Schachtman Law » statistical evidence

TORTINI

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

Statistical Deontology

March 2nd, 2018

In courtrooms across America, there has been a lot of buzzing and palavering about the American Statistical Association’s Statement on Statistical Significance Testing,¹ but very little discussion of the Society’s Ethical Guidelines, which were updated and promulgated in the same year, 2016. Statisticians and statistics, like lawyers and the law, receive their fair share of calumny over their professional activities, but the statistician’s principal North American professional organization is trying to do something about members’ transgressions.

The American Statistical Society (ASA) has promulgated ethical guidelines for statisticians, as has the Royal Statistical Society,² even if these organizations lack the means and procedures to enforce their codes. The ASA’s guidelines ³ are rich with implications for statistical analyses put forward in all contexts, including in litigation and regulatory rule making. As such, the guidelines are well worth studying by lawyers.

The ASA Guidelines were prepared by the Committee on Professional Ethics, and approved by the ASA’s Board in April 2016. There are lots of “thou shall” and “thou shall nots,” but I will focus on the issues that are more likely to arise in litigation. What is remarkable about the Guidelines is that if followed, they probably are more likely to eliminate unsound statistical practices in the courtroom than the ASA State on P-values.

Defining Good Statistical Practice

“Good statistical practice is fundamentally based on transparent assumptions, reproducible results, and valid interpretations.” Guidelines at 1. The Guidelines thus incorporate something akin to the Kumho Tire standard that an expert witness ‘‘employs 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, 152 (1999).

A statistician engaged in expert witness testimony should provide “only expert testimony, written work, and oral presentations that he/she would be willing to have peer reviewed.” Guidelines at 2. “The ethical statistician uses methodology and data that are relevant and appropriate, without favoritism or prejudice, and in a manner intended to produce valid, interpretable, and reproducible results.” Id. Similarly, the statistician, if ethical, will identify and mitigate biases, and use analyses “appropriate and valid for the specific question to be addressed, so that results extend beyond the sample to a population relevant to the objectives with minimal error under reasonable assumptions.” Id. If the Guidelines were followed, a lot of spurious analyses would drop off the litigation landscape, regardless whether they used p-values or confidence intervals, or a Bayesian approach.

Integrity of Data and Methods

The ASA’s Guidelines also have a good deal to say about data integrity and statistical methods. In particular, the Guidelines call for candor about limitations in the statistical methods or the integrity of the underlying data:

“The ethical statistician is candid about any known or suspected limitations, defects, or biases in the data that may impact the integrity or reliability of the statistical analysis. Objective and valid interpretation of the results requires that the underlying analysis recognizes and acknowledges the degree of reliability and integrity of the data.”

Guidelines at 3.

The statistical analyst openly acknowledges the limits of statistical inference, the potential sources of error, as well as the statistical and substantive assumptions made in the execution and interpretation of any analysis,” including data editing and imputation. Id. The Guidelines urge analysts to address potential confounding not assessed by the study design. Id. at 3, 10. How often do we see these acknowledgments in litigation-driven analyses, or in peer-reviewed papers, for that matter?

Affirmative Actions Prescribed

In the aid of promoting data and methodological integrity, the Guidelines also urge analysts to share data when appropriate without revealing the identities of study participants. Statistical analysts should publicly correct any disseminated data and analyses in their own work, as well as working to “expose incompetent or corrupt statistical practice.” Of course, the Lawsuit Industry will call this ethical duty “attacking the messenger,” but maybe that’s a rhetorical strategy based upon an assessment of risks versus benefits to the Lawsuit Industry.

Multiplicity

The ASA Guidelines address the impropriety of substantive statistical errors, such as:

“[r]unning multiple tests on the same data set at the same stage of an analysis increases the chance of obtaining at least one invalid result. Selecting the one “significant” result from a multiplicity of parallel tests poses a grave risk of an incorrect conclusion. Failure to disclose the full extent of tests and their results in such a case would be highly misleading.”

Guidelines at 9.

There are some Lawsuit Industrialists who have taken comfort in the pronouncements of Kenneth Rothman on corrections for multiple comparisons. Rothman’s views on multiple comparisons are, however, much broader and more nuanced than the Industry’s sound bites.⁴ Given that Rothman opposes anything like strict statistical significance testing, it follows that he is relatively unmoved for the need for adjustments to alpha or the coefficient of confidence. Rothman, however, has never deprecated the need to consider the multiplicity of testing, and the need for researchers to be forthright in disclosing the the scope of comparisons originally planned and actually done.

1 Ronald L. Wasserstein & Nicole A. Lazar, “The ASA’s Statement on p-Values: Context, Process, and Purpose,” 70 Am. Statistician, (2016). See “The American Statistical Association’s Statement on and of Significance” (March 17, 2016); “The ASA’s Statement on Statistical Significance – Buzzing from the Huckabees” (March 19, 2016).

2 Royal Statistical Society – Code of Conduct (2014); Steven Piantadosi, Clinical Trials: A Methodologic Perspective 609 (2d ed. 2005).

3 Shelley Hurwitz & John S. Gardenier, “Ethical Guidelines for Statistical Practice: The First 60 Years and Beyond,” 66 Am. Statistician 99 (2012) (describing the history and evolution of the Guidelines).

4 Kenneth J. Rothman, “Six Persistent Research Misconceptions,” 29 J. Gen. Intern. Med. 1060, 1063 (2014).

Posted in Data Sharing, Expert Witness Discovery, Scientific Misconduct, statistical evidence, Underlying Data | Comments Off on Statistical Deontology

The 5% Solution at the FDA

February 24th, 2018

The statistics wars rage on¹, with Bayesians attempting to take advantage of the so-called replication crisis to argue it is all the fault of frequentist significance testing. In 2016, there was an attempted coup at the American Statistical Association, but the Bayesians did not get what they wanted, with little more than a consensus that p-values and confidence intervals should be properly interpreted. Patient advocacy groups have lobbied for the availability of unapproved and incompletely tested medications, and rent-seeking litigation has argued and lobbied for the elimination of statistical tests and methods in the assessment of causal claims. The battle continues.

Against this backdrop, a young Harvard graduate student has published a a paper with a brief history of significance testing, and the role that significance testing has taken on at the United States Food and Drug Administration (FDA). Lee Kennedy-Shaffer, “When the Alpha is the Omega: P-Values, ‘Substantial Evidence’, and the 0.05 Standard at FDA,” 72 Food & Drug L.J. 595 (2017) [cited below as K-S]. The paper presents a short but entertaining history of the evolution of the p-value from its early invocation in 1710, by John Arbuthnott, a Scottish physician and mathematician, who calculated the probability that male births would exceed female births 82 consecutive years if their true proportions were equal. K-S at 603. Kennedy-Shaffer notes the role of the two great French mathematicians, Pierre-Simon Laplace and Siméon-Denis Poisson, who used p-values (or their complements) to evaluate empirical propositions. As Kennedy-Shaffer notes, Poisson observed that the equivalent of what would be a modern p-value about 0.005, was sufficient in his view, back in 1830, to believe that the French Revolution of 1830 had caused the pattern of jury verdicts to be changed. K-S at 604.

Kennedy-Shaffer traces the p-value, or its equivalent, through its treatment by the great early 20th century statisticians, Karl Pearson and Ronald A. Fisher, through its modification by Jerzy Neyman and Egon Pearson, into the bowels of the FDA in Rockville, Maryland. It is a history well worth recounting, if for no other reason, to remind us that the p-value or its equivalent has been remarkably durable and reasonably effective in protecting the public against false claims of safety and efficacy. Kennedy-Shaffer provides several good examples in which the FDA’s use of significance testing was outcome dispositive of approval or non-approval of medications and devices.

There is enough substance and history here that everyone will have something to pick at this paper. Let me volunteer the first shot. Kennedy-Shaffer describes the co-evolution of the controlled clinical trial and statistical tests, and points to the landmark study by the Medical Research Council on streptomycin for tuberculosis. Geoffrey Marshall (chairman), “Streptomycin Treatment of Pulmonary Tuberculosis: A Medical Research Council Investigation,” 2 Brit. Med. J. 769, 769–71 (1948). This clinical trial was historically important, not only for its results and for Sir Austin Bradford Hill’s role in its design, but for the care with which it described randomization, double blinding, and multiple study sites. Kennedy-Shaffer suggests that “[w]hile results were presented in detail, few formal statistical tests were incorporated into this analysis.” K-S at 597-98. And yet, a few pages later, he tells us that “both chi-squared tests and t-tests were used to evaluate the responses to the drug and compare the control and treated groups,” and that “[t]he difference in mortality between the two groups is statistically significant.” K-S at 611. Although it is true that the authors did not report their calculated p-values for any test, the difference in mortality between the streptomycin and control groups was very large, and the standards for describing the results of such a clinical trial were in their infancy in 1948.

Kennedy-Shaffer’s characterization of Sir Austin Bradford Hill’s use of statistical tests and methods takes on out-size importance because of the mischaracterizations, and even misrepresentations, made by some representatives of the Lawsuit Industry, who contend that Sir Austin dismissed statistical methods as unnecessary. In the United States, some judges have been seriously misled by those misrepresentations, which have their way into published judicial decisions.

The operative document, of course, is the publication of Sir Austin’s famous after-dinner speech, in 1965, on the occasion of his election to the Presidency of the Royal Society of Medicine. Although the speech is casual and free of scholarly footnotes, Sir Austin’s message was precise, balanced, and nuanced. The speech is a classic in the history of medicine, which remains important even if rather dated in terms of its primary message about how science and medicine move from beliefs about associations to knowledge of causal associations. As everyone knows, Sir Austin articulated nine factors or viewpoints through which to assess any putative causal association, but he emphasized that before these nine factors are assessed, our starting point itself has prerequisites:

“Disregarding then any such problem in semantics we have this situation. Our observations reveal an association between two variables, perfectly clear-cut and beyond what we would care to attribute to the play of chance. What aspects of that association should we especially consider before deciding that the most likely interpretation of it is causation?”

Austin Bradford Hill, “The Environment and Disease: Association or Causation?” 58 Proc. Royal Soc’y Med. 295, 295 (1965) [cited below as Hill]. The starting point, therefore, before the Bradford Hill nine factors come into play, is a “clear-cut” association, which is “beyond what we would care to attribute to the play of chance.”

In other words, consideration of random error is necessary.

Now for the nuance and the balance. Sir Austin acknowledged that there were some situations in which we simply do not need to calculate standard errors because the disparity between treatment and control groups is so large and meaningful. He goes on to wonder out loud:

“whether the pendulum has not swung too far – not only with the attentive pupils but even with the statisticians themselves. To decline to draw conclusions without standard errors can surely be just as silly? Fortunately I believe we have not yet gone so far as our friends in the USA where, I am told, some editors of journals will return an article because tests of significance have not been applied. Yet there are innumerable situations in which they are totally unnecessary – because the difference is grotesquely obvious, because it is negligible, or because, whether it be formally significant or not, it is too small to be of any practical importance. What is worse the glitter of the t table diverts attention from the inadequacies of the fare.”

Hill at 299. Now this is all true, but hardly the repudiation of statistical testing claimed by those who want to suppress the consideration of random error from science and judicial gatekeeping. There are very few litigation cases in which the difference between exposed and unexposed is “grotesquely obvious,” such that we can leave statistical methods at the door. Importantly, the very large differences between the streptomycin and placebo control groups in the Medical Council’s 1948 clinical trial were not so “grotesquely obvious” that statistical methods were obviated. To be fair, the differences were sufficiently great that statistical discussion could be kept to a minimum. Sir Austin gave extensive tables in the 1948 paper to let the reader appreciate the actual data themselves.

In his after-dinner speech, Hill also gives examples of studies that are so biased and confounded that no statistical method will likely ever save them. Certainly, the technology of regression and propensity-score analyses have progressed tremendously since Hill’s 1965 speech, but his point still remains. This point hardly excuses the lack of statistical apparatus in highly confounding or biased observations.

