Dayton Law Review

The Scientific Basis of Causality in Toxic Tort Cases

Appendix: The Logical Structure of Scientific Studies Relevant to Toxic Tort Cases

*Professor, Department of Orthopaedic Surgery and Department of Cellular Biology and Anatomy, Louisiana State University Medical Center, P.O. Box 33932 Shreveport, LA 71130-3932, Phone: 318-675-6177, Fax: 318-675-6186, e-mail:; Professor, Department of Bioengineering, Louisiana Tech University. Dr. Marino is the co-author of several books and has authored numerous papers and articles. Dr. Marino is admitted to practice in Louisiana and New York and holds a J.D., and a M.S. and Ph.D. in Biophysics, from Syracuse University and a B.S. in Physics from St. Joseph’s College.

**Associate, Oats Hudson, Lafayette, LA. Mr. Marino is admitted to practice in Louisiana and Texas and holds a J.D. from Tulane University and a B.S. from the University of Houston.

1. The masculine form is used in this article for both genders, except where it obviously applies to only one.

2. Since knowledge and opinion with respect to expert witness testimony may have overlapping meanings, it is worthwhile to adopt consistent definitions for these terms. An opinion is a statement colorably sounding as intellectual knowledge, and that the speaker accepts as true (that is, sufficiently justified), but which either is not accepted or has not yet been accepted by the listener. Knowledge is justified belief in the truth of a statement. Sensory knowledge (“I heard the crash”) is typically provided by the fact witness. Intellectual knowledge (“poison ivy causes a rash”), provided by the expert, is achieved when experience and understanding are focused on sensory knowledge. Scientific knowledge is intellectual knowledge of or pertaining to science. See infra note 8 and accompanying text. Despite the subjective certitude and passion with which statements reflecting sensory and intellectual knowledge are sometimes made, it is clear that the possibility of error cannot be eliminated. What permits putative knowledge to be characterized as knowledge is the nature and degree of the justification that can be provided to indicate that the statement is true.


4. Between the second and third of December 1984, winds blew a lethal gas, known as methyl isocyanate, from a chemical plant operated by Union Carbide India Limited, into densely populated areas of Bhopal, India. Over 2,000 people were killed and more than 200,000 were injured. See Union Carbide Corp. Gas Plant Disaster v. Union Carbide Corp., 809 F.2d 195 (2d Cir. 1987).

5. There are other legal theories under which scientific testimony involving causality might be used, such as fraud or misrepresentation, battery, nuisance, breach of fiduciary duty, and breach of contract. Nevertheless, if an expert can render an opinion regarding the causal issues in a toxic tort case, the expert also can do so with respect to any legal theory where scientific causality is a pertinent element. Thus, a toxic tort suit in negligence is an appropriate context within which to evaluate the legal implications of scientific testimony regarding causality.

6. W. PAGE KEETON ET AL, PROSSER AND KEETON ON THE LAW OF TORTS § 41, at 263-72 (5th ed.1984).

7. See RESTATEMENT (SECOND) OF TORTS § 9, at 16 (1965).

8. The dominant philosophical basis of modern science is the tradition known as logical empiricism. See RICHARD BOYD ET AL., THE PHILOSOPHY OF SCIENCE, (MIT Press 1991). The philosophical underpinnings of science are unimportant for the purposes of this Article, but it is important to recognize that modern scientific reasoning occurs within the context of a system of assumptions. A scientific statement, therefore, is not necessarily meaningful outside the context of that system.

9. For example, cigarette manufacturers urged this theory regarding inferences that smoking can lead to lung cancer. See Hearings on S. 772 Before the Senate Comm. on Labor and Human Resources, 98th Cong., 1st Sess. 98, at 253-56 (1983) (statement of Sheldon C. Sommers, M.D., consultant in pathology, Lenox Hill Hospital, New York, N.Y.).

10. In attempting to explain the notion of causality as it applies in toxic tort cases, many authors make exclusive use of examples involving physical laws, such as the law of gravity. These authors simply assume the notion of causality involved in such examples is directly applicable to the biological sciences. Such, however, is not the case. Biology is an autonomous science with its own methods and procedures, which do not necessarily depend on the paradigm of physics (including its conception of causality) for their ultimate rationale or validity This error results in the fostering of a falsely precise notion of the kind of scientific knowledge that is relevant to toxic tort cases.

11. Four different forces are recognized. The gravitational and electromagnetic forces are the causes of essentially all phenomena familiar to the layman. The other two forces are the strong force, which is responsible for the stability of atoms and for events that are observed in particle accelerators, and the weak force, which causes radioactive decay of atoms.

12. The goal of the physical sciences is the identification and quantification of the forces responsible for phenomena; this is accomplished by systematically varying the conditions of observation, and then formulating mathematical equations that can be used to predict future similar observations. This process has been extraordinarily successful—with the exception of esoteric situations such as those that existed at the time the universe began or that occur in supercolliders, the causes of all physical phenomena are known and their occurrence is predictable with mathematical precision. This knowledge of inanimate reality, which was achieved within the last three centuries, does not indicate that all consequences of physical laws are known; only that all known physical phenomena can be understood as consequences of known laws, and can be reproduced by anyone who cares to do so.

13. Since many factors could potentially influence any particular biological observation, the method of controlled observation is usually employed to study putative/causal relationships. The method consists of standardizing all pertinent environmental factors in a homogeneous population of living organisms except for a single factor, the effect of which is to be studied, and then varying that factor with respect to only some of the individuals (the experimental group). If a difference between the experimental group and the remaining subjects (the control group) is subsequently observed at the appropriate level of statistical certainty (greater than 95%), the factor that differed between the groups is accepted as the cause of the difference. See infra Appendix, at 58-62 for a discussion of the logical structure and principal types of biological studies.

14. Mechanistic causes are the Holy Grail of biological scientists and are equally difficult to find. They can never be precisely identified because it is impossible to prove that a particular mechanism is operative; the best that can be done is to produce evidence for or against a particular mechanism. Further, whenever evidence supporting a particular mechanism is found, it is always possible to ask: What is the mechanism of that mechanism? Thus, every mechanistic explanation is, at best, a partial explanation, and it is always possible to argue that the mechanism underlying a particular phenomenon is not known.

15. This means that an appropriate statistical test showed that the probability that the statement was true was greater than 95%. See discussion infra Appendix, at 58-62.

16. Not any specific amount or under any particular conditions of exposure.

17. Not any specific incidence or type of cancer. Throughout this Article, “x” designates a specific cause, “y” designates a specific effect, “X&lrdquo; designates a general cause, and “Y” designates a general effect. See infra Glossary.

18. Disagreements among scientists are foreseeable because scientists differ with regard to ability, personal values, and amount and type of experience. It is not true that scientists would necessarily agree on any particular judgment, if only they took the time and trouble to examine the data carefully.

19. When examining an expert, counsel must ensure that the preponderance of the evidence standard, and nothing more stringent, is applied.

20. The defendant will seek to prove that, although a causal inference could be true, one would not be justified in accepting it as true.

21. The logical relationship between positive and negative observations is a familiar feature of everyday life. For example, if 1,000 holes are drilled to varying depths at separate locations in a search for oil and no oil is found, a valid conclusion would be that there is no oil. But irrespective of the number of dry holes, if even a single hole results in the appearance of oil, the proposition that drilling a hole can lead to oil is established, and the evidentiary value of dry holes becomes reduced. Now, the dry holes indicate only that oil does not occur under a particular set of circumstances. As in science, even one positive observation rationalizes a positive conclusion even though there are numerous negative observations.

22. For example, in a personal injury case, if the breaking strength of an automobile fuel tank is at issue, the court might determine that an engineer who has experience with studies and measurements of the mechanical strength of fuel tanks is qualified to offer relevant testimony.

23. Typically, the academic attainment expected of a scientific expert involves completion of undergraduate and graduate courses designed to teach mastery of the knowledge, principles, and methods applicable to all biological sciences. In the United States, post-graduate education usually consists of approximately two years of classroom studies, followed by an apprentice period of three to five years devoted to the study and use of the methods of science employed for generating scientific knowledge. Performance of an independent scientific investigation, culminating in a dissertation deemed acceptable by the student’s mentor and advisory committee, as memorialized by the degree of Doctor of Philosophy (Ph.D.), is evidence that the principles of scientific methodology and reasoning have been mastered.

Traditional distinctions among various biological sciences have largely been blurred as a result of the rapid growth of biological science and increasing specialization within the past 20 years. Although the names and number of academic departments awarding the Ph.D. have not changed appreciably, the number of areas and amount of biological specialization has increased dramatically. At a meeting of one group of biological specialists (Experimental Biology ’94, April 24-28,1994, Anaheim, CA), more than 100 different specialized biological categories were necessary to classify the presentations. Other groups of biological specialists employ many additional categories. Multiple classifications within non-biological science are similarly numerous and diverse. For example, the American Society for Testing and Materials lists 280 categories of specialization. See AM. SOC’Y FOR TESTING AND MATERIALS, DIRECTORY OF SCIENTIFIC AND TECHNICAL CONSULTANTS AND EXPERT WITNESSES (1993-94). As a result, the name of the university department that awarded the expert’s Ph.D.—for example, physiology, biophysics, immunology, or biochemistry—is not a useful guide for determining whether the expert has the required training.

