Professor Bernstein, however, has now tempted us past endurance. In duplicate posts at Volokh and Point of Law, he has unveiled a new intellectual strategy for would-be excluders of expert evidence who don't find Daubert and its progeny sufficiently restrictive. Under this new strategy, it would be argued that the Supreme Court's decisions in Daubert and its progeny, which rested (it is claimed) on the Court's interpretation of a former version of Fed. R. Evid. 702, have now been superseded by the subsequent amendment of Rule 702. True, the amendments in question were actually intended to codify the teachings of the Daubert trilogy. But it is the plain meaning of Rule 702's text that should govern, to the exclusion of any such intent (or so goes the argument), and that amended text (according to Prof. Bernstein) is actually stricter than (at least some interpretations of) the Supreme Court's decisions.
We're not going to tackle this entire theory yet -- except to shiver in contemplation of the infinite regress in which the Supreme Court announces the proper interpretation of the rules, and the text of the rules is then amended to codify the Court's interpretation, and the Court then announces its new and different interpretation of the new and different text. We'll wait to see whether this theory dies aborning, or gains enough currency to warrant a rebuttal.
But we do want to address a subsidiary claim raised in Professor Bernstein's post:
In suggesting that courts have repeatedly sustained "differential diagnosis" testimony "in the absence of any proof that substance X CAN cause injury Y," Prof. Bernstein misportrays the state of decisional law. Almost uniformly, the federal appellate courts have characterized "differential diagnosis" as a two-step process, in which causal candidates are first "ruled in" -- i.e., shown to have the capability to cause the relevant health outcome (a.k.a. "generic causation") -- before it is then determined whether any can be "ruled out" by testing or other methods. See, e.g., Clausen v. M/V New Carissa, 339 F.3d 1049 (9th Cir. 2003).
[W]ay too many courts have relied on language from the trilogy in admitting phony "differential diagnosis" testimony on causation that in fact constitutes nothing more than "post hoc ergo propter hoc" reasoning. Allowing an expert to testify that exposure to substance X caused injury Y simply because the expert can't identify any other cause, and in the absence of any proof that substance X CAN cause injury Y, is not exactly reliable principles and methods being applied reliably to the facts of the case. It's more like, in many instances, allowing paid hired guns to speculate wildly to a conclusion mandated not by science but by the necessity of supporting a plaintiff's causation theory.
Thus defined, "differential diagnosis" involves an inferential process with a strong logical kinship to the venerable reductio ad absurdum (a.k.a. "the process of elimination"), and one that most of us recognize as valid in other contexts. (E.g., only Mary and Isabel were in the room when Louisa was stabbed, and Isabel is a quadriplegic; ergo, Mary stabbed Louisa.) The controversy, perhaps, should be over what level of support should be demanded before a causal candidate is "ruled in" or "ruled out." In particular, the defense bar typically argues that the existence of peer-reviewed epidemiological literature should be a sine qua non for "ruling in," whereas close temporal proximity between a bodily insult and a health outcome would be enough for many members of the plaintiffs' bar, especially in cases where epidemiology hasn't much gotten around to investigating the alleged toxin at issue.
Is reliance on mere close temporal proximity an example of naked "post hoc, ergo proper hoc" reasoning? Well, it could be, except for that qualifier "close." If you are doused with exotic chemicals on Day 1 and develop strange lesions on Day 2, there are strong grounds for suspicion, whereas only a considerably weaker causal case would be made out, from temporal sequence alone, if the lesions failed to develop until Day 743. It warrants mention, in this connection, that the timing of disease onset is one canonical factor that epidemiologists (and other medical investigators) traditionally consider in their analyses.
But too much space need not be devoted, here and now, to this last set of questions, because there is no actual epidemic of "naked temporal sequence" cases. They are actually pretty sparse. Our larger point is that "differential diagnosis" testimony unsupported by "any proof" of a toxin's capacity to cause harm is not, in fact, being admitted by "too many" courts. Maybe Prof. Bernstein simply wants to register his dissent from the prevailing judicial standards for what should count as "proof" of that capacity. He is, of course, entitled to do just that, just as we are entitled to express our profound skepticism that a dispositive answer on that issue can be located within the plain text of Rule 702.
We now return you to our previously scheduled hiatus.