In addressing the nine factors he identified, which presumed a “clear-cut” association, with random error ruled out, Sir Austin did opine that for the factors raised questions and that:

“No formal tests of significance can answer those questions. Such tests can, and should, remind us of the effects that the play of chance can create, and they will instruct us in the likely magnitude of those effects. Beyond that they contribute nothing to the ‘proof’ of our hypothesis.”

Hill at 299. Again, the date and the context are important. Hill is addressing consideration of the nine factors, not the required predicate association beyond the play of chance or random error. The date is important as well, because it would be foolish to suggest that statistical methods have not grown in the last half century to address some of the nine factors. The existence and the nature of dose-response are the subject of extensive statistical methods, and meta-analysis and meta-regression are used to assess and measure consistency between studies.

Kennedy-Shaffer might well have pointed out the great influence Sir Austin’s textbook on medical statistics had had on medical research and practice. This textbook, which went through numerous editions, makes clear the importance of statistical testing and methods:

“Are simple methods of the interpretation of figures only a synonym for common sense or do they involve an art or knowledge which can be imparted? Familiarity with medical statistics leads inevitably to the conclusion that common sense is not enough. Mistakes which when pointed out look extremely foolish are quite frequently made by intelligent persons, and the same mistakes, or types of mistakes, crop up again and again. There is often lacking what has been called a ‘statistical tact, which is rather more than simple good sense’. That fact the majority of persons must acquire (with a minority it is undoubtedly innate) by a study of the basic principles of statistical method.”

Austin Bradford Hill, Principles of Medical Statistics at 2 (4th ed. 1948) (emphasis in original). And later in his text, Sir Austin notes that:

“The statistical method is required in the interpretation of figures which are at the mercy of numerous influences, and its object is to determine whether individual influences can be isolated and their effects measured.”

Id. at 10 (emphasis added).

Sir Austin’s work taken as a whole demonstrates the acceptance of the necessity of statistical methods in medicine, and causal inference. Kennedy-Shaffer’s paper covers much ground, but it short changes this important line of influence, which lies directly in the historical path between Sir Ronald Fisher and the medical regulatory community.

Kennedy-Shaffer gives a nod to Bayesian methods, and even suggests that Bayesian results are “more intuitive,” but he does not explain the supposed intuitiveness of how a parameter has a probability distribution. This might make sense at the level of quantum physics, but does not seem to describe the reality of a biomedical phenomenon such as relative risk. Kennedy-Shaffer notes the FDA’s expression of willingness to entertain Bayesian analyses of clinical trials, and the rare instances in which such analyses have actually been deployed. K-S at 629 (“e.g., Pravigard Pac for prevention of myocardial infarction”). He concedes, however, that Bayesian designs are still the exception to the rule, as well as the cautions of Robert Temple, a former FDA Director of Medical Policy, in 2005, that Bayesian proposals for drug clinical trials were at that time “very rare.²” K-S at 630.

1 Deborah Mayo, Statistical Inference as Severe Testing: How to Get Beyond the Statistics Wars (2018).

2 Robert Temple, “How FDA Currently Makes Decisions on Clinical Studies,” 2 Clinical Trials 276, 281 (2005).

Posted in statistical evidence | Comments Off on The 5% Solution at the FDA

Scientific Evidence in Canadian Courts

February 20th, 2018

A couple of years ago, Deborah Mayo called my attention to the Canadian version of the Reference Manual on Scientific Evidence.¹ In the course of discussion of mistaken definitions and uses of p-values, confidence intervals, and significance testing, Sander Greenland pointed to some dubious pronouncements in the Science Manual for Canadian Judges [Manual].

Unlike the United States federal court Reference Manual, which is published through a joint effort of the National Academies of Science, Engineering, and Medicine, the Canadian version, is the product of the Canadian National Judicial Institute (NJI, or the Institut National de la Magistrature, if you live in Quebec), which claims to be an independent, not-for-profit group, that is committed to educating Canadian judges. In addition to the Manual, the Institute publishes Model Jury Instructions and a guide, Problem Solving in Canada’s Courtrooms: A Guide to Therapeutic Justice (2d ed.), as well as conducting educational courses.

The NJI’s website describes the Instute’s Manual as follows:

“Without the proper tools, the justice system can be vulnerable to unreliable expert scientific evidence.

* * *

The goal of the Science Manual is to provide judges with tools to better understand expert evidence and to assess the validity of purportedly scientific evidence presented to them. …”

The Chief Justice of Canada, Hon. Beverley M. McLachlin, contributed an introduction to the Manual, which was notable for its frank admission that:

“[w]ithout the proper tools, the justice system is vulnerable to unreliable expert scientific evidence.

****

Within the increasingly science-rich culture of the courtroom, the judiciary needs to discern ‘good’ science from ‘bad’ science, in order to assess expert evidence effectively and establish a proper threshold for admissibility. Judicial education in science, the scientific method, and technology is essential to ensure that judges are capable of dealing with scientific evidence, and to counterbalance the discomfort of jurists confronted with this specific subject matter.”

Manual at 14. These are laudable goals, indeed, but did the National Judicial Institute live up to its stated goals, or did it leave Canadian judges vulnerable to the Institute’s own “bad science”?

In his comments on Deborah Mayo’s blog, Greenland noted some rather cavalier statements in Chapter two that suggest that the conventional alpha of 5% corresponds to a “scientific attitude that unless we are 95% sure the null hypothesis is false, we provisionally accept it.” And he, pointed elsewhere where the chapter seems to suggest that the coefficient of confidence that corresponds to an alpha of 5% “constitutes a rather high standard of proof,” thus confusing and conflating probability of random error with posterior probabilities. Greenland is absolutely correct that the Manual does a rather miserable job of educating Canadian judges if our standard for its work product is accuracy and truth.

Some of the most egregious errors are within what is perhaps the most important chapter of the Manual, Chapter 2, “Science and the Scientific Method.” The chapter has two authors, a scientist, Scott Findlay, and a lawyer, Nathalie Chalifour. Findlay is an Associate Professor, in the Department of Biology, of the University of Ottawa. Nathalie Chalifour is an Associate Professor on the Faculty of Law, also in the University of Ottawa. Together, they produced some dubious pronouncements, such as:

Weight of the Evidence (WOE)

“First, the concept of weight of evidence in science is similar in many respects to its legal counterpart. In both settings, the outcome of a weight-of-evidence assessment by the trier of fact is a binary decision.”

Manual at 40. Findlay and Chalifour cite no support for their characterization of WOE in science. Most attempts to invoke WOE are woefully vague and amorphous, with no meaningful guidance or content.² Sixty-five pages later, if any one is noticing, the authors let us in a dirty, little secret:

“at present, there exists no established prescriptive methodology for weight of evidence assessment in science.”

Manual at 105. The authors omit, however, that there are prescriptive methods for inferring causation in science; you just will not see them in discussions of weight of the evidence. The authors then compound the semantic and conceptual problems by stating that “in a civil proceeding, if the evidence adduced by the plaintiff is weightier than that brought forth by the defendant, a judge is obliged to find in favour of the plaintiff.” Manual at 41. This is a remarkable suggestion, which implies that if the plaintiff adduces the crummiest crumb of evidence, a mere peppercorn on the scales of justice, but the defendant has none to offer, that the plaintiff must win. The plaintiff wins notwithstanding that no reasonable person could believe that the plaintiff’s claims are more likely than not true. Even if there were the law of Canada, it is certainly not how scientists think about establishing the truth of empirical propositions.

Confusion of Hypothesis Testing with “Beyond a Reasonable Doubt”

The authors’ next assault comes in conflating significance probability with the probability connected with the burden of proof, a posterior probability. Legal proceedings have a defined burden of proof, with criminal cases requiring the state to prove guilt “beyond a reasonable doubt.” Findlay and Chalifour’s discussion then runs off the rails by likening hypothesis testing, with an alpha of 5% or its complement, 95%, as a coefficient of confidence, to a “very high” burden of proof:

“In statistical hypothesis-testing – one of the tools commonly employed by scientists – the predisposition is that there is a particular hypothesis (the null hypothesis) that is assumed to be true unless sufficient evidence is adduced to overturn it. But in statistical hypothesis-testing, the standard of proof has traditionally been set very high such that, in general, scientists will only (provisionally) reject the null hypothesis if they are at least 95% sure it is false. Third, in both scientific and legal proceedings, the setting of the predisposition and the associated standard of proof are purely normative decisions, based ultimately on the perceived consequences of an error in inference.”

Manual at 41. This is, as Greenland and many others have pointed out, a totally bogus conception of hypothesis testing, and an utterly false description of the probabilities involved.

Later in the chapter, Findlay and Chalifour flirt with the truth, but then lapse into an unrecognizable parody of it:

“Inferential statistics adopt the frequentist view of probability whereby a proposition is either true or false, and the task at hand is to estimate the probability of getting results as discrepant or more discrepant than those observed, given the null hypothesis. Thus, in statistical hypothesis testing, the usual inferred conclusion is either that the null is true (or rather, that we have insufficient evidence to reject it) or it is false (in which case we reject it). 16 The decision to reject or not is based on the value of p if the estimated value of p is below some threshold value a, we reject the null; otherwise we accept it.”

Manual at 74. OK; so far so good, but here comes the train wreck:

“By convention (and by convention only), scientists tend to set α = 0.05; this corresponds to the collective – and, one assumes, consensual – scientific attitude that unless we are 95% sure the null hypothesis is false, we provisionally accept it. It is partly because of this that scientists have the reputation of being a notoriously conservative lot, given that a 95% threshold constitutes a rather high standard of proof.”

Manual at 75. Uggh; so we are back to significance probability’s being a posterior probability. As if to atone for their sins, in the very next paragraph, the authors then remind the judicial readers that:

“As noted above, p is the probability of obtaining results at least as discrepant as those observed if the null is true. This is not the same as the probability of the null hypothesis being true, given the results.”

Manual at 75. True, true, and completely at odds with what the authors have stated previously. And to add to the reader’s now fully justified conclusion, the authors describe the standard for rejecting the null hypothesis as “very high indeed.” Manual at 102, 109. Any reader who is following the discussion might wonder how and why there is such a problem of replication and reproducibility in contemporary science.

Conflating Bayesianism with Frequentist Modes of Inference

We have seen how Findlay and Chalifour conflate significance and posterior probabilities, some of the time. In a section of their chapter that deals explicitly with probability, the authors tell us that before any study is conducted the prior probability of the truth of the tested hypothesis is 50%, sans evidence. This an astonishing creation of certainty out nothingness, and perhaps it explains the authors’ implied claim that the crummiest morsel of evidence on one side is sufficient to compel a verdict, if the other side has no morsels at all. Here is how the authors put their claim to the Canadian judges:

“Before each study is conducted (that is, a priori), the hypothesis is as likely to be true as it is to be false. Once the results are in, we can ask: How likely is it now that the hypothesis is true? In the first study, the low a priori inferential strength of the study design means that this probability will not be much different from the a priori value of 0.5 because any result will be rather equivocal owing to limitations in the experimental design.”

Manual at 64. This implied Bayesian slant, with 50% priors, in the world of science would lead anyone to believe “as many as six impossible things before breakfast,” and many more throughout the day.

Lest you think that the Manual is all rubbish, there are occasional gems of advice to the Canadian judges. The authors admonish the judges to

“be wary of individual ‘statistically significant’ results that are mined from comparatively large numbers of trials or experiments, as the results may be ‘cherry picked’ from a larger set of experiments or studies that yielded mostly negative results. The court might ask the expert how many other trials or experiments testing the same hypothesis he or she is aware of, and to describe the outcome of those studies.”

Manual at 87. Good advice, but at odds with the authors’ characterization of statistical significance as establishing the rejection of the null hypothesis well-nigh beyond a reasonable doubt.

