08 Jul Causation on Trial: A Call For Change In How We Think About Causation In Tort
By: Richard C. Halpern January 26, 2021
“Every thing must be at rest which has no force to impell it; but as the least straw breaks the horse’s back, or a single sand will turn the beam of scales which holds weights as heavy as the world; so, without doubt, as minute causes may determine actions of men, which neither others, nor they themselves are sensible of; but certainly something must determine them, or else they could not be determined; and it is nothing to the purpose to say, that their choice determines them, if something else must determine that choice; for let it be what it will, the effect must be necessary”.
– John Trenchard (anonymously), Cato’s Letters London, 1724.
To say that an event is the cause of an effect sounds simple enough. It can be simple, but often it is not. An event, on its own, may not be enough to bring about an effect, and may need to follow or combine with another event to bring about an effect. More than one event may be needed to bring about an effect. Further, the magnitude of an event may be modest, yet sufficient to bring about the effect – “the least straw that breaks the horse’s back”.
We make causal connections each and every day, without so much as a second thought, sometimes with more and sometimes with less confidence in the conclusions we draw. If we explain the cause of an effect that has already occurred, our conclusion is an explanation for what has occurred. If we consider the future outcome from an event or series of events, we are engaged in causal prediction. It is this latter exercise that the law must engage in when determining causation, although the prediction will always be imagined, never real. A mere glance at the caselaw will quickly reveal how perplexing the notion of causation can be in tort law.
It is decidedly too ambitious, and assuredly futile, to attempt to craft a flawless paradigm for addressing causation in all scenarios. Afterall, for decades, and longer, every academic and every court that has set out to tackle the matter has failed in the endeavour. Indeed, these failed attempts have actually contributed more to the confusion and uncertainty surrounding causation than they have in resolving it. This paper is rather less ambitious and more realistic. By re-defining what is meant by the legal test for causation, and by clarifying how to apply that test in some more challenging factual scenarios, we may hope for some relief from the seemingly endless distractions and misdirections on the issue that currently exist in the caselaw.
If my contention that causation remains poorly understood and inadequately defined is accepted, despite extensive treatment in the legal literature and courts over the last few decades, one might still be justifiably skeptical about the prospect that this article might clear up anything. With no small degree of trepidation, the attempt will be made.
At the outset, I think we should accept that a comprehensive and all-encompassing definition of causation may remain elusive, but something less extensive will suffice for the vast majority of cases that confront lawyers and courts. There is a scarcity of reported cases where causation is truly a problem. Thus, the objective of this paper is practical. It is to clear up the true meaning of cause-in-fact in tort for virtually every fact scenario that most of us will encounter in our legal careers. To do so, I think it is important to get to the root of causation – the counterfactual. The counterfactual seeks to determine cause by hypothetically altering a fact, or facts, and asking if the harm would have been avoided.
Every tort lawyer knows that the traditional test for causation is the but-for test. The provenance of this test, however, is the counterfactual, and the way the but-for test has been used has been both too narrow and too limiting, ignoring its roots. It may be convenient to call the test by that name, but using the words “but-for” to define the test can lead to no end of difficulty, particularly as multiple factors often play a role in causing phenomena. A review of the cases covered in this paper will demonstrate, quite clearly, that a test for causation that ignores its roots in the counterfactual, that is the conventional recitation of the but-for test in modern caselaw, is a test that will fail to identify causation where it clearly exists. It is time to re-evaluate how we define that conventional test, and rework its current definition.
In its simplest terms, the counterfactual asks whether the harm suffered by the plaintiff would have been avoided had the wrongful act of “the specified” defendant not occurred. The hypothetical is asked in relation to a particular wrongdoer because tort liability is based on individual responsibility. That is, once a breach of duty is found against a wrongdoer, the causal question tends to ask whether the “specified” breach of duty by the “specified” wrongdoer caused the harm. To ask this question in relation to a one wrongdoer is to single out or “individuate” the causal question. The question is asked, inappropriately in some cases, in isolation from other possible contributors to the harm. Asking if the factor is “the cause” is to ignore the factor as “a cause”, the result of which would be a failure to capture some causes. Where a phenomenon is the result of a sequence of events, whether acting successively or cumulatively, the individuation of the first causal question inappropriately ignores how phenomena actually occur.
Thus, the test for causation becomes difficult to apply, if not impossible to apply, where factual scenarios are made more complex with the addition of more potential causative events attributed to more actors, whether innocent or wrongful. As individuation of liability breaks down with the conventional but-for test in these latter situations, courts and academics have suggested that an alternative to the but-for test is needed. To make this issue clear, as the law stands since the Supreme Court of Canada decision in Clements v. Clements, the compulsion for finding an alternative to but-for arises when the test fails to attribute causation where causation is known to exist. The test appears to fail, and an alternative test applied, where there are two or more wrongdoers, both parties defendant, and one or more must have caused the harm, but it is impossible to know which wrongdoer did. This result is untenable, if it is indeed how the test functions.