24. See discussion infra Appendix, at 58-62.

25. Children living beside powerlines were considered to be exposed to electromagnetic fields, in comparison with similar children who did not live beside powerlines. Nancy Wertheimer & Ed Leeper, Electrical Wiring Configurations and Childhood Cancer, 109 AM. J. EPIDMIOL. 273 (1979). Adults who lived near powerlines were considered to be exposed to electromagnetic fields in comparison to adults who did not live near powerlines. F. Stephen Perry et al., Environmental Power-Frequency Magnetic Fields and Suicide, 41 HEALTH PHYS. 267 (1981). People who lived near airport radars were considered to be exposed in comparison with others. John R. Lester & Dennis F. Moore, Cancer Incidence and Electromagnetic Radiation, I J. BIOELECTRICITY 59 (1982). Working in various electrical occupations including electricians, electrical engineers, and powerline workers was considered to represent increased exposure to electromagnetic fields, in comparison with non-electrical occupations. Michel Coleman et al., Leukemia Incidence in Electrical Workers, LANCET 982 (1983). Being a ham radio operator was considered to indicate increased exposure to electromagnetic fields, compared with other individuals who were not ham radio operators. Samuel Milham, Jr., Increased Mortality in Amateur Radio Operators Due to Lymphatic and Hematopoietic Malignancies, 127 AM. J. EPIDEMIOL. 50 (1988).

26. Dosimetry is the study of the amount of a toxic agent actually received by a subject under a specific set of conditions.

27. It is also possible that none of the authors of an epidemiological study is an expert in the area of doses and exposure levels.

28. Epidemiology is a nonlaboratory based specialty. Consequently, an epidemiologist is usually not qualified to testify on the ultimate issues in a toxic tort case if relevant laboratory evidence is available, unless the epidemiologist has acquired expertise regarding animal studies.

29. A physician is a specialist in the diagnosis and treatment of human disease. In the United States, a physician must graduate from a four-year post-graduate course of study at an accredited school, leading to the degree of Doctor of Medicine (M.D.) or Doctor of Osteopathy (D.O.). Typically, the first two years of medical school are spent in the classroom, and the last two years are devoted principally to learning the accepted forms of treatment for various clinical conditions, and to developing the ability to make diagnoses and administer treatment. Internship, the first year after medical school, is an apprentice year. Thereafter, the physician begins to exercise independent medical judgment in a clinical practice, or enters a residency program in a particular medical specialty. The latter path involves an additional three to five years of detailed study of the methods of diagnosis and treatment of a limited set of human diseases, such as those that occur in the musculoskeletal system (orthopaedic surgery), children (pediatrics), or women (gynecology).

30. It does not follow, however, that someone having an M.D. (or lacking a Ph.D.) is unqualified to testify regarding scientific matters because an earned Ph.D. is neither a necessary nor sufficient indicator of expertise.

31. The oncologist, for example, conducts tests and examinations to determine whether an identified mass is malignant or benign, but does not engage in a causal analysis to determine why the tumor mass occurred in the patient. Such an inquiry might be made by a scientist studying a group of similar patients to test a hypothesis about a cause, causal mechanism, or cure, but the clinical oncologist is ordinarily not trained for or concerned with such an inquiry. If the oncologist tested a hypothesis during the course of the patient’s treatment. then to the extent the oncologist followed the rules and procedures of scientific methodology, he would be functioning as a scientist rather than as a clinician.

32. In a medical malpractice action against an orthopaedic surgeon, 20,000 experts potentially could testify because there are approximately that many practicing orthopaedic surgeons. On the other hand, a claim that a toxic agent caused harm to the plaintiff’s bones can be sustained only by the testimony of an expert regarding the scientific knowledge of the effects produced by that agent, specifically regarding the effects on bone.

33. For example, in Cantrell v. GAF Corp., the court allowed a physician to make an inference of a causal relationship between asbestos and cancer partly based on anecdotal data gained through the witness’s clinical experience. 999 F.2d 1007,1013-14 (6th Cir. 1993). Courts have rarely recognized the impropriety of physicians testifying to causal links that are determinable only by the methods of science rather than medicine. See, e.g., Ferebee v. Chevron Chem. Co., 736 F.2d 1529,1535 (D.C. Cir.) (allowing physicians to testify regarding the ability of PCBs to cause pulmonary fibrosis), cert. denied, 469 U.S. 1062 (1984 ); Osburn v. Anchor Lab., Inc., 825 F.2d 908, 915 (5th Cir. 1987) (allowing physicians to testify regarding the ability of chloramphenicol to cause leukemia), cert. denied, 485 U.S. 1009 (1988); Sterling v. Velsicol Chem. Corp., 855 F.2d 1188, 1204 (6th Cir. 1988) (allowing physicians to testify that contaminated water caused the plaintiffs’ medical problems): Hines v. Consol. Rail Corp., 926 F.2d 262, 273 (3d Cir. 1991) (permitting physician to testify regarding ability of PCBs to cause cancer). But see Mason v. Texaco, Inc., 741 F.Supp. 1472,1497 (D. Kan. 1990) (noting that scientists are more qualified to testify regarding causation, and that a medical degree or training does not necessarily confer the ability to testify to causation), aff’d and remanded, 948 F.2d 1546 (10th Cir. 1991), cert. denied, 504 U.S. 910 (1992).

34. If an editor rejects a manuscript, but the author believes it has scientific value, the author remains free to submit it to another journal. Neither the substance nor the fact of previous reviews are disclosed to subsequent editors. When a manuscript involves arcane areas of biology, there may exist only a few journals that would be appropriate for its publication, but in other areas there might be several hundred journals that would consider publishing a manuscript. In neuroscience, for example, a manuscript might undergo numerous sequential independent peer reviews before it is finally published.

35. For example, suppose a study concluded that a food additive manufactured by ABC, Inc. produced no harmful effects in the gastrointestinal tract of animals, and consequently the author judged the additive to be safe for human use. Since peer review focuses solely on purely scientific considerations, it would be irrelevant to the reviewers whether the author was an independent scientist with an academic interest in the physiology of the gut, or an employed of ABC, Inc. seeking to allay concerns raised by the FDA.

36. For example, the National Institutes of Health and many private foundations require authors to acknowledge receipt of their grant support. Where the author of a study of food additives is an employee of a food additive company, such an affiliation would likely be disclosed when the report was actually published because it is the custom to publish the professional affiliations of the authors. If, however, the study was performed by a contractor, rather than an employee, the relationship would normally not be disclosed because the authors’ affiliations listed in the published report would be their employers, not the party that awarded the research contract. Some journals recently have begun requiring authors to disclose whether they have received, or will receive, personal or professional benefits from a commercial party related directly or indirectly to the subject or conclusions of the report. This practice, however, is not widespread.

37. Nevertheless, the report may be largely ignored for a variety of reasons, including a lack of importance of the results, or a judgment by scientists other than the peer reviewers that the work has no merit. Such is the fate of most published scientific reports.

38. See Symposium, Editorial Peer Review in Biomedical Publication. The First International Congress, 263 JAMA 1317, 1317-1444 (1990) for a detailed description of the inherent limitations and practical difficulties associated with peer review.

39. Scientific journals are overwhelmingly the most significant repository of the world’s scientific knowledge. The number of scientific journals worldwide is uncertain, but the number probably exceeds 100,000. A data base organized by the National Library of Medicine (NLM) subscribes to more than 4,000 biological journals and organizes the information therein to permit searching by topic or by key word, using either text or a computer. Both information-retrieval systems, known respectively as “Index Medicus” and “Medline,” are readily available, relatively inexpensive, and permit nearly instant access to knowledge concerning any topic in biology. In addition to the NLM data bases, many private, more specialized data bases permit access to journals not covered by the NLM.

40. Although independent peer review and publication in an archival scientific journal is the most common method of disseminating scientific information, there are other methods. For example, research may be performed, reviewed, and published by corporations, private research organizations, or governmental agencies. In such instances, the researchers, reviewers, editors, and publishers are employees of one organization, as opposed to work performed at academic institutions and published in scientific journals, where the respective parties are independent of each other.

