Viviano, J.

Johnson Law, PLC (by Ven R. Johnson and Christopher P. Desmond), for Michael A. Ray, Jaqueline M. Ray, and Kersch Ray.

Giarmarco, Mullins & Horton, PC (by Timothy J. Mullins and John L. Miller ), for Eric Swager.

BEFORE THE ENTIRE BENCH

Viviano, J.At issue is whether the conduct of defendant Eric Swager, who is a government employee, was—for purposes of the governmental tort liability act (GTLA), MCL 691.1401 et seq. —"the proximate cause" of the injuries suffered by plaintiff, Kersch Ray. We hold that the Court of Appeals failed to correctly analyze proximate cause because it only weighed various factual causes of plaintiff's injuries to determine whether one of them was a more direct cause of plaintiff's injuries than defendant's conduct. Under the GTLA, a proper proximate cause analysis must assess foreseeability and the legal responsibility of the relevant actors to determine whether the conduct of a government actor, or some other person, was "the proximate cause," that is, as our caselaw has described it, "the one most immediate, efficient, and direct cause" of the plaintiff's injuries. Accordingly, in lieu of granting leave to appeal, we vacate the Court of Appeals' decision and remand to the Court of Appeals for further proceedings not inconsistent with this opinion.

I. FACTS AND PROCEDURAL HISTORY

In the fall of 2011, the then-thirteen-year-old plaintiff was a member of the Chelsea High School cross-country team. Defendant was the coach. Shortly after the season began, defendant held an early morning practice; it was plaintiff's first morning practice as a member of the team. The practice began at 5:59 a.m., when it was still dark outside. At the beginning of the practice, defendant took the team off school grounds to run on public roads. During the run, the team approached an intersection with a two-lane highway. The "Do Not Walk" symbol was illuminated because the traffic light was green for the highway traffic. Defendant and the group of runners he was with, which consisted of most of the team, stopped at the intersection. Defendant saw a vehicle in the distance, but he determined that it was far enough away to safely cross. He instructed the runners to cross the intersection by stating, "Let's go." It is unclear whether all the team members, including plaintiff, heard the instruction. Although most of the team safely crossed the road, a few runners in the back of the group were still in or near the roadway when the vehicle entered the intersection. The vehicle hit plaintiff and one of his teammates as they were crossing the road. Plaintiff was severely injured, and he has no memory of the accident.

In his deposition, the driver of the vehicle testified that he suddenly saw the runners crossing the intersection off to the right side of the road but that he did not see plaintiff in the intersection or have time to brake. The driver testified that he was not distracted and had been driving below the posted speed limit. While the driver initially admitted that he sped up when the traffic light turned yellow, he later stated that he did not recall whether he did anything in reference to the yellow light. The Washtenaw County Sheriff's officer who conducted an accident reconstruction concluded that the driver was not responsible for the accident.

Plaintiffs sued defendant and the driver. Defendant moved for summary disposition under MCR 2.116(C)(7), asserting governmental immunity pursuant to the GTLA. Defendant also moved for summary disposition under MCR 2.116(C)(8) and (10). The trial court denied defendant's motion, stating that whether defendant's actions were grossly negligent and whether his actions were the proximate cause of plaintiff's injuries were questions of fact for the jury to decide.

Defendant appealed by right, and the Court of Appeals reversed in an unpublished per curiam opinion. The panel concluded that any factual disputes were not material because reasonable minds could not conclude that defendant was the proximate cause of plaintiff's injuries. The panel determined that the presence of the driver in the roadway and plaintiff's own actions were more immediate and direct causes of plaintiff's injuries and held that "the most proximate cause of [plaintiff's] injuries is the fact that he was struck by a moving vehicle."

Plaintiff filed an application for leave to appeal in this Court, and we ordered oral argument on plaintiff's application, directing the parties to address

whether a reasonable jury could determine that the defendant's conduct was "the proximate cause" of plaintiff Kersch Ray's injuries where the defendant's actions placed the plaintiff in the dangerous situation that resulted in the plaintiff's injuries. MCL 691.1407(2)(c) ; Robinson v. City of Detroit, 462 Mich. 439, 462, 613 N.W.2d 307 (2000) ; Beals v. Michigan, 497 Mich. 363, 871 N.W.2d 5 (2015).^{[ ]}

II. STANDARD OF REVIEW

The applicability of governmental immunity is a question of law that is reviewed de novo. We also review de novo a trial court's decision regarding a motion for summary disposition.

III. BACKGROUND

Under the GTLA, governmental agencies and their employees are generally immune from tort liability when they are engaged in the exercise or discharge of a governmental function. The act provides several exceptions to this general rule. One such exception is in MCL 691.1407(2), which provides in pertinent part:

[E]ach ... employee of a governmental agency ... is immune from tort liability for an injury to a person or damage to property caused by the ... employee ... while in the course of employment ... if all of the following are met:

(a) The ... employee ... is acting or reasonably believes he or she is acting within the scope of his or her authority.

(b) The governmental agency is engaged in the exercise or discharge of a governmental function.

(c) The ... employee's ... conduct does not amount to gross negligence that is the proximate cause of the injury or damage.

Defendant, as a governmental employee, has the burden "to raise and prove his entitlement to immunity as an affirmative defense." There is no dispute regarding whether defendant acted within the scope of his authority as a governmental employee for the school or whether he was engaged in the exercise or discharge of a governmental function. And the issue of whether defendant was grossly negligent is not presently before this Court. Accordingly, the only issue before us is whether there is a question of material fact regarding whether defendant was "the proximate cause" of plaintiff's injuries under MCL 691.1407(2)(c).

IV. ANALYSIS

Proximate cause, also known as legal causation, is a legal term of art with a long pedigree in our caselaw. Proximate cause is an essential element of a negligence claim. It "involves examining the foreseeability of consequences, and whether a defendant should be held legally responsible for such consequences." Proximate cause is distinct from cause in fact, also known as factual causation, which "requires showing that ‘but for’ the defendant's actions, the plaintiff's injury would not have occurred." Courts must not conflate these two concepts. We recognize that our own decisions have not always been perfectly clear on this topic given that we have used "proximate cause" both as a broader term referring to factual causation and legal causation together and as a narrower term referring only to legal causation. All this broader characterization recognizes, however, is that "a court must find that the defendant's negligence was a cause in fact of the plaintiff's injuries before it can hold that the defendant's negligence was the proximate or legal cause of those injuries." In a negligence action, a plaintiff must establish both factual causation, i.e., "the defendant's conduct in fact caused harm to the plaintiff," and legal causation, i.e., the harm caused to the plaintiff "was the general kind of harm the defendant negligently risked." If factual causation cannot be established, then proximate cause, that is, legal causation, is no longer a relevant issue.