When Greenland first called attention to this Manual, I reached to some people who had been involved in its peer review. One reviewer told me that it was a “living document,” and would likely be revised after he had the chance to call the NJI’s attention to the errors. But two years later, the errors remain, and so we have to infer that the authors meant to say all the contradictory and false statements that are still present in the downloadable version of the Manual.

1 See “‘Statistical Significance’ According to the U.S. Dept. of Health and Human Services (ii),” Error Statistics Philosophy (July 17, 2015). At the time, I wrote briefly about the Canadian Manual. See “Canadian Judges’ Reference Manual on Scientific Evidence” (July 24, 2015).

2 See “WOE-fully Inadequate Methodology – An Ipse Dixit By Another Name” (May 1, 2012); “Weight of the Evidence in Science and in Law” (July 29, 2017); see also David E. Bernstein, “The Misbegotten Judicial Resistance to the Daubert Revolution,” 89 Notre Dame L. Rev. 27 (2013).

Posted in Reference Manual on Scientific Evidence, Scientific Evidence, statistical evidence | Comments Off on Scientific Evidence in Canadian Courts

Ninth Circuit Quashes Harkonen’s Last Chance

January 8th, 2018

With the benefit of hindsight, even the biggest whopper can be characterized as a strategic choice for trial counsel. As are result of this sort of thinking, the convicted have a very difficult time in pressing claims of ineffective assistance of counsel. After the fact, a reviewing or an appellate court can always imagine a strategic reason for trial counsel’s decisions, even if they contributed to the client’s conviction.

In the Harkonen case, a pharmaceutical executive was indicted and tried for wire fraud and misbranding. His crime was to send out a fax with a preliminary assessment of a recently unblinded clinical trial. In his fax, Dr Harkonen described the trial’s results as “demonstrating” a survival benefit in study participants with mild and moderate disease. Survival (or mortality) was not a primary outcome of the trial, but it was a secondary outcome, and arguably the most important one of all. The subgroup of “mild and moderate” was not pre-specified, but it was highly plausible.

Clearly, Harkonen’s post hoc analysis would not be sufficient normally to persuade the FDA to approve a medication, but Harkonen did not assert or predict that the company would obtain FDA approval. He simply claimed that the trial “demonstrated” a benefit. A charitable interpretation of his statement, which was several pages long, would include the prior successful clinical trial, as important context for Harkonen’s statement.

The United States government, however, was not interested in the principle of charity, the context, or even its own pronouncements on the issue of statistical significance. Instead, the United States Attorney pushed for draconian sentences under the Wire Fraud Act, and the misbranding sections of the Food, Drug, and Cosmetics Act. A jury acquitted on the misbranding charge, but convicted on wire fraud. The government’s request for an extreme prison term and fines was rebuffed by the trial court, which imposed a term of six months of house arrest, and a small fine.¹ The conviction, however, effectively keeps Dr Harkonen from working again in the pharmaceutical industry.

In post-verdict challenges to the conviction, Harkonen’s lawyers were able to marshal support from several well-renown statisticians and epidemiologists, but the trial court was reluctant to consider these post-verdict opinions when the defense called no expert witness at trial. The trial situation, however, was complicated and confused by the government’s pre-trial position that it would not call expert witnesses on the statistical and clinical trial interpretative issues. Contrary to these representations, the government called Dr Thomas Fleming, as statistician, who testified at some length, and without objection, to strict criteria for assessing statistical significance and causation in clinical trials.

Having read Fleming’s testimony, I can say that the government got away with introducing a great deal of expert witness opinion testimony, without effective contradiction or impeachment. With the benefit of hindsight, the defense decision not to call an expert witness looks like a serious deviation from the standard of care. Fleming’s “facts” about how the FDA would evaluate the success or failure of the clinical trial were not relevant to whether Harkonen’s claim of a demonstrated benefit were true or false. More importantly, Harkonen’s claim involved an inference, which is not a fact, but an opinion. Fleming’s contrary opinion really did not turn Harkonen’s claim into a falsehood. A contrary rule would have many expert witnesses in civil and in criminal litigation behind bars on similar charges of wire or mail fraud.

After Harkonen exhausted his direct appeals,² he petitioned for a writ of coram nobis. The trial court denied the petition,³ and in a non-precedential opinion [sic], the Ninth Circuit affirmed the denial of coram nobis.⁴ United States v. Harkonen, slip op., No. 15-16844 (9th Cir., Dec. 4, 2017) [cited below as Harkonen].

The Circuit rejected Harkonen’s contention that the Supreme Court had announced a new rule with respect to statistical significance, in Matrixx Initiatives, Inc. v. Siracusano, 563 U.S. 27 (2011), which change in law required that his conviction be vacated. Harkonen’s lawyer, like much of the plaintiffs’ tort bar, oversold the Supreme Court’s comments about statistical significance, which were at best dicta, and not very well considered or supported dicta, at that. Still, there was an obvious tension, and duplicity, between positions that the government, through the Solicitor General’s office, had taken in Siracusano, and positions the government took in the Harkonen case.⁵ Given the government’s opportunistic double-faced arguments about statistical significance, the Ninth Circuit held that Harkonen’s proffered evidence was “compelling, especially in light of Matrixx,” but the panel concluded that his conviction was not the result of a “manifest injustice” that requires the issuance of the writ of coram nobis. Harkonen at 2 (emphasis added). Apparently, Harkonen had suffered an injustice of a less obvious and blatant variety, which did not rise to the level of manifest injustice.

The Ninth Circuit gave similarly short shrift to Harkonen’s challenge to the competency of his counsel. His trial lawyers had averred that they thought that they were doing well enough not to risk putting on an expert witness, especially given that the defense’s view of the evidence came out in the testimony of the government’s witnesses. The Circuit thus acquiesced in the view that both sides had chosen to forgo expert witness testimony, and overlooked the defense’s competency issue for not having objected to Fleming’s opinion trial testimony. Harkonen at 2-4. Remarkably, the appellate court did not look at how Fleming was allowed to testify on statistical issues, without being challenged on cross-examination.

1 See “The (Clinical) Trial by Franz Kafka”; “Further Musings on U.S. v. Harkonen”;

“Harkonen’s Appeal Updated.” See also Patti Zettler, “U.S. v. Harkonen: Should Scientists Worry About Being Prosecuted for How They Interpret Their Research Results?” Stanford Center for Law & the Biosciences (Oct. 7, 2013); William M. Briggs, “Use The Wrong P-value, Go To Jail: Not A Joke: Updated With Amicus Brief,” Statistician to the Stars (Oct. 1, 2013).

2 United States v. Harkonen, 510 F. App’x 633, 638 (9th Cir. 2013), cert. denied, 134 S. Ct. 824 (2013).

3 “District Court Denies Writ of Coram Nobis to Dr Harkonen.”

4 Dave Simpson, “9th Circuit Refuses To Rethink Ex-InterMune CEO’s Conviction,” Law360 (Dec. 5, 2017).

5 “Multiplicity versus Duplicity – The Harkonen Conviction”; “The Matrixx Motion in U.S. v. Harkonen.”

Posted in Causation, Scientific Evidence, statistical evidence | Comments Off on Ninth Circuit Quashes Harkonen’s Last Chance

Failed Gatekeeping in Ambrosini v. Labarraque (1996)

December 28th, 2017

The Ambrosini case straddled the Supreme Court’s 1993 Daubert decision. The case began before the Supreme Court clarified the federal standard for expert witness gatekeeping, and ended in the Court of Appeals for the District of Columbia, after the high court adopted the curious notion that scientific claims should be based upon reliable evidence and valid inferences. That notion has only slowly and inconsistently trickled down to the lower courts.

Given that Ambrosini was litigated in the District of Columbia, where the docket is dominated by regulatory controversies, frequently involving dubious scientific claims, no one should be surprised that the D.C. Court of Appeals did not see that the Supreme Court had read “an exacting standard” into Federal Rule of Evidence 702. And so, we see, in Ambrosini, this Court of Appeals citing and purportedly applying its own pre-Daubert decision in Ferebee v. Chevron Chem. Co., 552 F. Supp. 1297 (D.D.C. 1982), aff’d, 736 F.2d 1529 (D.C. Cir.), cert. denied, 469 U.S. 1062 (1984).¹ In 2000, the Federal Rule of Evidence 702 was revised in a way that extinguishes the precedential value of Ambrosini and the broad dicta of Ferebee, but some courts and commentators have failed to stay abreast of the law.

Escolastica Ambrosini was using a synthetic progestin birth control, Depo-Provera, as well as an anti-nausea medication, Bendectin, when she became pregnant. The child that resulted from this pregnancy, Teresa Ambrosini, was born with malformations of her face, eyes, and ears, cleft lip and palate, and vetebral malformations. About three percent of all live births in the United States have a major malformation. Perhaps because the Divine Being has sovereign immunity, Escolastica sued the manufacturers of Bendectin and Depo-Provera, as well as the prescribing physician.

The causal claims were controversial when made, and they still are. The progestin at issue, medroxyprogesterone acetate (MPA), was embryotoxic in the cynomolgus monkey², but not in the baboon³. The evidence in humans was equivocal at best, and involved mostly genital malformations⁴; the epidemiologic evidence for the MPA causal claim to this day remains unconvincing⁵.

At the close of discovery in Ambrosini, Upjohn (the manufacturer of the progestin) moved for summary judgment, with a supporting affidavit of a physician and geneticist, Dr. Joe Leigh Simpson. In his affidavit, Simpson discussed three epidemiologic studies, as well as other published papers, in support of his opinion that the progestin at issue did not cause the types of birth defects manifested by Teresa Ambrosini.

Ambrosini had disclosed two expert witnesses, Dr. Allen S. Goldman and Dr. Brian Strom. Neither Goldman nor Strom bothered to identify the papers, studies, data, or methodology used in arriving at an opinion on causation. Not surprisingly, the district judge was unimpressed with their opposition, and granted summary judgment for the defendant. Ambrosini v. Labarraque, 966 F.2d 1462, 1466 (D.C. Cir. 1992).

The plaintiffs appealed on the remarkable ground that Goldman’s and Strom’s crypto-evidence satisfied Federal Rule of Evidence 703. Even more remarkably, the Circuit, in a strikingly unscholarly opinion by Judge Mikva, opined that disclosure of relied-upon studies was not required for expert witnesses under Rules 703 and 705. Judge Mikva seemed to forget that the opinions being challenged were not given in testimony, but in (late-filed) affidavits that had to satisfy the requirement of Federal Rule of Civil Procedure 26. Id. at 1468-69. At trial, an expert witness may express an opinion without identifying its bases, but of course the adverse party may compel disclosure of those bases. In discovery, the proffered expert witness must supply all opinions and evidence relied upon in reach the opinions. In any event, the Circuit remanded the case for a hearing and further proceedings, at which the two challenged expert witnesses, Goldman and Strom, would have to identify the bases of their opinions. Id. at 1471.

Not long after the case landed back in the district court, the Supreme Court decided Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993). With an order to produce entered, plaintiffs’ counsel could no longer hide Goldman and Strom’s evidentiary bases, and their scientific inferences came under judicial scrutiny.

Upjohn moved again to exclude Goldman and Strom’s opinions. The district court upheld Upjohn’s challenges, and granted summary judgment in favor of Upjohn for the second time. The Ambrosinis appealed again, but the second case in the D.C. Circuit resulted in a split decision, with the majority holding that the exclusion of Goldman and Strom’s opinions under Rule 702 was erroneous. Ambrosini v. Labarraque, 101 F.3d 129 (D.C. Cir. 1996).

Although issued two decades ago, the majority’s opinion remains noteworthy as an example of judicial resistance to the existence and meaning of the Supreme Court’s Daubert opinion. The majority opinion uncritically cited the notorious Ferebee⁶ and other pre-Daubert decisions. The court embraced the Daubert dictum about gatekeeping being limited to methodologic consideration, and then proceeded to interpret methodology as superficially as necessary to sustain admissibility. If an expert witness claimed to have looked at epidemiologic studies, and epidemiology was an accepted methodology, then the opinion of the expert witness must satisfy the legal requirements of Daubert, or so it would seem from the opinion of the U.S. Court of Appeals for the District of Columbia.

