For your delectation and delight, desultory dicta on the law of delicts.

Multiplicity in the Third Circuit

September 21st, 2017

In Karlo v. Pittsburgh Glass Works, LLC, C.A. No. 2:10-cv-01283 (W. D. Pa.), plaintiffs claimed that their employer’s reduction in force unlawfully targeted workers over 50 years of age. Plaintiffs lacked any evidence of employer animus against old folks, and thus attempted to make out a statistical disparate impact claim. The plaintiffs placed their chief reliance upon an expert witness, Michael A. Campion, to analyze a dataset of workers agreed to have been the subject of the R.I.F. For the last 30 years, Campion has been on the faculty in Purdue University. His academic training and graduate degrees are in industrial and organizational psychology. Campion has served an editor of Personnel Psychology, and as a past president of the Society for Industrial and Organizational Psychology. Campion’s academic website page notes that he manages a small consulting firm, Campion Consulting Services1.

The defense sought to characterize Campion as not qualified to offer his statistical analysis2. Campion did, however, have some statistical training as part of his master’s level training in psychology, and his professional publications did occasionally involve statistical analyses. To be sure, Campion’s statistical acumen paled in comparison to the defense expert witness, James Rosenberger, a fellow and a former vice president of the American Statistical Association, as well as a full professor of statistics in Pennsylvania State University. The threshold for qualification, however, is low, and the defense’s attack on Campion’s qualifications failed to attract the court’s serious attention.

On the merits, the defense subjected Campion to a strong challenge on whether he had misused data. The defense’s expert witness, Prof. Rosenberger, filed a report that questioned Campion’s data handling and statistical analyses. The defense claimed that Campion had engaged in questionable data manipulation by including, in his RIF analysis, workers who had been terminated when their plant was transferred to another company, as well as workers who retired voluntarily.

Using simple z-score tests, Campion compared the ages of terminated and non-terminated employees in four subgroups, ages 40+, 45+, 50+, and 55+. He did not conduct an analysis of the 60+ subgroup on the claim that this group had too few members for the test to have sufficient power3Campion found a small z-score for the 40+ versus <40 age groups comparison (z =1.51), which is not close to statistical significance at the 5% level. On the defense’s legal theory, this was the crucial comparison to be made under the Age Discrimination in Employment Act (ADEA). The plaintiffs, however, maintained that they could make out a case of disparate impact by showing age discrimination at age subgroups that started above the minimum specified by the ADEA. Although age is a continuous variable, Campion decided to conduct z-scores on subgroups that were based upon five-year increments. For the 45+, 50+, and 55+ age subgroups, he found z-scores that ranged from 2.15 to 2.46, and he concluded that there was evidence of disparate impact in the higher age subgroups4. Karlo v. Pittsburgh Glass Works, LLC, C.A. No. 2:10-cv-01283, 2015 WL 4232600, at *11 (W.D. Pa. July 13, 2015) (McVerry, S.J.)

The defense, and apparently the defense expert witnesses, branded Campion’s analysis as “data snooping,” which required correction for multiple comparisons. In the defense’s view, the multiple age subgroups required a Bonferroni correction that would have diminished the critical p-value for “significance” by a factor of four. The trial court agreed with the defense contention about data snooping and multiple comparisons, and excluded Campion’s opinion of disparate impact, which had been based upon finding statistically significant disparities in the 45+, 50+, and 55+ age subgroups. 2015 WL 4232600, at *13. The trial court noted that Campion, in finding significant disparities in terminations in the subgroups, but not in the 40+ versus <40 analysis:

[did] not apply any of the generally accepted statistical procedures (i.e., the Bonferroni procedure) to correct his results for the likelihood of a false indication of significance. This sort of subgrouping ‘analysis’ is data-snooping, plain and simple.”

Id. After excluding Campion’s opinions under Rule 702, as well as other evidence in support of plaintiffs’ disparate impact claim, the trial court granted summary judgment on the discrimination claims. Karlo v. Pittsburgh Glass Works, LLC, No. 2:10–cv–1283, 2015 WL 5156913 (W. D. Pa. Sept. 2, 2015).

On plaintiffs’ appeal, the Third Circuit took the wind out of the attack on Campion by holding that the ADEA prohibits disparate impacts based upon age, which need not necessarily be on workers’ being over 40 years old, as opposed to being at least 40 years old. Karlo v. Pittsburgh Glass Works, LLC, 849 F.3d 61, 66-68 (3d Cir. 2017). This holding took the legal significance out of the statistical insignificance of Campion’s comparison 40+ versus <40 age-group termination rates. Campion’s subgroup analyses were back in play, but the Third Circuit still faced the question whether Campion’s conclusions, based upon unadjusted z-scores and p-values, offended Rule 702.

The Third Circuit noted that the district court had identified three grounds for excluding Campion’s statistical analyses:

(1) Dr. Campion used facts or data that were not reliable;

(2) he failed to use a statistical adjustment called the Bonferroni procedure; and

(3) his testimony lacks ‘‘fit’’ to the case because subgroup claims are not cognizable.

849 F.3d at 81. The first issue was raised by the defense’s claims of Campion’s sloppy data handling, and inclusion of voluntarily retired workers and workers who were terminated when their plant was turned over to another company. The Circuit did not address these data handling issues, which it left for the trial court on remand. Id. at 82. The third ground went out of the case with the appellate court’s resolution of the scope of the ADEA. The Circuit did, however, engage on the issue whether adjustment for multiple comparisons was required by Rule 702.

On the “data-snooping” issue, the Circuit concluded that the trial court had applied “an incorrectly rigorous standard for reliability.” Id. The Circuit acknowledged that

[i]n theory, a researcher who searches for statistical significance in multiple attempts raises the probability of discovering it purely by chance, committing Type I error (i.e., finding a false positive).”

849 F.3d at 82. The defense expert witness contended that applying the Bonferroni adjustment, which would have reduced the critical significance probability level from 5% to 1%, would have rendered Campion’s analyses not statistically significant, and thus not probative of disparate impact. Given that plaintiffs’ cases were entirely statistical, the adjustment would have been fatal to their cases. Id. at 82.