We take this opportunity to clarify the role that factual and legal causation play when analyzing whether a defendant's conduct was "the proximate cause" of a plaintiff's injuries under the GTLA. In any negligence case, including one involving a government actor's gross negligence, a court must determine whether "the defendant's negligence was a cause in fact of the plaintiff's injuries ...." But the court must also assess proximate cause, that is, legal causation, which requires a determination of whether it was foreseeable that the defendant's conduct could result in harm to the victim. A proper legal causation inquiry considers whether an actor should be held legally responsible for his or her conduct, which requires determining whether the actor's breach of a duty to the plaintiff was a proximate cause of the plaintiff's injury. It is not uncommon that more than one proximate cause contributes to an injury. However, under the GTLA, we have held that when assessing whether a governmental employee was "the proximate cause" of the plaintiff's injuries, a court must determine whether the defendant's conduct was "the one most immediate, efficient, and direct cause of the injury ...." Contrary to the dissent's assertion, "the proximate cause" is not determined by weighing factual causes. Such an approach distorts the meaning of "the proximate cause" by severing it from the concept of legal causation. There is no basis in our caselaw or our rules of statutory construction for interpreting "the proximate cause" as having nothing to do with the legal term of art "proximate cause." The Legislature amended the GTLA in 1986 to include the phrase "the proximate cause," and so we must determine what the term "the proximate cause" would have meant to the Legislature then. Before 1986, we can find no case that reads the term "proximate cause," even when preceded by a definite article, to mean "but-for cause." Although some of our earlier decisions refer to factual causation in connection with proximate cause, these cases merely reflect the unremarkable proposition that an actor cannot be a "proximate cause" without also being a "but-for cause." They do not contradict the well-established understanding of proximate cause, which, as we have long recognized, involves the foreseeability of the consequences of the conduct of human actors, regardless of whether "a proximate cause" or "the proximate cause" is at issue. Almost one hundred years ago this Court stated the rule regarding proximate cause as follows:

If a man does an act and he knows, or by the exercise of reasonable foresight should have known, that in the event of a subsequent occurrence, which is not unlikely to happen, injury may result from his act, and such subsequent occurrence does happen and injury does result, the

act committed is negligent, and will be deemed to be the proximate cause of the injury.^{[ ]}

This formulation, which is the "most general and pervasive approach" to proximate cause, has been repeatedly reaffirmed by this Court. We see no reason why this "peculiar and appropriate" legal meaning should not be ascribed to the Legislature's reference to proximate cause in the GTLA. Accordingly, we presume that by "proximate cause" the Legislature meant proximate cause.Most of our caselaw interpreting the GTLA is not to the contrary. In Robinson, we considered "whether the city of Detroit or individual police officers face[d] civil liability for injuries sustained by passengers in vehicles fleeing from the police when the fleeing car caused an accident." We held that the government officials in question were immune from suit because a different proximate cause—"the reckless conduct of the drivers of the fleeing vehicles"—was the proximate cause of the plaintiff's injuries. Robinson is therefore consistent with our holding today.

In Beals, we considered the defendant lifeguard's failure to intervene in the deceased's drowning. While our analysis in that case was somewhat opaque, we believe Beals is best understood as holding that the lifeguard could not have been "the proximate cause" of the decedent's drowning because the plaintiff failed to show even a genuine issue of factual causation. When a plaintiff attempts to establish factual causation circumstantially, that circumstantial proof must go beyond mere speculation. The plaintiff in Beals failed to make this showing. We emphasized that any connection between the lifeguard's breach of a duty and the drowning was only speculative. We also noted that "it [was] unclear that even a prudent lifeguard would have been able to observe and prevent the [deceased's] drowning," which further illustrated that the causal connection was "simply too tenuous." In other words, the plaintiff failed to show that the lifeguard was a but-for cause of the deceased's death. Accordingly, we held that the defendant lifeguard was not "the proximate cause" of the deceased's death for the purposes of the GTLA. The holding, if not all of the reasoning, of Beals is consistent with our understanding of the GTLA's use of "the proximate cause."

We recognize that our caselaw is not without its blemishes. In Dean v. Childs, the Court of Appeals held that the GTLA did not bar a claim against a firefighter who was alleged to be grossly negligent when fighting a house fire that killed the plaintiff's children. We issued an order summarily reversing for the reasons stated by the Court of Appeals' dissent, which had weighed the various factual causes of the children's deaths to conclude that the proximate cause "was the fire itself, not defendant's alleged gross negligence in fighting it."

This analysis was erroneous. Determining proximate cause under the GTLA, or elsewhere, does not entail the weighing of factual causes but instead assesses the legal responsibility of the actors involved. Moreover, because proximate cause is concerned with the foreseeability of consequences, only a human actor's breach of a duty can be a proximate cause. Consequently, nonhuman and natural forces, such as a fire, cannot be considered "the proximate cause" of a plaintiff's injuries for the purposes of the GTLA. Instead, these forces bear on the question of foreseeability, in that they may constitute superseding causes that relieve the actor of liability if the intervening force was not reasonably foreseeable. The dissenting Court of Appeals opinion in Dean failed to recognize these principles and erroneously concluded that the fire was the proximate cause of the children's deaths. Our subsequent order adopting the dissent was therefore in error. Accordingly, we overrule our order in Dean.

