Hostname: page-component-cd9895bd7-8ctnn Total loading time: 0 Render date: 2024-12-25T07:10:13.442Z Has data issue: false hasContentIssue false
  • English
  • Français

Science for the Environment: Examining the Allocation of the Burden of Uncertainty

Published online by Cambridge University Press:  20 January 2017

Elisa Vecchione
Elisa Vecchione*
Affiliation:
de Développement Durable, Sciences Po, Paris, Email:
Get access Rights & Permissions [Opens in a new window]

Extract

The aim of this paper is to review the basic literature on scientific uncertainty in its statistical paradigm in order to provide enlightenment on one pivotal facet of the precautionary principle, i.e. the allocation of the burden of proof to demonstrate that an activity is not harmful to the environment. The purpose is not to explain a new theory of statistical inference, but to show how regulatory policymaking that is properly informed by scientific expertise and designed to avoid one type of error, may actually make other errors more likely and thus expose the public to danger. This problem is explained in terms of the conceptual as well as operational conflicts that arise when knowledge about statistical-inferential methods is applied to policymaking. The paper argues that this issue can be resolved by first reconsidering the burden of proof as a burden of uncertainty.

Type
Articles
Information
European Journal of Risk Regulation , Volume 2 , Issue 2 , June 2011 , pp. 227 - 239
Copyright
Copyright © Cambridge University Press 2011

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

References

1 WTO Panel Report, 2006, European Communities – Measures Affecting the Approval and Marketing of Biotech Products, WT/ DS291/R, WT/DS292/R, WT/DS293/R, Corr.1 and Add.1, 2, 3, 4, 5, 6, 7, 8 and 9, adopted 21 November 2006.

2 Genetically Modified Organisms are products that have been altered using recombinant DNA technologies. Drawing from the European legislation on the subject (Directive 2001/18/EC on the Deliberate Release into the Environment of Genetically Modified Organisms, OJ 2001 L 106/3, and Regulation (EC) No 1829/2003 on Genetically Modified Food and Feed, OJ 2003 L 268/1) GMOs pertain to two fundamental categories: food and feed products, consisting or made from GMO, no longer considered living organisms themselves and intended for consumption; crops or plants in a living form, capable of and intended for growing.

3 The points of opposition raised by those who contest the usefulness of the precautionary principle can be summarised as follows: current regulatory provisions are inherently precautionary; the precautionary principle advocates making decisions without adequate scientific knowledge; the risk in implementing it is that technological innovation could be undermined as far as development risks associated with it would challenge the outright proof of safety of a specific product ( Holm, S. and Harris, J., “Precautionary principle stifles discovery”, 400 Nature (1999), pp. 398 et sqq. CrossRefGoogle ScholarPubMed).

4 In the age of post-positivism, Thomas Kuhn became one of the leaders of a critical sentiment toward the untouchable notion of science as inherently capable of solving the world's problems. “Normal” was the science that prevailed over competing theories, hence setting the scientific paradigm, but was not completely successful in solving problems. Kuhn, Thomas S., The Structure of Scientific Revolutions, 3rd ed. (Chicago: University Of Chicago Press 1996)CrossRefGoogle Scholar.

5 See Jasanoff, Sheila S., “Contested Boundaries in Policy-Relevant Science”, 17 Social Studies of Science (1987), pp. 195 et sqq CrossRefGoogle Scholar.; Jasanoff, Sheila, Designs on Nature. Science and democracy in Europe and the United States (Princeton and Oxford: Princeton University Press 2005)Google Scholar; Roqueplo, Philippe, “Entre savoir et décision, l’expertise scientifique”, in Sciences en questions (Paris: INRA 1996)Google Scholar; Joly, Pierre-Benoit and Barbier, Marc, “Séparation de l’évaluation et de la gestion des risques. Les leçons de la ‘guerre du boeuf’”, Actes du Colloque “L’organisation du recours a l’expertise scientifique en situation d’incertitude” (2002), pp. 1 et sqq Google Scholar.

6 Roqueplo, “Entre savoir et décision”, supra note 5, at p. 3.

7 Funtowicz, Silvio O. and Ravetz, Jerome R., “The worth of a songbird: Ecological economics as a post-normal science”, 10 Ecological Economics (1994), pp. 197 et sqq CrossRefGoogle Scholar.

8 Katherine Barrett and Carolyn Raffensperger, “Precautionary Science”, p. 5, presentation held at the Wingspread Conference on Implementing the Precautionary Principle, 26 January 1998.

9 See section IV.3 on “Uncertainty, The Information Paradigm And Risk Acceptability” about the difference between information and knowledge, the former being progressive for a matter of accumulation, the latter indicating closured information.

10 See Anderson, David R., Burnham, Kenneth P. and Thompson, William L., “Null hypothesis testing: Problems, prevalence, and an alternative”, 64 Journal of Wildlife Management (2000), pp. 912 et sqq. CrossRefGoogle Scholar; Parkhurst, David F., “Statistical Significance Tests: Equivalence and Reverse Tests Should Reduce Misinterpretation”, 51 Bioscience (2001), pp. 1051 et sqq CrossRefGoogle Scholar.