41. The following is an example of how the manner of disclosure of a study can affect its interpretation. Since the mid-1970s, investigators at Battelle Pacific Northwest Laboratories have performed contract research, partly funded by the Electric Power Research Institute (a consortium of U.S. electrical power companies) to show the safety of high voltage powerlines. One experiment involved the effects of long-term exposure to electromagnetic fields on the growth rate of mice. In the experiment, one group of animals was exposed to the field, and the other served as the comparison group to permit assessment of the effects of the field. The result was that the mice in the exposed group were smaller, on average, compared with the controls, and the difference could not be attributed to chance (less than a 5% possibility). The result was unexpected, and the experiment was repeated; this time, however, the exposed mice were found to be larger than their corresponding controls. Again, the results could not be attributed to chance. If the data from each study was evaluated separately, which was the initial plan, and the data properly interpreted according to the established rules of science, it would be concluded that exposure to electromagnetic fields can decrease or increase growth in mice, depending upon the presence or absence of other, unascertained factors. Instead, the investigators averaged the results of the two studies, and thus concluded that electromagnetic fields had no effect on growth in mice and, consequently, that the studies did not suggest a likelihood of harm to similarly exposed human subjects. R.D. PHILLIPS ET AL., U.S. DEPT. ENERGY, BIOLOGICAL EFFECTS OF HIGH STRENGTH ELECTRIC FIELDS ON SMALL LABORATORY ANIMALS, DOE/TIC- 10084 (1979) (available from National Technical Information Service (NTIS), U.S. Dept. of Commerce, 5285 Port Royal Road, Springfield, VA 22161); see also ROBERT O. BECKER & ANDREW A. MARINO, ELECTROMAGNETISM & LIFE 150 (1982); ANDREW A. MARINO & JOEL RAY, ELECTRIC WILDERNESS 98 (1986).

42. In 1994, federal spending for research and development in health was about $12 billion. Spending by industry was almost $16 billion. See Tim Beardsley, Big-Time Biology, SCI. AM.,Nov. 1994, at 90,92.

43. Such entities include Battelle, Midwest Research Institute in San Antonio, Texas, and the Oak Ridge National Laboratory in Oak Ridge, Tennessee.

44. Grants from the National Institutes of Health (NIH) or the National Science Foundation (NSF) are the backbone of independent science in the United States. These grants are awarded for one to five years. The sole consideration for funding is the scientific merit of the proposed work. Generally, it is irrelevant to the investigator and the funding agencies whether the implications of a particular study might tend to support or refute allegations of causal connections between particular toxins and particular diseases. Such research is performed pursuant to a specific plan, and the plan itself, as well as all data reported to the granting agency, is available under the federal Freedom-of-information Act. Freedom of Information Reform Act of 1986, 5 U.S.C. § 552 (1994). Moreover, both the Nl H and the NSF have promulgated policies directing that raw data and associated materials obtained during the research should be made available to all interested parties. No other federal agencies, state agencies, or private organizations have adopted such a policy.

Under the common law, work performed by an employee in the course of his employment is owned by the employer because it is work-for-hire. Thus, research data produced by a faculty member with institutional support is owned by the academic institution. Institutions also retain legal title to the scientific data produced by faculty members who are supported by NIH and NSF funds. See Administration of Grants, 45 C.F.R. part 74 (1993); Representation of Limited Rights Data and Restricted Computer Software, 48C.F.R. § 52.227-15 et seq. (1993). Since the institution owns the work, it has the copyright. 17 U.S.C. § 201 (b) (1994). Hence the institution is legally entitled to decide issues of publication and access. Although the academic institution is entitled to claim the copyright, it normally chooses not to do so. Instead, the academic tradition is that faculty members are permitted to claim the copyright for their research. Since there is no national registry of research pertinent to drugs or toxic torts, it is not possible to establish research that is occurring or has occurred.

45. The actual grantee is the academic institution which employs the scientist. The institution maintains the financial and administrative records, sets and implements purchasing procedures, retains title to equipment purchased with grant funds, and is vicariously liable for any scientific misconduct on the part of the grant’s principle investigator. Nevertheless: (I ) grant funds are provided for the services of a specific investigator; (2) expenditures of grant funds can be initiated only by the investigator; (3) the investigator has the unilateral an exclusive authority to vary the actual research conducted based on new information obtained subsequent to the award of the grant; (4) normally, the investigator has unilateral and unrestricted authority to determine access to the research data. Because of these factors, it appears the investigator is the grantee, and is commonly referred to as the grantee.

46. This subsection deals with whether an expert should rely on the work product of a blue-ribbon committee. See infra notes 99-101 and accompanying text for a discussion of the admissibility of the work product of a blue-ribbon committee into evidence.

47. Usually, blue-ribbon committees are established to allay public concerns regarding an issue, according to the collective judgment of the committee’s members. For example, the American National Standards Institute (ANSI) formed a committee to choose safe levels of electromagnetic fields. INST. OF ELEC. AND ELEC. ENG RS, AMERICAN NATIONAL STANDARD SAFETY LEVELS WITH RESPECT TO HUMAN EXPOSURE TO RADIO FREQUENCY ELECTROMAGNETIC FIELDS, 300 KHZ TO 1 00 GHz, ANSI C95. 1-1982 10 (1982) (available from The Institute of Electrical and Electronics Engineers, Inc., 345 East 47th Street, New York, NY 10017). It is historically true that privately appointed blue-ribbon committees called upon to evaluate chemical or physical agents present in the environment usually conclude that such agents do not pose a health risk. For example, the safe level chosen by ANSI is 200,000 times higher than the median exposure level in urban areas of the United States determined by the Environmental Protection Agency. In other words, according to the ANSI committee, essentially everyone is safe from harm due to electromagnetic fields. See U.S. ENVTL. PROTECTION AGENCY, POPULATION EXPOSURE TO VHF AND UHF BROADCAST RADIATION IN THE UNITED STATES, ORP/EAD 78-5 (1978) (available from United States Environmental Protection Agency, Office of Radiation Programs, Las Vegas Facility, P.O. Box 15027, Las Vegas, NV 89114).

48. For example, since the ANSI committee was not chosen randomly from among all qualified persons, its consensus is not likely to be similar to one that would have been reached by a representative group of qualified experts. The ANSI committee consisted of 53 members. Judging from the professional affiliations and degrees of its members, we estimate that there are more than half a million persons who could have qualified as members. Although the number of possible committees is unimaginably large, the one actually chosen is not representationally valid because at least some members were chosen because of their opinion. The theoretical rationale for a blue-ribbon committee is the same as that for a jury. The validity of a jury’s verdict is derived from its representative nature. Since the rationale for a blue-ribbon committee is that its conclusion is representative of those that would be reached by a larger group, and such is not the case with ANSI, adoption of the ANSI committee’s opinion as an authoritative statement of scientific fact is not justifiable. Instead, it is the opinion of the persons who chose the committee.

49. The ANSI committee, for example, consisted of employees of business organizations that manufactured devices that emit electromagnetic fields, but no representatives of those who are regularly exposed to electromagnetic fields. It is not reasonable to expect employees of companies that derive profit from manufacturing devices that emit electromagnetic fields to adequately represent the interests of those who are exposed to the emissions of these devices.

50. The National Academy of Sciences (NAS), in cooperation with the United States Navy, appointed a blue-ribbon committee to evaluate the safety of a large antenna that would emit electromagnetic fields similar to those emitted by power lines, except that the fields from the antenna would be 100,000 times weaker. Three experts, who previously testified that power line electromagnetic fields create no health risk were appointed to the NAS committee. Not surprisingly, the NAS committee found that the proposed antenna would be safe. NAT L ACADEMY OF SCIENCES, BIOLOGIC EFFECTS OF ELECTRIC AND MAGNETIC FIELDS ASSOCIATED WITH PROPOSED PROJECT SEAFARER: REPORT OF THE COMMITTEE ON BIOSPHERIC EFFECTS OF EXTREMELY-Low-FREQUENCY RADIATION (1977); Philip M. Boffey, Project Seafarer: Critics Attack National Academy’s Review Group, 192 Sci. 1213 (1976); see also ANDREW A. MARINO & JOEL RAY, supra note 41, at 98.

NAS committees are the most prestigious blue-ribbon committees in the United States. Although approximately 900 NAS committees are presently evaluating science policy in various areas, only about 100 of the more than 1600 NAS members serve on the committees. Critics of the NAS members’ absence argue that the quality of the reports would improve if more NAS members served on blue-ribbon committees. The president of the NAS, however, noting that most NAS members are older white males, suggested that any gain in wisdom might be offset by other factors. The mechanism by which individuals are chosen for the NAS committees has not been publicly disclosed. See Robert Langreth, Members Seek More Active Role, 263 SCI. 23 (1994).

51. A Florida state agency appointed a scientist who performed contract research on behalf of a national consortium of electric power companies to chair a blue-ribbon committee regarding powerline safety. The committee generally exonerated state regulatory practices, which did not require any special efforts to lessen exposure to electromagnetic fields or to apprise the public of the nature or extent of the exposure. Shortly thereafter, the chairman became the chief of staff for a law firm that represents power companies in legal actions involving allegations of health risks due to electromagnetic fields from powerlines. See FLA. ELEC. AND MAGNETIC FIELDS SCIENCE ADVISORY COMM’N, BIOLOGICAL EFFECTS OF 60-HZ POWER TRANSMISSION LINES, FLORIDA ELECTRIC AND MAGNETIC FIELDS SCIENCE ADVISORY COMMISSION REPORT (1985) (H.B. Graves, Chairman); see generally 8 MICROWAVE NEWS, Mar.-Apr, 1988.

