EXPERT SCIENTIFIC TESTIMONY IN COURTS:
THE IDEAL AND ILLUSION OF VALUE-FREE SCIENCE*
David S. Caudill
I. CONTOURS
The field of law, as a consumer of scientific knowledge, should be of particular interest to those who study science and argue about the role of values in scientific work. In agencies and legislatures (and in administrative and committee hearings), science is used to produce or support policy decisions; “scientific” conclusions about risk or harm, for example, are there intertwined with political arguments, cultural perceptions, and “value” judgments about the acceptability of evidence.[1] My own interest is in the use of science in litigation, and I am not alone – thousands of legal scholars over the last ten or fifteen years have sustained a debate about the appropriate standard for admissibility of expert scientific testimony in trials. One might even say that the “science wars” have arrived in law, but only a particular version of that debate is identifiable in law and science discourse.
With only slight exaggeration, until the U.S. Supreme Court’s Daubert[2] decision in 1993, the nature and limits of scientific discovery and practice was not a major issue in law. The so-called Frye[3] “general acceptance” test, from a 1923 federal court opinion, exemplified law’s deference to majoritarian views in the scientific community on questions of valid science. “Each scientific discipline had a different standard of reliability, which was accepted by the courts on its face.”[4] Concerns about bad or “junk” science in law were perennial, but legal scholarship often took the form of showing that laws were out of step with scientific progress. That is, no one questioned the colonization by the scientific community of the field of law.
Accounts vary as to why the U.S. Supreme Court took an interest in thinking about how science works, but it is clear that some wanted the Court to renew its commitment to mainstream science, so that we could avoid inconsistent verdicts in mass tort litigation[5] - Bendectin cannot cause birth defects in 1980, be safe in 1985, cause birth defects in 1990, etc. Obviously, judges and juries were being fooled in some of those cases. Others, however, highlighted the uncertainty of science and the need for courts to admit “novel” science that just might be valid despite its challenges to mainstream theories.[6] The Daubert decision satisfied neither side[7] - the Court acknowledged the slight possibility that novel science is good science, gave judges the role of deciding whether a theory is sound, and trusted the adversary system to keep bad science out of court. “General acceptance” remained a factor (but is not determinative), judges were allowed to admit conflicting scientific testimony, and juries were to determine which theory was correct.
The Daubert court’s definition of science was particularly interesting because of its representation of current views about science. Most of the amicus briefs, to educate the Court, contrasted an older, traditional view of science as certain with the new views of Kuhn and Popper, which briefs left the Court with the impression that not much has happened in the history and philosophy of science since 1970, and also with the impression that there is no such thing as sociology of science (or science studies, or SSK, or STS, etc.). Some who are knowledgeable in the “science wars” may think that it was appropriate for the Court to ignore social studies of science, but even critics of the sociology of scientific knowledge had to be baffled by the notion that Kuhn and Popper closed the book on how science works. In any event, a review of post-Daubert law and science discourse reveals that (i) a few scholars think that Daubert gives judges and juries too much power to evaluate science, (ii) a few scholars think that we should not defer to science since scientific facts are socially constructed, and (iii) most do not think at all about values in judicial determinations of scientific validity. If pressed, the latter group would acknowledge some discretion in science (methodological norms) and law (the methodology seems sound), but the ideal is good, value-free, society-free, legal-construction-free, science.
II. SCIENCE IN LAW
In one sense, the use of scientific expertise in courts of law to determine a fact is clearly an act of social construction – (a) the expert is always interested, having been hired to argue for a particular position; (b) the expert is presented by a lawyer as part of a rhetorical and adversarial project; (c) the reliability of the expert’s opinion is evaluated by a judge, whose role is highly institutionalized and whose training is rarely in science; and (d) the validity of the expert testimony is determined by a jury, another social institution within the social institution of law. Thus, for example, if there is uncontroverted evidence that high-level asbestos exposure causes harm, and an expert opines (in a particular case) that low-level exposure caused harm, a judge may admit that testimony and let the jury determine causation.[8] Was the plaintiff actually harmed by low-level asbestos exposure? Who knows? If the jury decides in favor of the plaintiff, the defendant will pay damages as if the asbestos caused the harm, since the legal fact of causation was constructed in court that day.
In another sense, the use of scientific expertise in courts is precisely an effort to avoid social constructions of scientific facts. We are concerned about the persuasiveness of junk or “courtroom” science, about inconsistent jury verdicts on causation, and about “authentic insights and innovations” that may not, for various reasons, make it into court.[9] Indeed, we pay a price, in terms of public confidence, for bad verdicts, even though they result from institutional limitations – courts must decide quickly, relying only on the evidence before them and on lawyers’ ingenuity. Bad verdicts are generally, however, viewed as lapses – as anomalies that do not call into question the ideal of science as a stable source of insights for law.