V. APPLICATION

The Court of Appeals' proximate cause inquiry confused proximate cause with cause in fact. The panel focused on whether plaintiff, the driver of the vehicle, and the vehicle itself were factual causes of plaintiff's injuries. This was a necessary inquiry because one's conduct cannot be the proximate cause without also being a factual cause. The panel's error was in its next step. Weighing these factual causes against defendant's actions, the Court of Appeals concluded that "there were obviously more immediate, efficient, and direct causes of [plaintiff's] injuries" than defendant's conduct. According to the panel, "clearly the most proximate cause of [plaintiff's] injuries is the fact that he was struck by a moving vehicle." The Court of Appeals' analysis failed to properly distinguish between factual causation and legal causation. The panel did not assess the legal responsibility of any of the actors involved, but instead attempted to discern whether any of the other factual causes was a more direct cause of plaintiff's injury than defendant's actions. This was error. Determining whether an actor's conduct was "the proximate cause" under the GTLA does not involve a weighing of factual causes. Instead, so long as the defendant is a factual cause of the plaintiff's injuries, then the court should address legal causation by assessing foreseeability and whether the defendant's conduct was the proximate cause.

To the extent the Court of Appeals' opinion attempted to analyze this issue, its analysis was incomplete. An appropriate proximate cause analysis should have considered the conduct and any legal responsibility therefor of defendant, plaintiff, and the driver of the vehicle that struck plaintiff. Further, before an actor can be a proximate cause, there must be the prerequisite determination that the actor was negligent—that is, that the actor breached a duty. In this case, the panel never determined whether the driver was negligent. Without that determination, his actions could not be a proximate cause of plaintiff's injuries. Similarly, the panel failed to correctly analyze whether plaintiff was negligent and a proximate cause of his own injuries. At the time of the accident, plaintiff was thirteen years old. Unlike adults, who are held to the reasonable person standard, determining whether a child was negligent requires application of a subjective standard. The court must assess whether the child acted with the degree of care that would reasonably be expected of a child of similar age, intelligence, capacity, and experience under the circumstances of the case. The Court of Appeals erred by singularly focusing on plaintiff's age without also considering plaintiff's subjective characteristics and the relevant factual context. Finally, even if the panel had determined that another actor was negligent and was a proximate cause of plaintiff's injuries, it still would have needed to determine whether defendant's actions were "the proximate cause." This would require considering defendant's actions alongside any other potential proximate causes to determine whether defendant's actions were, or could have been, "the one most immediate, efficient, and direct cause" of the injuries. If, on the basis of the evidence presented, reasonable minds could not differ on this question, then the motion for summary disposition should be granted. Because the Court of Appeals did not consider these issues in the first instance, we remand to that Court for reconsideration.

VI. RESPONSE TO THE DISSENT

Having read the dissent with care, we are simply perplexed. We agree with the dissent that one cannot be the or even a proximate cause without also being a cause in fact. Our opinion is very clear on this point. See, e.g., page 371 of this opinion ("All this broader characterization recognizes, however, is that a court must find that the defendant's negligence was a cause in fact of the plaintiff's injuries before it can hold that the defendant's negligencewas the proximate or legal cause of those injuries.") (quotation marks and citation omitted); page 371 of this opinion ("If factual causation cannot be established, then proximate cause, that is, legal causation, is no longer a relevant issue."); page 372 of this opinion ("[T]hese cases merely reflect the unremarkable proposition that an actor cannot be a ‘proximate cause’ without being a ‘but-for cause.’ "); page 373 of this opinion ("[O]ne's conduct cannot be the proximate cause without also being a factual cause."). We do not understand why the dissent repeatedly claims otherwise.

Neither can we follow the conclusion the dissent draws from our shared premise. We must remember that we are interpreting statutory language that the Legislature enacted in 1986. The question, therefore, is what the Legislature would have understood the phrase "the proximate cause" to mean in 1986. We believe that the answer to that question should draw on the decades of jurisprudence in this state, leading up to that date, defining "proximate cause." And, as our cases have uniformly held, one cannot be a or the"proximate cause" without being both a factual cause and a legal cause of the plaintiff's injuries.

The dissent's approach, however, would render legal cause irrelevant. Under the dissent's theory, any factual cause—even an inanimate one—can be "the proximate cause" if it is the "most immediate, efficient and direct" factual cause of the plaintiff's injuries. The dissent claims that this has been "the common-law meaning" attributed to the phrase " ‘the proximate cause’ ... in our jurisprudence since 1913." If that were true—if one's actions could, for now over one hundred years, have been "the proximate cause" without also being a legal cause—one would expect there to be volumes of cases from our Court standing for that proposition. But there is only one, our order in Dean, which we overrule today. Robinson did not hold that one's actions could be "the proximate cause" without also being a legal cause; indeed, we held in Robinson that the drivers' reckless conduct was the proximate cause of the accident. Neither did Beals ; but to extent that it relied on Dean, we disavow that portion of its reasoning. Not even Stoll, the 1913 case the dissent trumpets as the progenitor of the (elusive) "proximate cause without legal cause" line of cases, held any such thing. And the cases the dissent cites to support its unique pitch do not do so; they stand only for a proposition with which we fully agree: "Long before MCL 691.1407(2) was enacted in 1986—and many times since—our common-law has recognized that factual causation is ... an integral part of ‘proximate cause.’ " In fact, it is more than integral: factual causation is a condition precedent to proximate cause. That is, one's conduct cannot be a or the"proximate cause" of a plaintiff's injury without also being a factual cause thereof. But just because something is a factual cause of an injury does not mean it is a or the"proximate cause" thereof. "Proximate cause" has for a century in Michigan, like every other American jurisdiction, required both a finding of factual and legal cause. And that is undoubtedly what the Legislature would have understood when it used this legal term of art in the GTLA in 1986.

The dissent would jettison this understanding in favor of an approach that weighs factual causes to determine which among them is the "most immediate, efficient, and direct." At the outset, we are not sure how this theory can be derived from MCL 691.1407(2)(c). Why, one might wonder, would the Legislature choose language well-known in the law to denote one type of causation ("proximate cause") if it really meant another ("cause in fact")? And we struggle to understand what the dissent's approach would entail. Consider this case. What if: the coach had chosen a different route; the driver had arrived one minute earlier or one minute later to the intersection; plaintiff had not gone to cross-country practice that day; plaintiff had not joined the cross-country team; his parents had not chosen to live in this school district; his mother had not given birth to plaintiff; his parents had never met, etc. All of these, and more, are but-for causes of the accident. The "causes of an event go back to the dawn of human events, and beyond." How from those infinite causes a court, or anyone, could determine which is the most immediate, efficient, and direct but-for cause, we do not know. The dissent provides no guidance in this regard, instead finding it "easier to disparage the product of centuries of common law than to devise a plausible substitute ...." All the dissent offers is the legally unrecognizable assertion that defendant's alleged gross negligence was not the one most immediate, efficient, and direct cause of plaintiff's injuries because the injuries would not have occurred but-for plaintiff's conduct and that of the driver. But a test that allows the mere existence of other but-for causes—which are by definition present in every case—to immunize government actors from liability for their grossly negligent conduct is really no test at all.