52. A concept similar to the blue-ribbon committee, called the Science Court, was proposed by a Presidential advisory panel as a means of resolving scientific disputes. See Task Force of the Presidential Advisory Group on Anticipated Advances in Science and Technology, The Science Court Experiment: An Interim Report, 193 Sci. 653, 653-56 (1976). The basic idea of the Science Court was that scientists would be appointed as judges to resolve issues such as: Should hydrofluarocarbons be banned because of their impact on the ozone layer? Is red dye #40 safer than red dye #2? Should water supplies be fluoridated? Such questions are value-laden, and consequently can be resolved only if value judgments are incorporated into the decision making process. Since the values that must necessarily be applied are those of society as a whole, not those of science or particular scientists, the Science Court concept was fatally flawed. For a description of the failure of attempts to form a Science Court to consider whether electromagnetic fields from powerlines are a health hazard, see Allen Mazur et al., Separating Factual Disputes from Value Disputes in Controversies over Technology, I TECH. IN SOC’Y 229, 229-37 (1979).

53. The expert cannot rely on personal observation as the basis for asserting a causal relationship because the expert made no direct observations pertinent to the plaintiff or the conditions of his exposure. Even if the expert had observed the plaintiff continuously from the inception of exposure to the toxic agent until the plaintiff s disease was diagnosed, the expert still could not testify to the fact of causality based on direct observation because causation of human disease—as distinguished from its existence—is not amenable to direct observation.

54. See discussion infra Appendix, at 58-62. Test-tube studies are the premier method for studying causal mechanisms, but they constitute such a vast oversimplification of animal physiology, that they are essentially useless in direct assessment of the overall effects of human exposure to a putative toxin. Animal studies are capable of providing relatively clear demonstrations of cause-and-effect relationships and the results can be applicable to human exposure. Animal studies also have the following drawbacks: (1) relatively short exposure durations, often requiring large doses of the toxic agent; (2) biological endpoints that may not be directly relatable to a recognizable disease because they constitute physiological changes that are only precursor stages of disease; and (3) arguable inapplicability to human subjects based on physiological differences between laboratory animals and human beings. Epidemiological studies are directly relevant to toxic tort cases because they involve human beings who were actually exposed to the toxic agent and developed the disease. All epidemiological studies are confounded to some degree by the problem that unknown factors, rather than the toxic agent being studied, may have caused the observed change in the incidence of disease.

55. Many federal laws and accompanying regulations require the use of animal studies to assess human health risks. See Federal Pesticide Act of 1978, 7 U.S.C. § 136 et seq. (1994); Federal Hazardous Substances Act, 15 U.S.C. § 1261 et seq. (1994); Consumer Product Safety Act, 15 U .S.C. § 2051 et seq. (1994); Toxic Substance Control Act, 15 U.S.C. § 2601 et seq. (1994); Federal Food, Drug, and Cosmetic Act, 21 U.S.C. &ect; 301 et seq. (1988 & Supp. V 1993); Occupational Safety and Health Act of 1970, 29 U.S.C. § 651 et seq. (1988 & Supp. V 1993); Federal Water Pollution Control Act, 33 U.S.C. 1251 et seq. (1988 & Supp. V 1993); Safe Drinking Water Act, 42 U.S.C. § 300 et seq. (1988 amp;& Supp. V 1993); Resource Conservation and Recovery Act of 1976, 42 U.S.C. § 6901 et seq. (1988 & Supp. V 1993); Clean Air Act, 42 U.S.C. § 7401 et seq. (1988 & Supp. V 1993).

Federal public health authorities invariably consider both animal and epidemiological studies. See U.S. Envtl. Protection Agency, Final Guidelines for Developmental Toxicity Risk Assessment, 56 Fed. Reg. 63798, 63799 (1991) “[H]azard identification dose-response evaluation involves examining all available experimental animal and human data.”); U.S. Envtl. Protection Agency, Proposed Guidelines for Assessing Female Reproductive Risk, 53 Fed. Reg. 24834, 24836 (1988) (EPA consistently relies on “evaluation of toxicological data from humans and experimental animals“ in assessing reproductive and developmental risks.); U.S. Occupational Safety and Health Admin., Final Standard for Occupational Exposure to Ethylene Oxide, 49 Fed. Reg. 25734, 25743 (1984) (OSHA ruling rested on a “comprehensive review of the scientific evidence … based on information from many investigations in several species of experimental animals … as well as positive results from several human studies.”); U.S. Occupational Safety and Health Admin., ‘ Final Rule for the Identification, Classification, and Regulation of ’Potential Occupational Carcinogens, 45 Fed. Reg. 5002, 5040-59 (1980) (requiring data from other human studies or from experimental studies in test animals).

56. Banal as it may sound, the proffer of an expert witness’ opinion in a toxic tort case neither presupposes nor guarantees actual knowledge. Moreover, status does not imply knowledge as it does, for instance, in a medical malpractice action (a board-certified orthopaedic surgeon is presumed to have knowledge regarding orthopaedic surgery).

57. Truth, with respect to causal statements in science that are derived directly from controlled scientific studies, is a measure of the degree to which an effect measured in the study is attributable to the specific factor that was controlled. The degree of truth, or certainty, can be expressed on a numerical scale between O and 100%. By convention, a level of 95% is accepted as practical certainty. The degree of certainty can be specified numerically for individual scientific studies, but not for the causal conclusions directly pertinent to legal questions in toxic torts, because such conclusions are not directly based on scientific studies, but rather on inferences made from such studies.

58. See supra note 57.

59. Among the most important subjective considerations are the choice of studies to be considered and the weight to be given to individual studies. Additional considerations may be appropriate (whether the statistical test used in a particular study was conducted incorrectly) or inappropriate (the results of the study are adverse to the client’s interests).

60. For example, consider the assertion that “bouncing a basketball can cause it to pass through the floor of the basketball court.” While such a result is bizarre, according to the laws of modern physics there is an infinitesimally small but non-zero possibility that such an event could happen.

61. It is worthwhile to consider the process of generalization in greater detail to emphasize its routineness and societal importance. For example, suppose that a drug company believed a particular drug would be an effective antibiotic. In preparing the drug for market, the company would ordinarily perform test-tube studies to establish the drug’s effectiveness, and thereafter would test it using an appropriate animal model, such as rats that had purposely been infected. The studies would ascertain whether, to a 95% certainty, animals that were infected and then given the drug faired better than infected animals that did not receive the drug. Several studies would ordinarily be performed to test the drug under different circumstances (for example, three kinds of microbes, each placed into different organs). When the data is presented to the FDA to justify human experiments using the drug, the company experts will opine that the animal studies indicate the likelihood of efficacy of the drug in human subjects. That is, based on a consideration of the experimental design, the animal species used, and the unlikeliness that the observed correlation was due to chance, the company experts will conclude that the drug will likely be effective in human patients. The FDA may accept or reject this generalization. In the latter case, more animal studies may be required to permit rationalization of the drug company’s desired inference that the drug is likely to be effective in patients. No amount of animal studies could ever permit a stronger inference, such as “very likely” (merely a semantic, not a measurable scientific difference from the original opinion), or “certain.” Additional studies might give particular scientists more confidence in a “likely” conclusion, but this is a judgment upon which individual scientists may differ—in the present example, the FDA’s opinion, which would be determinative. Assuming the requisite number of animal studies needed to justify “likely” had been performed, the FDA staff would grant the drug company permission to use the drug in a limited human study (that is, under particular circumstances and in specific types of patients at specific institutions) for the purpose of scientifically testing the drug company’s belief that the drug is effective. Following completion of these studies, if the drug company experts observe a statistical association between use of the drug and clinical improvement, the FDA will consider allowing the drug to be marketed, not because it will help a particular patient or because it is likely to help, but because it might help individual patients. In other words, the FDA staff will permit use of the drug for the treatment of infection if the human studies show that a likelihood that at least some patients who receive the antibiotic will improve, compared with the fate of the same subjects if they either had not been treated, or had been treated with a standard drug (actually, the applicable law provides that the new drug need only be as effective as the drug already in use). A parallel set of studies and analyses must be conducted to characterize and delimit the nature of side effects that likely will be associated with the drug.

The process of scientific generalization is routine in areas that directly affect society at large, and generally accepted rules and procedures exist for implementing a generalization. It is unnecessary to specifically discuss the rules here. It is sufficient to recognize that the rules exist. The purpose for which the rules are applied to the scientific data (to persuade FDA staff in this example or to persuade a jury in a toxic tort case) is irrelevant to the issue of the validity of the rules and procedures. For example, if four animal studies involving rats performed with a particular standard of care and analyzed with a specific statistical test are sufficient to form the basis for a generalization that the drug under study will likely be effective in patients, then a similar number of comparable studies can also form the basis for a generalization that the test agent will likely cause an undesirable effect in patients. The rules for generalization are the same in both cases. The expert witness must use these rules in a toxic tort case in generalizing or refusing to generalize. Application of he norms for scientific generalization cuts both ways in a toxic tort case. If a method of reasoning does not exist, an expert cannot validly create it in a courtroom, and if the process does exist, an expert cannot ignore it merely because its applicability would lead to a result adverse to his client’s interest.