Despite the majority’s hand waving, a careful reader will discern that there must have been substantial gaps and omissions in the explanations and evidence cited by plaintiffs’ expert witnesses. Seeing anything clearly in the Circuit’s opinion is made difficult, however, by careless and imprecise language, such as its descriptions of studies as showing, or not showing “causation,” when it could have meant only that such studies showed associations, with more or less random and systematic error.

Dr. Strom’s report addressed only general causation, and even so, he apparently did not address general causation of the specific malformations manifested by the plaintiffs’ child. Strom claimed to have relied upon the “totality of the data,” but his methodologic approach seems to have required him to dismiss studies that failed to show an association.

“Dr. Strom first set forth the reasoning he employed that led him to disagree with those studies finding no causal relationship [sic] between progestins and birth defects like Teresa’s. He explained that an epidemiologist evaluates studies based on their ‘statistical power’. Statistical power, he continued, represents the ability of a study, based on its sample size, to detect a causal relationship. Conventionally, in order to be considered meaningful, negative studies, that is, those which allege the absence of a causal relationship, must have at least an 80 to 90 percent chance of detecting a causal link if such a link exists; otherwise, the studies cannot be considered conclusive. Based on sample sizes too small to be reliable, the negative studies at issue, Dr. Strom explained, lacked sufficient statistical power to be considered conclusive.”

Id. at 136⁷.

Putting aside the problem of suggesting that an observational study detects a “causal relationship,” as opposed to an association in need of further causal evaluation, the Court’s précis of Strom’s testimony on power is troublesome, and typical of how other courts have misunderstood and misapplied the concept of statistical power. Statistical power is a probability of observing an association of a specified size at a specified level of statistical significance. The calculation of statistical power turns indeed on sample size, the level of significance probability preselected for “statistical significance, an assumed probability distribution of the sample, and, critically, an alternative hypothesis. Without a specified alternative hypothesis, the notion of statistical power is meaningless, regardless of what probability (80% or 90% or some other percentage) is sought for finding the alternative hypothesis. Furthermore, the notion that the defense must adduce studies with “sufficient statistical power to be considered conclusive” creates an unscientific standard that can never be met, while subverting the law’s requirement that the claimant establish causation.

The suggestion that the studies that failed to find an association cannot be considered conclusive because they “lacked sufficient statistical power” is troublesome because it distorts and misapplies the very notion of statistical power. No attempt was made to describe the confidence intervals surrounding the point estimates of the null studies; nor was there any discussion whether the studies could be aggregated to increase their power to rule out meaningful associations.

The Circuit court’s scientific jurisprudence was thus seriously flawed. Without a discussion of the end points observed, the relevant point estimates of risk ratios, and the confidence intervals, the reader cannot assess the strength of the claims made by Goldman and Strom, or by defense expert Simpson, in their reports. Without identifying the study endpoints, the reader cannot evaluate whether the plaintiffs’ expert witnesses relied upon relevant outcomes in formulating their opinions. The court viewed the subject matter from 30,000 feet, passing over at 600 mph, without engagement or care. A strong dissent, however, suggested serious mischaracterizations of the plaintiffs’ evidence by the majority.

The only specific causation testimony to support plaintiff’s claims came from Goldman, in what appears to have been a “differential etiology.” Goldman purported to rule out a genetic cause, even though he had not conducted a critical family history or ordered a state-of-the-art chromosomal study. Id. at 140. Of course, nothing in a differential etiology approach would allow a physician to rule out “unknown” causes, which, for birth defects, make up the most prevalent and likely causes to explain any particular case. The majority acknowledged that these were short comings, but rhetorically characterized them as substantive, not methodologic, and therefore as issues for cross-examination, not for consideration by a judicial gatekeeping. All this is magical thinking, but it continues to infect judicial approaches to specific causation. See, e.g., Green Mountain Chrysler Plymouth Dodge Jeep v. Crombie, 508 F. Supp. 2d 295, 311 (D.Vt. 2007) (citing Ambrosini for the proposition that “the possibility of uneliminated causes goes to weight rather than admissibility, provided that the expert has considered and reasonably ruled out the most obvious”). In Ambrosini, however, Dr. Goldman had not ruled out much of anything.

Circuit Judge Karen LeCraft Henderson dissented in a short, but pointed opinion that carefully marshaled the record evidence. Drs. Goldman and Strom had relied upon a study by Greenberg and Matsunaga, whose data failed to show a statistically significant association between MPA and cleft lip and palate, when the crucial issue of timing of exposure was taken into consideration. Ambrosini, 101 F.3d at 142.

Beyond the specific claims and evidence, Judge Henderson anticipated the subsequent Supreme Court decisions in Joiner, Kumho Tire, and Weisgram, and the year 2000 revision of Rule 702, in noting that the majority’s acceptance of glib claims to have used a “traditional methodology” would render Daubert nugatory. Id. at 143-45 (characterizing Strom and Goldman’s methodologies as “wispish”). Even more importantly, Judge Henderson refused to indulge the assumption that somehow the length of Goldman’s C.V. substituted for evidence that his methods satisfied the legal (or scientific) standard of reliability. Id.

The good news is that little or nothing in Ambrosini survives the 2000 amendment to Rule 702. The bad news is that not all federal judges seem to have noticed, and that some commentators continue to cite the case, as lovely.

Probably no commentator has promiscuously embraced Ambrosini as warmly as Carl Cranor, a philosopher, and occasional expert witness for the lawsuit industry, in several publications and presentations.⁸ Cranor has been particularly enthusiastic about Ambrosini’s approval of expert witness’s testimony that failed to address “the relative risk between exposed and unexposed populations of cleft lip and palate, or any other of the birth defects from which [the child] suffers,” as well as differential etiologies that exclude nothing.⁹ Somehow Cranor, as did the majority in Ambrosini, believes that testimony that fails to identify the magnitude of the point estimate of relative risk can “assist the trier of fact to understand the evidence or to determine a fact in issue.”¹⁰ Of course, without that magnitude given, the trier of fact could not evaluate the strength of the alleged association; nor could the trier assess the probability of individual causation to the plaintiff. Cranor also has written approvingly of lumping unrelated end points, which defeats the assessment of biological plausibility and coherence by the trier of fact. When the defense expert witness in Ambrosini adverted to the point estimates for relevant end points, the majority, with Cranor’s approval, rejected the null findings as “too small to be significant.”¹¹ If the null studies were, in fact, too small to be useful tests of the plaintiffs’ claims, intellectual and scientific honesty required an acknowledgement that the evidentiary display was not one from which a reasonable scientist would draw a causal conclusion.

1Ambrosini v. Labarraque, 101 F.3d 129, 138-39 (D.C. Cir. 1996) (citing and applying Ferebee), cert. dismissed sub nom. Upjohn Co. v. Ambrosini, 117 S.Ct. 1572 (1997) See also David E. Bernstein, “The Misbegotten Judicial Resistance to the Daubert Revolution,” 89Notre Dame L. Rev. 27, 31 (2013).

2 S. Prahalada, E. Carroad, M. Cukierski, and A.G. Hendrickx, “Embryotoxicity of a single dose of medroxyprogesterone acetate (MPA) and maternal serum MPA concentrations in cynomolgus monkey (Macaca fascicularis),” 32 Teratology 421 (1985).

3 S. Prahalada, E. Carroad, and A.G. Hendrick, “Embryotoxicity and maternal serum concentrations of medroxyprogesterone acetate (MPA) in baboons (Papio cynocephalus),” 32 Contraception 497 (1985).

4 See, e.g., Z. Katz, M. Lancet, J. Skornik, J. Chemke, B.M. Mogilner, and M. Klinberg, “Teratogenicity of progestogens given during the first trimester of pregnancy,” 65 Obstet Gynecol. 775 (1985); J.L. Yovich, S.R. Turner, and R. Draper, “Medroxyprogesterone acetate therapy in early pregnancy has no apparent fetal effects,” 38 Teratology 135 (1988).

5 G. Saccone, C. Schoen, J.M. Franasiak, R.T. Scott, and V. Berghella, “Supplementation with progestogens in the first trimester of pregnancy to prevent miscarriage in women with unexplained recurrent miscarriage: a systematic review and meta-analysis of randomized, controlled trials,” 107 Fertil. Steril. 430 (2017).

6 Ferebee v. Chevron Chemical Co., 736 F.2d 1529, 1535 (D.C. Cir.), cert. denied, 469 U.S. 1062 (1984).

7 Dr. Strom was also quoted as having provided a misleading definition of statistical significance: “whether there is a statistically significant finding at greater than 95 percent chance that it’s not due to random error.” Ambrosini at 101 F.3d at 136. Given the majority’s inadequate description of the record, the description of witness testimony may not be accurate, and error cannot properly be allocated.

8 Carl F. Cranor, Toxic Torts: Science, Law, and the Possibility of Justice 320, 327-28 (2006); see also Carl F. Cranor, Toxic Torts: Science, Law, and the Possibility of Justice 238 (2d ed. 2016).

9 Carl F. Cranor, Toxic Torts: Science, Law, and the Possibility of Justice 320 (2006).

10 Id.

11 Id. ; see also Carl F. Cranor, Toxic Torts: Science, Law, and the Possibility of Justice 238 (2d ed. 2016).

Posted in Causation, Expert Witness Discovery, Rule 702, Rule 703, Scientific Evidence, statistical evidence, Statistical Power | Comments Off on Failed Gatekeeping in Ambrosini v. Labarraque (1996)

Ferebee Revisited

December 28th, 2017

The following post was originally published on November 8, 2012, but was hacked, no doubt by the lawsuit industry, and replaced with mindless fluff as is its wont. It is now restored.

Ferebee Revisited

I used to think of the infamous Ferebee decision as the Dred Scott decision of scientific evidence; in essence, declaring that science has no validity issues that the law is bound to respect. Ferebee v. Chevron Chem. Co., 552 F. Supp. 1297 (D.D.C. 1982), aff’d, 736 F.2d 1529 (D.C. Cir.), cert. denied, 469 U.S. 1062 (1984). The rhetoric on expert witnesses, from the district and circuit courts in this case is sometimes jarring, but the facts of the case make the holding, rather than the expansive dicta, not so unreasonable, under all the facts and circumstances of the case.

On rereading Ferebee, I was struck by several aspects of the case that rarely are discussed when Ferebee is cited. On sober second thought, Ferebee may not be such a bad decision, especially considering that it has no continuing validity as a rule of decision for expert witness admissibility in federal court.

1. Ferebee is a government negligence case.

The plaintiff worked for the federal government when he was exposed to the herbicide paraquat. Richard Ferebee began working for the Department of Agriculture’s Beltsville Agricultural Research Center (BARC), in Beltsville, Maryland. He started spraying paraquat in the summer of 1977, and used the herbicide regularly through the time he was diagnosed with pulmonary fibrosis, in November 1979. 736 F.2d at 1531-32. Ferebee brought a failure to warn claim against the supplier of paraquat, Chevron Chemical Company. The allegations of actual or constructive knowledge of a hazard, however, could just as readily be asserted against the federal government, which owned the BARC facility, employed Ferebee, controlled and supervised his use of paraquat, and failed to comply with Chevron’s instructions. The federal government further regulated the sale and use of paraquat extensively, first by the Department of Agriculture, and later by the Environmental Protection Agency. Id. at 1532.