Finally, in addition to being unsupported by the plain language of the statute or our caselaw, the dissent's interpretation fails to consider the statutory history of the GTLA. We have cautioned that "courts must pay particular attention to statutory amendments, because a change in statutory language is presumed to reflect either a legislative change in the meaning of the statute itself or a desire to clarify the correct interpretation of the original statute." Before 1986, the GTLA did not address whether government actors were immune from tort liability. After our 1984 decision in Ross v. Consumers Power Co. (On Rehearing) afforded qualified immunity from all tort liability to government actors if they met certain conditions, the Legislature amended the GTLA to create a narrow exception to this broad grant of immunity. The exception, which is at issue in this case, does not extend immunity to those government actors whose conduct amounts to "gross negligence that is the proximate cause of the injury or damage."

The dissent's approach, under which even inanimate objects could be the proximate cause, fails to give meaning to the 1986 amendment of the GTLA. That is, the dissent's approach would immunize government actors for every harm that is a foreseeable result of their gross negligence. Under the dissent's reasoning, as best we understand it, a government actor whose gross negligence foreseeably causes a fire that burns the plaintiff's house to the ground could avoid liability by blaming the fire, notwithstanding that the resulting harm is precisely that which the actor's conduct foreseeably risked. Likewise, the plaintiff foreseeably injured by the grossly negligent discharge of a government actor's firearm would have no recourse—it is the bullet that is always the more direct cause. Under this approach, government actors do not injure people, the implements they use do. The dissent provides no explanation to the contrary. We believe the dissent's theory fails to give meaning to the 1986 amendment by eliminating the narrow exception to government immunity created by MCL 691.1407(2)(c). If the Legislature had intended governmental officials to be absolutely immune from liability, we cannot make sense of the words they chose to accomplish that result. The 1986 amendment undoubtedly creates an exception, albeit a narrow one, to governmental immunity. To read it otherwise is to ignore the words of the statute. "Where the [statutory] language is unambiguous, we presume that the Legislature intended the meaning clearly expressed—no further judicial construction is required or permitted, and the statute must be enforced as written." Here, the Legislature amended the GTLA to provide a narrow exception to governmental immunity for grossly negligent acts that were "the proximate cause" of a plaintiff's injuries. "Proximate cause" has for a hundred years in this state, and elsewhere, been a legal term of art; one's actions cannot be a or the"proximate cause" without being both a factual and a legal cause of the plaintiff's injuries. We give the text its plain meaning.

VII. CONCLUSION

Proximate cause requires determining whether the defendant's negligence foreseeably caused the plaintiff's injuries. That negligence (or gross negligence in the case of the GTLA) cannot have been a proximate cause of a plaintiff's injury if it is not both a factual and legal cause of the injury. A court should take all possible proximate causes into account when determining whether the defendant was "the proximate cause," i.e., "the one most immediate, efficient, and direct cause of the injury ...." In this case, the Court of Appeals erred by instead attempting to discern whether the various but-for causes of plaintiff's injuries were a more direct cause of those injuries than defendant's alleged gross negligence, without first determining whether any of the asserted but-for causes were proximate causes. Accordingly, we vacate the Court of Appeals' decision and remand to the Court of Appeals for further proceedings not inconsistent with this opinion.

Bridget M. McCormack, Richard H. Bernstein, Joan L. Larsen, JJ., concur.

Wilder, J. (dissenting).I respectfully dissent. In my judgment, the Court of Appeals reached the correct outcome by duly applying our germane holdings in Robinson v. Detroit, 462 Mich. 439, 613 N.W.2d 307 (2000), and Beals v. Michigan, 497 Mich. 363, 871 N.W.2d 5 (2015). Because the Court of Appeals committed no error in its application of Robinson and Beals to the facts of the instant case, I would affirm.

I

Robinson is our seminal case interpreting the "employee provision" of the governmental tort liability act (GTLA) — MCL 691.1407(2) —which provides an exception to the GTLA's "broad" grant of tort immunity to governmental actors. Robinson, 462 Mich. at 455, 458, 613 N.W.2d 307. In pertinent part, MCL 691.1407(2) provides:

Except as otherwise provided in this section, and without regard to the discretionary or ministerial nature of the conduct in question, each officer and employee of a governmental agency, each volunteer acting on behalf of a governmental agency, and each member of a board, council, commission, or statutorily created task force of a governmental agency is immune from tort liability for an injury to a person or damage to property caused by the officer, employee, or member while in the course of employment or service ... while acting on behalf of a governmental agency if all of the following are met:

(a) The officer, employee, member, or volunteer is acting or reasonably believes he or she is acting within the scope of his or her authority.

(b) The governmental agency is engaged in the exercise or discharge of a governmental function.

(c) The officer's, employee's, member's, or volunteer's conduct does not amount to gross negligence that is the proximate cause of the injury or damage. [Emphasis added.]

In construing the phrase "the proximate cause" in Robinson, 462 Mich. at 459, 613 N.W.2d 307, we relied on several well-settled principles of statutory interpretation:

Because the Legislature is presumed to understand the meaning of the language it enacts into law, statutory analysis must begin with the wording of the statute itself. Each word of a statute is presumed to be used for a purpose, and, as far as possible, effect must be given to every clause and sentence. The Court may not assume that the Legislature inadvertently made use of one word or phrase instead of another. Where the language of the statute is clear and unambiguous, the Court must follow it.

These rules of statutory construction are especially germane in the cases now before us because Michigan strictly construes statutes imposing liability on the state in derogation of the common-law rule of sovereign immunity. This Court has repeatedly acknowledged that governmental immunity legislation evidences a clear legislative judgment that public and private tortfeasors should be treated differently. [Quotation marks and citations omitted.]