62. See supra notes 20-21 and accompanying text.

63. For example, based on testimony that a police officer used a particular type of radar gun for a certain number of years, and that the gun was held in certain positions in particular circumstances for varying lengths of time during the typical work day, an expert could conclude the plaintiff was actually exposed to electromagnetic fields emanating from the gun. Additionally, based on measurement data, such as the electrical characteristics of the gun, the angle of the spread of the beam as it exited the barrel of the gun, and the amount of reflection the beam produced when it strikes metal and glass surfaces, the expert could estimate the dosage the plaintiff experienced.

64. HART & HONORÉ, supra note 3, at 428-30.

65. Daubert v. Merrell Dow Pharmaceuticals, Inc., 113 S. Ct. 2786 (1993).

66. Suppose an expert offers to opine “X can cause Y,” where “X” is a toxic agent and “Y” is the plaintiff’s disease. For example. “asbestos can cause cancer” or “Bendectin can cause birth defects.” One possible approach to the issue of reliability would be for the court to evaluate the scientific evidence and decide whether the inductive generalization was probably true according to applicable scientific principles, as the expert purports. An affirmative answer equates to a finding of reliability. But a court is incapable of determining whether cause-and-effect relationships reported in particular studies were validly inferred according to orthodox rules of scientific procedure and analysis because these matters are purely within the scientific domain. Therefore, only scientists can initially resolve such issues.

67. An example of a standard greater than reasonably possible is a determination by the court whether the testimony is likely to be true.

68. 293 F. 1013 (D.C. Cir. 1923).

69. 113 S. Ct. 2786 (1993).

70. Id. at 2795.

71. United States v. Baller, 519 F.2d 463, 466 (4th Cir. 1975), cert. denied, 423 U.S. 1019 (1975) (“[B]ecause of its apparent objectivity, an opinion that claims a scientific basis is apt to carry undue weight with the trier of fact.”); United States v. Addison, 498 F.2d 741, 744 (D.C. Cir. 1974) (“[S]cientific evidence may assume a posture of mystic infallibility in the eyes of a jury of laymen.”); United States v. Amaral, 488 F.2d 1148, 1152 (9th Cir. 1973) (noting scientific testimony has an “aura of special reliability and trustworthiness.”); D’Arc v. D’Arc, 385 A.2d 278 (N.J. Super. Ct. Ch. Div. 1978), aff’d, 421 A.2d 602 (N.J. Super. Ct. App. Div. 1980) (stating scientific evidence has an “aura of mystic infallibility.”), cert. denied, 451 U.S.971 (1981).

72. Frye v. United States, 293 F. 1013,1014 (D.C. Cir 1923).

73. Id. at 1013.

74. That is the inductive generalization.

75. Frye, 293 F. at 1014.

76. Id.

77. An article on the Frye rule published in 1980 did not mention civil cases. Paul C. Gianelli, The Admissibility of Novel Scientific Evidence: Frye v. United States a Hay Century Later, 80 COLUM. L. REV. 1197 (1980). Frye was not applied by a federal appellate court to exclude testimony until 1984. Barrel of Fun, Inc. v. State Farm Fire & Casualty Co., 739 F.2d 1028, 1031 (5th Cir. 1984).

78. Historically, the absence of acute effects nearly always provided the basis for regarding toxic agents as safe. It is frequently impossible, however, to evaluate causes of chronic effects based on data related to acute effects. In Ferebee v. Chevron Chem. Co., the court recognized the distinction between acute and chronic effects of toxic chemicals by refusing to exclude the plaintiffs’ expert testimony on chronic effects regardless of defendants’ ability to prove the plaintiff could not have suffered acute effects. 736 F.2d 1529, 1536 (D.C. Cir.), cert. denied, 469 U.S. 1062 (1984).

79. For example, when testimony was first given that electromagnetic fields from powerlines could cause disease it was true that the theory was not “generally accepted.” See Minutes of Public Hearing from Public Service Commission of New York, Common Record Hearings on the Health and Safety of 765 kV Transmission Lines, Cases 26529 & 26559 (Oct. 1974) (pre-filed testimony of Robert O. Becker, M.D.). Subsequently, in legal actions founded on the theory of harm induced by electromagnetic fields, the adverse party sought to exclude expert testimony under the Frye rule.

80. Not surprisingly, no court has provided pertinent guidance, even though particular choices could greatly enhance the chances of finding “general acceptance.” For example, radar guns are generally accepted as safe by scientists who are employed by radar-gun manufacturers.


82. Contrary testimony has been presented in cases involving asbestos, Agent Orange, PCBs, cigarettes, drugs, and electromagnetic fields.

83. Generally, the defendant advances the affirmative defense that a controversy over “general acceptance” exists.

84. The issue was argued extensively prior to Daubert. In the following cases the expert was allowed to introduce a scientific principle in court. The court assessed reliability by determining how the expert arrived at the opinion. See United States v. Jakobetz, 955 F.2d 786 (2d Cir. 1991), cert. denied, 121 L. Ed. 2d63 (1992); United States v. Williams, 583 F.2d 1194 (2dCir. 1978), cert. denied, 439 U.S. 1117 (1979); DeLuca v. Merrell Dow Pharmaceuticals, Inc., 911 F.2d 941 (3d Cir. 1990), cert. denied, 114 S. Ct. 691 (1994); United States v. Ferri, 778 F.2d 985 (3d Cir. 1985), cert. denied, 476 U.S. 1172 (1986); United States v. Downing, 753 F.2d 1224 (3d Cir. 1985); Clinchfield R.R. v. Lynch, 784 F.2d 545 (4th Cir. 1986); United States v. Baller, 519 F.2d 463 (4th Cir. 1975), cert. denied, 423 U.S. 1019 (1975); Spryncynatyk v. General Motors,771 F.2d 1112 (8thCir.1985), cert. denied, 475 U.S. 1046 (1986); United States v. Luschen, 614 F.2d 1164 (8th Cir. 1980), cert. denied, 446 U.S. 817 (1980); United States v. Bennett, 539 F.2d 45 (10th Cir. 1976), cert. denied, 429 U.S. 925 (1976); Mustafa v. Brown, 22 M.J. 165 (C.M.A.), cert. denied, 479 U.S. 953 (J986); Carter v. St. Vincent Infirmary, 690 S.W.2d 741 (Ark. Ct. App. 1985); State v. Hall, 297 N.W.2d 80 (Iowa Ct. 1980), cert. denied, 450 U.S. 927 (1981); Andrews v. State, 533 So.2d 841 (Fla. Dist. Ct. App. 1988); State v. Williams, 388 A.2d 500 (Me. 1978); State ex rel. Elg v. Erickson, 363 N.W.2d 859 (Minn. Ct. App. 1985); Barmeyer v. Montana Power Co., 657 P.2d 594 (Mont 1983); State v. Dorsey, 539 P.2d 204 (N.M. 1975); Minot Sand & Gravel Co. v. Hjelle, 231 N.W.2d 716 (N.D. 1975); State v. Johnston, 529 N.E.2d 898 (Ohio 1988); State v. Brown, 687 P.2d 751 (Or. 1984); State v. Walstad, 351 N.W.2d 469 (Wis. 1984).

The following are cases in which the court required prior to trial that the scientific principle be introduced by other scientists and to be generally accepted by scientists. The court deferred to its own perception of what scientists believe or accept. United States v. Shorter, 809 F.2d 54 (D.C. Cir.), cert. denied, 484 U.S. 817 (1987); United States v. McDaniel, 538 F.2d 408 (D.C. Cir. 1976); United States v. McBride, 786 F.2d 45 (2d Cir. 1986); Christophersen v. Allied-Signal Corp., 939 F.2d 1106 (5th Cir. 1991), cert. denied, 503 U.S. 912 (1992); Barrel of Fun, Inc. v. State Farm Fire & Casualty Co., 739 F.2d 1028 (5th Cir. 1984); United States v. Kozminski, 821 F.2d 1186 (6th Cir. 1987), aff’d, 487 U.S. 931 (1988); United States v. Distler, 671 F.2d 954 (6th Cir. 1981), cert. denied, 454 U.S. 827 (1981); United States v. Brady, 595 F.2d 359 (6th Cir. 1979), cert. denied, 444 U.S. 862 (1979); United States v. Smith, 869 F.2d 348 (7th Cir. 1989); United States v. Carmel, 801 F.2d 997 (7th Cir. 1986); Daubert v. Merrell Dow Pharmaceuticals, Inc., 951 F.2d 1128 (9th Cir. 1991), vacated and remanded, 113 S. Ct. 2786 (1993); United States v. Kilgus, 571 F.2d 508 (9th Cir. 1978); Lynn v. Helitec Corp., 698 P.2d 1283 (Ariz. Ct. App. 1984); People v. Wochnick, 219 P.2d 70 (Cal. Ct. App. 1950); KN Energy, Inc. v. Great Western Sugar Co., 698 P.2d 769 (Colo. 1985), cert. denied, 472 U.S. 1022 (1985); State v. Atwood, 479 A.2d 258 (Conn. Super. Ct. 1984); State v. Marks, 647 P.2d 1292 (Kan. 1982); Reed v. State, 391 A.2d 364 (Md. 1978); People v. Pullins, 378 N.W.2d 502 (Mich. Ct. App. 1985); State v. Danielski, 350 N.W.2d 895 (Minn. Ct. App. 1984); State v. Maule, 667 P.2d 96 (Wash. Ct. App. 1983); State v. Bohner, 246 N.W. 314 (Wis. 1933).

