Five Issues of Causation and Proof
By: Randy Barnett
Cato Journal, vol. 2 n. 1 Spring 1982
In any discussion of causation in torts cases in general and pollution-engendered torts cases in particular it is necessary to distinguish two types of inquiry. The first is the nature of causation in torts; that is, what do we mean when we say that one person causes harm to another? Given an answer to this question, the second inquiry is how we demonstrate the causal connection in practice; that is, how do we prove that a particular activity by one person has caused a certain harm to another? It seems fair to say that while causal questions of the first sort are not more vexing in pollution cases than in most other types of torts cases, the second kind of inquiry can present quite serious and sometimes intractable difficulties.
With this distinction in mind, Professor Shindell's paper can be seen as a provocative exploration of some problems faced by those attempting to demonstrate a causal link between a particular pollutant and a subsequent health effect. In my comments I will attempt to put his analysis in the context of five common legal issues of causation and proof. The first is the concept of causation employed in the analysis. The second is the nature of the evidence used by Professor Shindell. The third is the standard of proof that should be applied. The fourth is the question of who should bear the burden of proof. Finally, I shall briefly consider the need for a causal requirement. Though I will not here attempt to conclusively resolve these issues, much confusion will be avoided if they are kept in mind when discussing the problems of environmentally engendered torts.
The Nature of Causation
In the course of his excellent discussion, Professor Shindell makes an observation that if taken seriously can only result in causal problems becoming more acute than necessary. Concerning the nature of the causal relationship between a health hazard and a health effect, he says: A health effect is only causally related if: (a) it would not be expected to occur in the absence of the hazard, or (b) it would not have occurred with the frequency experienced had it not been for the presence of the hazard.
This characterization of the causal issue is a form of the well-known "but for" test of causality. Professor Shindell might have said, "But for the hazard, a health effect would not have occurred as frequently." Notice, as Professor Epstein has pointed out, that this expression "is in counterfactual form and requires an examination of what would have been the case if things had been otherwise." It does not ask the more straightforward question of what caused the health effect being considered.
The problem created by such a formulation is that it fails to identify the agent causally responsible for the health effect. On this view any occurrence that was a necessary condition for the health effect is a cause of the effect. Thus, in the case of water pollution, were it not for the consumption of the contaminated water no health effect would have occurred. Is the drinking of contaminated water to be considered a "cause" of the same order as the existence of the contamination? The "but for" test cannot distinguish among necessary conditions, and efforts to identify the "proximate" cause have met with little success.
This type of counterfactual concept of causality is unnecessary. The statement, "asbestos-caused mesothelioma" needs no more linguistic reformulation than arsenic-caused death" or "the blow caused a broken nose." Though the factual chain of causation could be very long indeed, the causal paradigm operating in pollution cases is simply, "A poisoned B." It should be stressed that such a change in the causal language Professor Shindell has employed would not affect the merit of his analysis. What he is observing in this passage is the fact that the only evidence of a causal link between the hazard and the harmful effect may be a statistical association. It is to this issue that I now turn.
The Nature of Causal Evidence
The causal link between the defendant's conduct and the plaintiff's injury is a fact that must be proved in every action in tort. The evidence offered in proof may be of two kinds: Direct and circumstantial. Direct evidence proves the fact in issue without reference to any intermediate inference. Most often it takes the form of eyewitness testimony to the effect of "I saw A hit B in the nose." This is direct evidence of the identity of B's assailant and the nature of the attack. Of course it may be believed or disbelieved by the trier of fact. If accepted as true, however, these facts are, without more, proved. Circumstantial or indirect evidence is evidence of facts that together with other proof or reasonable inferences can be indicative of a fact in issue — here causation. An example of circumstantial evidence is a fingerprint found at a crime scene that might indicate presence, but not time or circumstances. It should not be assumed that one form of evidence is inherently more reliable than the other. Often circumstantial proof will be preferred to eyewitness testimony.