At the trial level and on appeal, plaintiffs and Campion had objected to the data-snooping charge on ground that

(1) he had engaged in only four subgroups;

(2) virtually all subgroups were statistically significant;

(3) his methodology was “hypothesis driven” and involved logical increments in age to explore whether the strength of the evidence of age disparity in terminations continued in each, increasingly older subgroup;

(4) his method was analogous to replications with different samples; and

(5) his result was confirmed by a single, supplemental analysis.

Id. at 83. According to the plaintiffs, Campion’s approach was based upon the reality that age is a continuous, not a dichotomous variable, and he was exploring a single hypothesis. A.240-241; Brief of Appellants at 26. Campion’s explanations do mitigate somewhat the charge of “data snooping,” but they do not explain why Campion did not use a statistical analysis that treated age as a continuous variable, at the outset of his analysis. The single, supplemental analysis was never described or reported by the trial or appellate courts.

The Third Circuit concluded that the district court had applied a ‘‘merits standard of correctness,’’ which is higher than what Rule 702 requires. Specifically, the district court, having identified a potential methodological flaw, did not further evaluate whether Campion’s opinion relied upon good grounds. 849 F.3d at 83. The Circuit vacated the judgment below, and remanded the case to the district court for the opportunity to apply the correct standard.

The trial court’s acceptance that an adjustment was appropriate or required hardly seems a “merits standard.” The use of a proper adjustment for multiple comparisons is very much a methodological concern. If Campion could reach his conclusion only by way of an inappropriate methodology, then his conclusion surely would fail the requirements of Rule 702. The trial court did, however, appear to accept, without explicit evidence, that the failure to apply the Bonferroni correction made it impossible for Campion to present sound scientific argument for his conclusion that there had been disparate impact. The trial court’s opinion also suggests that the Bonferroni correction itself, as opposed to some more appropriate correction, was required.

Unfortunately, the reported opinions do not provide the reader with a clear account of what the analyses would have shown on the correct data set, without improper inclusions and exclusions, and with appropriate statistical adjustments. Presumably, the parties are left to make their cases on remand.

Based upon citations to sources that described the Bonferroni adjustment as “good statistical practice,” but one that is ‘‘not widely or consistently adopted’’ in the behavioral and social sciences, the Third Circuit observed that in some cases, failure to adjust for multiple comparisons may “simply diminish the weight of an expert’s finding.”5 The observation is problematic given that Kumho Tire suggests that an expert witness must use “in the courtroom the same level of intellectual rigor that characterizes the practice of an expert in the relevant field.” Kumho Tire Co. v. Carmichael, 526 U.S. 137, 150, (1999). One implication is that courts are prisoners to prevalent scientific malpractice and abuse of statistical methodology. Another implication is that courts need to look more closely at the assumptions and predicates for various statistical tests and adjustments, such as the Bonferroni correction.

These worrisome implications are exacerbated by the appellate court’s insistence that the question whether a study’s result was properly calculated or interpreted “goes to the weight of the evidence, not to its admissibility.”6 Combined with citations to pre-Daubert statistics cases7, judicial comments such as these can appear to be a general disregard for the statutory requirements of Rules 702 and 703. Claims of statistical significance, in studies with multiple exposure and multiple outcomes, are frequently not adjusted for multiple comparisons, without notation, explanation, or justification. The consequence is that study results are often over-interpreted and over-sold. Methodological errors related to multiple testing or over-claiming statistical significance are commonplace in tort litigation over “health-effects” studies of birth defects, cancer, and other chronic diseases that require epidemiologic evidence8.

In Karlo, the claimed methodological error is beset by its own methodological problems. As the court noted, adjustments for multiple comparisons are not free from methodological controversy9. One noteworthy textbook10 labels the Bonferroni correction as an “awful response” to the problem of multiple comparisons. Aside from this strident criticism, there are alternative approaches to statistical adjustment for multiple comparisons. In the context of the Karlo case, the Bonferroni might well be awful because Campion’s four subgroups are hardly independent tests. Because each subgroup is nested within the next higher age subgroup, the subgroup test results will be strongly correlated in a way that defeats the mathematical assumptions of the Bonferroni correction. On remand, the trial court in Karlo must still make his Rule 702 gatekeeping decision on the methodological appropriateness of whether Campion’s properly considered the role of multiple subgroups, and multiple anaslyses run on different models.

1 Although Campion describes his consulting business as small, he seems to turn up in quite a few employment discrimination cases. See, e.g., Chen-Oster v. Goldman, Sachs & Co., 10 Civ. 6950 (AT) (JCF) (S.D.N.Y. 2015); Brand v. Comcast Corp., Case No. 11 C 8471 (N.D. Ill. July 5, 2014); Powell v. Dallas Morning News L.P., 776 F. Supp. 2d 240, 247 (N.D. Tex. 2011) (excluding Campion’s opinions), aff’d, 486 F. App’x 469 (5th Cir. 2012).

2 See Defendant’s Motion to Bar Dr. Michael Campion’s Statistical Analysis, 2013 WL 11260556.

3 There was no mention of an effect size for the lower aged subgroups, and a power calculation for the 60+ subgroup’s probability of showing a z-score greater than two. Similarly, there was no discussion or argument about why this subgroup could not have been evaluated with Fisher’s exact test. In deciding the appeal, the Third Circuit observed that “Dr. Rosenberger test[ed] a subgroup of sixty-and-older employees, which Dr. Campion did not include in his analysis because ‘[t]here are only 14 terminations, which means the statistical power to detect a significant effect is very low’. A.244–45.” Karlo v. Pittsburgh Glass Works, LLC, 849 F.3d 61, 82 n.15 (3d Cir. 2017).

4 In the trial court’s words, the z-score converts the difference in termination rates into standard deviations. Karlo v. Pittsburgh Glass Works, LLC, C.A. No. 2:10-cv-01283, 2015 WL 4232600, at *11 n.13 (W.D. Pa. July 13, 2015). According to the trial court, Campion gave a rather dubious explanation of the meaning of the z-score: “[w]hen the number of standard deviations is less than –2 (actually–1.96), there is a 95% probability that the difference in termination rates of the subgroups is not due to chance alone” Id. (internal citation omitted).