85. Daubert v. Merrell Dow Pharmaceuticals, Inc., 113 S. Ct. 2786 (1993).

86. See id. at 2791.

87. Id. at 2791-92.

88. The FDA was required by law to consider all the pertinent research, some of which the pharmaceutical industry produced.

89. Brock v. Merrell Dow Pharmaceuticals, Inc., 874 F.2d 307, 310 (5th Cir. 1989).

90. Daubert, 113 S. Ct. at 2791-92.

91. Id. at 2791.

92. Daubert v. Merrell Dow Pharmaceuticals, Inc., 727 F. Supp. 570 (S.D. Cal. 1989).

93. Daubert v. Merrell Dow Pharmaceuticals, Inc., 951 F.2d 1128 (9th Cir. 1991).

94. Daubert, 113 S. Ct. at 2792.

95. Over the objections of Chief Justice Rehnquist and Justice Stevens, the Court made “general observations” regarding the characteristics of “scientific knowledge,” and it listed two factors that were discussed supra. Id. at 2796. First, the technique or theory being offered by the expert should have been tested. Id. at 2796-97. In practical terms, this means the induction itself should have been considered by experts who then prospectively conducted experiments to test the theory. Id. Second, the existence of peer review is a factor that should be considered in the preliminary hearing to determine whether an opinion is scientific knowledge. Id. at 2797.

The remaining two factors the Court mentioned were unartfully described, and are not scientifically significant in comparison with the other two factors discussed in the opinion and the other factors described supra that were not mentioned by the Court. The Court’s third factor was “the known or potential error rate.” Id. In the context of causal analysis, error refers to statistical probabilities. See WILLIAM MENDENHALL, INTRODUCTION TO PROBABILITY AND STATISTICS 202-03 (7th ed. 1987). In the context of scientific measurement, error refers to precision and accuracy. See id. Since the most common substantive characteristic of the peer-reviewed scientific literature is the use of statistics to analyze data, it is unlikely that proffered testimony satisfying the Court’s first two factors would fail to meet its third factor. Finally, although the Court said that “general acceptance” cannot serve as a rule of exclusion, it suggested that it might continue as a rule of inclusion. Daubert, 113 S. Ct. at 2799. But it is difficult to visualize a situation in which this suggestion would have any practical meaning. If the theory was disputed by the parties, then the applicable test would be “scientific … knowledge;” otherwise, the theory could be put into evidence via judicial notice. See id at 2795. Consequently, there seems to remain little room for an independent rule of inclusion.

The intrinsic validity of scientific knowledge derives from the method by which the knowledge was inferred, and is independent of the purpose for which it was inferred (the motivation of the investigators), and for which it may be employed. Compare with “scientific validity for one purpose is not necessarily scientific validity for other. unrelated purposes,” where the Court majority confused validity with relevance. Daubert, 113 S. Ct. at 2796.

96. See infra notes 99-102 and accompanying text for a discussion of the admissibility of expert testimony under the hearsay rule.

97. United States v. Baller, 519 F.2d 463, 466 (4th Cir.), cert. denied, 423 U.S. 1019 (1975) (“[B]ecause of its apparent objectivity, an opinion that claims a scientific basis is apt to carry undue weight with the trier of fact.”); United States v. Addison, 498 F.2d 741, 744 (D.C. Cir. 1974) (“[S]cientific evidence may assume a posture of mystic infallibility in the eyes of a jury of laymen.”); United States v. Amaral, 488 F.2d 1148, 1152 (9th Cir. 1973) (noting scientific testimony has an “aura of special reliability and trustworthiness”); D’Arc v. D’Arc, 385 A.2d 278 (N.J. Super. Ct. Ch. Div. 1978), aff’d, 421 A.2d 602 (N.J. Super. Ct. App. Div. 1980) (noting scientific evidence has an “aura of mystic infallibility”), cert. denied, 451 U.S. 971 (1981).

The supposed aura of infallibility of scientific evidence and its consequent impact on a lay jury, notwithstanding the fact that the evidence might be specious, is often cited in arguments for a relatively high judicial threshold for admissibility of scientific testimony. The perceived abuse, however, would not be remedied by raising the threshold for admission because a judge is also a layman with respect to science. Hence, a judge is also susceptible to the layman’s perception of infallibility. If an expert undeservedly creates an aura of infallibility, the fault resides with the opposing counsel for failing to make effective use of the panoply of powerful tools available to the cross-examiner.

It may be worthwhile to consider the situation from the viewpoint of the expert. An expert witness is a highly, but narrowly, educated specialist called upon to translate knowledge from an arcane scientific specialty to the real world of ordinary people so that its potential importance can be evaluated. Frequently, scientists perceive the legal system as unfamiliar and intimidating, with its complex rules, non-intuitive procedures and central authority figure who exercises a near dictatorial control over events in the courtroom.

98. Essentially every toxic tort case involves two schools of thought regarding the scientific evidence, one of which is generally favorable toward each party. Science on the opposing side of the dispute is called “junk science” to distinguish it from the “good science” advanced by the other side, thereby illustrating the contentious nature of science; “good science” is my science, and “junk science” is the other guy’s science. Science probably never deserved a free ride with regard to determining what the law should accept as truth, even if it did, that day has passed.

The writings of Peter W. Huber indirectly promote the idea that scientific issues having economic consequences are inherently controversial. PETER W. HUBER, GALILEO’S REVENGE: JUNK SCIENCE IN THE COURTROOM (1991). Huber popularized the term “junk science.” Id. Huber argued that tort law, particularly toxic torts, had run amok and was becoming a system for arbitrary redistribution of wealth. He opposed the use of science in the courtroom unless the scientific issues had been completely settled by the scientists themselves (because, otherwise, it was “junk”). Id. Huber’s thesis was based on his economic and political views, not on legal or scientific analysis or fact. Kenneth J. Chesebro, Galileo ’s Retort: Peter Huber’s Junk Scholarship. 42 AM. U. L. REV. 1637 (1993). Huber’s writings emphasized the controversial nature of science, and this tended to bring about exactly what he opposed, namely recognition of the need for an expanded role for the courts.

99. Rule 810(c) of the Federal Rules of Evidence defines hearsay as “ a statement, other than one made by the [person making the statement] while testifying at [a] trial or hearing, offered into evidence to prove the truth of the matter asserted.” A statement is defined as an &ddquo;oral or written assertion or … [the] non-verbal conduct of a person, if it is intended by the person as an assertion.” FED. R. VID. 810(a).

100. See supra notes 47-52 and accompanying text.

101. Blue-ribbon committees are usually created by an agency or organization with a vested interest in a particular solution to a scientific question. This interest may influence the choice of the individual committee members, resulting in a quasi judicial body that lacks the detachment and independence of a court. It is not surprising that particular organizations will attempt to portray controversial scientific issues in a light most favorable to their interests. It cannot be plausibly maintained that the work product resulting from such efforts is unbiased, and should therefore be accepted by courts. Blue-ribbon committees often represent the opinion of one set of interests. Although those represented may be the most powerful or prominent in the industry, it does not automatically follow that the opinion is the common wisdom within the area, the best grounded in scientific facts and knowledge, or that which best serves the interests of society as a whole.

Furthermore, not all members of blue-ribbon committees are intellectually free to analyze pertinent data and reach an appropriate conclusion based solely on the scientific data. An employee of a particular industry would not normally be expected to enjoy the prerogative of freely commenting on scientific evidence when acting as a member of a blue-ribbon committee. Rather, the employee would be expected to represent the company’s interest on the committee, which normally includes non-scientific considerations. Similarly, a government employee lacks academic freedom as a member of a blue-ribbon committee because such an individual is bound by the views of his agency. A blue-ribbon committee member who is an employee of a governmental organization would be expected to advance the agency’s view during committee deliberations. These views necessarily incorporate value judgments and political considerations in formulating any position, including positions involving potentially controversial scientific matters.

102. For examples of judicial acceptance of opinions of other experts see Johnston v. United States, 597 F. Supp. 374, 410-11 (D. Kan. 1984) (stating that plaintiff’s experts are not credible because they disagree with a blue-ribbon committee that contained “the most eminent scientists in the radiation community”); In re Agent Orange Prod. Liab. Litig., 611 F. Supp. 1223, 1240 (E.D.N.Y. 1985) (stating that government studies are reliable and impartial), aff ’d, 818 F.2d 187 (2d Cir. 1987), cert. denied, 487 U.S. 1234 (1988).

