Concept
The But-For Test
But for and substantial causes
The But-For Test
Negligence is a legal cause of harm only if it made a difference to whether the harm occurred. The Restatement (Second) of Torts writes that requirement as a rule of elimination: under section 432(1), and subject to an exception in Subsection (2), negligent conduct is not a substantial factor in bringing about harm if the harm would have been sustained even had the actor not been negligent.1 Because section 431(a) makes a substantial factor a requirement of legal cause, a defendant who fails the elimination question is not a legal cause of the harm.2
The rule takes this shape because of what a judgment does. Shifting a loss onto a defendant is justified when the defendant's wrong changed what happened to the plaintiff. If the harm would have come anyway, the judgment charges the defendant for a loss the carelessness did not produce, and it gives no one a reason to take more care, since the skipped precaution would not have prevented the injury. The but-for question therefore sets a floor: a defendant who fails it leaves the case, and a plaintiff who passes it must still satisfy the other requirements of legal cause.
Applying the rule is harder than stating it, for two reasons. The comparison runs between what happened and what would have happened had the defendant used reasonable care, and no one can observe that second world. And accidents seldom have only one necessary antecedent, so several negligent acts can satisfy the test for the same harm. The three subsections below take up the rule itself, the proof of the counterfactual, and the multiple-cause cases.
The Necessary Antecedent Rule
Courts have traditionally analyzed cause in fact in negligence cases with the “but for” test, and the Restatement (Third) of Torts states it in the same terms: conduct is a factual cause of harm when the harm would not have occurred absent the conduct. Cause in fact will almost always be established when the test is met. A few quirky situations, taken up at the end of this chapter, produce results that courts find unsatisfactory, but they are the exception.3
Applying the test means replaying the accident. The factfinder looks at what did happen and compares it with what would have happened had the defendant not been negligent. If the injury would not have resulted without the negligence, the negligence is a sine qua non of the injury, a necessary antecedent, an event “without which not,” and the cause-in-fact element is met.4 The replay removes only the negligence and leaves the rest of the defendant's conduct in place. A test that deleted the whole course of conduct would make every participant a cause, since a driver who never left the garage never collides, and it would screen no one out. The factfinder therefore substitutes the care a reasonable person would have used and asks whether the harm still follows.
The failures show the test most clearly, because in them the replay ends where the real events did. Phillips may have been negligent in removing a saw's blade guard, but Mancuso was hurt by fumes that presumably would have reached him with the guard in place. The accident happens the same way, so removing the guard is not a cause in fact of his injury.5
Statutory violations show the gap between breach and causation at its widest, because a statute can set the standard of care while its violation still fails the test. McGee might sue ChemHaul for delivering a tank during working hours in violation of a statute that requires deliveries outside them. The statute was probably aimed at keeping deliveries out of busy periods, when accidents are likelier, so it may establish a relevant standard of care. But ChemHaul had finished, and the tank was all hooked up, when the accident happened, so presumably the accident would have occurred the same way had the tank arrived the night before instead of ten minutes before.6
A missing fire escape works the same way, and the evidence of where the victim died decides the case. If Swensson is found in the middle of the hall, his family can argue that the lack of a fire escape “changed history”: perhaps he reached the back window, could not get out, and ran back toward the front. That would show that he died because of the missing fire escape and that, but for the landlord's negligence, he would be alive today. Establishing that link is the plaintiff's burden.7 Had he never left his bed, a fire escape would have changed nothing.
The question travels well beyond collisions, because it asks only whether a wrong made a difference. When a driver crashes into a traffic warning sign whose flashing arrow is not working and injures the plaintiff, recovery is barred if the driver, being asleep, would not have seen a working arrow.8 Professional negligence follows the same logic. A lawyer owes the client the care a reasonable professional would provide, and a malpractice claim for negligence closely resembles a medical malpractice claim.9 Causation of damages is not presumed, however serious the lapse: the client must prove that he or she would have achieved a more favorable result but for the negligence.10 A lawyer who misses the deadline to answer a complaint therefore causes no harm when the client had no defense, because the client would have lost anyway.11 Products cases are no different. Negligence remains the classic claim in American products liability law, and it is still the principal theory of recovery in states that did not adopt strict products liability, so a plaintiff who sues a manufacturer in negligence faces the same causal showing.12
A literal but-for link can also arise by coincidence, which is why some courts and scholars have suggested that negligence must also increase the probability that the plaintiff will be harmed before it counts as a cause in fact.13 A shipper negligently delays a load of computers, and a tornado destroys the warehouse where the delay left them. Without the delay the computers would have been elsewhere, so the test is literally met, yet the delay did nothing to make the loss more likely. In nearly every ordinary case a but-for cause does raise the probability of the harm, so the caveat leaves the elimination analysis undisturbed, although it overlaps with the scope-of-liability limits previewed earlier in this chapter.