11 For the sake of clarity, it should be specified that this action is not intended either to prove causation between events, nor to give a direct measure of, i.e. to quantify, the probability that the null hypothesis is true ( Adelman, David E., “Scientific activism and restraint: The interplay of statistics, judgment, and procedure in environmental law”, 79 Notre Dame Law Review (2004), pp. 497 et sqq. Google Scholar).

12 This criticism is clearly expounded in Shrader-Frechette, Kristin and Lemons, John, “Methodological Rules for Four Classes of Scientific Uncertainty”, in Lemons, John (ed.), Scientific Uncertainty and Environmental Problem Solving (USA: Blackwell Science 1996)Google Scholar; cf. also Kriebel, David et al., “The Precautionary Principle in Environmental Science”, 109 Environmental Health Perspectives (2001), pp. 871 et sqq CrossRefGoogle ScholarPubMed.

13 Shrader-Frechette, “Four Classes of Scientific Uncertainty”, supra note 12, at p. 6; Barrett and Raffensperger, “Precautionary Science” supra note 8, at p. 4; Kriebel et al., “The Precautionary Principle”, supra note 12, at p. 6.

14 Kriebel et al., “The Precautionary Principle”, at p. 875, supra note 12, at p. 6. Indeed, in epidemiology and biotechnology trials the language of “safety” assessment incorporates this kind of concern by requiring a sufficient statistical power or b probability (cf. EFSA, “Guidance document of the scientific panel on genetically modified organisms for the risk assessment of genetically modified plants and derived food and feed”, 99 EFSA Journal (2006), pp. 1 et sqq.).

15 Environmental science is considered a “soft” science as opposed to other “hard” sciences such as physics and chemistry. The two categories differ in that the latter have more predictive power than the former (Jordan and Miller, 1996). This is fundamentally due to the fact that inductivism is the methodology affiliated to experimentation. It is ampliative in that the conclusion has a content that goes beyond the content of its premises; it is not necessarily truthpreserving, in that there could be true premises and false conclusions; it is not erosion-proof, in that new premises can completely undermine the argument; any combination of premises and conclusions (be they true and/or false) is possible for the validity of the argumentation to test, which is why inductive arguments have different degrees of strength ( Salmon, Merrilee H., Earman, John, Glymour, Clark and Lennox, James, Introduction to the Philosophy of Science, 1st ed. (Indianapolis and Cambridge: Hackett Pub Co Inc 1999)Google Scholar).

16 To the same extent, methodological problems concerning the size and composition of the sample and the time extension of the experiment have been specifically advanced for GMO safety studies. On this point, see Séralini, Gilles-Eric, Cellier, Dominique and de Vendômois, Joël Spiroux, “New analysis of a rat feeding study with a genetically modified maize reveals signs of hepatorenal toxicity”, 52 Archives of Environmental Contamination and Toxicology (2007), pp. 596 et sqq CrossRefGoogle ScholarPubMed.

17 For the purposes of this paper, only critiques concerning the choice of the error to minimise have been reported. However, more technical critiques exist that focus on the intrinsic simplification of the sources of uncertainty against hypothesis testing, denouncing its uninformative structural character and highlighting two main points: the acknowledged sources of uncertainty in statistical testing mainly converge on sampling variability through the standard p-value and confidence intervals; and the high chance of incurring systematic errors can originate from both selection bias and the actual measurement of the levels of specific factors according to the availability of detecting technologies. On these issues, see Anderson, Burnham and Thompson, “Null hypothesis testing”, supra note 10, at p. 5.

18 Hume, David, A treatise of human nature, 2nd ed. (Oxford: Oxford University Press 1978)Google Scholar.

19 These circumstances are better known as the “fallacy of affirming the consequent”, in which a logic failure originates from the fact that inference processes are not necessarily truth-preserving, which means that even if premises are true, conclusions can be false, and vice versa (Salmon et al., Philosophy of Science, supra note 15, at p. 7).

20 Salmon et al., Philosophy of Science, supra note 15, at p. 7.

21 Popper, Karl, “Probability Magic or Knowledge out of Ignorance”, 11 Dialectica (1957), pp. 354 et sqq CrossRefGoogle Scholar.; Popper, Karl, The Logic of Scientific Discovery (New York: Basic Books 1959)Google Scholar.

22 It is not automatic to disprove a hypothesis as soon as an observation to the contrary occurs, especially if it has long been confirmed by evidence. This choice in fact depends on the scientific group conducting the analysis, which legitimately may deem an “alien” observation as a simple anomaly. This happens mainly in epidemiologic studies and generally in scientific disciplines where improvements occur through the criteria of preponderance of evidence.

23 Rothman, Kenneth J. and Greenland, Sander, Modern Epidemiology (Philadelphia, U.S: Lippincott Williams & Wilkins 1998), at p. 8 Google Scholar.

24 Within scientific theories, there exists an established principle of due care which is the principle of parsimony (or lex parsimoniae), also known as the Ockham's razor principle. According to which, among competing theories only those that are based on the fewest assumptions should be retained, and conversely those that are useless (entia non sunt multiplicanda, translated: entities are not to proliferate) should be eliminated, or “shaved off”.