An alternative to the but-for test has been referred to as the “material contribution” test or, more recently, the “material contribution to risk” test, as the test was revised and articulated in Clements. The material contribution test is based on the notion that it would be unfair to deprive the plaintiff of a remedy where causation can be clearly attributed to one or more defendants. In this sense, however, there are two points to make: first, in the scenario described, one or more of the defendants caused the loss; and, second, the alternative is not a test for cause-in-fact at all, but merely “proves” what has already been decided.
With regard to proof of cause-in-fact, this paper makes three essential arguments. First, the material contribution test, as defined by the Supreme Court of Canada in Clements, is incapable of addressing the shortcomings of the traditional but-for test. Second, material contribution, as defined in Clements, is not a test of causation at all. This fundamental interpretation needs to be explicitly recognized. Rather, it is a policy-based rule that entirely relieves the plaintiff of the burden to prove causation. This makes a principled application of such a test too open-ended and random. Third, material contribution, used properly, is an adjunct to an appropriately formulated counterfactual, to be subsumed within what we understand to be the but-for test. In that sense, material contribution should be taken as defining the threshold of proof, in more difficult factual scenarios, required to satisfy a test of cause-in-fact. That is, where cause-in-fact is determined by inference, a contributing factor is causal only when its role in creating a phenomenon is more than de minimis. That the finding by inference is made on the basis of material contribution is no less cause-in-fact than when used in the simplest factual scenario.
With regard to the first point, that an alternative to a but-for does not work, this can be demonstrated using various factual scenarios that will be described. Taking each wrongdoer as a separate legal entity, as one must do in tort law, requires that their conduct, alone, be evaluated against the rights of the person said to be injured. Where two or more tortfeasors might be to blame for the plaintiff’s injury, it may be possible to prove only that one or more in the group caused the loss but identifying the individual or individuals responsible is impossible. In such a scenario, the but-for test is satisfied insofar as the group is concerned, or “globally”. The point here is that material contribution, as defined, will shed no more light on the issue than the but-for test has already done. Thus, it is hopeless to think that material contribution can ferret out the actual wrongdoer. The matter becomes more problematic when there are both innocent and guilty factors causing the same phenomenon.
In Clements, McLachlin CJ’s notion of “impossibility” in relation to wrongdoers who are named parties to a lawsuit as the trigger required to invoke an alternative to the but-for test simply does not work as a test of causation and is not really expressed as an alternative, but is to be used in conjunction with but-for.
When the reference is made to “global” proof of cause in Clements, requiring that cause be attributed to “one or more tortfeasors”, in reality that amounts to but-for having been proven as against the group of defendants. “Impossibility” in this sense is only about attribution once but-for is proven. Of course, this breaks down when impossibility includes both wrongful and innocent factors, and the approach is thus lacking in principle.
If the exception finds its origins in the English case of McGhee, which I argue against, and is expanded in the English case of Fairchild, which I also refute, then “impossibility” would relate only to the wrongful or specified factor, regardless of whether innocent factors are also at play. It should not matter, however, as illustrated in another English case, per Lord Hoffmann in Barker, that one cause is innocent and the other guilty. Such a test, to be invoked in a principled way, must operate when “the impossibility of proving that the defendant caused the damage arises out of the existence of another potential agent which operated in the same way”. This impossibility arises, as do the arguments in support of an exception, whether the other potential cause is innocent or guilty. This points out a rather dramatic flaw in the reasoning in Clements, which I will discuss when I take a closer look at the case.
As to the second point, if material contribution is incapable of identifying cause-in-fact, then attribution of liability to a defendant is made in the absence of proof of cause-in-fact. It is high time that we recognize that if there are scenarios where attribution of cause, and therefore liability, is imposed where traditional but-for or even an expanded but-for fails, it is because the plaintiff has been relieved of the onus of proof entirely. That relief from the onus to prove causation arises out of exceptional circumstances that will be described when discussing Clements. The failure to recognize that relieving the plaintiff of the burden of proof of causation, which is actually the practical effect of the alternative test described in Clements, is at the core of our confusion about causation in the case law. It is for this reason that I contend that material contribution is not a test of cause-in-fact at all and should, therefore, be neither proclaimed nor mistaken as such. If there is a justification for departing from the but-for test, in the form reviewed in this paper, and that must be seen to occur in the rarest of circumstances, the preferrable approach, I suggest, would have been to follow the reasoning of Justice Sopinka in Snell v. Farrell, by shifting the burden of proof. Before shifting the burden of proof, one must have regard to the robust and pragmatic approach to causation, described by Justice Sopinka in Snell, which gives rise to my third point.
The third point asserts that material contribution is merely part and parcel of a cause-in-fact analysis of causation, which we describe imprecisely as “but-for”, but which has its roots in the counterfactual. For the most part, the notion that the but-for test might be extended or that an alternative might be found has its genesis in McGhee. The facts in McGhee were challenging, and the case will be dealt with extensively later in this paper. It is my view that McGhee did not extend the but-for test in any sense. It is the overly rigid application of the traditional but-for test, touched on in Snell, that is at the root of the problem. It is also the failure to properly adapt the causation question for the more complicated factual scenarios that has muddled the notion of cause-in-fact. We must have resort to the counterfactual to sort through these issues.