Tuesday, April 17, 2007

Return of Ramirez

The Colorado Supreme Court has issued an amended opinion in the Ramirez case (see our posts of 3/27/07, 3/28/07, 4/1/07, & 4/4/07). In footnote 10, the court now cites its earlier decision in Wilkerson, distinguishing it on the basis that the Wilkerson expert offered a quantitative conclusion, whereas the Ramirez expert did not.

In a related vein, Jim Beck and Mark Herrmann comment at Drug and Device Law on the proposal, floated in comment e to section 28 of Tentative Draft #5 of the Restatement (Third) of Torts, Liability for Physical and Emotional Harm, to abolish the "reasonable medical certainty" requirement. They take umbrage at the proposal, whereas we do not. We'll have more to say in a day or so.


Wednesday, April 04, 2007

Even More on Ramirez

To our post of 4/1/07, James England has written in response:
Although I agree with much of your analysis, I remain troubled by the following aspects of Ramirez.

You have come up with an elegant way to attempt reconciliation of Ramirez and Wilkerson, but I find it significant that the Colorado Supreme Court never itself hints at such a reconciliation. Having decided Wilkerson less than two years ago, I would have thought it necessary for that Court to have at least mentioned that decision in Wilkerson, and to provide some scraps of rationale (especially to this state’s trial judges) as to how the two cases do not conflict. Without any such attempt, I fear future readers will reach the same conclusion you do as to a de facto way to harmonize the decisions: prosecutors win, and criminal defendants don’t. (One of the more interesting aspects of both holdings is that I would be surprised to see eiither one reached in a civil case.) As to your suggestion that Wilkerson would or should be restricted to advocates who expressly present their expert opinions couched in “quantitative, statistical meaning,” the opinion rejected by the Wilkerson court was merely a conclusion characterized by the expert as “more probable than not,” a formulation which rolls off the tongues of Colorado litigators as many trial judges will in fact require every expert opinion to carry such a characterization. This is what leads to my concern that if the Wilkerson court really meant what it seemed to say, then almost every expert opinion offered in this state could be attacked for lacking empirical or methodological support for that claimed probability.

As to your discussion of the differences between Rule 705’s provisions concerning data underlying expert opinions which need not (always) be laid before the jury and the Rule 104 necessity for establishing a sufficient evidentiary foundation for the judge, I am again very troubled by the fact that such a distinction is not hinted at in Ramirez. This is especially troubling because there was no pretrial (or other out-of-the-jury’s-presence) Daubert hearing in Ramirez: the presentation of the expert’s testimony to the jury was the only evidentiary foundation presented to or considered by the trial court. Given that the issue before the Supreme Court was the reliability of that evidence for admission, it seems to me the Court here relies on Rule 705, not to justify not disclosing underlying data to the jury, but to excuse the lack of a decent evidentiary foundation before the trial judge. That appears to me to abandon Rule 104 and Daubert gatekeeping requirements, or at least to carry a high risk that later trial courts will see this reliance on Rule 705 to excuse meaningful Daubert foundations (at least, as you suggest, by prosecutors).

Finally, as to the somewhat run-of-the-mill nature of the nurse’s testimony here, I would be less bothered if this had been a run-of-the-mill decision by our intermediate state appellate court. Here, however, that intermediate court had issued a fairly innocuous, unpublished decision which seems to me to soberly and analytically apply familiar Daubert principles to the very shaky evidentiary foundation laid before the trial court. What makes this case more significant is that the Colorado Supreme Court then exercised its discretionary certiorari review (my impression is that it grants less than 5 percent of the certiorari petitions it receives) to reach out and reverse that innocuous opinion. And it can’t be (can it?) that it did so in order to clarify the “medical certainty” issue, as that was not the decisional basis for either the trial court nor the Court of Appeals. Given that, it is distressing that the Court seized upon this case in one of its relatively infrequent revisitations of Daubert, only to blur the standards for opinion admissibility generally, confusingly resort to Rule 705, and put its stamp of approval on non-objective, “suspiciously” reliable and poorly supported “expert” opinion.