2. The exposure.

Ferebee filed suit in 1981, he died in 1982. His case was tried twice. In the first trial, the jury deadlocked; in the second trial, the jury returned a verdict in favor of his estate, and for his family, for $60,000. In his deposition testimony, Ferebee described how sprayed paraquat, in the summer of 1977. The chemical was diluted for use, per Chevron’s instructions. There was no evidence that Ferebee ever had direct contact with undiluted paraquat, or that the paraquat he was exposed to was not diluted according to the proportions recommended on Chevron’s label. 552 F. Supp. at 1295 & n. 3. Ferebee frequently got the chemical on his hands. 552 F. Supp. at 1294-95. Ferebee further described an occasion when he was drenched with paraquat when he walked behind a tractor that was spraying the chemical, and another incident when he used a defective sprayer that leaked paraquat “all over his pants.” 736 F.2d at 1532. On both occasions, Ferebee did not wash, and apparently went home contaminated, where he fell asleep, tired and dizzy, without showering. Id. As we will see, the exposure that Ferebee described would not have occurred had his federal employer followed the instructions on the label that it mandated. In 1978, the federal Occupational Health & Safety Administration published Guidelines on the need for protective clothing, respirators, immediate washing of contaminated skin, etc. Ferebee’s federal employer recklessly disregarded its own guidelines.

3. The warnings.

Paraquat could be sold in the United States only when labeled in accordance with EPA regulations, promulgated pursuant to the Federal Insecticide, Fungicide, and Rodenticide Act, 7 U.S.C. § 136, et seq. (FIFRA) The statute bars EPA from allowing sale of regulated herbicides, such as paraquat, unless the chemicals, as labeled, will not cause “unreasonable adverse effects on the environment.” 7 U.S.C. § 136a(c)(5)(C). Such effects are in turn defined as any unreasonable risk to man or the environment, taking into account the economic, social, and environmental costs and benefits of the use of [the] pesticide. 7 U.S.C. § 136(bb). FIFRA further requires the EPA to require labeling that is “adequate to protect health and the environment” and that is “likely to be read and understood.” 7 U.S.C. § 136(q)(1)(E). 736 F.2d at 1539-40.

Unfortunately, the courts failed to provide the complete warning label and the material data safety sheets. There are “snippets” provided, which make clear that the federal government was largely to blame for failing to comply with the directions required under FIFRA. For instance, the district court, in a footnote, acknowledged:

“For example, the label advised the user spraying paraquat to wear waterproof clothing and goggles, to avoid working in spray mist, and to wash splashes on the skin or eyes immediately with water.”

552. F. Supp. at 1304 n.40. The Court of Appeals reported that “the label, in large bold letters states:

DANGER

CAN KILL IF SWALLOWED

HARMFUL TO THE EYES AND SKIN

736 F.2d at 1536. The label also informed users to wash any exposed areas immediately, and to remove contaminated clothing. Id.

4. The Stipulation.

A key fact, rarely described or explained in discussions of the Ferebee case, is the parties’ stipulation

“that Mr. Ferebee’s only significant exposure to paraquat was on his intact skin; i.e., there was no evidence that Mr. Ferebee swallowed or inhaled paraquat, or that he spilled or sprayed it on an area of his skin upon which he had any apparent cuts or scrapes. The jury was not, of course, precluded from concluding that a person engaged in Mr. Ferebee’s line of work could have had some, or even many, minor cuts or abrasions not readily discernible to the naked eye or likely to be remembered some time later.”

552. F. Supp. at 1295 & n. 3.

Why did the plaintiffs try to present their case solely as a dermal exposure cases? As we will see, this stratagem made their medical causation case more difficult, but it avoided serious misuse and lack of proximate cause issues. Ferebee had been instructed by his co-workers and supervisors that paraquat was extremely dangerous if swallowed, and probably also if inhaled. The warning label was unequivocal in detailing the dangers and the need to avoid ingestion. (Without the full label, it is difficult to evaluate how well the label warned against inhalation, but the 1978 OSHA guidelines address the use of a proper respirator for situations in which paraquat may be inhaled.) On the other hand, the label had a weakness, which could be exploited, as long as the preemption defense could be held at bay: the label urged protective clothing, goggles, and immediate washing of contaminated skin, but it failed to describe the consequence of dermal exposure other than irritation. Ferebee could thus avoid his culpable conduct, as well as a sophisticated intermediary defense, by claiming that his exposure was only dermal.

Why did Chevron agree to the stipulation? The defendant probably felt sanguine about its preemption defense, and thus also about the adequacy of its warnings overall. The stipulation limited the plaintiff’s medical causation case to a route of exposure that put it into an arguable “first instance” case report. Chevron stood to gain a claim of “lack of notice,” and thus lack of actual or constructive knowledge of the risk of lung disease from dilute dermal exposure. The clinical presentation itself differed from many of the cases of known paraquat poisoning, see infra, and Chevron probably believed that it could deal with the medical causation claim better if exposure was limited to transdermal absorption. Curiously, Chevron did not argue that Ferebee must have had some inhalational exposure, which he almost certainly did. I suspect that Chevron’s position on inhalation was hedged because its warning label did not specify respirator usage for ordinary work exposures of applicators (as opposed to workers who handled undiluted paraquat, worked in confined spaces, etc.).

5. Medical causation

Chevron took a strident position, standing on the fact that there had been no previous documented cases of pulmonary fibrosis in workers exposed to diluted paraquat through their skin. The following facts were uncontroverted:

Paraquat causes pulmonary fibrosis in humans.
The evidence that established paraquat as a cause of pulmonary fibrosis was largely case series of acute onset of pulmonary fibrosis after ingestion.
Paraquat induces pulmonary fibrosis relatively rapidly.
Paraquat can be absorbed through the skin.
The parties agreed that any type of exposure – ingestion, inhalation, or dermal absorption – could cause lung damage. 552. F. Supp. at 1300 & n.28.
Once paraquat is ingested, inhaled, or absorbed, it can travel to the lungs.
Lung fibrosis caused by dermal absorption of paraquat had been described previously only with skin lesions before or after the injury. 736 F.2d at 1538.
The lungs are the target organ for paraquat.
There are numerous causes of pulmonary fibrosis (such as asbestosis, scleroderma, rheumatoid arthritis, etc.).
The variants of pulmonary fibrosis do not all look alike, present alike, or progress alike.
Mr. Ferebee had no known other disease or exposure that could account for his pulmonary fibrosis.
There is are cases of pulmonary fibrosis with no identifiable cause, known as idiopathic pulmonary fibrosis (IPF).
IPF is relatively rare; it too has a rapid onset and progression, although not as fast as the cases described after exposure to undiluted paraquat.
Mr. Ferebee’s medical history was largely unhelpful in explaining his clinical course.
Ferebee had some shortness of breath before starting to use paraquat. 552. F. Supp. at 1295.
Ferebee used paraquat occasionally over three years before he was diagnosed with pulmonary fibrosis.

Some observations about these facts. General causation in a sense was not contested. Paraquat causes pulmonary fibrosis. The issue was whether dilute dermal exposure over three years causes pulmonary fibrosis. Chevron stridently asserted that the “scientific method” required controlled experimental or observational (epidemiologic) studies. The problem with Chevron’s position was that general causation had already been established, and not by analytical epidemiologic studies.

6. The expert witnesses.

Ferebee was initially treated by Dr. Muhammed Yusuf, a pulmonary specialist, who diagnosed pulmonary fibrosis. Dr. Yusef referred Ferebee to the National Institutes of Health (NIH), where he came under the care of Dr. Ronald G. Crystal of the Heart, Lung, and Blood Institute. (Dr. Crystal is now at Cornell-Weill, where he is Chairman of Genetic Medicine, and he practices pulmonary medicine.)

Chevron called Dr. Carrington, who diagnosed Ferebee with IPF. Dr. Carrington challenged the plaintiffs’ expert witnesses’ opinions for lacking reliance upon controlled observational or experimental studies. 552. F. Supp. at 1301. Dr. Carrington, however, acknowledged that dermal cases are too rare for observational epidemiologic analysis, but emphasized that no animal studies of sufficient size had been done to support plaintiffs’ hypothesis. Chevron also called a Dr. Fisher, who presented a toxicokinetic (TK) analysis of Ferebee’s dermal absorption. Based upon his TK analysis, Dr. Fisher concluded that the maximal amount of paraquat absorbed by Ferebee was too small, based upon known cases and animal studies, to have caused paraquat toxicity. Id.

7. Chevron’s challenge to plaintiffs’ expert witnesses’ causation opinion.

None of the defendant’s expert witnesses examined Ferebee. The courts thought this was relevant, but they never articulated what would have been observed on physical examination that was important to resolving the differential diagnosis of paraquat toxicity versus IPF. There was no dispute that Ferebee had rapidly progressing pulmonary fibrosis. The expert witnesses on both sides evaluated Ferebee’s clinical data, presentation, clinical course, and arrived at different diagnoses. The plaintiffs’ expert witnesses’ diagnosis, however, involved a causal attribution to paraquat exposure.

The Ferebee case was litigated under Maryland law because federal statutory law requires state law to control in a wrongful death action arising out of the neglect or wrongful act of another on a federal enclave. 16 U.S.C. § 457. 736 F.2d at 1533. (Maryland law is actually favorable to a sophisticated intermediary defense, although the key decisions post-date Ferebee.) Chevron appears to have relied upon Maryland’s articulation of the Frye general acceptance doctrine, and the courts analyzed Chevron’s arguments as a Frye challenge. 552 F. Supp. at 1301; 736 F.2d at 1535. Although the use of Maryland law to determine an evidentiary issue seems suspect, Chevron pressed apparently pressed its challenge in terms of Maryland’s version of Frye, and not based upon Federal Rule of Evidence 702. The infamous language used by both the district and the circuit courts was, therefore, not an interpretation of federal law. Rule 702 was never cited or discussed in either the trial or the appellate court’s opinion.

My re-reading of Ferebee has softened my criticisms of state courts that had relied upon the case, even after the Supreme Court’s decision in Daubert. Softened but not eliminated my criticism — Ferebee is still a case largely confined to its facts, and the language quoted as a standard of admissibility is really a statement of the appellate standard of review for the jury’s determination of medical causation.

8. The judicial resolution of Chevron’s Frye challenge

The district court insightfully recognized that Chevron was demanding a level of evidence, which had never been required to establish paraquat’s generally accepted ability to cause pulmonary fibrosis. This recognition led to the district court’s colorful language:

“It is true that medical expert testimony must be grounded in proper scientific methodology, but the extremely stringent standard that defendant suggests is beyond reason. Product liability law, especially as it relates to relatively new products or those with a relatively rare yet significant danger, would be rendered next to meaningless if a plaintiff could prove he was injured by a product only after a ‘statistically significant’ number of other people were also injured. A civilized legal system does not require that much human sacrifice before it can intervene. The fact that this is the first case of this exact type-or at least the first of its exact type in which the involvement of paraquat was discovered by alert doctors — cannot be enough by itself to shield defendant from liability. Defendant’s experts were not able to fault Dr. Crystal for his basic diagnostic methodology; in fact, they used the same kinds of test results, consultations, and other tools that he did. What they disagreed with chiefly were his conclusions.”

552 F. Supp. at 1301. The important observation is that general causation had been established case series and reports of human exposure. There never was statistical evidence that had been evaluated for “significance,” to establish general causation for undiluted paraquat, and the trial court refused, under Maryland law, to require such evidence for general causation for diluted paraquat. In this context, we can see that the trial court’s suggestion that statistical significance was not required has little bearing upon, cases in which general causation could only be established using epidemiologic evidence, with its attendant statistical inferences.

Of course, the matter only became worse when Chevron persisted in its argument and presented it to a liberal panel of the D.C. Circuit. (Judge Mikva wrote the opinion for a panel that included Judge Wald, and Senior Judge Bazelon.) The panel’s decision ratcheted up the rhetoric:

“Thus, a cause-effect relationship need not be clearly established by animal or epidemiological studies before a doctor can testify that, in his opinion, such a relationship exists. As long as the basic methodology employed to reach such a conclusion is sound, such as use of tissue samples, standard tests, and patient examination, product liability does not preclude recovery until a ‘statistically significant’ number of people have been injured or until science has had the time and resources to complete sophisticated laboratory studies of the chemical. In a courtroom, the test for allowing a plaintiff to recover is not scientific certainty, but legal sufficiency; if reasonable jurors could conclude from the expert testimony that paraquat more likely than not caused Ferebee’s injury, the fact that another jury might reach the opposite conclusion or that science would require more evidence before conclusively considering the causation question resolved is irrelevant. That Ferebee’s case may have been the first of its exact type, or that his doctors may have been the first alert enough to recognize such a case, does not mean that the testimony of those doctors, who are concededly well qualified in their fields, should not have been admitted.”

