FABE, Justice.
I.
INTRODUCTION
After experiencing near fatal injuries in a car accident, Kimberly Farnsworth sued General Motors (GM) in strict liability for designing a defective seat restraint system. Following a fiveweek trial, the jury returned a verdict for Farnsworth, awarding her $2,138,973 in compensatory damages and $5,600,000 in punitive damages. The superi- or court denied GM’s motion for a judgment notwithstanding the verdict and a new trial, and GM appeals. We conclude that the superior court erred in rejecting a comparative negligence instruction and refusing to require the jury to allocate fault to the driver who caused the accident. Because they may have affected the jury’s views of GM’s liability, these two errors also impugn the finding of causation and certain aspects of the damages awards. However, we find no error tainting the jury’s verdict that the seat restraint system was defective or its calculation of compensatory damages. We therefore remand for a new trial limited to the issues of comparative negligence, allocation of fault, causation, and punitive damages.
II.
FACTS AND PROCEEDINGS
On the evening of February 3, 1989, Farnsworth was riding as a passenger in a 1984 GMC Jimmy truck driven by Jon Fen-nie. On their way back to Anchorage from dinner in Girdwood, Farnsworth and Fennie
were driving north on the Seward Highway. As they rounded a curve, Charles Walters, who was driving in the wrong lane, crashed his Honda Accord head on into Fennie’s truck.
Although Fennie walked away from the accident with minor injuries, Farnsworth nearly died. She lingered in a coma for five weeks, underwent ten operations, was believed to be a candidate for a lung transplant, and had her spine reassembled using two metal rods. Among other complications, she lost the ability to walk and use her hands and had to relearn these skills through physical therapy. Three months after the accident, the hospital released her on supplemental oxygen and with a walker. Farnsworth continues to suffer from permanent restrictive lung disease and limited spinal mobility, which hamper her ability to participate in athletic activities. Due to massive internal adhesions, she also faces a high risk pregnancy if she should choose to have a child.
A.
Farnsworth’s Theory of the Case
At trial, Farnsworth contended that a defect in the Jimmy’s restraint system caused all of her significant injuries.
According to Farnsworth, the accident was a moderate, frontal collision that should not have resulted in serious injuries if her restraint system had worked properly. Her expert testified that each of the cars was moving approximately 30 m.p.h. when they collided, speeds that he said are more typical of accidents occurring in residential areas rather than on highways. He also testified that the change in velocity (delta-v) for the Jimmy was 21-25 m.p.h. Farnsworth argued that GM’s own studies showed that an individual in a properly designed seat restraint system should not suffer significant injuries even in a 32-33 m.p.h. delta-v collision.
Farnsworth claimed that the injuries she experienced resulted from “submarining” under her lap belt.
According to Farnsworth, “[submarining occurs when a lap belt moves up over the top of the load-bearing pelvis and into a person’s abdomen. During submarin-ing, the lap belt can cause massive abdominal injuries and fracture the spine.” Farnsworth argued that she had submarined under the lap belt because GM’s defective design only protected individuals the size of an average man or larger. Therefore, she claimed, Fen-nie, who at 175 pounds and 5'11" was slightly larger than an average sized man, walked away from the accident, whereas she, at 129 pounds and 5'3", nearly died.
Farnsworth argued that improper testing of the Jimmy had led to the belt’s defective design. According to Farnsworth, GM had used only a 50th percentile male dummy in testing the Jimmy’s restraint system. Farnsworth claimed that such testing was insufficient because information well known by the automotive industry establishes that submarining tendency increases as occupant size decreases. Thus, she asserted, GM’s
tests proved only that the Jimmy would protect individuals the size of an average man or larger from submarining. Farnsworth argued that GM should have conducted sled tests
using the 5th percentile female dummy, which is traditionally used by the automotive industry to test the belt’s safety for small occupants. If GM had conducted these tests, she contended, the defects in its seat restraint system would have been obvious.