5 See 849 F.3d 61, 83 (3d Cir. 2017) (citing and quoting from Paetzold & Willborn § 6:7, at 308 n.2) (describing the Bonferroni adjustment as ‘‘good statistical practice,’’ but ‘‘not widely or consistently adopted’’ in the behavioral and social sciences); see also E.E.O.C. v. Autozone, Inc., No. 00-2923, 2006 WL 2524093, at *4 (W.D. Tenn. Aug. 29, 2006) (‘‘[T]he Court does not have a sufficient basis to find that … the non-utilization [of the Bonferroni adjustment] makes [the expert’s] results unreliable.’’). And of course, the Third Circuit invoked the Daubert chestnut: ‘‘Vigorous cross-examination, presentation of contrary evidence, and careful instruction on the burden of proof are the traditional and appropriate means of attacking shaky but

admissible evidence.’’ Daubert, 509 U.S. 579, 596 (1993).

6 See 849 F.3d at 83 (citing Leonard v. Stemtech Internat’l Inc., 834 F.3d 376, 391 (3d Cir. 2016).

7 See 849 F.3d 61, 83 (3d Cir. 2017), citing Bazemore v. Friday, 478 U.S. 385, 400 (1986) (‘‘Normally, failure to include variables will affect the analysis’ probativeness, not its admissibility.’’).

8 See Hans Zeisel & David Kaye, Prove It with Figures: Empirical Methods in Law and Litigation 93 & n.3 (1997) (criticizing the “notorious” case of Wells v. Ortho Pharmaceutical Corp., 788 F.2d 741 (11th Cir.), cert. denied, 479 U.S. 950 (1986), for its erroneous endorsement of conclusions based upon “statistically significant” studies that explored dozens of congenital malformation outcomes, without statistical adjustment). The authors do, however, give an encouraging example of a English trial judge who took multiplicity seriously. Reay v. British Nuclear Fuels (Q.B. Oct. 8,1993) (published in The Independent, Nov. 22,1993). In Reay, the trial court took seriously the multiplicity of hypotheses tested in the study relied upon by plaintiffs. Id. (“the fact that a number of hypotheses were considered in the study requires an increase in the P-value of the findings with consequent reduction in the confidence that can be placed in the study result … .”), quoted in Zeisel & Kaye at 93. Zeisel and Kaye emphasize that courts should not be overly impressed with claims of statistically significant findings, and should pay close attention to how expert witnesses developed their statistical models. Id. at 94.

9 See David B. Cohen, Michael G. Aamodt, and Eric M. Dunleavy, Technical Advisory Committee Report on Best Practices in Adverse Impact Analyses (Center for Corporate Equality 2010).

10 Kenneth J. Rothman, Sander Greenland, and Timoth L. Lash, Modern Epidemiology 273 (3d ed. 2008); see also Kenneth J. Rothman, “No Adjustments Are Needed for Multiple Comparisons,” 1 Epidemiology 43, 43 (1990)

 Another Haack Article on Daubert

October 14th, 2016

In yet another law review article on Daubert, Susan Haack has managed mostly to repeat her past mistakes, while adding a few new ones to her exegesis of the law of expert witnesses. See Susan Haack, “Mind the Analytical Gap! Tracing a Fault Line in Daubert,” 654 Wayne L. Rev. 653 (2016) [cited as Gap].  Like some other commentators on the law of evidence, Haack purports to discuss this area of law without ever citing or quoting the current version of the relevant statute, Federal Rule of Evidence 703. She pours over Daubert and Joiner, as she has done before, with mostly the same errors of interpretation. In discussing Joiner, Haack misses the importance of the Supreme Court’s reversal of the 11th Circuit’s asymmetric standard of Rule 702 trial court decisions. Gap at 677. And Haack’s analysis of this area of law omits any mention of Rule 703, and its role in Rule 702 determinations. Although you can safely skip yet another Haack article, you should expect to see this one, along with her others, cited in briefs, right up there with David Michael’s Manufacturing Doubt.

A Matter of Degree

“It may be said that the difference is only one of degree. Most differences are, when nicely analyzed.”[1]

Quoting Holmes, Haack appears to complain that the courts’ admissibility decisions on expert witnesses’s opinions are dichotomous and categorical, whereas the component parts of the decisions, involving relevance and reliability, are qualitative and gradational. True, true, and immaterial.

How do you boil a live frog so it does not jump out of the water?  You slowly turn up the heat on the frog by degrees.  The frog is lulled into complacency, but at the end of the process, the frog is quite, categorically, and sincerely dead. By a matter of degrees, you can boil a frog alive in water, with a categorically ascertainable outcome.

Humans use categorical assignments in all walks of life.  We rely upon our conceptual abilities to differentiate sinners and saints, criminals and paragons, scholars and skells. And we do this even though IQ, and virtues, come in degrees. In legal contexts, the finder of fact (whether judge or jury) must resolve disputed facts and render a verdict, which will usually be dichotomous, not gradational.

Haack finds “the elision of admissibility into sufficiency disturbing,” Gap at 654, but that is life, reason, and the law. She suggests that the difference in the nature of relevancy and reliability on the one hand, and admissibility on the other, creates a conceptual “mismatch.” Gap at 669. The suggestion is rubbish, a Briticism that Haack is fond of using herself.  Clinical pathologists may diagnose cancer by counting the number of mitotic spindles in cells removed from an organ on biopsy.  The number may be characterized by as a percentage of cells in mitosis, a gradational that can run from zero to 100 percent, but the conclusion that comes out of the pathologist’s review is a categorical diagnosis.  The pathologist must decide whether the biopsy result is benign or malignant. And so it is with many human activities and ways of understanding the world.