736 F.2d at 1535-36 (emphasis in original).

Again, the dismissive attitude towards statistically significant evidence is limited to the context of a causal analysis that had been made, to everyone’s satisfaction, for undiluted paraquat, without the need for epidemiologic, statistical evidence. Statistical significance was never at issue. In this way, Ferebee resembles the untoward language on statistical significance from Matrixx Initiatives Inc. v. Siracusano. In both cases, statistical significance was never really at issue. In Ferebee, there was no statistical evidence needed or used to reach causal conclusions about paraquat’s ability to induce pulmonary fibrosis. In Matrixx Initiatives, allegations of statistical significance and causation were not necessary because the plaintiffs needed only to allege materiality of the facts suppressed by the company in order to plead a securities fraud case. Materiality could be established without causation, and thus neither causation nor statistical significance needed to be alleged.

As for Chevron’s Frye challenge, the district court rejected the implied call for a vote on the general acceptance of Dr. Crystal’s reasoning. Frye may require “vote counting” of some sort, but the process becomes irrelevant when virtually no one has registered to vote. Otherwise, the defense and the plaintiffs’ expert witnesses appeared to be using the same technique of arguing by analogy to accepted cases of paraquat poisoning or IPF. Dr. Crystal opined that Ferebee’s case was “similar” to three other cases he had identified. Dr. Carrington argued that Ferebee’s case was more like IPF cases, although IPF cases themselves have some clinical heterogeneity as well. Paraquat cases described onset to death as a very rapid process. Ferebee did not present with significant symptoms for three years after his first exposure, and then he survived for another two plus years. Ferebee did not report skin lesions, which had been reported in previous cases of dermal exposure leading up to pulmonary fibrosis. The case presented, on the diagnostic level, a difficult call, but it is easy to see the courts’ impatience with the defendant’s insistence upon more stringent criteria and evidence than was used to establish the causal connection with undiluted paraquat.

9. Expert witness qualifications.

Chevron never challenged Dr. Yusuf’s or Dr. Crystal’s qualifications. The oft-quoted comments about expert witness qualifications were made in the context of describing the appellate court’s standard of review, and the court’s role in not assessing credibility or weighing the evidence:

“These admonitions apply with special force in the context of the present action, in which an admittedly dangerous chemical is alleged through long-term exposure to have caused disease. Judges, both trial and appellate, have no special competence to resolve the complex and refractory causal issues raised by the attempt to link low-level exposure to toxic chemicals with human disease. On questions such as these, which stand at the frontier of current medical and epidemiological inquiry, if experts are willing to testify that such a link exists, it is for the jury to decide whether to credit such testimony.”

736 F.2d at 1534.

This procedural posture is obviously very different from the initial determination of admissibility. As far as credentials are concerned, Drs. Yusuf and Crystal were hardly “hired guns”; both physicians were well qualified. Dr. Crystal had outstanding qualifications, and Chevron wisely never challenged them. Remarkably, this language has been mistakenly invoked as a standard for trial courts to use in determining the admissibility of expert witness opinion testimony. It is no such thing.

10. Preemption and Warnings Causation.

Ultimately, Chevron’s preemption defense was rejected by both the district and the circuit court. FIFRA preemption has had its ups and downs; no surprise there. More interesting is the emphasis that both courts gave to the important role of the employer in the case. The evidence overwhelming showed that Ferebee had never read the warning label, and thus the element of proximate causation between allegedly inadequate warning and harm was in jeopardy of going unproved. The courts, however, emphasized the role that the employer, through its supervisors and responsible co-workers, play in the complex organizational situation of a modern workplace:

“Mr. Ferebee’s situation was quite different, however. He did not purchase paraquat for his personal use; rather, it was provided to him by his employer for use on the job. The evidence showed that his principal source of information about paraquat was the oral instructions of his supervisors and co-workers, not the written label. He learned from them how to mix the product and how to spray it. It was also from this source that he learned of the danger of getting the product in his mouth: one of his co-workers warned him that if he accidently swallowed paraquat, it would ‘get in his blood’ and poison him. This is a common pattern of instruction and use of occupational materials in the workplace. Learning by doing and learning by oral instruction are tried and true methods of educating manual workers in their jobs. Therefore, although it is crucial to plaintiff’s case that someone would have read the label, it was not necessary for Mr. Ferebee to have done so. And it is obvious that one or more employees at BARC did read the label, since information did reach Mr. Ferebee about the proportions for diluting the product and about the dangers about which the label did warn. It was appropriate for the jury to infer that a warning about the danger of fatal lung disease from dermal exposure would also have been communicated to Mr. Ferebee. See Restatement (Second) of Torts § 388 comment n (seller normally entitled to assume that adequate warning will be passed on by purchaser to ultimate user); cf. Chambers v. G.D. Searle & Co., 441 F.Supp. at 381 (in product liability case involving prescription drug, relevant warning is the one given to doctor, not patient).”

552 F. Supp. at 1303-04 (internal citations omitted). So here we have Ferebee, the subject of so much derision and aspersion from defense counsel, embracing the Section 388, comment n, as well as applying learned intermediary principles to a case not involving prescription drugs. The appellate court was waxed enthusiastic about the principles of Section 388, and went so far as to cite Victor Schwartz in support:

“We live in an organizational society in which traditional common-law limitations on an actor’s duty must give way to the realities of society. *** In this case, Mr. Ferebee did not purchase the paraquat for his personal use, and there was substantial evidence that workplace communication about the dangers associated with various chemicals usually took the form of oral instructions from supervisors to workers, the latter of whom then retransmitted the information to co-workers. This, rather than individual reading of product warnings, is a typical method by which information is disseminated in the modern workplace. See Schwartz & Driver, “Warnings in the Workplace: The Need for a Synthesis of Law and Communication Theory,” 52 U. Cinn. L. Rev. 38, 66-83 (1983). The requirement that an improper warning proximately ‘cause’ the injury should be elaborated against this background. We believe Maryland would construe its tort law in this case to require only that someone in the workplace have read the label, not that Mr. Ferebee personally have read it. Because there is no dispute that one or more employees at BARC did read the label, we hold that the jury could properly have inferred that, had a warning about the danger of disease from dermal exposure been included on the label, that warning would have been communicated to Mr. Ferebee and that he would as a result have acted differently. Alternatively, the jury could have inferred that an adequate warning would have led Ferebee’s employers to undertake steps that would have protected him from paraquat poisoning-for example, provision of showers for use after spraying.”

736 F.2d at 1539 (emphasis in original; internal citation omitted). Judge Mikva’s prediction, of course, was absolutely accurate; Maryland tort law did, soon thereafter, embrace the sophisticated intermediary defense to exculpate the defendant in such remote supplier situations. See, e.g., Kennedy v. Mobay Corp., 84 Md. App. 397 (1990) (applying sophisticated user defense to bar claims against manufacturers of toluene diisocyanate), aff’d, 325 Md. 385 (1992); Higgins v. E.I. DuPont de Nemours, Inc., 671 F. Supp. 1055 (D. Md. 1987) (Maryland law; holding that manufacturer of paint was in better position than bulk supplier to communicate warnings to customers’ employees), aff’d, 863 F.2d 1162 (4th Cir. 1988). The principle invoked to excuse plaintiff from reading the warning label also works to exculpate the defendant when that warning label is otherwise adequate, or when the intermediary knows of the hazard in any event.

Posted in Causation, Rule 702, Scientific Evidence, statistical evidence | Comments Off on Ferebee Revisited

Some High-Value Targets for Sander Greenland in 2018

December 27th, 2017

A couple of years ago, Sander Greenland and I had an interesting exchange on Deborah Mayo’s website. I tweaked Sander for his practice of calling out defense expert witnesses for statistical errors, while ignoring whoopers made by plaintiffs’ expert witnesses. See “Significance Levels Made a Whipping Boy on Climate-Change Evidence: Is p < 0.05 Too Strict?” Error Statistics (Jan. 6, 2015).¹ Sander acknowledged that he received a biased sample of expert reports through his service as a plaintiffs’ expert witness, but protested that defense counsel avoided him like the plague. In an effort to be helpful, I directed Sander to an example of bad statistical analysis that had been proffered by Dr Bennett Omalu, in a Dursban case, Pritchard v. Dow Agro Sciences, 705 F. Supp. 2d 471 (W.D. Pa. 2010), aff’d, 430 F. App’x 102, 104 (3d Cir. 2011).²

Sander was unimpressed with my example of Dr. Omalu; he found the example “a bit disappointing though because [Omalu] was merely a county medical examiner, and his junk analysis was duly struck. The expert I quoted in my citations was a full professor of biostatistics at a major public university, a Fellow of the American Statistical Association, a holder of large NIH grants, and his analysis (more subtle in its transgressions) was admitted” (emphasis added). Sander expressed an interest in finding “examples involving similarly well-credentialed, professionally accomplished plaintiff experts whose testimony was likewise admitted… .”

Although it was heartening to read Sander’s concurrence in the assessment of Omalu’s analysis as “junk,” Sander’s rejection of Dr. Omalu as merely a low-value target was disappointing, given that Omalu also has a master’s degree in public health, from the University of Pittsburgh, where he claims he studied with Professor Lew Kuller. Omalu has also gained some fame and notoriety for his claim to have identified the problem of chronic traumatic encephalopathy (CTE) among professional football players. After all, even Sander Greenland has not been the subject of a feature-length movie (Concussion), as has Omalu.

I lost track of our exchange in 2015, until recently I was reminded of it when reading an expert report by Professor Martin Wells. Unlike Omalu, Wells meets all the Greenland criteria for high-value targets. He is not only a full, chaired professor but also the statistics department chairman at an ivy-league school, Cornell University. Wells is a fellow of both the American Statistical Association and the Royal Statistical Society, but most important, Wells is a frequent plaintiffs’ expert witness, who is well known to Sander Greenland. Both Wells and Greenland served, side by side, as plaintiffs’ expert witnesses in the pain pump litigation.

So here is the passage in the Wells’ report that is worthy of Greenland’s attention:

“If a 95% confidence interval is specified, the range encompasses the results we would expect 95% of the time if samples for new studies were repeatedly drawn from the same population.”

In re Testosterone Replacement Therapy Prods. Liab. Litig., Declaration of Martin T. Wells, Ph.D., at 2-3 (N.D. Ill., Oct. 30, 2016). Unlike the Dursban litigation involving Bennett Omalu, where the “junk analysis” was excluded, in the litigation against AbbVie for its manufacture and selling of prescription testosterone supplementation, Wells’ opinions were not excluded or limited. In re Testosterone Replacement Therapy Prods. Liab. Litig., No. 14 C 1748, MDL No. 2545, 2017 WL 1833173 (N.D. Ill. May 8, 2017) (denying Rule 702 motions).

Now this statement by Wells surely offends the guidance provided by Greenland and colleagues.³ And it was exactly the sort of misrepresentation that led to a confabulation of the American Statistical Association, and that Association’s consensus statement on statistical significance.⁴

And here is another example, which occurs not in a distorting litigation forum, but on the pages of an occupational health journal, where the editor in chief, Anthony L. Kiorpes, ranted about the need for better statistical editing and writing in his own journal. See Anthony L Kiorpes, “Lies, damned lies, and statistics,” 33 Toxicol. & Indus. Health 885 (2017). Kiorpes decried he misuse of statistics:

“I am not implying that it is the intent of the scientists who publish in these pages to mislead readers by their use of statistics, but I submit that the misuse of statistics, whether intentional or otherwise, creates confusion and error.”