Sunday, April 01, 2007

Still More on Ramirez

Stimulated by James England's thoughtful analysis in this space, we've devoted more thought to the Colorado Supreme Court's decision last week in People v. Ramirez, and we have the following questions and observations, all of them untainted by any deep acquaintance with Colorado evidence law:

(1) What is the real scope of any tension between Ramirez and the same court's 2005 decision in People v. Wilkerson?

Wilkerson, it will be recalled, involved an ergonomics expert who proposed to testify, on behalf of a criminal defendant, that a shooting was unintentional. At a pretrial hearing, the expert opined that the shooting was "more likely than not" accidental. He explained, on cross-examination, that by that opinion, he meant there was more than a 50% probability that the shooting was unintended. In arguing for the testimony's admissibility, defense counsel likewise characterized it as an opinion that under conditions described by the witness, an accidental shooting could happen 51% of the time. Understanding the expert to be offering a statistical characterization, and concluding that the statistical characterization did not enjoy reliable support, the trial court barred the witness from offering testimony that the shooting was more probably than not accidental. The trial court did not rule on the defendant's alternative proffer -- qualitative testimony to the general effect, as defense counsel put it, that "the best minds that study this sort of thing and study stress, and study systems and human factors and study handguns, and have written articles about this sort of thing say that, yes, accidental shootings can happen."

On interlocutory review, the Colorado Supreme Court upheld the trial court's ruling, concluding that the trial court could permissibly have excluded the "more likely than not" opinion under Colo. R. Evid. 403, and also that the expert had offered insufficient support, under Colo. R. Evid. 702, for a quantitative assessment. The high court said:
Given the preciseness of the district court’s order, rejecting an expert conclusion expressed in terms of quantifiable, scientific test results, the court’s emphasis on the expert’s failure to test a specific weapon and the absence of peer reviewed literature concerning standards for measurement and error rates did not evidence a misunderstanding of the flexible test we described in [People v. Shreck, 22 P.3d 68, 78-79 (Colo. 2001) (rejecting Frye in favor of a more Daubertian approach)]. Rather than excluding expert opinion merely because it was incapable of quantification, the district court was faced with an opinion expressed in numeric terms, and it appears to have considered precisely those factors most applicable to the admissibility of such a numeric test result.
Given the trial court's failure to rule on the defendant's alternative proffer, the Wilkerson court expressed "no opinion concerning the reliability or relevance of any other proffered testimony of the defense expert."

We1 voiced some dismay when the Wilkerson opinion was first issued. The basis for that dismay was our perception that the trial and appellate courts in Wilkerson may have fallen prey to a fairly common intellectual parlor trick, in which the "probability" concept goes into the hat in the form of a general qualitative standard for expert confidence or the burden of persuasion (the handkerchief), only to emerge later as a term of technical statistical art (the rabbit). This maneuver is sometimes employed in toxic tort litigation to insinuate a defense-friendly requirement that individual causation be proved by odds ratios reaching some specific numerical threshold. It might be called the Probability Pivot, and it generally proceeds in several well-defined steps:
(a) The expert states some opinion.

(b) The expert is asked on cross (at deposition or during a pretrial hearing on admissibility, or perhaps even at trial) to confirm that the expert holds the opinion to the requisite degree of confidence under the law of the relevant state (e.g., "a reasonable medical probability") or that the expert's conclusion satisfies2 the preponderance test for sufficiency of the evidence (e.g., "more likely than not").

(c) The expert complies, often without the questioning having placed the expert on clear notice that the expert's response will later be taken as importing some numerical or statistical standard.

(d) The objecting party then complains that the expert's opinion is not supported by evidence that would establish the opinion as "probable" or "likely" in any numerical or statistical sense.

