Bias in privately funded scientific research has been much in the news recently.(1) Research bias causes problems in various contexts, from medical decisionmaking(2) to regulatory evaluations, (3) but the problems it presents are particularly acute in litigation. Scientific evidence is difficult for lay fact finders to assess under the best of circumstances. Bias exacerbates this difficulty because the same unfamiliarity with the practice of science that makes it difficult for laypersons to assess scientific evidence also makes it difficult for them to appreciate how bias can corrupt that evidence. Therefore, even though bias in general is a problem with which the adversary process is familiar, bias in scientific research may require an approach tailored to the particular difficulties that arise in that context.
Of course, the problem of bias in scientific evidence has not gone unnoticed by courts and commentators.(4) But most efforts to address the problem have focused on biased witnesses rather than on biased researchers.(5) These efforts, in other words, have focused on bias introduced in the process of testifying and have assumed that the underlying research record itself is unbiased.(6) An important example, because of its influence, is the Ninth Circuit's opinion in Daubert v. Merrell Dow Pharmaceuticals, Inc.(7) In that opinion--to which this Article will refer as Daubert II--the court asked "whether the experts are proposing to testify about matters growing naturally and directly out of research they have conducted independent of the litigation, or whether they have developed their opinions expressly for purposes of testifying."(8) The court then went on to treat research conducted independent of litigation as reliable, ignoring the many other, nonlitigation sources of research bias.(9)
The weakness in this approach is apparent when one considers that a court would admit, apparently without further scrutiny, testimony regarding research funded by parties to litigation, so long as the research was not conducted in connection with the litigation. For example, in a suit challenging the safety of a drug, it would admit--again, without further scrutiny--research on the safety of the drug that was funded by its manufacturer, if the research predated the litigation. Yet the scientist conducting that research would be no less aware of the result desired by the manufacturer than would a scientist conducting similar research in connection with the litigation. Hence, Daubert II's litigation-based test does not draw an appropriate line for admissibility.
Daubert II is surely correct, though, in treating bias as part of the more general problem of evidentiary reliability that was the focus of the Supreme Court in its Daubert opinion.(10) In Daubert, the Court held that, when considering scientific evidence, "evidentiary reliability will be based on scientific validity."(11) Bias certainly is an element of reliability; therefore, although Daubert did not discuss bias problems specifically, the opinion at least suggested that the Court would have bias issues decided in court as science decides them. Indeed, science has adopted procedures for dealing with bias that are in some respects similar to other scientific practices--such as peer review--to which the Court directed judges to defer.(12)
This Article argues, however, that a reading of Daubert that requires a scientific approach to conflicts would be inappropriate. Bias is no more a valid part of scientific fact finding than it is of legal fact finding. The approaches of scientists to managing conflicts of interest therefore are not themselves scientific, but are simply efforts to preserve the objectivity of science. There is no particular reason to think that the same methods would be appropriate for ensuring objectivity in litigation, which is a very different discipline.
Moreover, even in litigation, different sorts of conflicts present different problems. …