Id. at 885. Kiorpes then proceeded to hold himself up as Exhibit A to his screed:

“Remember that p values are estimates of the probability that the null hypothesis (no difference) is true.”

Id. Uggh; we seem to be back sliding after the American Statistical Association’s consensus statement.

“Almost all scientists have stated (or have been tempted to state) something like ‘the mean of Group A was greater than that of Group B, but the difference was not statistically significant’. With very few exceptions (which I will mention below), this statement is nonsense.”

* * * * *

“What the statistics are indicating when the p-value is greater than 0.05 is that there is ‘no difference’ between group A and group B.”

Id. at 886.

Let’s hope that this gets Sander Greenland away from his biased sampling of expert witnesses, off the backs of defense expert witnesses, and on to some of the real culprits out there, in the new year.

2 See also “Pritchard v. Dow Agro – Gatekeeping Exemplified” (Aug. 25, 2014); “Omalu and Science — A Bad Weld” (Oct. 22, 2016); Brian v. Association of Independent Oil Distributors, No. 2011-3413, Westmoreland Cty. Ct. Common Pleas, Order of July 18, 2016 (excluding Dr. Omalu’s testimony on welding and solvents and Parkinson’s disease).

3 See, e.g., Sander Greenland, Stephen J. Senn, Kenneth J. Rothman, John B. Carlin, Charles Poole, Steven N. Goodman, and Douglas G. Altman, “Statistical tests, P values, confidence intervals, and power: a guide to misinterpretations,” 31 Eur. J. Epidem. 337 (2016).

4 Ronald L. Wasserstein & Nicole A. Lazar, “American Statistical Association Statement on statistical significance and p values,” 70 Am. Statistician 129 (2016)

Posted in Cognitive Biases, Conflicts of Interest, Expert Witnesses, Rule 702, statistical evidence | Comments Off on Some High-Value Targets for Sander Greenland in 2018

Statistical Gobbledygook Goes to the Supreme Court

October 20th, 2017

Back in July, my summer slumber was rudely interrupted by an intemperate, ad hominem rant from statistican Sander Greenland. Greenland’s rant concerned my views of the the Supreme Court’s decision in Matrixx Initiatives v. Siracusano, 563 U.S. 27 (2011).

Greenland held forth, unfiltered, on Deborah Mayo’s web blog, where he wrote:

“Glad to have finally flushed out Schachtman, whose blog did not allow my critical dissenting comments back when this case first hit. Nice to see him insult the intellect of the Court too, using standard legal obfuscation of the fact that the Court is entitled to consider science, ordinary logic, and common sense outside of that legal framework to form and justify its ruling – that reasoning is what composes the bulk of the opinion I linked. Go read it and see what you think without the smokescreen offered by Schachtman.”

“A megateam of reproducibility-minded scientists look to lowering the p-value,” Error Statistics (July 25, 2017).

Oh my! It is true that my blog does not have comments enabled, but as I have written on several occasions, I would gladly welcome requests to post opposing views, even those of Sander Greenland. On Deborah Mayo’s blog, I had the opportunity to explain carefully why Greenland has been giving a naïve, mistaken characterization of the holding of Matrixx Initiatives, in his expert witness reports for plaintiffs’ counsel, as well as in his professional publications. Ultimately, Greenland ran out of epithets, lost his enthusiasm for the discussion, and slunk away into cyber-silence.

I was a bit jarred, however, by Greenland’s accusation that I had insulted the Court. Certainly, I did not use any of the pejorative adjectives that Greenland had hurled at me; rather, I simply have given legal analysis of the Court’s opinions and a description of the legal, scientific, and statistical errors therein.¹ And, to be sure, other knowledgeable writers and evidence scholars, have critiqued the Court’s decision and some of the pronouncements of the parties and the amici in Matrixx Initiatives².

This week, John Pfaff, a professor at Fordham Law School, published an editorial in the New York Times, to argue that “The Supreme Court Justices Need Fact-Checkers,” N.Y. Times (Oct. 18, 2017). No doubt, Greenland would consider Pfaff’s editorial to be “insulting” to the Court, unless of course, Greenland thinks criticism can be insulting only if it challenges views he wants to see articulated by the Court.

In support of his criticism of the Court, Pfaff adverted to the Chief Justice’s recent comments in the oral argument of a gerrymandering case, Gill v. Whitford. In a question critical of the gerrymander challenge, Chief Justice Roberts described the supporting evidence:

“it may be simply my educational background, but I can only describe as sociological gobbledygook.”

Oral Argument before the U.S. Supreme Court at p.40, in Gill v. Whitford, No. 16-1161 (Oct. 3, 2017). The Chief Justice’s dismissive comments about gobble may well have been provoked by an amicus brief filed on behalf of 44 election law, scientific evidencce, and empirical legal scholars, who explored the legal and statistical basis for striking down the Wisconsin gerrymander. See Brief of Amici Curiae, of 44 Election Law, Scientific Evidence, and Empirical Legal Scholars, filed in Gill v. Whitford, No. 16-1161 (Sept. 1, 2017).

As with Greenland’s obsequious respect for the Matrixx Initiatives opinion, no one is likely to have been misled by Chief Justice Roberts’ false modesty. John Roberts was graduated summa cum laude from Harvard College in three years, although with a major in a “soft” discipline, history. He went on to Harvard Law School, where he was the managing editor of the Harvard Law Review, and was graduated magna cum laude. As a lawyer, Roberts has had an extraordinarily successful career. And yet, the Chief Justice went out of his way to disparage the mathematical and statistical models used to show gerrymandering in the Gill case, as “gobbledygook.” Odds are that the Chief Justice was thus not deprecating his own education; yet, inquiring minds might wonder whether that education was deficient in mathematics, statistics, and science.

Policy is a major part of the court’s docket now, whether the Justices likes it or not. The Justices cannot avoid adapting to the technical requirements of scientific and statistical issues, and they cannot simply dismiss evidence they do not understand as “gobbledygook.” Referencing a recent ProPublica report, Professor Pfaff suggests that the Supreme Court might well employ independent advisors to fact check their use of descriptive statistics³.

The problem identified by Pfaff, however, seems to implicate a fundamental divide between the “two cultures” of science and the humanities. See C.P. Snow, The Rede Lecture 1959. Perhaps Professor Pfaff might start with his own educational institution. The Fordham University School of Law does not offer a course in statistics and probability; nor does it require entering students to have satisfied a requirement of course work in mathematics, science, or statistics. The closest offering at Fordham is a course on accounting for lawyer, and the opportunity to take a one-credit course in “quantitative methods” at the graduate school.

Fordham School of Law, of course, is hardly alone. Despite cries for “relevancy” and experiential learning in legal education, some law schools eschew courses in statistics and probability for legal applications, sometimes on the explicit acknowledgement that such courses are too “hard,” or provoke too much student anxiety. The result, as C.P. Snow saw over a half century ago, is that lawyers and judges cannot tell gobbledygook from important data analysis, even when it smacks them in the face.

1 With David Venderbush of Alston & Bird LLP, I published my initial views of the Matrixx case, in the the form of a Washington Legal Foundation Legal Backgrounder, available at the Foundation’s website. See Schachtman & Venderbush, “Matrixx Unbounded: High Court’s Ruling Needlessly Complicates Scientific Evidence Principles,” 26 (14) Legal Backgrounder (June 17, 2011). I expanded on my critique in several blog posts. See, e.g., “Matrixx Unloaded” (Mar. 29, 2011); “The Matrixx Oversold” (Apr. 4, 2011); “The Matrixx – A Comedy of Errors” (Apr. 6, 2011); “De-Zincing the Matrixx” (Apr. 12, 2011); “Siracusano Dicta Infects Daubert Decisions” (Sept. 22, 2012).

2 See David Kaye, “The Transposition Fallacy in Matrixx Initiatives, Inc. v. Siracusano: Part I” (Aug. 19, 2011), and “The Transposition Fallacy in Matrixx Initiatives, Inc. v. Siracusano: Part II” (Aug. 26, 2011); David Kaye, “Trapped in the Matrixx: The U.S. Supreme Court and the Need for Statistical Significance,” BNA Product Safety & Liability Reporter 1007 (Sept. 12, 2011).

3 Ryan Gabrielson, “It’s a Fact: Supreme Court Errors Aren’t Hard to Find; A ProPublica review adds fuel to a longstanding worry about the nation’s highest court: The justices can botch the truth, sometimes in cases of great import,” ProPublica (Oct. 17, 2017).

Posted in Scientific Evidence, statistical evidence | Comments Off on Statistical Gobbledygook Goes to the Supreme Court

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 Services ¹.

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 power³. Campion 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)

Posted in Expert Witnesses, Rule 702, Rule 703, statistical evidence | Comments Off on Multiplicity in the Third Circuit

WOE — Zoloft Escapes a MDL While Third Circuit Creates a Conceptual Muddle

July 31st, 2017

Multidistrict Litigations (MDLs) can be “muddles” that are easy to get in, but hard to get out of. Pfizer and subsidiary Greenstone fabulously escaped a muddle through persistent lawyering and the astute gatekeeping of a district judge, in the Eastern District of Pennsylvania. That judge, the Hon. Cynthia Rufe, sustained objections to the admissibility of plaintiffs’ epidemiologic expert witness Anick Bérard. When the MDL’s plaintiffs’ steering committee (PSC) demanded, requested, and begged for a do over, Judge Rufe granted them one more chance. The PSC put their litigation industry eggs in a single basket, carried by statistician Nicholas Jewell. Unfortunately for the PSC, Judge Rufe found Jewell’s basket to be as methodologically defective as Bérard’s, and Her Honor excluded Jewell’s proffered testimony. Motions, paper, and appeals followed, but on June 2, 2017, the Third Circuit declared that the PSC and its clients had had enough opportunities to get through the gate. Their baskets of methodological deplorables were not up to snuff. In re Zoloft Prod. Liab. Litig., No. 16-2247 , __ F.3d __, 2017 WL 2385279, 2017 U.S. App. LEXIS 9832 (3d Cir. June 2, 2017) (affirming exclusion of Jewell’s dodgy opinions, which involved multiple methodological flaws and failures to follow any methodology faithfully) [Slip op. cited below as Zoloft].

Plaintiffs Attempt to Substitute WOE for Depressingly Bad Expert Witness Opinion

The ruse of conflating “weight of the evidence,” as used to describe the appellate standard of review for sustaining or reversing a trial court’s factual finding with a purported scientific methodology for inferring causation, was on full display by the PSC in their attack on Judge Rufe’s gatekeeping. In their appellate brief in the Court of Appeals for the Third Circuit, the PSC asserted that Jewell had used a “weight of the evidence method,” even though that phrase, “weight of the evidence” (WOE) was never used in Jewell’s litigation reports. The full context of the PSC’s argument and citations to Milward make clear a deliberate attempt to conflate WOE as an appellate judicial standard for reviewing jury fact finding and a purported scientific methodology. See Appellants’ Opening Brief at 54 (Aug. 10, 2016) [cited as PSC] (asserting that “[a]t all times, the ultimate evaluation of the weight of the evidence is a jury question”; citing Milward v. Acuity Specialty Products Group, Inc., 639 F.3d 11, 20 (1st Cir. 2011), cert. denied, 133 S. Ct. 63 (2012).

Having staked the ground that WOE is akin to a jury’s factual finding, and thus immune to any but the most extraordinary trial court action or appellate intervention, the PSC then pivoted to claim that Jewell’s WOE-ful method was nothing much more than an assessment of “the totality of the available scientific evidence, guided by the well-accepted Bradford-Hill criteria.” PSC at 3, 4, 7. This maneuver allowed the PSC to argue, apparently with a straight face, that WOE methodology as used by Jewell, had been generally accepted in the scientific community, as well as by the Third Circuit, in previous cases in which the court accepted the use of Bradford Hill’s considerations as a reliable method for establishing general causation. See PSC at 4 (citing Gannon v. United States, 292 F. App’x 170, 173 n.1 (3d Cir. 2008)). Jewell then simply plugged in his expertise and “40 years of experience,” and the desired conclusion of causation popped out. Id. Quod erat demonstrandum.