(e) Alternatively, the expert declines the cross-examiner's invitation to characterize his or her opinion in terms of likelihood or probability, in which event the objecting party complains that the opinion does not satisfy the operative legal requirements for confidence or sufficiency (a step in which those requirements are sometimes conflated with Daubert's reliability standard).
We call this a parlor trick because neither expert confidence requirements nor tests for sufficiency have traditionally been understood in statistical terms, if occasional metaphorical glosses are left to one side. There may be tenable policy arguments for adopting quantitative tests as the substantive standard for certain elements of proof in particular areas of the law. But if so, those policy arguments should be presented to the appropriate governmental branch (the legislature, we're inclined to think) in candid and transparent form -- not smuggled into the decisional law via category mistakes abetted through free-wheeling use of the thesaurus (whether by the judiciary or by attorneys conducting cross). So too, there may be some forms of expert testimony that should not be counted as reliable if specific standards of quantitative analysis go unmet. These may include testimony from experts whose discipline normally requires the use of quantitative methods. Most would also agree that testimony presenting the expert's conclusions in quantitative terms (like the ergonomist's opinion in Wilkerson, as the trial and appellate courts understood it) falls in this category. Our dismay over Wilkerson arose from a worry that the expert was lured unwittingly into quantitative testimonial territory through sneaky cross-examination -- a possibility to which the trial and appellate courts in Wilkerson seemed to us to give unduly short shrift.

But unlike us, the courts in Wilkerson did enjoy access to the record, and they concluded that the opinion was being offered in terms intended to carry quantitative, statistical meaning -- indeed, that proponent's counsel had defended the opinion in those very terms. This being so, the holding in Wilkerson would seem rather narrow. Where a party offers testimony held out as supporting a statistical or quantitative conclusion or inference, the testimony should enjoy a reliable basis in methods or techniques supporting the relevant statistical or quantitative conclusion or inference. Neither the trial court nor the Colorado Supreme Court in Wilkerson, remember, ruled on how the reliability of the alternative, qualitative proffer should be assessed. Neither court in Wilkerson purported to hold, in other words, that common law standards for expert confidence, whether or not they survived adoption of Rule 702, carried any inherent implications for the modern reliability calculus.3

We remain unconvinced, in short, that Ramirez particularly implicates Wilkerson, except in the thin sense that all decisions under an evidentiary regime informed by Daubertian principles are grist for debate over how stringently reliability should be tested, and via what methods. In both cases, we note, a trial court's reliability decision was ultimately affirmed, in keeping with the principle (familiar from federal evidentiary jurisprudence and apparently followed in Colorado as well) that trial courts' reliability determinations are discretionary and entitled to substantial deference. Both decisions also happen to comport with the general de facto principle that in criminal cases, the prosecution's expert testimony almost invariably comes in, whereas the defendant's quite possibly doesn't.

(2) Is the Ramirez court's invocation of Rule 705 really suspect?

Lacking access to the record or briefing in Ramirez, we cannot be sure what argument the defendant/appellant may have raised on appeal about the foundation of the nurse practitioner's testimony. The word "foundation" can be used to denote two slightly different things: (a) the support shown for an expert's opinion in the context of an admissibility inquiry (e.g., during voir dire); or (b) "foundation" testimony offered at trial for consumption by the trier of fact. It is unclear to us, from the Ramirez opinion, how much of the nurse practitioner's testimony defending the reliability of her opinion occurred in each of these respective contexts. But the two contexts are not generally coextensive. If the defendant's argument on appeal was that the testimony offered to the jury did not sufficiently establish the "foundational" predicates for reliability under Rule 702, then Rule 705 would seem a correct (if incomplete) rejoinder. Admissibility determinations are for the court, not the jury. They need not be based on record evidence, and record evidence, for its part, need not recapitulate the predicates for admissibility. Where "reliability" issues have been determined to pose questions of weight rather than admissibility, Rule 705 sensibly permits the trial court to leave their detailed exploration to cross-examination. For the federal courts, at least, requiring that Rule 702 "reliability" be established on direct examination at trial would represent a substantial innovation -- one that would seemingly require amendment of several evidentiary rules, including Rule 104(a).

3. How thin, actually, was the support for the reliability of the nurse practitioner's opinion?

We don't pretend to expertise in this area, and to repeat, we haven't studied the briefing or the record in Ramirez. So it's best we tread cautiously. That said, we find little occasion to doubt the reliability of a nurse practitioner's conclusion that the results of a physical examination were "suspicious," where he or she: (a) examined a ten-year-old child 41 days after an alleged sexual assault and found a healing laceration; (b) had conducted about 90 similar previous examinations; and (c) had reviewed at least some literature on the subject. To us, this seems like the exactly the sort of report, reflecting the educated clinical judgment of a medical professional, that doctors and patients routinely credit, to one degree or another, in nonlitigious contexts, even where the report enjoys no stronger bases than the nurse practitioner's opinion here. It also seems to us that the failure of an earlier examination to find signs of sexual assault is the sort of point that juries are quite competent to assess.