The Problems with Daubert (in Haack’s View)

Atomism versus Holism

Haack repeats a litany of complaints about Daubert, but she generally misses the boat.  Daubert was decisional law, in 1993, which interpreted a statute, Federal Rule of Evidence 702.  The current version of Rule 702, which was not available to, or binding on, the Court in Daubert, focuses on both validity and sufficiency concerns:

A witness who is qualified as an expert by knowledge, skill, experience, training, or education may testify in the form of an opinion or otherwise if:

(a) the expert’s scientific, technical, or other specialized knowledge will help the trier of fact to understand the evidence or to determine a fact in issue;

(b) the testimony is based on sufficient facts or data;

(c) the testimony is the product of reliable principles and methods; and

(d) the expert has reliably applied the principles and methods to the facts of the case.

Subsection (b) renders most of Haack’s article a legal ignoratio elenchi.

Relative Risks Greater Than Two

Modern chronic disease epidemiology has fostered an awareness that there is a legitimate category of disease causation that involves identifying causes that are neither necessary nor sufficient to produce their effects. Today it is a commonplace that an established cause of lung cancer is cigarette smoking, and yet, not all smokers develop lung cancer, and not all lung cancer patients were smokers.  Epidemiology can identify lung cancer causes such as smoking because it looks at stochastic processes that are modified from base rates, or population rates. This model of causation is not expected to produce uniform and consistent categorical outcomes in all exposed individuals, such as lung cancer in all smokers.

A necessary implication of categorizing an exposure or lifestyle variable as a “cause,” in this way is that the evidence that helps establish causation cannot answer whether a given individual case of the outcome of interest was caused by the exposure of interest, even when that exposure is a known cause.  We can certainly say that the exposure in the person was a risk for developing the disease later, but we often have no way to make the individual attribution.  In some cases, more the exception than the rule, there may be an identified mechanism that allows the detection of a “fingerprint” of causation. For the most part, however, risk and cause are two completely different things.

The magnitude of risk, expressed as a risk ratio, can be used to calculate a population attributable risk, which can in turn, with some caveats, be interpreted as approximating a probability of causation.  When the attributable risk is 95%, as it would be for people with light smoking habits and lung cancer, treating the existence of the prior risk as evidence of specific causation seems perfectly reasonable.  Treating a 25% attributable risk as evidence to support a conclusion of specific causation, without more, is simply wrong.  A simple probabilistic urn model would tell us that we would most likely be incorrect if we attributed a random case to the risk based upon such a low attributable risk.  Although we can fuss over whether the urn model is correct, the typical case in litigation allows no other model to be asserted, and it would be the plaintiffs’ burden of proof to establish the alternative model in any event.

As she has done many times before, Haack criticizes Judge Kozinski’s opinion in Daubert,[2] on remand, where he entered judgment for the defendant because further proceedings were futile given the small relative risks claimed by plaintiffs’ expert witnesses.  Those relative risks, advanced by Shanna Swan and Alan Done, lacked reliability; they were the product of a for-litigation juking of the stats that were the original target of the defendant and the medical community in the Supreme Court briefing.  Judge Kozinski simplified the case, using a common legal strategem of assuming arguendo that general causation was established.  With this assumption favorable to plaintiffs made, but never proven or accepted, Judge Kozinski could then shine his analytical light on the fatal weakness of the specific causation opinions.  When all the hand waving was put to rest, all that propped up the plaintiff’s specific causation claim was the existence of a claimed relative risk, which was less than two. Haack is unhappy with the analytical clarity achieved by Kozinski, and implicitly urges a conflation of general and specific causation so that “all the evidence” can be counted.  The evidence of general causation, however, does not advance plaintiff’s specific causation case when the nature of causation is the (assumed) existence of a non-necessary and non-sufficient risk. Haack quotes Dean McCormick as having observed that “[a] brick is not a wall,” and accuses Judge Kozinski of an atomistic fallacy of ruling out a wall simply because the party had only bricks.  Gap at 673, quoting from Charles McCormick, Handbook of the Law of Evidence at 317 (1954).

There is a fallacy opposite to the atomistic fallacy, however, namely the holistic “too much of nothing fallacy” so nicely put by Poincaré:

“Science is built up with facts, as a house is with stones. But a collection of facts is no more a science than a heap of stones is a house.”[3]

Poincaré’s metaphor is more powerful than Haack’s call for holistic evidence because it acknowledges that interlocking pieces of evidence may cohere as a building, or they may be no more than a pile of rubble.  Poorly constructed walls may soon revert to the pile of stones from which they came.

Haack proceeds to criticize Judge Kozinski for his “extraordinary argument” that

“(a) equates degrees of proof with statistical probabilities;

(b) assesses each expert’s testimony individually; and

(c) raises the standard of admissibility under the relevance prong to the standard of proof.”

Gap at 672.

Haack misses the point that a low relative risk, with no other valid evidence of specific causation, translates into a low probability of specific causation, even if general causation were apodictically certain. Aggregating the testimony, say between  animal toxicologists and epidemiologists, simply does not advance the epistemic ball on specific causation because all the evidence collectively does not help identify the cause of Jason Daubert’s birth defects on the very model of causation that plaintiffs’ expert witnesses advanced.

All this would be bad enough, but Haack then goes on to commit a serious category mistake in confusing the probabilistic inference (for specific causation) of an urn model with the prosecutor’s fallacy of interpreting a random match probability as the evidence of innocence. (Or the complement of the random match probability as the evidence of guilt.) Judge Kozinski was not working with random match probabilities, and he did not commit the prosecutor’s fallacy.

Take Some Sertraline and Call Me in the Morning

As depressing as Haack’s article is, she manages to make matters even gloomier by attempting a discussion of Judge Rufe’s recent decision in the sertraline birth defects litigation. Haack’s discussion of this decision illustrates and typifies her analyses of other cases, including various decisions on causation opinion testimony on phenylpropanolamine, silicone, bendectin, t-PA, and other occupational, environmental, and therapeutic exposures. Maybe 100 mg sertraline is in order.