When the U.S. Supreme Court in Daubert rejected the “generally accepted” standard or Frye test for admissibility of expert scientific testimony, many were troubled. Some thought that this old idea, that courts should not accept scientific evidence until the institutions of science accept it, was a good one. Indeed, the Daubert opinion recognized “general acceptance” as an important, but not exclusive, factor in reliability determinations. On the other hand, some thought that we should stop deferring to mainstream science, since good novel theories are by definition not generally accepted. The Daubert court, however, only acknowledged that in the rare case a novel theory may be right, and viewed with suspicion any theory that has not yet been able to garner support in the scientific community.
The Daubert opinion concluded with some general observations about the nature of science that have become the so-called four-part test: scientific knowledge is testable, its techniques have low rates of error, it is usually published and subject to peer review, and it is usually generally accepted.[10] Somewhat problematically, these characteristics of science have become in later opinions a set of non-exclusive factors to consider, since there are other relevant indices of reliability, such as whether research was done for trial or independently, whether a conclusion is not justified by the premises or data of an expert, whether alternative explanations have been considered, whether an expert is being as rigorous in court as he or she is in his or her regular work, and whether a field of expertise is itself known to be unreliable.[11] After Daubert, the problem remained that some judges took the four-part test literally as a definition of science, while others felt free to pick and choose from the four factors, or other factors, to decide whether an expert was being scientific.[12] Most importantly, Daubert assigned to judges the role of gatekeeper to make reliability determinations. Instead of deferring to science, judges have considerable leeway not only in reliability determinations, but in how each decides to make reliability determinations.[13] That is why some judges and legal scholars have proposed science panels whose pronouncements would be law-like, rather than factual, since judges theoretically defer to law but decide (or instruct the jury to decide) facts.[14] Daubert, in contrast, trusts the adversary system to ensure that if bad science gets past the judicial gatekeeper, good cross-examination and compelling arguments can guard against bad verdicts.[15]
Literally thousands of law review articles since 1994 have considered and reconsidered Daubert and later opinions, often critically, and the federal rules of evidence have now been amended to correct and clarify the standard of admissibility for expert scientific testimony.[16] The new three-part test admits testimony that (1) is based on sufficient facts or data, (2) is the product of reliable principles and methods, and (3) applies the principles and methods reliably to the facts of the case.[17] In short, it is methodology that characterizes science, though the data must be sufficient and the application at hand reliable.
Again, the turn to methodology can be seen as another failed attempt to define science. Professor Gary Edmond, for example, argues that:
Because various sciences maintain different approaches, theories, criteria, canons of practice, metaphysics . . . , levels of relevance, levels of abstraction, and so on, it would be highly naďve to suggest that we could expect some basic or universal criteria which could be applied consistently to determine “reliability.”[18]
Thus, for
The efforts of the Advisory Committee on the Federal Rules of Evidence to set new standards for admission of expert testimony are, on the one hand, a social constructivist’s dream. Judicial discretion as to (a) whether an expert’s technique or theory has been objectively tested, (b) when an error rate is acceptable, (c) the weight given to peer review and publications, (d) the degree of acceptance necessary for “general acceptance” within a field, (e) whether the testimony is based on “science for litigation” and not on independent research, (f) whether the “analytical gap” between premise and conclusion is “too great,” (g) whether alternative explanations are “obvious,” (h) how careful the expert is in court with respect to his or her usual professional work, (i) whether a discipline generally lacks reliability, (j) whether evidence will confuse the jury, (k) the perceived dangers of novel evidence, and (l) how much weight to give any of the above determinations, all point to the manipulability of evidentiary standards and categories. Of course, if a trial judge makes a mistake in evaluating reliability, the decision can be appealed, but the standard of review on appeal is whether there was “abuse of discretion.”[24] This standard is generally viewed as deferential to the trial judge, so much so that in Joiner v. General Electric, Justice Stevens pointed out that while it was not an abuse of discretion to disallow certain expert testimony, it would not have been reversible error to allow it.[25] The “abuse of discretion” standard is not friendly to those bringing appeals.[26]
On the other hand, the recent efforts to stabilize reliability standards also provide an opportunity for critics of social constructivism to demand a more rigorous scientific standard in law. The notion that courts will admit contradictory evidence is an invitation for contradictory verdicts, and arguably we should not allow juries to adopt scientific theories that are contrary to those of the scientific community. Moreover, allowing a judge to determine the reliability of scientific techniques and propositions betrays an unfounded confidence in the judiciary. So far, such criticism has been unsuccessful, but not because the Advisory Committee recognizes the social construction of scientific facts. To the contrary, the Committee never suggests that science itself is interested, rhetorical, or biased due to institutional training, practices, and standards, except in rare instances. Rather, its goal is to ensure that socio-legal constructions of science mirror real science.