Proving the Counterfactual
Section 432(1) asks what would have happened without the negligence, and no one can observe the answer.15 The accident occurred once, with the defendant's carelessness inside it, so the factfinder must reconstruct a second history in which the defendant used reasonable care and judge how it would have ended. A motorcyclist is killed after pulling out in front of a bus travelling seven miles per hour over the limit, and the bare facts do not say whether the bus could have stopped at the lawful speed. A patient with chest pains is not diagnosed as having a heart attack, is not given suitable emergency treatment, and dies of the attack, and the bare facts do not show whether prompt treatment would have saved him.16
The comparison cannot be skipped, because the defendant's argument has force: even a negligent defendant should not be held liable unless the negligence made a difference.17 The law keeps the exercise workable by setting the standard of proof short of certainty. The plaintiff need only convince the jury that it is more probable than not that she would not have been injured but for the defendant's negligence; she need not show that the harm could not possibly have happened without it.18 A demand for certainty would defeat nearly every claim, since no evidence can show with certainty what would have happened in a world that never existed. A willingness to accept whatever the plaintiff can imagine would let verdicts rest on nothing and fall on defendants whose carelessness may have changed nothing, the result section 432(1) exists to prevent. The burden of establishing causation stays with the plaintiff, and the later section on burden, court, and jury develops who decides what.19
What makes the proof manageable is evidence that supplies facts for the comparison. In the bus case, the distance between the bus and the motorcycle when the rider entered the road, together with expert testimony on stopping distances, would let the jury work the question out. In the heart-attack case, medical evidence might show that prompt treatment probably would have saved the patient, or that death was inevitable even with immediate care. Certainty never arrives, but a jury can usually make a reasoned judgment about whether the negligence contributed to the outcome.20
The same comparison arises when the missing history turns on what a person would have done. Smith shoots Doe, and the defendant's negligence is tied to Smith's gun permit. The jury might conclude that Smith would have carried a gun and shot Doe regardless. But the evidence might instead lead the jury to conclude that Smith would not have carried a gun without a permit, and that Doe would therefore not have been shot. Doe need not show that Smith could not possibly have shot her without the permit. Some speculation is unavoidable, but one can imagine facts on which a jury could reasonably find it more probable than not that she would not have been injured but for the negligence.21
At the other pole, a court will not let a jury fill a gap in the plaintiff's evidence by guessing. In one reported case, the plaintiff argued that the condition of a set of steps caused a man named Bacon to stumble and fall. Bacon testified that he had stumbled on the same steps before through his own clumsiness, not because of their condition.22 The most that could be inferred was that he was about to climb the stairs when he stumbled and fell. Why he stumbled would still require speculation, and neither suspicion nor speculation is enough to defeat a motion for summary judgment.23
The line between inference and conjecture runs through the evidence. An inference is a conclusion a reasonable person may draw from facts the record establishes, and conjecture is a possibility offered with no facts behind it. Words like “might” and “conceivably” warn that an argument has left the evidence.
Between full proof and conjecture, the law lets a plaintiff reach the jury without direct evidence of the counterfactual in one kind of case. When a car collides more than an hour after sundown with an unseen buggy proceeding without lights, a causal connection may be inferred between the collision and the lack of signals. If nothing else is shown to break the connection, the case is prima facie sufficient on negligence contributing to the result, as in Martin v. Herzog.24 Zuchowicz states the principle generally. If (a) a negligent act was wrongful because it increased the chances that a particular type of accident would occur, and (b) a mishap of that very sort did happen, the trier of fact may find that the negligence caused the harm. The negligent party must then bring forward evidence denying but-for cause and suggesting that, in the actual case, the wrongful conduct was not a substantial factor.25 The reason the conduct was negligent already supplies the link between the conduct and the harm, and the defendant whose carelessness created the very risk that materialized is the party best placed to show that the risk played no part this time. Both halves of the premise must be present, so the inference is unavailable when the harm came from a different kind of accident than the one that made the conduct negligent.