Haack criticizes what she perceives to be the conflation of admissibility and sufficiency issues in how the sertraline MDL court addressed the defendants’ motion to exclude the proffered testimony of Dr. Anick Bérard. Gap at 683. The conflation is imaginary, however, and the direct result of Haack’s refusal to look at the specific, multiple methodological flaws in plaintiffs’ expert witness Anick Bérard’s methodologic approach taken to reach a causal conclusion. These flaws are not gradational, and they are detailed in the MDL court’s opinion[4] excluding Anick Bérard. Haack, however, fails to look at the details. Instead Haack focuses on what she suggests is the sertraline MDL court’s conclusion that epidemiology was necessary:

“Judge Rufe argues that reliable testimony about human causation should generally be supported by epidemiological studies, and that ‘when epidemiological studies are equivocal or inconsistent with a causation opinion, experts asserting causation opinions must thoroughly analyze the strengths and weaknesses of the epidemiological research and explain why [it] does not contradict or undermine their opinion’. * * *

Judge Rufe acknowledges the difference between admissibility and sufficiency but, when it comes to the part of their testimony he [sic] deems inadmissible, his [sic] argument seems to be that, in light of the defendant’s epidemiological evidence, the plaintiffs’ expert testimony is insufficient.”

Gap at 682.

This précis is a remarkable distortion of the material facts of the case. There was no plaintiffs’ epidemiology evidence and defendants’ epidemiologic evidence.  Rather there was epidemiologic evidence, and Bérard ignored, misreported, or misrepresented a good deal of the total evidentiary display. Bérard embraced studies when she could use their risk ratios to support her opinions, but criticized or ignored the same studies when their risk ratios pointed in the direction of no association or even of a protective association. To add to this methodological duplicity, Anick Bérard published many statements, in peer-reviewed journals, that sertraline was not shown to cause birth defects, but then changed her opinion solely for litigation. The court’s observation that there was a need for consistent epidemiologic evidence flowed not only from the conception of causation (non-necessary, not sufficient), but from Berard’s and her fellow plaintiffs’ expert witnesses’ concessions that epidemiology was needed.  Haack’s glib approach to criticizing judicial opinions fails to do justice to the difficulties of the task; nor does she advance any meaningful criteria to separate successful from unsuccessful efforts.

In attempting to make her case for the gradational nature of relevance and reliability, Haack acknowledges that the details of the evidence relied upon can render the evidence, and presumably the conclusion based thereon, more or less reliable.  Thus, we are told that epidemiologic studies based upon self-reported diagnoses are highly unreliable because such diagnoses are often wrong. Gap at 667-68. Similarly, we are told that in consider a claim that a plaintiff suffered an adverse effect from a medication, that epidemiologic evidence showing a risk ratio of three would not be reliable if it had inadequate or inappropriate controls,[5] was not double blinded, and lacked randomization. Gap at 668-69. Even if the boundaries between reliable and unreliable are not always as clear as we might like, Haack fails to show that the gatekeeping process lacks a suitable epistemic, scientific foundation.

Curiously, Haack calls out Carl Cranor, plaintiffs’ expert witness in the Milward case, for advancing a confusing, vacuous “weight of the evidence” rationale for the methodology employed by the other plaintiffs’ causation expert witnesses in Milward.[6] Haack argues that Cranor’s invocation of “inference to the best explanation” and “weight of the evidence” fails to answer the important questions at issue in the case, namely how to weight the inference to causation as strong, weak, or absent. Gap at 688 & n. 223, 224. And yet, when Haack discusses court decisions that detailed voluminous records of evidence about how causal inferences should be made and supported, she flies over the details to give us confused, empty conclusions that the trial courts conflated admissibility with sufficiency.

[1] Rideout v. Knox, 19 N.E. 390, 392 (Mass. 1892).

[2] Daubert v. Merrell Dow Pharm., Inc., 43 F.3d 1311, 1320 (9th Cir. 1995).

[3] Jules Henri Poincaré, La Science et l’Hypothèse (1905) (chapter 9, Les Hypothèses en Physique)( “[O]n fait la science avec des faits comme une maison avec des pierres; mais une accumulation de faits n’est pas plus une science qu’un tas de pierres n’est une maison.”).

[4] In re Zoloft Prods. Liab. Litig., 26 F. Supp. 3d 466 (E.D. Pa. 2014).

[5] Actually Haack’s suggestion is that a study with a relative risk of three would not be very reliable if it had no controls, but that suggestion is incoherent.  A risk ratio could not have been calculated at all if there had been no controls.

[6] Milward v. Acuity Specialty Prods., 639 F.3d 11, 17-18 (1st Cir. 2011), cert. denied, 132 S.Ct. 1002 (2012).

Judge Bernstein’s Criticism of Rule 703 of the Federal Rules of Evidence

August 30th, 2016

Federal Rule of Evidence Rule 703 addresses the bases of expert witness opinions, and it is a mess. The drafting of this Rule is particularly sloppy. The Rule tells us, among other things, that:

“[i]f experts in the particular field would reasonably rely on those kinds of facts or data in forming an opinion on the subject, they need not be admissible for the opinion to be admitted.”

This sentence of the Rule has a simple grammatical and logical structure:

If A, then B;

where A contains the concept of reasonable reliance, and B tells us the consequence that the relied upon material need not be itself admissible for the opinion to be admissible.

But what happens if the expert witness has not reasonably relied upon certain facts or data; i.e., ~A?  The conditional statement as given does not describe the outcome in this situation. We are not told what happens when an expert witness’s reliance in the particular field is unreasonable.  ~A does not necessarily imply ~B. Perhaps the drafters meant to write:

B if and only if A.

But the drafters did not give us the above rule, and they have left judges and lawyers to make sense of their poor grammar and bad logic.

And what happens when the reliance material is independently admissible, say as a business record, government report, and first-person observation?  May an expert witness rely upon admissible facts or data, even when a reasonable expert would not do so? Again, it seems that the drafters were trying to limit expert witness reliance to some rule of reason, but by tying reliance to the admissibility of the reliance material, they managed to conflate two separate notions.