In light of such principles, we held that "[t]he Legislature's use of the definite article ‘the’ " in the phrase "the proximate cause" clearly demonstrated "an intent to focus on one cause." Id. at 458–459, 613 N.W.2d 307. Recognizing that our "duty is to give meaning to the Legislature's choice of one word over the other," we decided that the phrase " the proximate cause" must not be interpreted as synonymous with "a proximate cause." Id. at 461, 613 N.W.2d 307. Therefore, we afforded the phrase "the proximate cause" the common-law meaning that it has held in our jurisprudence since 1913:

We are helped by the fact that this Court long ago defined "the proximate cause" as "the immediate efficient, direct cause preceding the injury." Stoll v. Laubengayer, 174 Mich. 701, 706, 140 N.W. 532 (1913). The Legislature has nowhere abrogated this, and thus we conclude that ... the Legislature provided

tort immunity for employees of governmental agencies unless the employee's conduct amounts to gross negligence that is the one most immediate, efficient, and direct cause of the injury or damage, i.e., the proximate cause. [ Id. at 462, 613 N.W.2d 307 (emphasis added).]

The Stoll decision on which we relied in Robinson treated the question of proximate "cause" as one necessarily involving the consideration of factual (i.e., but-for) causation. Stoll, 174 Mich. at 706, 140 N.W. 532 ("But for this act of [the decedent] (subsequent to defendant's alleged negligent act, and therefore proximate to the injury) no accident could have occurred.").

In Beals, 497 Mich. at 365–366, 871 N.W.2d 5, we applied the principles set forth in Robinson to a factual scenario in which the plaintiff, autistic 19-year-old William Beals, drowned in an indoor swimming pool while in the presence of the defendant lifeguard, William Harman, who was a governmental employee. "Applying this Court's rationale in Robinson," we held that

Harman's failure to intervene in Beals's drowning cannot reasonably be found to be "the one most immediate,

efficient, and direct cause" of Beals's death. While it is unknown what specifically caused Beals to remain submerged under the water, the record indicates that Beals voluntarily entered the pool and voluntarily dove under the surface of the shallow end into the deep end without reemerging. Although plaintiff alleges that Harman's inattentiveness prevented him from attempting a timely rescue of Beals, in our view, it is readily apparent that the far more "immediate, efficient, and direct cause" of Beals's death was that which caused him to remain submerged in the deep end of the pool without resurfacing.

... That we lack the reason for Beals's prolonged submersion in the water does not make that unidentified reason any less the "most immediate, efficient, and direct" cause of his death. Consequently, while Harman's failure to intervene may be counted among the myriad reasons that Beals did not survive this occurrence, it certainly was not "the proximate cause" of his death for purposes of MCL 691.1407(2)(c). [ Beals, 497 Mich. at 373–374, 871 N.W.2d 5.]

In other words, even though Harman's conduct would undoubtedly have been recognized as "a" potential legal cause of the drowning under our ordinary negligence jurisprudence, in Beals we compared a number of potential factual (i.e., but-for) causes of Beals's drowning and determined that the behavior of the governmental employee was not "the" proximate cause.

It was unnecessary in Beals to analyze foreseeability or legal causation to conclude that the governmental actor was not the proximate cause of Beals's drowning—we were able to determine that Harman's conduct was not the one most immediate, efficient, and direct cause of Beals's death simply by comparing the but-for causes. That is because it is well settled in our negligence jurisprudence that the phrase "proximate cause" is "a legal term of art that incorporates both cause in fact and legal (or ‘proximate’) cause." Craig v. Oakwood Hosp, 471 Mich. 67, 86, 684 N.W.2d 296 (2004)emphasis added). See also Weymers v. Khera, 454 Mich. 639, 647, 563 N.W.2d 647 (1997) ("To establish proximate cause, the plaintiff must prove the existence of both cause in fact and legal cause."). Indeed, while the term "proximate cause" has also been used as a synonym for "legal cause," see, e.g., Haliw v. Sterling Heights, 464 Mich. 297, 310, 627 N.W.2d 581 (2001) ("[L]egal cause or ‘proximate cause’^{[ ]} normally involves examining the foreseeability of consequences, and whether a defendant should be held legally responsible for such consequences") (quotation marks and citation omitted), well before 1986 we recognized that the "sine qua non" of proximate cause is "cause in fact." See Glinski v. Szylling, 358 Mich. 182, 196–197, 99 N.W.2d 637 (1959) (plurality opinion by SMITH , J.). See also Moning v. Alfono, 400 Mich. 425, 439–440, 440 n. 13, 254 N.W.2d 759 (1977) (explaining that "[p]roximate cause encompasses a number of distinct problems," one of which is "[t]he problem of causation in fact"), quoting Prosser, Torts (4th ed.), § 42, p. 244. "As a matter of logic, a court must find that the defendant's negligence was a cause in fact of the plaintiff's injuries before it can hold that the defendant's negligence was the ... legal cause of those injuries." Craig, 471 Mich. at 87, 684 N.W.2d 296. See also 4 Harper, James & Gray, Torts (2d ed.), § 20.2, pp. 89–91 ("Through all the diverse theories of proximate cause runs a common thread; almost all agree that defendant's wrongful conduct must be a cause in fact of plaintiff's injury before there is liability. This notion is not a metaphysical one but an ordinary, matter-of-fact inquiry into the existence or nonexistence of a causal relation as lay people would view it."). In other words, but-for causation is an indispensable part of the "proximate cause" inquiry. If something is not a factual cause of an injury, it is not a "cause" of the injury at all, and ergo it cannot, as a matter of logic, be a legal cause. To hold otherwise is to accept the paradoxical outcome that an actor who did not cause an injury may nevertheless be deemed its "proximate cause."

The test announced by Robinson and reaffirmed by Beals fits within this analytical framework and satisfies this Court's duty to narrowly construe the exception to immunity provided by MCL 691.1407(2)(c). One cannot determine whether a governmental actor's conduct was "the one most immediate, efficient, and direct cause" of an injury without considering factual causation. In other words, to determine which cause—among more than one—was "most immediate, efficient, and direct," one must consider the panoply of but-for causes and weigh their immediacy, efficiency, and directness. Legal causation, on the other hand, only becomes a relevant consideration in GTLA litigation when "the one most immediate, efficient, and direct cause" of the injury is a cause to which the law will not assign liability. See Skinner v. Square D Co., 445 Mich. 153, 163, 516 N.W.2d 475 (1994) ("A plaintiff must adequately establish cause in fact in order for legal cause ... to become a relevant issue."). Put differently, legal cause acts only as a limitation to the scope of liability under the GTLA, not a means of assigning liability. See Moning, 400 Mich. at 439, 254 N.W.2d 759 ("Proximate cause encompasses a number of distinct problems including the limits of liability for foreseeable consequences.") (emphasis added).