In pressing its point, the PSC took full advantage of loose, inaccurate language from the American Law Institute’s Restatement’s notorious comment C:

“No algorithm exists for applying the Hill guidelines to determine whether an association truly reflects a causal relationship or is spurious.”

PSC at 33-34, citing Restatement (Third) of Torts: Physical and Emotional Harm § 28 cmt. c(3) (2010). Well true, but the absence of a mathematical algorithm hardly means that causal judgments are devoid of principles and standards. The PSC was undeterred, by text or by shame, from equating an unarticulated use of WOE methodology with some vague invocation of Bradford Hill’s considerations for evaluating associations for causality. See PSC at 43 (citing cases that never mentioned WOE but only Bradford Hill’s 50-plus year old heuristic as somehow supporting the claimed identity of the two approaches)¹.

Pfizer Rebuffs WOE

Pfizer filed a comprehensive brief that unraveled the PSC’s duplicity. For unknown reasons, tactical or otherwise, however, Pfizer did not challenge the specifics of PSC’s equation of WOE with an abridged, distorted application of Bradford Hill’s considerations. See generally Opposition Brief of Defendants-Appellees Pfizer Inc., Pfizer International LLC, and Greenstone LLC [cited as Pfizer]. Perhaps given page limits and limited judicial attention spans, and just how woefully bad Jewell’s opinions were, Pfizer may well have decided that a more directed approach of assuming arguendo WOE’s methodological appropriateness was a more economical, pragmatic approach. A close reading of Pfizer’s brief, however, makes clear that it never conceded the validity of WOE as a scientific methodology.

Pfizer did point to the recasting of Jewell’s aborted attempt to apply Bradford Hill considerations as an employment of WOE methodology. Pfizer at 46-47. The argument reminded me of Abraham Lincoln’s famous argument:

“How many legs does a dog have if you call his tail a leg?

Four.

Saying that a tail is a leg doesn’t make it a leg.”

Allen Thorndike Rice, Reminiscences of Abraham Lincoln by Distinguished Men of His Time at 242 (1909). Calling Jewell’s supposed method WOE or Bradford Hill or WOE/Bradford Hill did not cure the “fatal methodological flaws in his opinions.” Pfizer at 47.

Pfizer understandably and properly objected to the PSC’s attempt to cast Jewell’s “methodology” at such a high level of generality that any consideration of the many instances of methodological infidelity would be relegated to mere jury questions. Acquiescence in the PSC’s rhetorical move would constitute a complete abandonment of the inquiry whether Jewell had used a proper method. Pfizer at 15-16.

Interestingly, none of the amici curiae addressed the slippery WOE arguments advanced by the PSC. See generally Brief of Amici Curiae American Tort Reform Ass’n & Pharmaceutical Research and Manufacturers of America (Oct. 18, 2016); Brief of Washington Legal Fdtn. as Amicus Curiae (Oct. 18, 2016). There was no meaningful discussion of WOE as a supposedly scientific methodology at oral argument. See Transcript of Oral Argument in In re Zoloft Prod. Liab. Litig., No. 16-2247 (Jan. 25, 2017).

The Third Circuit Acknowledges that Some Methodological Infelicities, Flaws, and Fallacies Are Properly the Subject of Judicial Gatekeeping

Fortunately, Jewell’s methodological infidelities were easily recognized by the Circuit judges. Jewell treated multiple studies, which were nested within one another, and thus involved overlapping and included populations, as though they were independent verifications of the same hypothesis. When the population at issue (from the Danish cohort) was included in a more inclusive pan-Scandivanian study, the relied-upon association dissipated, and Jewell utterly failed to explain or account for these data. Zoloft at 5-6.

Jewell relied upon a study by Anick Bérard, even though he later had to concede that the study had serious flaws that invalidated its conclusions, and which flaws caused him to have a lack of confidence in the paper’s findings.² In another instance, Jewell relied innocently upon a study that purported to report a statistically significant association, but the authors of this paper were later required by the journal, The New England Journal of Medicine, to correct the very calculated confidence interval upon which Jewell had relied. Despite his substantial mathematical prowess, Jewell missed the miscalculation and relied (uncritically) upon a finding as statistically significant when in fact it was not.

Jewell rejected a meta-analysis of Zoloft studies for questionable methodological quibbles, even though he had relied upon the very same meta-analysis, with the same methodology, in his litigation efforts involving Prozac and birth defects. Not to be corralled by methodological punctilio, Jewell conducted his own meta-analysis with two studies Huybrechts (2014) and Jimenez-Solem (2012), but failed to explain why he excluded other studies, the inclusion of which would have undone his claimed result. Zoloft at 9. Jewell purported to reanalyze and recalculate point estimates in two studies, Jimenez-Solem (2012) and Huybrechts (2014), without any clear protocol or consistency in his approach to other studies. Zoloft at 9. The list goes on, but in sum, Jewell’s handling of these technical issues did not inspire confidence, either in the district or in the appellate court.

WOE to the Third Circuit

The Circuit gave the PSC every conceivable break. Because Pfizer had not engaged specifically on whether WOE was a proper, or any kind of, scientific method, the Circuit treated the issue as virtually conceded:

“Pfizer does not seem to contest the reliability of the Bradford Hill criteria or weight of the evidence analysis generally; the dispute centers on whether the specific methodology implemented by Dr. Jewell is reliable. Flexible methodologies, such as the “weight of the evidence,” can be implemented in multiple ways; despite the fact that the methodology is generally reliable, each application is distinct and should be analyzed for reliability.”

Zoloft at 18. The Court acknowledged that WOE arose only in the PSC’s appellate brief, which would have made the entire dubious argument waived under general appellate jurisdictional principles, but the Court, in a footnote, indulged the assumption, “for the sake of argument,” that WOE was Jewell’s purported method from the inception. Zoloft at 18 n. 39. Without any real evidentiary support or analysis or concession from Pfizer, the Circuit accepted that WOE analyses were “generally reliable.” Zoloft at 21.

The Circuit accepted, rather uncritically, that Jewell used a combination of WOE analysis and Bradford Hill considerations. Zoloft at 17. Although Jewell had never described WOE in his litigation report, and WOE was not a feature of his hearing testimony, the Circuit impermissibly engrafted Carl Cranor’s description of WOE as involving inference to the best explanation. Zoloft at 17 & n.37, citing Milward v. Acuity Specialty Prods. Grp., Inc., 639 F.3d 11, 17 (1st Cir. 2011) (internal quotation marks and citation omitted).

There was, however, a limit to the Circuit’s credulousness and empathy. As the Court noted, there must be some assurance that the purported Bradford Hill/WOE method is something more than a “mere conclusion-oriented selection process.” Zoloft at 20. Ultimately, the Court put its markers down for Jewell’s putative WOE methodology:

“there must be a scientific method of weighting that is used and explained.”

Zoloft at 20. Calling the method WOE did not, in the final analysis, exclude Jewell from Rule 702 gatekeeping. Try as the PSC might, there was just no mistaking Jewell’s approach as anything other than a crazy patchwork quilt of numerical wizardry in aid of subjective, result-oriented conclusion mongering.

In the Court’s words:

“we find that Dr. Jewell did not 1) reliably apply the ‘techniques’ to the body of evidence or 2) adequately explain how this analysis supports specified Bradford Hill criteria. Because ‘any step that renders the analysis unreliable under the Daubert factors renders the expert’s testimony inadmissible’, this is sufficient to show that the District Court did not abuse its discretion in excluding Dr. Jewell’s testimony.”

Zoloft at 28. As heartening as the Circuit’s conclusion is, the Court’s couching its observation as a finding (“we find”) is disheartening with respect to the Third Circuit’s apparent inability to distinguish abuse-of-discretion review from de novo appellate findings. Equally distressing is the Court’s invocation of Daubert factors, which were dicta in a Supreme Court case that was superseded by an amended statute over 17 years ago, in Federal Rule of Evidence 702.

On the crucial question whether Jewell had engaged in an unreliable application of methods or techniques that superficially, at a very high level of generality, claim to be generally accepted, the Court stayed on course. The Court “found” that Jewell had applied techniques, analyses, and critiques so obviously inconsistently that no amount of judicial indulgence, assumptions arguendo, or careless glosses could save Jewell and his fatuous opinions from judicial banishment. Zoloft 28-29. Returning to the correct standard of review (abuse of discretion), but the wrong governing law (Daubert instead of Rule 702), the Court announced that:

“[b]ecause ‘any step that renders the analysis unreliable under the Daubert factors renders the expert’s testimony inadmissible’, this is sufficient to show that the District Court did not abuse its discretion in excluding Dr. Jewell’s testimony.”

Zoloft at 21 n.50 (citation omitted). The Court found itself unable to say simply and directly that “the MDL trial court decided the case well within its discretion.”

The Zoloft case was not the Third Circuit’s first WOE rodeo. WOE had raised its unruly head in Magistrini v. One Hour Martinizing Dry Cleaning, 180 F. Supp. 2d 584, 602 (D.N.J. 2002), aff’d, 68 F. App’x 356 (3d Cir. 2003), where an expert witness, David Ozonoff, offered what purported to be a WOE opinion. The Magistrini trial court did not fuss with the assertion that WOE was generally reliable, but took issue with how Ozonoff tried to pass off his analysis as a comprehensive treatment of the totality of the evidence. In Magistrini, Judge Hochberg noted that regardless of the rubric of the methodology, the witness must show that in conducting a WOE analysis:

“all of the relevant evidence must be gathered, and the assessment or weighing of that evidence must not be arbitrary, but must itself be based on methods of science.”

Magistrini, 180 F. Supp. 2d at 602. The witness must show that the methodology is more than a “mere conclusion-oriented selection process,” and that it has a “a scientific method of weighting that is used and explained.” Id. at 607. Asserting the use of WOE was not an excuse or escape from judicial gatekeeping as specified by Rule 702.

Although the Third Circuit gave the Zoloft MDL trial court’s findings a searching review (certainly much tougher than the prescribed abuse-of-discretion review), the MDL court’s finding that Jewell “failed to consistently apply the scientific methods he articulates, has deviated from or downplayed certain well-established principles of his field, and has inconsistently applied methods and standards to the data so as to support his a priori opinion” were ultimately vindicated by the Court of Appeals. Zoloft at 10.

All’s well that ends well. Perhaps. It remains unfortunate, however, that a hypothetical method, WOE — which was never actually advocated by the challenged expert witnesses, which lacks serious support in the scientific community, and which was merely assumed arguendo to be valid — will be taken by careless readers to have been endorsed the Third Circuit.

1 Among the cases cited without any support for the PSC’s dubious contention were Gannon v. United States, 292 F. App’x 170, 173 n.1 (3d Cir. 2008); Bitler v. A.O. Smith Corp., 391 F.3d 1114, 1124-25 (10th Cir. 2004); In re Joint E. & S. Dist. Asbestos Litig., 52 F.3d 1124, 1128 (2d Cir. 1995); In re Avandia Mktg., Sales Practices & Prods. Liab. Litig., No. 2007-MD-1871, 2011 WL 13576, at *3 (E.D. Pa. Jan. 4, 2011) (“Bradford-Hill criteria are used to assess whether an established association between two variables actually reflects a causal relationship.”).

2 Anick Bérard, Sertraline Use During Pregnancy and the Risk of Major Malformations, 212 Am. J. Obstet. Gynecol. 795 (2015).

Posted in Expert Witnesses, Rule 702, Scientific Evidence, statistical evidence | Comments Off on WOE — Zoloft Escapes a MDL While Third Circuit Creates a Conceptual Muddle