103. Where “X” is the toxic agent in the case, “x” is the dose received by the plaintiff, “Y” is the type of disease manifested by the plaintiff, and “y” is the plaintiff’s disease.

104. Acceptability of scientific data by scientists is determined by standards involving compliance with various methodological and statistical rules. See supra notes 34-45 and accompanying text. In contrast, acceptability of scientific data by laymen is determined by an examination of the circumstances attendant its production, distribution, and general use within the scientific community. If a study were of a type routinely performed by the investigator irrespective of considerations involving toxic tort cases, the study methods and statistical procedures were both routine and common within the specialty, and the results of the experiment were published in the open peer-reviewed scientific literature, the trier of fact would be justified in accepting an expert’s confidence in the published data. The plaintiff must provide such evidence in order to carry his burden of proof with respect to the assertion “X can cause Y.”

105. An expert may opine that many, or all, of the scientific studies relied upon by another expert were inferior, defective, or otherwise not suitable as a basis for generalization. Such conflicting opinions raise questions of fact and credibility. Suppose, for example, the defendant’s expert uncovered a fatal defect in the statistical analysis of a study relied upon by the plaintiff’s expert. In such a case, one element of the plaintiff’s expert’s opinion is refuted because he relied on faulty data. Additionally, the credibility of the plaintiff’s expert concerning his opinions of other studies might be diminished because the trier of fact could reasonably conclude that if the expert relied on one study that was seriously flawed, he could have done so in other instances.

106. Minutes of Public Hearing from Public Service Commission of New York, Common Record Hearing on the Health and Safety of 765 kV Transmission Lines, Cases 26529 & 26559 (1976) (testimony of Hermann Schwan at 6731).

107. Id.

108. Id.

109. The expert further testified that when the experiment concluded that there was no effect, “I was not further interested in digging into the material.” Id.

110. For example, in a proceeding involving the health hazards of high-voltage powerlines, an epidemiologist working for a consulting company hired by the defendant power company filed a detailed report that analyzed studies of the health hazards of powerlines, and exonerated them as a possible cause of disease. See Andrew A. Marino, Editorial, Trust Me, I’m a Doctor, 8 J. BIOELECTRICITY v-vi (1989). The expert’s knowledge regarding the subject of her testimony, however, was acquired within the six-week period between the time she began working for the company and the day she testified, and her report was similar to that provided in other forums by other company experts. Id.

111. Some courts have held that only epidemiological studies may be used to assess the causal role of toxic agents in human disease. See. e.g., Ealy v. Richardson-Merrell, Inc., 897 F.2d 1159,1161-62 (D.C. Cir.), cert. denied, 498 U.S. 950 (1990); Brock v. Merrell Dow Pharmaceuticals, Inc., 874 F.2d 307, 313 (5th Cir. 1989), cert. denied, 494 U.S. 1046 (1990); Richardson v. Richardson-Merrel, Inc., 857 F.2d 823, 830-31 (D.C. Cir. 1988), cert. denied, 493 U.S. 882 (1989); In re Agent Orange Prod. Liab. Litig., 611 F. Supp. 1223, 1240-41 (E.D.N.Y. 1985), aff’d, 818 F.2d 187 (2d Cir. 1987), cert. denied, 487 U.S. 1234 (1988). In Dauber v. Merrell Dow Pharmaceuticals, Inc., the respondent urged this serious error on the Supreme Court (see Respondent’s Brief at 41-44) but the Court rejected the concept in favor of the “scientific … knowledge” standard, which includes but is not limited to epidemiological knowledge. 113 S. Ct. 2786 (1993).

112. Why are animal studies ethically preferable? The possible types of relevant studies are animal studies, human experiments (studies employing experimental designs normally used with laboratory animals), and epidemiological studies (human studies in which statistical associations are sought in the absence of investigator control over the behavior of the study subjects). A human experiment could test a hypothesis that a toxic agent causes disease. In such a study, healthy subjects would be randomly assigned to either the exposed or control groups, and the percentage of subjects in each group that developed disease would be compared. Obvious ethical and practical factors prevent such a study: (I) no authority exists which is capable of assigning subjects to particular groups; (2) the putative study is inherently costly because great numbers of subjects must be followed to ascertain the existence of a relatively small number of subjects that develop the disease; (3) it is impossible to maintain the comparability of the two groups because the daily activities of the subjects cannot be controlled for the time needed to perform the study.

Because of these difficulties, the less logically powerful but more practical designs for epidemiological studies were developed. See discussion infra Appendix, at 58-62. It is reasonable to perform epidemiological studies to gain insight into whether human beings have suffered disease as a result of an agent whose danger was, in good faith, not initially appreciated by the party responsible for its dissemination. It would be quite a different matter, however, to simply presume that an agent has no long-term side effects, with that presumption tested only retrospectively, if at all, in naive and nonconsenting subjects. Such premeditated reliance on epidemiological studies to assess toxic effects is unethical because it amounts to using human beings as subjects in scientific experimentation without their informed consent. The applicable ethical principle is that of personal autonomy: each person possesses a set of individual rights, and among them is the right to determine how and in what manner one’s own body will be used or employed. This principle should ordinarily exclude a purposeful reliance on epidemiological studies as the primary means for evaluating the risks due to toxic agents. Additionally, it is obviously better to ascertain whether agents are harmful before human beings suffer and die as a result of exposure to the agents. Thus, (1) use of animals is the ethically preferable method for generating scientific knowledge pertinent to toxic tort cases, and (2) the animal studies ought to be performed prior to use of the toxic agent.

113. For the same reasons, animal studies often employ high doses, as compared with human exposure to determine whether the agent could have any effect on human beings.

114. Biological effects in test animals are probative with regard to the principal inductive opinion because all disease-causing agents cause biological changes that do not themselves indicate the presence of cancer. It could be reasoned that, because the putative toxic agent was capable of causing various immunological, hematological, endocrine, and other types of changes in test animals, and since such changes are the prius of the cancerous state, the animal studies are probative with regard to the expert’s principal inductive opinion. If the law is to permit the use of animal research in toxic tort cases, it must sanction scientific reasoning from biological effects produced in animals that are something other than the type of disease manifested by the plaintiff.

115. Zappavigna v. New York, Claim No. 74085, at 80 (N.Y. Ct. Cl. 1988) (testimony of Richard Bockman, Oct. 11, 1988) (stating that powerlines are safe based on animal studies) (unreported).

116. Id. at 5 (testimony of Margaret Tucker, Oct. 1 3, 1 988) (powerlines are safe based on epidemiological studies).

117. Id.

118. See supra note 61 and accompanying text for a description of the approval process.

119. The same scientific rules of interpretation and extrapolation apply to other animal studies to determine whether an inference of harm, under other circumstances, is warranted.

120. Minutes of Public Hearing from Public Service Comm’n of New York, Common Record Hearing on the Health and Safety of 765 kV Transmission Lines, Cases 26529 & 26559, at 5921 (Apr. 28, 1976) (testimony of Morton Miller); see also MARINO & RAY, supra note 41, at 41.

121. The epidemiological studies linking cigarettes and cancer do not list the actual cigarette brands. The epidemiological studies linking electromagnetic fields and cancer do not (for the most part) identify the frequency of the field as being that of powerlines, broadcast towers, cellular telephones, or radar. It would be result-oriented advocacy to argue that only evidence that a specific brand, for instance Camels, or that a specific frequency, for instance radar, can cause cancer should be considered with regard to the link with cancer because there is no proper scientific basis for the distinction. Hutchison v. Kustom Signals, Inc., Civ. No. C91 1174 BAC (N. D. Cal. May 29, 1992) (deposition of Linda Erdreich, at 12) (arguing that only data from epidemiological studies of police radar guns (of which none exist) are relevant to the question whether they can cause cancer) (unreported).

122. Rausch v. School Board of Palm Beach County, Civ. No. CL 8810772 AD (D Fla. 1989) (testimony of Phillip Cole), aff’d, 582 So. 2d 631 (1991). For further details see Andrew A. Marino, Negative Studies and Common Sense, 8 J. BIOELECTRICITY v (1989). Courts have frequently relied on the volume of negative data to reject the existence of an asserted causal link. See, e.g., Richardson v. Richardson-Merrell, Inc., 857 F.2d 823, 831 (D.C. Cir. 1988), cert. denied, 493 U.S. 882 (1989); Ealy v. Richardson-Merrell, Inc., 897 F.2d 1159, 1162 (D.C. Cir. 1990). But see Ongmore v. Merrell Dow Pharmaceuticals, Inc., 717 F. Supp. 1117, 1120 (D. Idaho 1990) (holding that the court’s focus should not be on the number of negative studies but on the soundness of the methodology employed by plaintiff’s expert).