Applying the above principles to the documentary evidence and well-pleaded allegations in this case, it is evident that defendant Eric Swager's conduct was not "the proximate cause" of plaintiff Kersch Ray's injuries for purposes of MCL 691.1407(2). Even assuming, for the sake of argument, that Swager was grossly negligent in directing the team to cross the road, and further assuming that Ray in fact heard Swager and entered the roadway on Swager's directive, no reasonable juror could conclude that Swager's conduct was the one most immediate, efficient, and direct cause of Ray's injuries. At a minimum, the 13-year-old Ray both could and should have verified that it was safe to enter the roadway before he voluntarily did so. See, e.g., Ackerman v. Advance Petroleum Transp., Inc., 304 Mich. 96, 106, 107, 7 N.W.2d 235 (1942) (ruling that an eight-year-old child had the mental capacity to understand the dangers of crossing the street); Pratt v. Berry, 37 Mich.App. 234, 238, 194 N.W.2d 465 (1971) ("[A]n ordinarily prudent seven-year-old child must look while crossing the street."). See also Stoll, 174 Mich. at 706, 140 N.W. 532 (holding that a five-year-old child's act of sledding into the street was the proximate cause of her death because but for that act she would not have been injured). As the Court of Appeals aptly reasoned:

The contention that Ray essentially ran into the road in blind obedience to a verbal instruction from Swager simply does not alter the undeniable reality that Ray entered the road under his own power and he was then struck by a moving vehicle driven by someone other than Swager. Had Ray himself verified that it was safe to enter the roadway, as did many of his fellow teammates, the accident would not have occurred. Likewise, had Platt not been driving on the roadway that morning, or had he otherwise avoided Ray, the accident would not have occurred. In these circumstances, there were obviously more immediate, efficient, and direct causes of Ray's injuries than Swager's oral remarks. [ Ray v. Swager, unpublished per curiam opinion of the Court of Appeals, issued October 15, 2015 (Docket No. 322766), pp. 3–4, 2015 WL 6087208.]

Put differently, it was either Ray's conduct or that of the driver—but not Swager's conduct—that was the one most immediate, efficient, and direct cause of Ray's injuries. But for Ray's conduct (his act of running into the roadway without assessing traffic despite a do-not-cross signal) and that of the driver (failing to yield to a pedestrian or failing to reasonably survey the roadway), Ray would not have sustained his injuries. Because Swager's alleged gross negligence was not "the" proximate cause of Ray's injuries, Swager was immune from suit under the GTLA and thus entitled to summary disposition.

II

Largely characterizing proximate cause as legal cause only, the majority incorrectly claims that "[o]ur first characterization of ‘proximate cause’ as meaning both ‘proximate cause’^{[ ]} and ‘but-for cause’ " did not occur until our decision in Skinner, 445 Mich. at 162–163, 516 N.W.2d 475. This ignores several earlier precedents, including our 1977 decision in Moning, 400 Mich. at 439, 440 n. 13, 254 N.W.2d 759 (explaining that "[p]roximate cause encompasses a number of distinct problems," one of which is "[t]he problem of causation in fact"), quoting Prosser, Torts (4th ed.), § 42, p. 244, the 1959 plurality decision in Glinski, 358 Mich. at 196–197, 99 N.W.2d 637 (opinion by SMITH , J.) (explaining that cause in fact is the "sine qua non" of proximate of cause), and our 1913 decision in Stoll, 174 Mich. at 706, 140 N.W. 532 ("But for this act of [the decedent] (subsequent to defendant's alleged negligent act, and therefore proximate to the injury) no accident could have occurred."). The majority asserts that over time this Court has improperly conflated factual and legal causation, but the very cases that the majority cites in support, including Craig, Moning, Skinner, and CharlesReinhart Co. v. Winiemko, demonstrate that this Court has on numerous occasions recognized that a proper proximate cause analysis cannot take place without consideration of the sine qua non, cause in fact. The Legislature is presumed to have been aware of such decisions when it amended the GTLA to afford a narrow exception to governmental immunity, and the majority cites no authority for the proposition that factual causation was not considered to be part of the proximate cause analysis under our common law in 1986. It is in this temporal context that we must consider the proper construction of the phrase "the proximate cause."

Because it is the Legislature's intent that controls, we are charged under MCL 8.3a with the duty of according the phrase "the proximate cause" the "peculiar and appropriate" legal meaning that it had at the time MCL 691.1407(2)(c) was enacted, not the meaning that we might wish to ascribe to that phrase today. See Sam v. Balardo, 411 Mich. 405, 425, 308 N.W.2d 142 (1981). Consequently, this Court's power to redefine over the years the phrase "proximate cause" for purposes of the common law does not grant it the constitutional authority to retroactively alter the meaning of that phrase when interpreting MCL 691.1407(2)(c).

Nevertheless, that is precisely what the majority now does. Long before MCL 691.1407(2)(c) was enacted in 1986—and many times since—our common law has recognized that factual causation is, like legal causation, an integral part of "proximate cause." See, e.g., Craig, 471 Mich. at 86, 684 N.W.2d 296 (" ‘Proximate cause’ is a legal term of art that incorporates both cause in fact and legal (or ‘proximate’) cause."); Moning, 400 Mich. at 439, 440 n. 13, 254 N.W.2d 759 ; Glinski, 358 Mich. at 196–197, 99 N.W.2d 637 (opinion by SMITH , J.); Stoll, 174 Mich. at 706, 140 N.W. 532 ("But for this act of [the decedent] (subsequent to defendant's alleged negligent act, and therefore proximate to the injury) no accident could have occurred.") (emphasis added). The majority, however, conflates these two prongs of proximate causation, treating "legal cause" as if it has something to do with causation, which it does not. See Dobbs, § 185, p. 622 ("Scope of liability, formerly termed proximate cause, is not about causation at all but about the significance of the defendant's conduct or the appropriate scope of liability in light of moral and policy judgments about the very particular facts of the case."). The majority's conflation in this regard leads it to construe the phrase "the proximate cause"—i.e., the one most immediate, efficient, and direct cause—as if it is centered mainly on legal causation (i.e., scope of liability), with cause in fact acting only as a threshold consideration. From a purely logical standpoint, however, it is not possible to use the foreseeability-based limitations on liability provided by the concept of legal cause to affirmatively decide which cause, among several, was most immediate, efficient, and direct. Only by comparing factual causes can one discern their relative immediacy, efficiency, and directness.