25 Sunstein, Cass, Laws of Fear. Beyond the Precautionary Principle (Cambridge: Cambridge University Press 2005)CrossRefGoogle Scholar.

26 For a discussion of how individuals construct their preferences, see Kahneman, Daniel and Tversky, Amos, Choices, Values, and Frames, 1st ed. (New York: Cambridge University Press 2000)Google Scholar.

27 See supra note 24, at p. 10.

28 Kahneman and Tversky, Choices, Values, and Frames, supra note 26, at p. 10.

29 Parkhurst, “Statistical Significance Tests”, supra note 10, at p. 1051.

30 This formulation is simplified. Generally, a certain chemical substance is compared with another one already in usage, and it is assumed that there is no difference between the two in terms of some parameters. For instance, if the parameter is the toxicity of the chemical, the null hypothesis is that the chemical being tested is no more toxic than the conventional one.

31 Cf. section III.1.a “For more precautionary statistics”.

32 The concept of substantial equivalence is the key for a comparative assessment, in which traditionally cultivated crops have gained a history of safe use upon which they provide the baseline for determining any substantial difference between the GMO and its wild counterpart (cf. EFSA, “Scientific panel on genetically modified organisms”, supra note 14, at p. 7).

33 This means that in a welfare function we suppose that the coefficient of the marginal utility of the potential victims (or weakest parties) is higher than any other coefficient of other categories.

34 Popper, The Logic of Scientific Discovery, supra note 21, at p. 9.

35 Barrett and Raffensperger, “Precautionary Science”, supra note 8, at p. 7.

36 In fact, as explained in the section below, to shift the burden of proof is not just a matter of deciding who should prove or bear the risk of uncertainty about a possibly dangerous event, but it is also a matter of how the proof (i.e. preponderance of evidence) about the occurrence of that event is constructed to err on the side of more or less precaution.

37 Walker, Vern R., “The Myth of Science as a ‘Neutral Arbiter’ for Triggering Precautions”, 26 Boston College International and Comparative Law Review (2003), pp. 197 et sqq Google Scholar.

Walker's statement was made in reference to lawmakers, to define the factual predicate for taking precaution. Nonetheless, scientists face the same dilemma when selecting the relevant element for conducting their studies.

38 Interview with Gérard Pascal, Chairman of the Scientific Steering Committee of the European Union, Paris, France (April 8, 2008). Gérard Pascal remembers just one instance, during the years of the mad cow crisis when he served at the European Commission, DG SANCO, as president of the scientific committee, where he experienced policy confrontation with the responsible Commissioner.

39 As some authors have warned, as well as Type I and Type II errors, there exists another type which demands a more fundamental question: what is the problem? Hence, Type III error accounts for the risk to produce an accurate answer for the wrong question (see Kriebel et al., “The Precautionary Principle”, supra note 12, at p. 5; Sanderson, H. and Solomon, K. R., “Precautionary Limits to Environmental Science and Risk Management: Three Types of Errors”, 2 The Journal of Transdisciplinary Environmental Studies (2003), pp. 1 et sqq Google Scholar.).

40 Cf. with section below on the burden the potential injurer is supposed to discharge.

41 Shapiro, Sidney A, “Keeping the Baby and Throwing Out the Bathwater: Justice Breyer's Critique of Regulation”, 8 Administrative Law Journal (1995), pp. 721 et sqq Google Scholar.

42 GMOs are presumed safe to the extent that they are substantially equivalent to their natural counterparts. See supra note 32, at p. 13.

43 Parkhurst, “Statistical Significance Tests”, supra note 10, at p. 5.

44 See supra note 42, at p. 18.

45 McBride, Graham B., “Equivalence tests can enhance environmental science and management”, 41 Australian & New Zealand Journal of Statistics (1999), pp. 19 et sqq CrossRefGoogle Scholar.

46 Parkhurst, “Statistical Significance Tests”, supra note 10, at p. 5.

47 McBride, “Equivalence tests”, at p. 3, supra note 45.

48 The point estimate is zero because it informs that the point hypothesis is a no-effect one.

49 Remember that failing to reject the null hypothesis does not entail that it is to be accepted or confirmed to be true.

50 SirAustin, Hill, Bradford, “The Environment and Disease: Association or Causation?”, 58 Proceedings Royal Society of Medicine (1965), pp. 295 et sqq Google Scholar., available on the Internet at <http://www.edwardtufte.com/tufte/hill>.

51 Phillips, Carl and Goodman, Karen, “The missed lessons of Sir Austin Bradford Hill”, 1 Epidemiologic Perspectives & Innovations (2004), pp. 3 et sqq CrossRefGoogle Scholar.

52 In this statement, Hill (see supra note 50, at p. 21) was presenting the example of introducing a drug for early-morning sickness in pregnant women. As he said, the doctor can decide to restrict the use of the drug even on relatively slight evidence, the fact being that the “good lady and the pharmaceutical industry will doubtless survive.”