It is unlikely that causation in cases of poisoning or pollution could ever be other than circumstantial. Even if a witness saw A pour arsenic into B's tea, the fact that B's subsequent death was caused by the arsenic would need to be inferred from an examination of B's remains. And the degree of certainty with which such a conclusion can be made will depend upon the strength of the direct and circumstantial evidence as well as upon the extent of our understanding of the effect of arsenic on the human body that permits us to make inferences from the circumstances. It should not be thought that causal questions are unique in this regard. A person's state of mind must almost always be inferred from that person's behavior, though factual presumptions are also employed.
The evidence that Professor Shindell advances as proof of causation in pollution cases takes the form of statistical associations, which are given certain epistemic assumptions) one kind of circumstantial evidence. When should this type of evidence be allowed to supplement or substitute for direct proof of causation? If our answer is never, then we are in effect granting immunity to most polluters for whatever harm they may cause. If our answer is that any association will always be dispositive, then we will inevitably enjoin industrial activities that are not in fact harmful. Assuming that statistical associations may be a way of discovering the existence of a causal link, and further, that the strength of any such inference will depend on the degree of association, then we must determine what degree is sufficient for a finder of fact to conclude the existence of the causal link. What should be the standard of proof?
The Standard of Proof
The standard of proof to be adopted in a given type of case will depend on the nature of the interests that would be affected by an incorrect decision. So even if we believe, as I do not, that it is sometimes justified to protect persons from harming themselves by the consumption of some chemical like nicotine or saccharin, we may be unwilling to do so on the basis of a "mere" statistical association between consumption and the incidence of cancer in humans or more attenuated inferences drawn from the results of tests on other animals. In the case of an involuntary ingestion of the same chemical, however, our standard might differ and such evidence be allowed. When the interest affected is individual liberty, as it is when the sanction is imprisonment for a crime, the standard adopted is a high one, but not one that guarantees no error.
The problem posed by pollution cases, indeed all private tort actions, is that a decision in favor of one party will be a decision against the other. Assuming our law is based on individual rights, an erroneous judgment will unjustly restrict the losing party's liberty. To place a greater burden, then, on one party than the other would seem to provide a greater protection for the rights of the party with the lesser burden. All things being equal, we should place the same standard of proof on all parties. But all things are not equal. The standard of proof will depend on factors apart from the seriousness of the interests at stake.
Standards of proof that amount to a (rebuttable) presumption in favor or against finding a particular fact to be true will also depend on our common understanding of the world. Hence, if we believe that in most instances chemicals released into the environment will not cause adverse health effects or if we believe that the chances of a particular act of pollution being responsible for a given injury is quite low, we may adopt a presumption to that effect. In doing so we are saying that recovery may be allowed provided the plaintiff demonstrates the truth of his or her claim beyond a certain level of doubt.
This type of analysis can help account for and justify a different standard of proof in judging requests for damages after a harm has been sustained than is used in cases where injunctive or preventive relief is sought. Surely the standard of substantial risk of imminent harm" applied to pleas for injunctive relief reflects both our uncertainty about whether a harm will in fact occur if the activity is allowed and an even more fundamental doubt about the feasibility of demonstrating such a fact.
Allocation of the Burden of Proof
Yet a fourth issue must be identified in pollution-engendered torts cases and that is: Who shall be required to produce evidence of a certain fact? Put bluntly, must a plaintiff prove that the defendant's emissions caused or will cause a harm or must the defendant prove this is not or will not be the case? Allocation of this burden will involve both factors mentioned in regard to the standard of proof, but will also depend upon the principle of justice that prohibits individuals from being forced to rectify injuries they haven't caused. Before acting a court will need a reason to single out the defendant, and this reason has traditionally been demanded of the requesting party. More will be said on this point in the next section.
