Monday, August 12, 2013
Frederick Vars, an Associate Professor at the University of Alabama School of Law, has an interesting chapter, Behavioral Economics and Evidence Law, in the forthcoming Oxford Manual on Behavioral Economics and Law. Here's the abstract to the chapter:
Behavioral economics has been applied to evidence law in at least four different ways: (1) to explain or justify current law; (2) to argue that current law counteracts or fails to counteract the findings; (3) to advocate changes in practice under existing law; and (4) to argue for law reform. On the surface, the first two applications appear descriptive and the second two normative. Below the surface all four missions share common assumptions about the power and generalizability of the psychological findings. Even superficially explanatory assertions are premised on the normative view that psychology should inform our understanding of evidence law. Whether it should, and how law and practice should be changed, are incredibly complex questions. Illustrating the complexity and suggesting ways forward are the goals of this Chapter.
Professor Vars considers several types of evidence in his chapter, including character evidence, expert evidence, and eyewitness testimony. In this post, however, I want to focus on his discussion of evidence of subsequent remedial measures.Federal Rule of Evidence 407 states that
When measures are taken that would have made an earlier injury or harm less likely to occur, evidence of the subsequent measures is not admissible to prove:
- culpable conduct;
- a defect in a product or its design; or
- a need for a warning or instruction.
But the court may admit this evidence for another purpose, such as impeachment or — if disputed — proving ownership, control, or the feasibility of precautionary measures.
Rule 407 is one of the specialized relevance rules contained in Federal Rules of Evidence 407-411. These Rules deem a certain category of evidence (e.g., subsequent remedial measures) per se inadmissible to prove certain purposes (e.g., negligence) but potentially admissible if offered for certain other purposes (e.g., ownership if it disputed).
People overestimate the predictability of past events. This tendency is known as “hindsight bias” (Fischhoff 1975). One set of commentators argue that excluding evidence of remedial measures taken after an injury to prove negligence before the injury (Federal Rules of Evidence 407) “represents an adaptation to the effects of the hindsight bias” (Guthrie et al. 2001). It appears that the commentators are making both descriptive and normative claims: this is how the law came to be, and it is a good thing. Logic provides some support for both claims. The hindsight bias is a robust psychological phenomenon and would predictably lead factfinders sometimes to conclude that a defendant should have predicted an injury because the defendant took measures after it to prevent its recurrence. The defendant should have seen it coming sooner, the biased factfinder might reason. To be sure, updating probability estimates based on new data is perfectly rational, if done correctly (Sanchirico 2003, pp. 1197-1200).
So far so good, but there’s a missing step, as one of the authors previously observed: “suppressing subsequent remedial measures in accident cases slightly reduces the pool of relevant evidence that the fact-finder has available to decide the case. If the effect of the hindsight bias is small, these adaptations might be overreactions to the problem of judging liability in hindsight” (Rachlinski 2000, p. 73). Dan Kahan picked up on this important observation, concluding that the excluding or admitting subsequent remedial measures on a case-specific basis may be optimal (Kahan 2010).
Kahan’s argument is significant on its own terms, but also as an exemplar of careful application of psychological findings to law. The critical insight is that evidence of subsequent remedial measures makes negligence somewhat more likely, even if not by as much as the hindsight bias leads factfinders to believe, so a per se rule of exclusion will generate erroneous outcomes. Kahan, while expressly eschewing explanation, offers the following prescription: “If a judge excludes [subsequent remedial measures] proofs in cases in which the plaintiff’s case is otherwise weak, and is receptive to admission of such proofs in cases in which the issue of the defendant’s substandard conduct is otherwise close, she will minimize the sum total of erroneous outcomes---either mistaken findings of liability or mistaken findings of nonliability--relative to a rule that categorically excludes or admits such proofs” (p. 1637).
But Kahan cannot resist the gravitational pull of current practice when he argues that this is what judges do all the time in balancing probative value and prejudicial effect under Federal Rule of Evidence 403, which basically requires that the value of evidence be net neutral or positive to be admissible (pp. 1638-39). Here Kahan falls short. He effectively assumes that the subsequent remedial measures evidence is the last piece of evidence offered so that “the full evidentiary context” is otherwise complete (p. 1639). This may not always be true. Relatedly, Kahan also assumes that judges measure probative value in terms of how likely evidence is to change the outcome (“contribution . . . to accurate factfinding” (p. 1649)). More plausibly, judges ask how far new evidence moves the meter. If hindsight bias is strong enough and probative value fixed, then unfair prejudice may substantially outweigh probative value in every case: weak, close, or overwhelming. Kahan may or may not be correct in all of his conclusions, but at least he frames the question welll.
This is a fascinating discussion and begs the question of whether the specialized relevance rules are a hatchet when a scalpel is needed. The interesting thing with Rule 407, though, is that there is actually a second layer of analysis, and this second layer is especially intertwined with behavioral economics. As noted in the Advisory Committee's Note to Rule 407,
The rule rests on two grounds. (1) The conduct is not in fact an admission, since the conduct is equally consistent with injury by mere accident or through contributory negligence. Or, as Baron Bramwell put it, the rule rejects the notion that “because the world gets wiser as it gets older, therefore it was foolish before.”...Under a liberal theory of relevancy this ground alone would not support exclusion as the inference is still a possible one. (2) The other, and more impressive, ground for exclusion rests on a social policy of encouraging people to take, or at least not discouraging them from taking, steps in furtherance of added safety.
Many of the specialized relevance rules, while justifiable on accuracy grounds, are more naturally classified as policy-based rules. For example, one can justify Rule 407, which excludes subsequent remedial measures, as eliminating evidence that does not necessarily prove negligence and that may be overweighed by a jury. However, subsequent remedies can often be probative, and courts can police unfair prejudice on a case-by-case basis under Rule 403. A far more compelling justification for Rule 407's blanket exclusion is to encourage (or at least not discourage) prosocial improvements. Its most famous treatment, Justice Mosk's opinion in Ault v. International Harvester Company, drips of such incentive-based arguments.
Thus, are the specialized relevance rules more properly labeled the specialized policy rules because they categorically exclude certain types of evidence for certain purposes primarily on policy grounds? That certainly seems to be the case with Rule 407. The proof of this can be found in the fact that courts consistently find that Rule 407 does not exclude evidence of subsequent remedial measures by third parties. If Paul falls on the steps leading into Dan's Diner, Rule 407 would not exclude evidence that Ned, the owner of a neighboring store, fixed the light above the steps. This distinction is not based on relevance grounds and can only be based on policy grounds because third parties don't need Rule 407 to remove the disincentive to act.
Of course, all of this begs the question of whether Rule 407 actually encourages repairs by defendants, and I'm not sure that an empirical study has ever been done on the subject. In any event both the chapter by Professor Vars and the article by Professor Cheng are fascinating reading and highly recommended.