Stated another way, while accusing the dissent of "distort[ing] the meaning of ‘the proximate cause’ by severing it from the concept of legal causation," the majority seeks to divorce that same phrase largely from the concept of factual causation. But the majority does not explain how courts might determine what constitutes "the one most immediate, efficient, and direct cause" of an injury without weighing factual causes. Such a test—asking judges to determine what caused an injury without considering its potential causes—is akin to asking jurors to determine damages without any consideration of harm. The majority asks this state's jurists to take the "cause" out of causation. Yet, the majority provides no roadmap regarding how this state's jurists might go about accomplishing that feat.

Additionally, after holding that lower courts are not permitted to weigh factual causes when determining "the proximate cause" under MCL 691.1407(2)(c), the majority goes on to state that a proper analysis does"require considering defendant's actions alongside any other potential proximate causes to determine whether defendant's actions were, or could have been, ‘the one most immediate, efficient, and direct cause’ of the injuries." (Citations omitted). This statement makes no logical sense, however, unless factual causation is considered to be part of the "proximate cause" analysis. By "proximate cause," the majority actually means "legal cause." And as already explained, "legal cause" is a misnomer insomuch as it has nothing at all to do with causation; instead, it "involves examining the foreseeability of consequences, and whether a defendant should be held legally responsible for such consequences." Skinner, 445 Mich. at 163, 516 N.W.2d 475 ; see also Dobbs, § 185, p. 622 (explaining that legal cause "is not about causation at all ...."). It is impossible to compare the responsibility of different actors on the basis of foreseeability without comparing their conduct—i.e., the potential causes in fact. Put differently, one cannot determine legal cause without comparing factual causes. How "weighing" such causes differs from "considering" them "alongside" one another, the majority does not explain.

Nor does the majority explain how its expansive interpretation of MCL 691.1407(2)(c) comports with our well-established duty to construe exceptions to governmental immunity narrowly. See Yono v. Dep't of Transp., 499 Mich. 636, 641, 885 N.W.2d 445 (2016) ("[W]e are to narrowly construe exceptions to governmental immunity. ..."); Beals, 497 Mich. at 370, 871 N.W.2d 5 (noting that the GTLA's exceptions to the "general rule" of governmental immunity "must be narrowly construed"); Wilson v. Alpena Co. Rd. Comm., 474 Mich. 161, 166, 713 N.W.2d 717 (2006) ("Statutory exceptions to the immunity of governmental agencies are to be narrowly construed."); Mack v. Detroit, 467 Mich. 186, 196 n. 10, 649 N.W.2d 47 (2002) ("[A]s this Court has consistently held since its seminal case" on the subject, "exceptions to governmental immunity are narrowly construed."). In concert, Craig, Skinner, Moning, Glinski, and Stoll clearly demonstrate that before 1986, during 1986, and after 1986 factual causation was a well-established element of "proximate cause" under our common law. Thus, because the Legislature is presumed to be aware of judicial decisions interpreting the common law and to legislate in accordance with them, it must be presumed that when the Legislature used the phrase "the proximate cause" in MCL 691.1407(2)(c), it intended that phrase to include both factual causation and legal causation as described in Stoll, 174 Mich. at 706, 140 N.W. 532. Nevertheless, the majority treats "proximate cause" as if it merely denotes "legal cause," thereby construing MCL 691.1407(2)(c) as broadly as possible. By so ruling, the majority massively expands the exception to governmental immunity provided by MCL 691.1407(2)(c).

Indeed, the majority tacitly acknowledges this impact of its construction, stating that the current approach, which permits factual causation to be considered, "fails to give meaning to the 1986 amendment of the GTLA," leaving little room for exceptions to governmental immunity. That is, however, precisely what a narrow ly drawn exception does—it leaves little room for exceptions to the general rule.

Moreover, contrary to the majority's assertions, the statutory history of the GTLA does not compel the result reached by the majority. The majority aptly observes that legislative history and statutory history are distinct concepts. The cardinal rule of statutory interpretation, however, is that "[w]here the [statutory] language is unambiguous, ‘we presume that the Legislature intended the meaning clearly expressed—no further judicial construction is required or permitted, and the statute must be enforced as written.’ " Pohutski v. City of Allen Park, 465 Mich. 675, 683, 641 N.W.2d 219 (2002), quoting DiBenedetto v. West Shore Hosp., 461 Mich. 394, 402, 605 N.W.2d 300 (2000). One need not delve into statutory history, as the majority does, to discern that MCL 691.1407(2)(c) was intended to create an exception to governmental immunity. See Conroy v. Aniskoff, 507 U.S. 511, 519, 113 S.Ct. 1562, 123 L.Ed.2d 229 (1993) (Scalia, J., concurring). As Beals recognized, the plain language of the GTLA proves the point quite nicely standing alone. See Beals, 497 Mich. at 365, 871 N.W.2d 5 ("While governmental agencies and their employees are generally immune from tort liability under the [GTLA], MCL 691.1407(2)(c) provides an exception to this general rule ...."). Because nothing in the statutory history of the GTLA contravenes the plain meaning of the statutory language, the majority's discussion of statutory history, while perhaps academically interesting, is ultimately superfluous.

Nor does the obvious legislative intent to create an exception to governmental immunity obviate our duty to construe that exception narrowly. The majority disregards the presumption that the Legislature was aware, when it added MCL 691.1407(2)(c) to the GTLA, of our earlier decisions recognizing that exceptions to governmental immunity are narrowly construed. Had the Legislature wished to provide a broad exception, it would have employed language manifesting that intent. It did not.