123. See Kenneth R. Foster et al., Science and the Toxic Tort, 261 SCI. 1509 (1993). The authors argued that it would be unreasonable to conclude that video display terminals or Bendectin could cause disease because “the epidemiologic evidence regarding miscarriage and the use of video display terminals or birth defects and the morning sickness drug Bendectin includes a sprinkling of positive results in a body of overwhelmingly negative findings.” Id. But overall conclusions cannot be reached simply by counting and classifying studies—the studies themselves must be considered. If there were only one positive study involving video display terminals, it would then be correct to conclude that they can cause disease. Suppose, further, the negative studies were performed by scientists working for the manufacturers of video display terminals. It would be a question of fact whether the studies could reasonably be relied upon. This example illustrates both the futility of an argument based on counting studies, and the excessive naivete inherent in the failure to consider the origins of the studies relied upon.

124. See, e.g., Alabama Power Co. v. Western Pocahontas Props., No. CU88-676 (Ala. Cir. Ct. Apr. 17, 1992) (deposition of Mary Ellen O’Connor, at 213) (stating that knowledge of underlying mechanisms is required to show a causal relation between electromagnetic fields and health risks) (unreported).

125. For some toxic agents, such as pesticides, human risks must be estimated from animal studies employing dosages that far exceed those ordinarily encountered in the environment because low-dose animal studies are incapable of yielding biological effects sufficient to be observable over a reasonable period of time. But other toxic agents produce biological effects at doses far less than those routinely present in the environment. For example, electromagnetic fields far weaker than those produced by ordinary powerlines altered calcium levels in animal brains, affected human body rhythms, human brain electrical activity, and increased the risk of childhood leukemia. See S.M. Bawin & W.R. Adey, Sensitivity of Calcium Binding in Cerebral Tissue to Weak Environmental Electric Fields Oscillating at Low Frequency, 73 PROC. NAT. ACAD. SCI. 1999 (1976); R. Wever, The Effects of Electric Fields on Circadian Rhythmsin Man, 8 LIFE SCI. SPACE RES.177(1970); Glen Bell et al., Human Sensitivity to Weak Magnetic Fields, 338 LANCET 1521(1991); Nancy Wertheimer & Ed Leeper, Electrical Wiring Configurations and Childhood Cancer, 109 AM. J. EPIDEMIOL. 273 (1979).

126. For example, all members of the public are exposed to electromagnetic fields. Thus, the relationship between the level to which the plaintiff was exposed and the average exposure level experienced by the public is a consideration in a suit alleging injury due to field exposure. When the difference between the two is large, the possible causal role of background exposure can be rejected. See, e.g., DONALD L. LAMBDIN, U.S. ENVTL. PROTECTION AGENCY, AN INVESTIGATION OF ENERGY DENSITIES IN THE VICINITY OF VEHICLES WITH MOBILE COMMUNICATIONS EQUIPMENT AND NEAR A HAND-HELD WALKIE TALKIE, ORP/EAD-79-2 (1979) (available from U.S. Environmental Protection Agency, Office of Radiation Programs, Electromagnetic Radiation Analysis Branch, P.O. Box 15027, Las Vegas, NV 89114); RICHARD A. TELL & EDWIN D. MANTIPLY, U.S. ENVTL. PROTECTION AGENCY, POPULATION EXPOSURE TO VHF AND UHF BROADCAST RADIATION IN THE UNITED STATES, ORP/EAD-78-5 (1978) (showing that a typical walkie-talkie or cellular telephone user is exposed to a dosage 100,000 times greater than background) (available from U.S. Environmental Protection Agency, Office of Radiation Programs, Electromagnetic Radiation Analysis Branch, P.O. Box 15027, Las Vegas, NV 89114).

127. For example, if the plaintiff lived beside a high-voltage powerline, operated a radar gun, smoked, had extensive exposure to diagnostic x-rays, and worked in the petrochemical industry, the resulting Gordian knot of toxic exposures might obscure the extent of legal liability of particular parties.

128. Oncology is the treatment of various forms of cancer using drugs.

129. Although the studies’ durations were short compared with typical human lifetimes, the affiant states that since they represent typical lifetimes for the respective species, they are comparable to human lifetimes. In other words, exposure for one year in rodents or two years in rabbits is equivalent to lifetime exposure in human beings.

130. The studies performed on the tissues of the exposed animals were highly technical in nature. For example, they included measurement of the amount of acetylcholine released at the neuromuscular junction, and the blood and tissue levels of various substances termed “cytokines.” Apparently, normal levels for the different parameters have been established, and although it is possible to assess whether deviations from normal have occurred, frequently it is not possible to objectively characterize the change as good or bad—only that it differs from the norm. Since blue dye #2 is not supposed to produce any biological changes in the person wearing socks that were colored with the dye, any change that it does produce must be presumed to be adverse. Therefore any predicate changes in the animals must also be viewed as being adverse.

131. See Ayers v. Township of Jackson, 525 A.2d 287, 301-02 (N.J. 1987) (defining “cause” in relation to cancer).

132. The definitions are not meant to be exhaustive, but rather to convey the essential notion that is denoted when the words are used in the text (unless the context obviously requires a different meaning).

133. This method of scientific decision-making is known as significance testing at the 95% level. Less frequently employed methods include the use of confidence intervals, meta-analysis, Bayesian analysis, and significance testing using levels less than 95%. Significance testing at the 95% level is the established and generally applicable norm for scientific reasoning in present-day science, but one or more of the various alternatives may be appropriate in particular cases. Courts have sometimes accepted alternatives to standard significance testing. See In re Paoli R Yard PCB Litig., 916 F.2d 829, 857 (3d Cir. 1990) (accepting testimony based on meta-analysis), cert. denied, 494 U.S. 1046 (1990); Brock v. Merrell Dow Pharmaceuticals, Inc., 874 F.2d 307, 311-12 (5th Cir. 1989) (accepting confidence-interval testing), cert. denied, 499 U.S. 961 (1991).

134. There are many possibilities in a scientific study for bias to occur, even when it would not be reasonable to expect that a prudent scientist would have recognized the problem initially and changed the procedure to eliminate it. Thus, scientific bias may involve a kind of negligence, but it does not involve intent. Intentional bias, in contrast, is a species of fraud; it is an attempt to shape the inference of a study so that some desired conclusion is reached, irrespective of that which naturally flows from the data. Suppose, for example, an investigator conducted two experiments, one of which supported a particular hypothesis and one of which did not. If the results of only the first were disseminated, the inference drawn within the scientific community would be quite different than if both studies were published. This discussion focuses solely on scientific bias.

135. This is similar to the process of distinguishing prior cases in law that are apparently similar to the case at bar. There are nearly always differences that may arguably justify or require a contrary view.

136. Three types of epidemiological studies are important for toxic tort cases. In a case-control study, subjects having the disease chosen for study are identified, and the proportion of the diseased subjects that were exposed to the toxic agent is determined. A control group is chosen (some of whom, unknown to the investigator, may also have been exposed to the toxic agent), and the proportions of exposed subjects in the two groups are compared to determine whether those who had the disease were more likely to have had exposure to the toxic agent. If the control subjects were disease-free, then the hypothesis tested when the data was subjected to the appropriate statistical test would be whether exposure was more likely among diseased subjects compared with healthy subjects; thus, the tendency of the toxic agent to cause disease in healthy subjects—which is the basic issue of concern—would be assessable. If the study subjects had a particular disease, such as leukemia, and the control subjects had non-leukemia cancer, then the hypothesis actually tested would be whether exposure was more likely among leukemia subjects compared with subjects having other forms of cancer.

A proportional mortality (or morbidity) study (PMR) permits a determination of whether a particular disease was more likely among dead exposed subjects, than among dead subjects generally. Since a PMR study includes only dead subjects, and not subjects who were at risk of dying, no direct inferences are possible regarding similarly exposed but healthy subjects. A single PMR study is therefore capable of justifying only the conclusion that an association between a toxic agent and a disease was stronger than the association between the agent and other diseases. The standardized mortality (or morbidity) study (SMR) is a third practical design. Subjects exposed to a toxic agent are identified and the proportion that developed a particular disease is determined. Comparison with the corresponding proportion in the control group permits assessment of whether the disease was more likely among the exposed subjects. Unfortunately, an unascertained number of subjects in the control group usually will also have been exposed to the toxic agent, and those subjects might contribute disproportionately to the fraction of the control group that develop disease, thereby complicating interpretation of the results. The inherent limitations of PMR and SMR studies render them less useful than case-control studies for evaluating risks of toxic agents; their major advantage is that they are usually much cheaper and easier to perform.

137. Suppose the control group was disease free and it is found that those who had the disease were more likely to have had exposure to the toxic agent: What conclusions follow? The authors will argue that since the agent and the disease were associated, the result suggests the possibility of a causal link, and therefore that more studies are warranted. Those opposed to the authors will identify a possible bias and argue that the two groups were not properly comparable as a result of the presence of the bias, and therefore that there is no evidence of an association, and consequently no suggestion of a causal link. Thus, they will argue, there is no evidence rendering it proper or beneficial to inquire further into the existence of such a link.

138. In contrast, one laboratory study would be sufficient to support such an inference, although more than one would be needed before the result could be generalized and used to support a medical, business, or regulatory action.


HOME | Contact Marino |