And why is reliance judged by the expert witness’s particular field?  Fields of study and areas of science and technology overlap. In some fields, it is common place for putative experts to rely upon materials that would not be given the time of day in other fields. Should we judge the reasonableness of homeopathic healthcare providers’ reliance by the standards of reasonableness in homeopathy, such as it is, or should we judge it by the standards of medical science? The answer to this rhetorical question seems obvious, but the drafters of Rule 703 introduced a Balkanized concept of science and technology by introducing the notion of the expert witness’s “particular field.” The standard of Rule 702 is “knowledge” and “helpfulness,” both of which concepts are not constrained by “particular fields.”

And then Rule 703 leaves us in the dark about how to handle an expert witness’s reliance upon inadmissible facts or data. According to the Rule, “the proponent of the opinion may disclose [the inadmissible facts or data] to the jury only if their probative value in helping the jury evaluate the opinion substantially outweighs their prejudicial effect. And yet, disclosing inadmissible facts or data would always be highly prejudicial because they represent facts and data that the jury is forbidden to consider in reaching its verdict.  Nonetheless, trial judges routinely tell juries that an expert witness’s opinion is no better than the facts and data on which the opinion is based.  If the facts and data are inadmissible, the jury must disregard them in its fact finding; and if an expert witness’s opinion is based upon facts and data that are to be disregarded, then the expert witness’s opinion must be disregarded as well. Or so common sense and respect for the trial’s truth-finding function would suggest.

The drafters of Rule 703 do not shoulder all the blame for the illogic and bad results of the rule. The judicial interpretation of Rule 703 has been sloppy, as well. The Rule’s “plain language” tells us that “[a]n expert may base an opinion on facts or data in the case that the expert has been made aware of or personally observed.”  So expert witnesses should be arriving at their opinions through reliance upon facts and data, but many expert witnesses rely upon others’ opinions, and most courts seem to be fine with such reliance.  And the reliance is often blind, as when medical clinicians rely upon epidemiologic opinions, which in turn are based upon data from studies that the clinicians themselves are incompetent to interpret and critique.

The problem of reliance, as contained within Rule 703, is deep and pervasive in modern civil and criminal trials. In the trial of health effect claims, expert witnesses rely upon epidemiologic and toxicologic studies that contain multiple layers of hearsay, often with little or no validation of the trustworthiness of many of those factual layers. The inferential methodologies are often obscure, even to the expert witnesses, and trial counsel are frequently untrained and ill prepared to expose the ignorance and mistakes of the expert witnesses.

Back in February 2008, I presented at an ALI-ABA conference on expert witness evidence about the problems of Rule 703.[1] I laid out a critique of Rule 703, which showed that the Rule permitted expert witnesses to rely upon “castles in the air.” A distinguished panel of law professors and judges seemed to agree; at least no one offered a defense of Rule 703.

Shortly after I presented at the ALI-ABA conference, Professor Julie E. Seaman published an insightful law review in which she framed the problems of rule 703 as constitutional issues.[2] Encouraged by Professor Seaman’s work, I wrote up my comments on Rule 703 for an ABA publication,[3] and I have updated those comments in the light of subsequent judicial opinions,[4] as well as the failure of the Third Edition of the Reference Manual of Scientific Evidence to address the problems.[5]


Judge Mark I. Bernstein is a trial court judge for the Philadelphia County Court of Common Pleas. I never tried a case before Judge Bernstein, who has announced his plans to leave the Philadelphia bench after 29 years of service,[6] but I had heard from some lawyers (on both sides of the bar) that he was a “pro-plaintiff” judge. Some years ago, I sat next to him on a CLE panel on trial evidence, at which he disparaged judicial gatekeeping,[7] which seemed to support his reputation. The reality seems to be more complex. Judge Bernstein has shown that he can be a critical consumer of complex scientific evidence, and an able gatekeeper under Pennsylvania’s crazy quilt-work pattern of expert witness law. For example, in a hotly contested birth defects case involving sertraline, Judge Bernstein held a pre-trial evidentiary hearing and looked carefully at the proffered testimony of Michael D. Freeman, a chiropractor and self-styled “forensic epidemiologist, and Robert Cabrera, a teratologist. Applying a robust interpretation of Pennsylvania’s Frye rule, Judge Bernstein excluded Freeman and Cabrera’s proffered testimony, and entered summary judgment for defendant Pfizer, Inc. Porter v. Smithkline Beecham Corp., 2016 WL 614572 (Phila. Cty. Ct. Com. Pl.). SeeDemonstration of Frye Gatekeeping in Pennsylvania Birth Defects Case” (Oct. 6, 2015).

And Judge Bernstein has shown that he is one of the few judges who takes seriously Rule 705’s requirement that expert witnesses produce their relied upon facts and data at trial, on cross-examination. In Hansen v. Wyeth, Inc., Dr. Harris Busch, a frequent testifier for plaintiffs, glibly opined about the defendant’s negligence.  On cross-examination, he adverted to the volumes of depositions and documents he had reviewed, but when defense counsel pressed, the witness was unable to produce and show exactly what he had reviewed. After the jury returned a verdict for the plaintiff, Judge Bernstein set the verdict aside because of the expert witness’s failure to comply with Rule 705. Hansen v. Wyeth, Inc., 72 Pa. D. & C. 4th 225, 2005 WL 1114512, at *13, *19, (Phila. Ct. Common Pleas 2005) (granting new trial on post-trial motion), 77 Pa. D. & C. 4th 501, 2005 WL 3068256 (Phila. Ct. Common Pleas 2005) (opinion in support of affirmance after notice of appeal).

In a recent law review article, Judge Bernstein has issued a withering critique of Rule 703. See Hon. Mark I. Bernstein, “Jury Evaluation of Expert Testimony Under the Federal Rules,” 7 Drexel L. Rev. 239 (2015). Judge Bernstein is clearly dissatisfied with the current approach to expert witnesses in federal court, and he lays almost exclusive blame on Rule 703 and its permission to hide the crucial facts, data, and inferential processes from the jury. In his law review article, Judge Bernstein characterizes Rules 703 and 705 as empowering “the expert to hide personal credibility judgments, to quietly draw conclusions, to individually decide what is proper evidence, and worst of all, to offer opinions without even telling the jury the facts assumed.” Id. at 264. Judge Bernstein cautions that the subversion of the factual predicates for expert witnesses’ opinions under Rule 703 has significant, untoward consequences for the court system. Not only are lawyers allowed to hire professional advocates as expert witnesses, but the availability of such professional witnesses permits and encourages the filing of unnecessary litigation. Id. at 286. Hear hear.