In the alternative, the majority contends that a narrow interpretation of MCL 691.1407(2)(c) is untenable because it "would immunize government actors for every harm that is a foreseeable result of their gross negligence." This argument is a veiled reliance on the so-called "absurd results" doctrine. See generally Johnson v. Recca, 492 Mich. 169, 821 N.W.2d 520 (2012). "To properly invoke the ‘absurd results' doctrine," however, it must be shown "that it is quite impossible that the Legislature could have intended" the result in question. Id. at 193, 821 N.W.2d 520 (emphasis added). A proper analysis in this vein cannot rest merely on what seems to be the "obvious" legislative intent but should instead delve into substance and, in a given case, "might require a serious-minded analysis of the Legislature's policy objectives in enacting the statutes, the political realities and disagreements within the Legislature that adopted the statutes, the necessity for compromise and negotiation leading to enactment of the statutes, and the public impetus behind the statutes ...." Id. at 194, 821 N.W.2d 520. Because "this Court is not empowered to act as the people's lawmaker-in-chief," "it must be assumed that the language and organization of the statute better embody the ‘obvious intent’ of the Legislature than does some broad characterization surmised or divined by judges." Id. at 196–197, 821 N.W.2d 520. Here, the majority has failed to demonstrate that it is "quite impossible" that the Legislature intended the exception to governmental immunity set forth by MCL 691.1407(2)(c) to be a narrow one.

Furthermore, although it engages in little stare decisis analysis, the majority's holding is patently inconsistent with Robinson and Beals. A mere two years ago, the Beals majority of six—including two members of the instant majority—did precisely what the instant majority assigns as error in this case, comparing a number of factual causes of the drowning and determining that the behavior of the governmental employee was not the one most immediate, efficient, and direct cause. Beals, 497 Mich. at 373–374, 871 N.W.2d 5. Nor is Beals an outlier in that respect. The dissenting Court of Appeals opinion on which we relied in Dean v. Childs, 474 Mich. 914, 705 N.W.2d 344 (2005), employed the same analytical framework. See Dean v. Childs, 262 Mich.App. 48, 684 N.W.2d 894 ( GRIFFIN , J., concurring in part and dissenting in part.) Notably, the Beals majority cited Dean favorably, remarking that Dean and Beals were "analogous." Beals, 497 Mich. at 375, 871 N.W.2d 5.

Nevertheless, the instant majority overrules Dean while merely criticizing our conspicuously similar decisions in Robinson and Beals and disavowing Beals 's reliance on Dean. "[W]e should be consistent rather than manipulative in" our application of stare decisis. See Lawrence v. Texas, 539 U.S. 558, 587, 123 S.Ct. 2472, 156 L.Ed.2d 508 (2003) (Scalia, J., dissenting). To maintain such consistency, it is vital that this Court openly recognizes when it is issuing a holding that is inconsistent with settled precedent, even if only in part. It is poor practice for "this Court to simply ignore precedents with which it disagrees." Beasley v. Michigan, 483 Mich. 1025, 1029, 765 N.W.2d 608 (2009) ( CORRIGAN , J., dissenting). Doing so permits conflicting lines of caselaw to develop, which yields jurisprudential uncertainty and variations in outcome that turn solely on which holding—among our several inconsistent holdings—a lower court chooses to follow. Ultimately, this undermines the rule of law.

As this Court unanimously recognized earlier this term, "Generally, in order to ‘avoid an arbitrary discretion in the courts, it is indispensable that [courts] should be bound down by strict rules and precedents which serve to define and point out their duty in every particular case that comes before them[.]’ " Coldwater v. Consumers Energy Co., ––– Mich. ––––, ––––, 895 N.W.2d 154, 161 (2017), quoting The Federalist No. 78 (Hamilton) (Rossiter ed., 1961), p. 471 (second and third alterations in original). " ‘Liberty finds no refuge in a jurisprudence of doubt.’ " Lawrence, 539 U.S. at 586, 123 S.Ct. 2472 (Scalia, J., dissenting), quoting Planned Parenthood of Southeastern Pa. v. Casey, 505 U.S. 833, 844, 112 S.Ct. 2791, 120 L.Ed.2d 674 (1992). Therefore, "principles of law deliberately examined and decided by a court of competent jurisdiction should not be lightly departed," Coldwater, ––– Mich. at ––––, 895 N.W.2d at 161 (quotation marks and citation omitted).

In the instant case, if the majority disagrees with the holdings in Robinson and Beals (as its criticism of those decisions seemingly indicates), it should perform a stare decisis analysis and conclusively decide whether to overrule them. As former Chief Justice CORRIGAN once noted in a different context,

If it intends to alter legal principles embedded in this Court's decisions, then the ... majority should explain its reasons clearly and intelligibly. Instead, the ... majority overrules by indirection, or at least leaves the impression that it is doing so, thereby sowing the seeds of confusion and making it difficult for the citizens of this state to comprehend precisely what our caselaw requires. [ Beasley, 483 Mich. at 1030, 765 N.W.2d 608 ( CORRIGAN , J., dissenting).]

Here, because the majority avoids the stare decisis question altogether, treating its instant holding and the holding in Beals as if they are consistent, two lines of cases will almost inevitably arise in lower courts, one following the Beals analysis and another attempting to follow, to the extent that it might be possible to do so, the analysis set forth by the majority today. Then again, given the inherent unworkability of the purported "test" set forth by the majority, in the future lower courts may simply avoid the proximate cause issue altogether, instead focusing on the "gross negligence" requirement of MCL 691.1407(2)(c). Finally, in response to the majority's assertion that this opinion is "unsupported by the plain language of the statute or our caselaw," I simply note three things. First, it is this opinion that relies on the language of MCL 691.1407(2)(c) to support its interpretation of the GTLA, while the majority is compelled to resort to an analysis of statutory history. See pages 391–92 of this opinion. Second, I would construe MCL 691.1407(2)(c) by following Robinson, Dean, and Beals. Third, in stark contrast to the majority opinion, this opinion is not forced to overrule, disavow, or reconcile prior pronouncements from this Court in order to explain its result or avoid tension with existing precedent.

III

For all of those reasons, I dissent. Because the Court of Appeals reached the correct outcome for the correct reasons by following Robinson and Beals, I would affirm.

Stephen J. Markman, C.J., Brian K. Zahra, J., agrees.