Multiple But-For Causes
Section 431(a) asks whether the actor's conduct is a substantial factor in bringing about the harm, and the indefinite article carries weight.27 Every occurrence has many causative antecedents, and the cause for which the defendant is responsible need not be the sole cause, nor the last or nearest one.28 Because section 432(1) eliminates a defendant only when the harm would have been sustained without the negligence,29 every defendant whose negligence was necessary stays in the case, whatever else was also necessary.
The rule reads this way because of how many conditions any injury requires. A collision needs two travelers whose paths cross, and the defendant's careless act is only one entry on that list. If a negligent defendant could escape by pointing to any other necessary condition, whether another person's carelessness or a patch of wet pavement, almost nobody would be liable: each careless actor would point at the other, and the plaintiff would lose against both. Treating every negligent necessary antecedent as a cause preserves the incentive to take care, because any of those actors could have prevented the accident by taking the precaution owed. How to divide the loss among the contributors is a separate question of apportionment.
A two-driver collision shows the rule at work. Sullivan drives without looking carefully, and Pei runs a stop sign coming out of a side street. Sullivan hits Pei, and his car careens into a pedestrian. The jury finds that the accident would not have happened without Pei's running the stop sign, and also that, even assuming Pei's negligence, it would not have happened had Sullivan kept a proper lookout. Neither act alone was enough to cause the accident, but each was a necessary antecedent, so each is a sine qua non and both are but-for causes. Neither driver can answer that the other's negligence also contributed, because the defendant's act need be a but-for cause and not the sole one.30
The same reasoning holds when the negligent acts are separated in time. If A negligently spills gasoline and hours later B negligently ignites it, injuring P, A and B are joint tortfeasors because their independent negligence concurred to cause P's single, indivisible harm.31
An innocent contributor does not excuse a negligent defendant either. The defendant may not escape responsibility because an “innocent” occurrence such as an “act of God,” or other tortious conduct, may also have been a concurrent cause of the injury.32 Suppose Corbusier neglects his brakes, and on a rainy day his car fails to stop at a spot where a puddle has made the road slick. The crash would not have happened without the puddle, but it also would not have happened with properly maintained brakes, so the puddle's part does not excuse him.
A defendant's activity can be a necessary antecedent without the defendant's negligence being one. Dobbs drives carefully to the store for milk, and Fletcher negligently pulls out of a side street; the cars collide and Schwartz is hurt. Without Dobbs's trip there would be no collision, so his driving is a necessary antecedent in the bare sense. But a negligent actor is liable only if the conduct is negligent toward the other and the negligence is a legal cause of the other's harm.33 The replay swaps Fletcher's careless pullout for a careful one, and the collision disappears; Dobbs's trip contains no negligence to swap, so the claim against Dobbs fails for want of negligence, an element separate from causation. Zuchowicz draws the same line in a medical setting: that the drug Danocrine was likely responsible for the plaintiff's disease did not suffice, and the factfinder had to conclude that the overdose, the negligent part, was the cause.34
The usual consequence of finding several but-for causes is joint liability for an indivisible injury. If Joshua and Jessica negligently drive cars that combine to break Alyssa's hip, traditional law holds both liable for the whole. If Joshua's car injures her right arm and Jessica's car injures her left leg, the injury divides, and each answers only for his or her own portion.35 Where no fair method of apportioning the loss exists, the common-law result is joint and several liability for the total. Many jurisdictions have narrowed that rule by apportioning damages according to fault, so the local rule on dividing the loss must be checked separately.36 Each of these rules operates after the causal finding and leaves it undisturbed.
Every example so far involves causes that were each necessary, which is why the test works and every negligent contributor passes it. A different problem arises when each of two negligent acts would alone have produced the whole harm, as when two negligently set fires combine to destroy a building. Neither fire is necessary, so the sine qua non rule is inapplicable, and both defendants may be held jointly and severally liable for the whole loss.37 That problem belongs to the substantial factor test, which comes next, and it should be kept apart from the cooperating-cause cases above, where removing either act ends the accident.
The plaintiff's own negligence can be one of the necessary antecedents. The factual analysis does not change when it is: the harm would not have occurred without the plaintiff's carelessness, and would not have occurred without the defendant's, so the plaintiff's negligence is one more but-for cause. Whether it bars or reduces recovery is a separate question for the jurisdiction's contributory or comparative negligence rules, and the answer does not alter the finding that both people's negligence caused the harm.