From Russell Brown,* “Cause-in-Fact at the Supreme Court of Canada: Developments in Tort Law in 2012-13” (2014), 64 S.C.L.R. (2d) 327 at 331
(* now a Justice of the Alberta Court of Appeal and Honorary Professor of Law at the University of Alberta; Justice of the Court of Queen’s Bench of Alberta at the time of publication)
It is worth making two observations about the drawing of a causal inference. First, there is nothing unorthodox about the proposition that a fact-finder may draw causal inferences, even in the absence of verifying scientific evidence. Inference-drawing is inherent not only in all causal fact-finding, but in all fact finding. Just as one might infer from evidence that an automobile made a screeching noise immediately prior to a collision that the driver saw the obstruction just before striking it, one may also infer (or not infer) that evidence of risk, of harm and of the surrounding circumstances is so arrayed so as to support the drawing of an inference of a causal relationship between the risk and the harm – and all without the reassurance of expert evidence. And, just as the absence of direct evidence that the driver saw the obstruction before hitting it does not preclude an inference that he or she did, the absence of direct evidence verifying a causal link between risk and harm is similarly not determinative. Fact-finders infer, or they do not infer, and they may do so on evidence that bears only indirectly to the proposition being advanced.
The second observation is that causal inference-drawing is not presented in Snell as some form of alternative or watered-down version of the but for test. If all fact-finding is necessarily inferential, then fact-finding to the but for standard is also inferential. So long as the inference is drawn on a balance of probabilities, causal inference is no alternative to the but for test, but is an instantiation of it. It is how the but for test is to be applied – and not merely in some cases, but in all cases.
 Russell Brown, “The Possibility of ‘Inference Causation’: Inferring Cause-in-Fact and the Nature of Legal Fact-Finding” (2010) 55 McGill L.J. 1, at 30-35 [Brown, “Inference Causation”].
 This was the criticism of Lewis Klar, in Tort Law, 3d ed. (Toronto: Thomson Carswell, 2003) at 403: “[w]hile [inference-drawing] may produce a pragmatic solution to a plaintiff’s dilemma in difficult causation cases, it does depart from the traditional “but for” test, and the balance of probability standard.” This statement does not appear in subsequent editions of Professor Klar’s book.
Let’s return to Justice Brown’s second observation: that making causal inferences is the manner in which the but-for test is to be applied in all cases: whether there is direct evidence or not; whether there is expert evidence or not.
Here’s the short version of the Supreme Court of Canada authority for that proposition, in reverse chronological order.
Ediger v Johnston, 2013 SCC 18,  2 SCR 98:
 This Court recently summarized the legal test for causation in Clements v. Clements, 2012 SCC 32,  2 S.C.R. 181. Causation is assessed using the “but for” test (Clements, at paras. 8 and 13; Resurfice Corp. v. Hanke, 2007 SCC 7,  1 S.C.R. 333, at paras. 21-22). That is, the plaintiff must show on a balance of probabilities that “but for” the defendant’s negligent act, the injury would not have occurred (Clements, at para. 8). “Inherent in the phrase ‘but for’ is the requirement that the defendant’s negligence was necessary to bring about the injury — in other words that the injury would not have occurred without the defendant’s negligence” (para. 8 (emphasis deleted)).
Clements v. Clements, 2012 SCC 32,  2 SCR 181:
 The test for showing causation is the “but for” test. The plaintiff must show on a balance of probabilities that “but for” the defendant’s negligent act, the injury would not have occurred. Inherent in the phrase “but for” is the requirement that the defendant’s negligence was necessary to bring about the injury ― in other words that the injury would not have occurred without the defendant’s negligence. This is a factual inquiry. If the plaintiff does not establish this on a balance of probabilities, having regard to all the evidence, her action against the defendant fails.
 The “but for” causation test must be applied in a robust common sense fashion. There is no need for scientific evidence of the precise contribution the defendant’s negligence made to the injury. See … Snell v. Farrell,  2 S.C.R. 311.
 A common sense inference of “but for” causation from proof of negligence usually flows without difficulty. Evidence connecting the breach of duty to the injury suffered may permit the judge, depending on the circumstances, to infer that the defendant’s negligence probably caused the loss. See Snell and Athey v. Leonati,  3 S.C.R. 458. …
 Where “but for” causation is established by inference only, it is open to the defendant to argue or call evidence that the accident would have happened without the defendant’s negligence, i.e. that the negligence was not a necessary cause of the injury, which was, in any event, inevitable. As Sopinka J. put it in Snell, at p. 330:
The legal or ultimate burden remains with the plaintiff, but in the absence of evidence to the contrary adduced by the defendant, an inference of causation may be drawn although positive or scientific proof of causation has not been adduced. If some evidence to the contrary is adduced by the defendant, the trial judge is entitled to take account of Lord Mansfield’s famous precept [that “all evidence is to be weighed according to the proof which it was in the power of one side to have produced, and in the power of the other to have contradicted” (Blatch v. Archer (1774), 1 Cowp. 63, 98 E.R. 969, at p. 970)]. This is, I believe, what Lord Bridge had in mind in Wilsher when he referred to a “robust and pragmatic approach to the . . . facts” (p. 569). [Emphasis added.]
[Emphasis in para. 11 added by McLachlin C.J. in Clements; some non-Canadian authorities omitted]
Let’s agree to overlook the implications of the “where ‘but-for’ causation is established by inference only” statement that opens para. 11 because, with all due respect, it’s never the case that factual causation is established by inference only. Stating the problem with this statement another way, it’s impossible to provide a certain, useful, explanation of what it means, in law, to make a conclusion of factual causation using “only” whatever one means by a a process which uses only conclusions drawn from inferences.
Resurfice Corp. v. Hanke, 2007 SCC 7,  1 SCR 333 at para. 9
 The plaintiff submits that the trial judge discounted the evidence of expert witnesses called by the plaintiff on the design of gas delivery systems and the behaviour of workers. It is true that the trial judge placed no reliance on these witnesses. However, a trial judge is not obliged to consider the opinions of expert witnesses if he can arrive at the necessary conclusions on issues of fact and responsibility without doing so: R. v. Mohan,  2 S.C.R. 9, at pp. 23-24.
 First, the basic test for determining causation remains the “but for” test. This applies to multi-cause injuries. …
 This fundamental rule has never been displaced and remains the primary test for causation in negligence actions. As stated in Athey v. Leonati, at para. 14, per Major J., “[t]he general, but not conclusive, test for causation is the ‘but for’ test, which requires the plaintiff to show that the injury would not have occurred but for the negligence of the defendant”. …