Rule 703’s practical consequence of eliminating the hypothetical question has enabled the expert witness qua advocate, and has up-regulated the trial as a contest of opinions and opiners rather than as an adversarial procedure that is designed to get at the truth. Id. at 266-67. Without having access to real, admissible facts and data, the jury is forced to rely upon proxies for the truth: qualifications, demeanor, and courtroom poise, all of which fail the jury and the system in the end.

As a veteran trial judge, Judge Bernstein makes a persuasive case that the non-disclosure permitted under Rule 703 is not really curable under Rule 705. Id. at 288.  If the cross-examination inquiry into reliance material results in the disclosure of inadmissible facts, then judges and the lawyers must deal with the charade of a judicial instruction that the identification of the inadmissible facts is somehow “not for the truth.” Judge Bernstein argues, as have many others, that this “not for the truth” business is an untenable fiction, either not understood or ignored by jurors.

Opposing counsel, of course, may ask for an elucidation of the facts and data relied upon, but when they consider the time and difficulty involved in cross-examining highly experienced, professional witnesses, opposing counsel usually choose to traverse the adverse opinion by presenting their own expert witness’s opinion rather than getting into nettlesome details and risking looking foolish in front of the jury, or even worse, allowing the highly trained adverse expert witness to run off at the mouth.

As powerful as Judge Bernstein’s critique of Rule 703 is, his analysis misses some important points. Lawyers and judges have other motives for not wanting to elicit underlying facts and data: they do not want to “get into the weeds,” and they want to avoid technical questions of valid inference and quality of data. Yet sometimes the truth is in the weeds. Their avoidance of addressing the nature of inference, as well as facts and data, often serves to make gatekeeping a sham.

And then there is the problem that arises from the lack of time, interest, and competence among judges and jurors to understand the technical details of the facts and data, and inferences therefrom, which underlie complex factual disputes in contemporary trials. Cross examination is reduced to the attempt to elicit “sound bites” and “cheap shots,” which can be used in closing argument. This approach is common on both sides of the bar, in trials before judges and juries, and even at so-called Daubert hearings. See David E. Bernstein & Eric G. Lasker,“Defending Daubert: It’s Time to Amend Federal Rule of Evidence 702,” 57 William & Mary L. Rev. 1, 32 (2015) (“Rule 703 is frequently ignored in Daubert analyses”).

The Rule 702 and 703 pretrial hearing is an opportunity to address the highly technical validity questions, but even then, the process is doomed to failure unless trial judges make adequate time and adopt an attitude of real intellectual curiosity to permit a proper exploration of the evidentiary issues. Trial lawyers often discover that a full exploration is technical and tedious, and that it pisses off the trial judge. As much as judges dislike having to serve as gatekeepers of expert witness opinion testimony, they dislike even more having to assess the reasonableness of individual expert witness’s reliance upon facts and data, especially when this inquiry requires a deep exploration of the methods and materials of each relied upon study.

In favor of something like Rule 703, Bernstein’s critique ignores that there are some facts and data that will never be independently admissible. Epidemiologic studies, with their multiple layers of hearsay, come to mind.

Judge Bernstein, as a reformer, is wrong to suggest that the problem is solely in hiding the facts and data from the jury. Rules 702 and 703 march together, and there are problems with both that require serious attention. See David E. Bernstein & Eric G. Lasker,“Defending Daubert: It’s Time to Amend Federal Rule of Evidence 702,” 57 William & Mary L. Rev. 1 (2015); see alsoOn Amending Rule 702 of the Federal Rules of Evidence” (Oct. 17, 2015).

And we should remember that the problem is not solely with juries and their need to see the underlying facts and data. Judges try cases too, and can butcher scientific inference with any help from a lay jury. Then there is the problem of relied upon opinions, discussed above. And then there is the problem of unreasonable reliance of the sort that juries cannot discern even if they see the underlying, relied upon facts and data.

[1] Schachtman, “Rule 703 – The Problem Child of Article VII”; and “The Effective Presentation of Defense Expert Witnesses and Cross-examination of Plaintiffs’ Expert Witnesses”; at the ALI-ABA Course on Opinion and Expert Witness Testimony in State and Federal Courts (February 14-15, 2008).

[2] See Julie E. Seaman, “Triangulating Testimonial Hearsay: The Constitutional Boundaries of Expert Opinion Testimony,” 96 Georgetown L.J. 827 (2008).

[3]  Nathan A. Schachtman, “Rule of Evidence 703—Problem Child of Article VII,” 17 Proof 3 (Spring 2009).

[4]RULE OF EVIDENCE 703 — Problem Child of Article VII” (Sept. 19, 2011)

[5] SeeGiving Rule 703 the Cold Shoulder” (May 12, 2012); “New Reference Manual on Scientific Evidence Short Shrifts Rule 703,” (Oct. 16, 2011).

[6] Max Mitchell, “Bernstein Announces Plan to Step Down as Judge,” The Legal Intelligencer (July 29, 2016).

[7] See Schachtman, “Court-Appointed Expert Witnesses,” for Mealey’s Judges & Lawyers in Complex Litigation, Class Actions, Mass Torts, MDL and the Monster Case Conference, in West Palm Beach, Florida (November 8-9, 1999). I don’t recall Judge Bernstein’s exact topic, but I remember he criticized the Pennsylvania Supreme Court’s decision in Blum v. Merrill Dow Pharmaceuticals, 534 Pa. 97, 626 A.2d 537 ( 1993), which reversed a judgment for plaintiffs, and adopted what Judge Bernstein derided as a blending of Frye and Daubert, which he called Fraubert. Judge Bernstein had presided over the Blum trial, which resulted in the verdict for plaintiffs.