We do admit to some unease, especially in the criminal context, over the presentation of expert opinion that some fact -- although not sufficient, by itself or in combination with other facts, to support an actual finding by the expert on the point -- is "suspicious." Our unease would be still greater if the "suspicious" datum were that the accused had been seen outside the bank a few days before the robbery, twirling his mustache. Our disquiet on this front, however, springs mostly from anxiety over the idea of sending people to jail for being suspects. That, perhaps, is more a sufficiency concern, to be evaluated in individual cases by reference to the entire record, than a reliability one.

1 The editorial "we" refers here to the voice of Blog 702. We mean to ascribe no views, one way or the other, to James England.

2 Sufficiency, of course, is actually a distinct issue from reliability, and one properly evaluated by reference to the entire record, not the testimony of one single witness.

3 The conclusion that the rules of evidence displaced common law standards for expert confidence seems eminently defensible to us, but the issue does not appear to have been presented or addressed in Wilkerson.


Wednesday, March 28, 2007

More on the Colorado Supreme Court's Ramirez Decision

Yesterday, we posted about the Colorado Supreme Court's decision in People v. Ramirez, focusing on its abandonment of "reasonable medical certainty" as a necessary condition for admissibility. We paid less attention to the opinion's treatment of Daubert's requirements under Colorado law.

James M. England, our Colorado correspondent, now writes and fills that gap:
Back in July of 2005, you carried my summary of a new Daubert opinion from the Colorado Supreme Court, which I speculated might be "far-reaching." Never mind; this week that same court unanimously departed as far as possible from that case, without even mentioning it.

In the 2005 case, People v. Wilkerson, the Court upheld the exclusion of defense expert testimony that a shooting was more likely accidental than not, based on the lack of "empirical or methodological justification in the record" for the expert's supposed "numeric calculation of the possibility that the shooting in this case was accidental." In extreme contrast, its new decision, People v. Ramirez, approves the admission of prosecution expert testimony expressly based only upon "suspicion" and unaccompanied by any indicia of reliability or sound methodology.

Ramirez was tried for sexual assault on a ten-year-old child. A day or two after the last alleged assault, hospital medical examination detected no evidence of sexual assault. Some 40 days later, the child was re-examined by a clinic's nurse practitioner, who testified at a preliminary hearing (but not at trial) that she found signs of a "healing laceration." This nurse was presented at trial as an expert witness "in the area of child sexual assault examinations," for which she was found qualified by having "done" some 91 such examinations. (Aside from two week-long conferences, the nurse testified she had no formal training or education in evaluating children for evidence of sexual assault, and that her knowledge in this area was derived from "a lot of reading" of journal articles which she did not identify, and in assisting an unidentified pediatrician who performed sexual assault examinations.)

On voir dire, the nurse said she had not reached conclusions to a "reasonable degree of medical certainty," but had reached findings which "are suspicious, but not diagnostic." The nurse further described her resort to a four-tier hierarchy of possible examination findings ranging from "normal" to "non-specific" to "suspicious" to "definitive," and said that she found the results of her examination of this child to be "suspicious," which "raises my index of suspicion," and which "may well have been caused by sexual abuse but there could have been another possibility." The trial court overruled defense reliability objections, stating that the nurse's testimony "doesn't really show any sexual penetration, but it just shows that there was something there that was suspicious and I think that is useful." In testifying, the nurse never did describe to the jury what her "suspicious" finding was, nor provide any basis for her belief that it was "suspicious."

The Colorado Court of Appeals reversed the resulting conviction, finding (in an unpublished opinion) that the opinion testimony addressed "only possibilities" and amounted to "no more than conjecture and speculation." In passing, that Court also noted the general requirement in Colorado law that a medical opinion be based on "reasonable medical probability."