The Doctrine of Res Ipsa Loquitur and the Need for a Causal Requirement
Res ipsa loquitur as traditionally formulated is not a doctrine that speaks to causal issues but one which specifies certain circumstances where the defendant's conduct is rebuttably presumed to be negligent. In these cases where it is shown that the defendant's conduct harmed the plaintiff the burden will shift to the defendant, who then must prove that he was not negligent. A case subject to this doctrine is one of strict liability,  and if this presumption is held to be legally unrebuttable then the theory is one of absolute liability. In neither case are causal barriers lowered except in one sense: If a plaintiff must show that defendant acted negligently, it must also be shown that it was the defendant's negligence that caused the harm to the plaintiff. To illustrate this, suppose that while driving my car at twice the speed limit I run you over. If it can be shown that had I been operating the car at normal speed, (i.e., non-negligently), I still could not have avoided hitting you — because of lighting, your clothing, etc. — then on a negligence theory there is no causal link between my speeding (i.e., the negligence) and your injury even though my actions were concededly the cause-in-fact of your injuries, and you will be denied recovery.
In this way res ipsa loquitur may limit the more extensive and complicated causal inquiry required by a theory of negligence, but it does not eliminate the need to prove that the defendant's conduct caused the plaintiff's injury.
Persons should not be held liable for harms that we don't or can't know they caused. The problem of limited knowledge is an inherent quality of human existence. We must always operate on less than perfect information. A free society that recognizes individual rights consistent with individual liberty must adopt certain principles of justice, the most important of which is that a person will be free to use and enjoy his or her person and property unless that use invades the equal rights of his or her neighbors. Any proposal that preventively restricts human action or retroactively penalizes individuals for conduct that has not infringed on another's rights has seriously undermined the very concept of individual liberty. In the absence of proof to the contrary, we must presume that one person's conduct is not invasive of another's rights if we wish to remain a free society. There is no middle course.
It is important to be aware that decisions concerning the five issues of causation and proof raised here must be made and that any decision reached will favor one party at the expense of the other. Ultimately, these issues must be resolved by an appeal to notions of how the world works, how we come to understand it, and where justice lies.
 Sidney Shindell, "Evidentiary Problems in Pollution-Engendered Torts" Cato Journal 2 (Spring 1982): 137-155.
 Ibid., p. 153.
 See generally William L. Prosser, Handbook of the Law of Torts, 4th ed. (St.Paul,
 Richard A. Epstein, "A Theory of Strict Liability” Journal of Legal Studies 2 (1973):
 See ibid., pp. 160-189; idem, "Causation and Corrective Justice: A Reply to Two
Critics, "Journal of Legal Studies 8 (1979): 477; see generally H.L.A. Hart and A.M.
Honore, Causation of the Law (London: Oxford University Press, 1959); Prosser, Law of Torts, pp. 244-289.
 We say, for example, that a person is presumed to intend the natural and ordinary consequences of his or her actions as a matter, perhaps, of introspective insight.
 E.g., poisoning cases of which pollution cases may be considered a species.
 Two aspects of "burden of proof" have been distinguished: The burden of going forward with evidence; i.e., which party must raise an issue by competent evidence and the burden of persuasion; i.e., once raised, which party must satisfy the trier of fact by proof beyond a certain level of doubt. It may, for example, be the rule that a plaintiff must present some evidence indicating that the defendant caused an injury and then the defendant must prove beyond a preponderance of the evidence that he did not. An allocational choice must be made here that will involve much the same concerns outlined throughout this comment.
 See Epstein, "A Theory of Strict Liability"; idem, "Defenses and Subsequent Pleas
in a System of Strict Liability," Journal of Legal Studies 4 (1975): 391.
 In theory, that is. The bias in favor of strict liability in such cases is so great that
even within a nominally negligence system a defendant would be lucky to escape
liability absent serious misconduct by plaintiff.
 Contra Robert Best and James Collins, "Legal Issues in Pollution-Engendered
Torts," Cato Journal 2 (Spring 1982): 101-136.