Systematic Reviews and Meta-Analyses in Litigation, Part 2

February 11th, 2016

Daubert in Knee’d

In a recent federal court case, adjudicating a plaintiff’s Rule 702 challenge to defense expert witnesses, the trial judge considered plaintiff’s claim that the challenged witness had deviated from PRISM guidelines[1] for systematic reviews, and thus presumably had deviated from the standard of care required of expert witnesses giving opinions about causal conclusions.

Batty v. Zimmer, Inc., MDL No. 2272, Master Docket No. 11 C 5468, No. 12 C 6279, 2015 WL 5050214 (N.D. Ill. Aug. 25, 2015) [cited as Batty I]. The trial judge, the Hon. Rebecca R. Pallmeyer, denied plaintiff’s motion to exclude the allegedly deviant witness, but appeared to accept the premise of the plaintiff’s argument that an expert witness’s opinion should be reached in the manner of a carefully constructed systematic review.[2] The trial court’s careful review of the challenged witness’s report and deposition testimony revealed that there had mean no meaningful departure from the standards put forward for systematic reviews. SeeSystematic Reviews and Meta-Analyses in Litigation” (Feb. 5, 2016).

Two days later, the same federal judge addressed a different set of objections by the same plaintiff to two other of the defendant’s, Zimmer Inc.’s, expert witnesses, Dr. Stuart Goodman and Dr. Timothy Wright. Batty v. Zimmer, Inc., MDL No. 2272, Master Docket No. 11 C 5468, No. 12 C 6279, 2015 WL 5095727, (N.D. Ill. Aug. 27, 2015) [cited as Batty II]. Once again, plaintiff Batty argued for the necessity of adherence to systematic review principles. According to Batty, Dr. Wright’s opinion, based upon his review of the clinical literature, was scientifically and legally unreliable because he had not conducted a proper systematic review. Plaintiff alleged that Dr. Wright’s review selectively “cherry picked” favorable studies to buttress his opinion, in violation of systematic review guidelines. The trial court, which had assumed that a systematic review was the appropriate “methodology” for Dr. Vitale, in Batty I, refused to sustain the plaintiff’s challenge in Batty II, in large part because the challenged witness, Dr. Wright, had not claimed to have performed a systematic or comprehensive review, and so his failure to follow the standard methodology did not require the exclusion of his opinion at trial. Batty II at *3.

The plaintiff never argued that Dr. Wright misinterpreted any of his selected studies upon which he relied, and the trial judge thus suggested that Dr. Wright’s discussion of the studies, even if a partial, selected group of studies, would be helpful to the jury. The trial court thus left the plaintiff to her cross-examination to highlight Dr. Wright’s selectivity and lack of comprehensiveness. Apparently, in the trial judge’s view, this expert witness’s failure to address contrary studies did not render his testimony unreliable under “Daubert scrutiny.” Batty II at *3.

Of course, it is no longer the Daubert judicial decision that mandates scrutiny of expert witness opinion testimony, but Federal Rule of Evidence 702. Perhaps it was telling that when the trial court backed away from its assumption, made in Batty I, that guidelines or standards for systematic reviews should inform a Rule 702 analysis, the court cited Daubert, a judicial opinion superseded by an Act of Congress, in 2000. The trial judge’s approach, in Batty II, threatens to make gatekeeping meaningless by deferring to the expert witness’s invocation of personal, idiosyncratic, non-scientific standards. Furthermore, the Batty II approach threatens to eviscerate gatekeeping for clinical practitioners who remain blithely unaware of advances in epidemiology and evidence-based medicine. The upshot of Batty I and II combined seems to be that systematic review principles apply to clinical expert witnesses only if those witness choose to be bound by such principles. If this is indeed what the trial court intended, then it is jurisprudential nonsense.

The trial court, in Batty II, exercised a more searching approach, however, to Dr. Wright’s own implant failure analysis, which he relied upon in an attempt to rebut plaintiff’s claim of defective design. The plaintiff claimed that the load-bearing polymer surfaces of the artificial knee implant experienced undue deformation. Dr. Wright’s study found little or no deformation on the load bearing polymer surfaces of the eight retrieved artificial joints. Batty II at *4.

Dr. Wright assessed deformation qualitatively, not quantitatively, through the use of a “colormetric map of deformation” of the polymer surface. Dr. Wright, however, provided no scale to define or assess how much deformation was represented by the different colors in his study. Notwithstanding the lack of any metric, Dr. Wright concluded that his findings, based upon eight retrieved implants, “suggested” that the kind of surface failing claimed by plaintiff was a “rare event.”

The trial court had little difficulty in concluding that Dr. Wright’s evidentiary base was insufficient, as was his presentation of the study’s data and inferences. The challenged witness failed to explain how his conclusions followed from his data, and thus his proffered testimony fell into the “ipse dixit” category of inadmissible opinion testimony. General Electric v. Joiner, 522 U.S. 136, 146 (1997). In the face of the challenge to his opinions, Dr. Wright supplemented his retrieval study with additional scans of surficial implant wear patterns, but he failed again to show the similarity of previous use and failure conditions in the patients from whom these implants were retrieved and the plaintiff’s case (which supposedly involved aseptic loosening). Furthermore, Dr. Wright’s interpretation of his own retrieval study was inadequate in the trial court’s view because he had failed to rule out other modes of implant failure, in which the polyethylene surface would have been preserved. Because, even as supplemented, Dr. Wright’s study failed to support his proffered opinions, the court held that his opinions, based upon his retrieval study had to be excluded under Rule 702. The trial court did not address the Rule 703 implications for Dr. Wright’s reliance upon a study that was poorly designed and explained, and which lacked the ability to support his contention that the claimed mode of implant failure was a “rare” event. Batty II at *4 – 5.

[1] See David Moher , Alessandro Liberati, Jennifer Tetzlaff, Douglas G. Altman, & The PRISMA Group, “Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA Statement,” 6 PLoS Med e1000097 (2009) [PRISMA].

[2] Batty v. Zimmer, Inc., MDL No. 2272, Master Docket No. 11 C 5468, No. 12 C 6279, 2015 WL 5050214 (N.D. Ill. Aug. 25, 2015).