III. POSITIONS
The three positions in the ongoing legal debate about expert scientific testimony can be characterized as the “current view” (represented by the new Rules of Evidence), the reaction by social constructivists, and the reaction by defenders of the scientific establishment. The current view combines a traditional commitment to science (as a producer of stable, useful knowledge) with a commitment to a mediating or filtering role for judges, the latter of which is simply viewed as necessary. Law has its own jurisdiction, its own problems, and its own procedures – law cannot wait for science when a case arises, so it must create its own science. The ideal, nevertheless, is that science is science, and that “legal science” should always be “good science.”
In reaction, social constructivists reject the idealization of science and concede that all science, in court or otherwise, is a social, institutional, cultural, and rhetorical affair. Cynically, we might say that the social constructivist has no problem with creating “legal science” to serve the ends of justice, without regard to mainstream science, because mainstream science is itself serving social ends all the time.
Instead of rejecting the idealization of science in the current view, the second set of critics rejects the necessity of mediation by judges, and doubts the capacity of lawyers and juries to challenge and reject junk science. Scientific knowledge is not a proposed fact to be accepted or discredited, like a witness’s testimony, but a law-like reality to be acknowledged at the outset of trial. Just as courts announce a law (“If you caused the injury, you are liable”), courts should announce the findings of mainstream science (“Bendectin does not cause birth defects”). Nor should reliability be decided by a preponderance-of-evidence (i.e., a 51% standard); the burden of proof among scientists is more like beyond-a-reasonable-doubt (i.e., “95 percent accuracy.”)[27]
At first glance, it appears that only the proponents of law-like scientific testimony are requesting a change in practice. Social constructivists should be happy with the current state of affairs. Yet social constructivists are concerned that the commitment to an idealized view of science will cause judges, in particular cases, to exercise discretion to get the results that the proponents of law-like science want. All of that discretion means, potentially, that judges who simply do not believe that two theories in conflict can both be admissible will be suspicious of novel science. The result will be that the ideal of value-free science will be dominant, and we will reinstate law’s unjustified deference to science. In the words of Professor Paul Rice:
The reality is that judges are compelled to return to the same relevant scientific communities for answers. Daubert is little more than Frye in drag. Judges simply retool Frye by anointing a single expert and substituting that expert’s opinions for those of the relevant scientific community. The only difference is that the judge lays out the criticism and concerns of the expert as if they were the judge’s own . . . .[28]
Professor Rice, by the way, recommended a public debate on these issues; he suggested that “Congress should reassert its authority over the rules” of evidence, since the Advisory Committee did not provide sufficient guidance to judges as to standards of scientific reliability.[29] I agree, but I hope that in future debates, lawyers and judges would acknowledge and consult with science studies scholars. I noticed, however, that the institutions of law – the U.S. Supreme Court, the Advisory Committee – were in a hurry to settle things. We want a simple story of the past, simple representation of the “new” view of science, and a “progressive” standard. The ongoing debates in science studies, once revealed, make that discipline seem unstable. We have so many problems in law, it seems, that we do not need to import new ones.
* This paper was delivered at the
Conference on “Value-Free Science: Ideal or Illusion?” held February
23-25, 2001, at the
[1] See generally Acceptable Evidence: Science and Values in Risk Management (Deborah Mayo & Rachelle Hollander eds., New York: Oxford Univ. Press, 1991).
[2] See Daubert v. Merrell
Dow Pharmaceuticals, Inc., 509
[3] See Frye v.
[4] Paul R. Rice, Truth in Test Tubes, Legal Times, Oct. 16, 2000, at 85.
[5] See Michael D. Green, Bendectin and Birth Defects: The Challenges of Mass Toxic Substances Litigation 306-11 (Philadelphia: U. Penn. Press, 1996).
[6] See Edward J. Imwinkelried,
[7] See Green, supra note 5, at 310: “The [Daubert] Court’s opinion contains something for everyone.”
[8]
[9] The
[10] See Daubert,
509
[11] See Communication From the Chief Justice, The Supreme Court of the United States, Transmitting Amendments to the Federal Rules of Evidence . . ., May 2, 2000 (U.S. Gov’t Printing Office), at 42-44.
[12] See David
[13] See Kumho Tire v. Carmichael,
526
[14] See Michael J. Saks, The Aftermath of Daubert: An Evolving Jurisprudence of Expert Evidence, 40 Jurimetrics 229 (2000).
[15] See Daubert, 509
[16] See Fed R. Evid. 702.
[17]
[18] See Gary Edmond, Judicial Representations of Scientific Evidence, 63 Mod. L. Rev. 216, 251 (2000).
[19]
[20] See Fed R. Evid. 702.
[21] See Communication, supra note 11, at 44-45.
[22]
[23]
[24] See Joiner v. General
Electric, 522
[25] 522
[26] See Ruiz-Troche v.
Pepsi Cola of
[27] See Rice, supra note 4, at 84.
[28]
[29]