validity and reliability. We do not 55
Spencer S. Hsu, FBI Admits Flaws in Hair
http://bit.ly/proportionality_nov17
offer opinions concerning legal standards.”) Analysis Over Decades, Wash. Post, Apr. 18,
48 2015. • Standards and Best Practices For Large
Kumho Tire Co. v. Carmichael, 526 U.S. 137
56 and Mass-Tort MDLs:
(1999). Nat’l Acad. of Scis., Forensic Analysis:
49 Weighing Bullet Lead Evidence (2004). http://bit.ly/MDLbestpractices
Ziva Kunda, Motivated Inference: Self-serving
Generation and Evaluation of Causal Theories, 53 57
Joe Palazzolo, Texas Commission Recommends Ban
J. Pers. & Soc. Psych. 636 (1987). on Bite-Mark Evidence, Wall St. J., Feb. 12 FORTHCOMING:
50 2016. • EDRM Guidelines for Using Technology-
Examples include the Ptolemaic model of
58 Assisted Review (TAR) in Discovery
the solar system (which had the earth at the Office of the Inspector Gen., Oversight
center), the flat earth theory, and even the & Review Div., U.S. Department of • Guidelines and Best Practices
recently discredited theory that ulcers are Justice, A Review of the FBI’s Handling Implementing 2018 Amendments to
caused by stress (they are caused by bacteria). of the Brandon Mayfield Case 1–4 (2006), Rule 23 – Class-Action Settlement Provision
51
Frye v. U.S., 293 F. 1013 (D.C. Cir. 1923). https://oig.justice.gov/special/s0601/final.pdf.
52
59
Itiel E. Dror, David Charlton, & Ailsa E.
• Standards and Best Practices for
William Thompson, John Black, Anil Increasing Diversity in MDL and Class-
Jain, & Joseph Kadane, Forensic Science Peron, Contextual Information Renders Experts
Vulnerable to Making Erroneous Identifications, Action Leadership Positions
Assessments: A Quality and Gap Analysis:
Latent Fingerprint Examination 1 (2017) 156 Forensic Sci. Int’l 74 (2006). • Guidelines and Best Practices Addressing
(“Serious questions have been raised, however, 60
Itiel E. Dror & Greg Hampikian, Subjectivity Issues in Securities Class Actions
about how well judges have performed this and Bias in Forensic DNA Mixture Interpretation,
[gatekeeping] role”); Simon A. Cole, Toward 51 Sci. & Justice 204 (2011); Jan W. de The Duke Conference series at the Bolch Judicial
Evidence-Based Evidence: Supporting Forensic Keijser, Marijke Malsch, Egge T. Luining, et Institute prepares best practices in areas of impor-
Knowledge Claims in the Post-Daubert Era, 43 al., Differential Reporting of Mixed DNA Profiles tance to the judiciary and legal profession. Find
Tulsa L. Rev. 263, 277; Peter J. Neufeld, The and its Impact on Jurists’ Evaluation of Evidence,
them at judicialstudies.duke.edu/conferences/
(Near) Irrelevance of Daubert to Criminal Justice: 23 Forensic Sci. Int’l: Genetics 71 (2016);
And Some Suggestions for Reform, 95 (Supp. 1) publications.
Sarah V. Stevenage & Alice Bennett, A Biased