On certiorari review, the Colorado Supreme Court first examined the evolution of the standard of "reasonable medical certainty," before (helpfully) concluding that it is not a valid requirement under CRE 702. Implicitly at least, the Court then also rejected the requirement of expert testimony based on "reasonable probability," without mentioning Wilkerson (let alone trying to reconcile with it). (Wilkerson was apparently never cited in the briefs filed with the Supreme Court.) The Supreme Court then went on to consider whether this expert testimony was sufficiently reliable under CRE 702 and sufficiently grounded "in the methods and procedures of science rather than subjective belief or unsupported speculation." Remarkably, the Court concluded that the testimony of a "suspicious" finding was sufficiently reliable because it was the product of a "reliable scientific principle" (the medical examination) and that the "four-point scale" was also based on sufficiently reliable scientific principles, resorting to the following unpersuasive analysis:
One would expect to find "normal" and "definitive" at the polar ends of a scale measuring signs of sexual abuse. Although it is difficult given the record before us to describe in any detail the difference between a "non-specific" finding and a "suspicious" finding, we nonetheless find the scale used by [the nurse] to be reasonably reliable. It logically follows that "suspicious," being closer on the scale to "definitive" than "normal," indicates a finding that is more indicative of sexual abuse than a "non-specific" finding, which is closer to "normal" than "definitive" on the scale. We find that the jury could have reasonably inferred from [the nurse's] testimony that "suspicious," while not being an absolute finding of sexual abuse, was nevertheless a strong indicator of sexual abuse. Thus, the scale was sufficiently reliable under CRE 702 for [the nurse] to use.
In this regard, a startling aspect of the Court's opinion is that it appears to considerably dilute the extent of any foundation required for any expert opinion. The Court of Appeals (not surprisingly) had noted that the prosecution had not established any basis for its witness' conclusion that her examination had resulted in a "suspicious" finding. The Supreme Court, however, justified the absence of such foundation under CRE 705's provision that expert opinion does not first require testimony as to "the underlying facts or data, unless the court requires otherwise." The Court's opinion not only glosses over the discretionary nature of this provision, but also disregards Daubert's reliability gatekeeping requirements. Especially where an expert couches a conclusory opinion in terms of a suspicion which is ranked on an idiosyncratic and objective "scale," a basic Daubert foundation would require display and consideration of the methodology supposedly utilized by that expert in order to determine if sufficient reliability exists, entirely without regard to Rule 705.

Unless this opinion is significantly modified on rehearing, the state of Daubert in Colorado is now substantially confused. An attorney wanting to exclude virtually any expert opinion (at least where the expert is unable to support an opinion of probability with "empirical or methodological justification") need only cite Wilkerson. Conversely, an attorney wanting to admit even the flimsiest and equivocal expert opinion need only cite Ramirez (and can further use it to justify not laying more than a minimal foundation for that opinion). And the conscientious trial judge does what?


Tuesday, March 27, 2007

"Reasonable Medical Certainty" Not Required for Admissibility, Colorado High Court Rules

The admissibility of a medical expert's opinion does not depend, in form or substance, on whether the opinion is held to a reasonable medical certainty or probability, the Colorado Supreme Court has ruled in a major decision on the subject. From the opinion:
We granted certiorari in this case to review the court of appeals' decision that the expert testimony of a certified pediatric nurse practitioner in a sexual assault case should not have been admitted by the trial court. The nurse described her findings as "suspicious" and testified that her conclusions were not based on a reasonable degree of medical certainty. After reviewing the cases on which the court of appeals based its decision, we find that the standards of admissibility set forth in those cases, including the requirement that expert medical testimony be offered with a reasonable degree of medical certainty or probability, predate the Colorado Rules of Evidence, which is the modern standard for determining the admissibility of expert testimony.
See People v. Ramirez, No. 06SC71 (Colo. Mar. 26, 2007).


Fed. R. Evid. 702: If scientific, technical, or other specialized knowledge will assist the trier of fact to understand the evidence or to determine a fact in issue, a witness qualified as an expert by knowledge, skill, experience, training, or education, may testify thereto in the form of an opinion or otherwise, if (1) the testimony is based upon sufficient facts or data, (2) the testimony is the product of reliable principles and methods, and (3) the witness has applied the principles and methods reliably to the facts of the case.