Farnsworth’s design expert testified that the Jimmy’s seat restraint system failed to protect small occupants from submarining because it was defective in three ways: (i) it had an excessively shallow outer lap belt angle; (ii) the lap/shoulder belt buckle junction was too close to the seat occupant’s centerline (the “CJ” length was too short); and (iii) the seat did not contain an anti-submarining ramp. As proof, Farnsworth pointed to sled tests conducted on seat restraint systems analogous to the system in Fennie’s Jimmy in which the small female dummy consistently submarined. Farns-worth then introduced evidence showing that the small female dummy did not submarine during sled tests of newer Jimmy seats that incorporated the anti-submarining features suggested by her design expert.
Farnsworth asked the jury to award her compensatory as well as punitive damages. She claimed that punitive damages were appropriate in this case because, by testing the Jimmy’s restraint system with only the 50th percentile male dummy, GM knowingly exposed most of the population to the risk of submarining. Farnsworth suggested that the jury calculate punitive damages by multiplying the cost of installing anti-submarining features on each Jimmy, $3.50, by the number of trucks sold in all fifty states, 1.6 million.
B. GM’s
Theory of the Case
GM argued that Farnsworth’s injuries resulted not from a defect in the seat restraint system, but from the severity of the accident and her own misuse of the belt.
First, GM’s experts testified that the delta-v for Farns-worth was 27.9 m.p.h. rather than 24 m.p.h. as Farnsworth’s expert had concluded. Second, and more importantly, GM claimed that Farnsworth had misused the belt by wearing it under her arm. GM theorized that due to this misuse, Farnsworth had “jackknifed” over the belt, which in turn had caused all of her significant injuries. GM claimed that the pattern of Farnsworth’s physical injuries was more consistent with its theory of injury than with Farnsworth’s.
GM further claimed that its seat restraint system was not defective. In general, GM argued that no single design can protect all occiipants equally well and that a design that optimizes safety for the average sized man will decrease safety for women and children. Similarly, it argued, a design that gives slightly less protection against submarining may give slightly more protection against head injury. In particular, GM claimed that the three alleged defects identified by Farns-worth’s expert were not defects at all. GM argued: (i) that the outer lap belt angle was actually steeper on Farnsworth than it would be on an average sized man; (ii) that although the “CJ” length was shorter on Farnsworth than on an average sized man, Farnsworth’s design expert failed to show that the CJ length was significant in this accident; and (iii) that the seat did in fact have a 15 degree upward angle that functioned like an anti-submarining ramp.
Throughout the trial, GM also contested Farnsworth’s ability to prove causation. It argued that Farnsworth’s design expert failed to specify how different and “better” designs could have prevented Farnsworth’s injuries in this accident.
Additionally, GM stressed that even if its seat restraint system were defective, Farns-
worth’s injuries were largely the fault of Fennie and Walters. GM argued that Farnsworth did not contest that Walters was under the influence of cocaine at the time of the accident and that his negligence in driving in the wrong lane was the immediate cause of the collision. In fact, GM claimed, Walters had admitted that he was drug impaired, and he was tried and convicted of criminal assault in the fourth degree for causing personal injury to Fennie and Farns-worth. GM also attributed fault to Fennie, arguing that the collision would have been less severe if he had braked more quickly. GM claimed that Fennie’s own driving had been impaired by alcohol; the evidence at trial showed that he had drunk a double gin before dinner and shared two bottles of champagne with Farnsworth and another friend during dinner.
Finally, GM claimed that punitive damages were inappropriate. It argued that the crash tests it conducted on the 1984 Jimmy were the most representative and meaningful tests available. GM explained that sled tests are the usual testing tool because full car crash tests cost more and because it is unusual for a complete car to be available at the time that the seat restraint system is tested. The 1984 Jimmys, however, were fully completed when the seats were ready for evaluation. As a result, GM argued, its engineers took advantage of the unusual opportunity and conducted the full car crash tests. The engineers conducted the tests with the 50th percentile male dummy because federal standards so required. GM claimed that the tests did not show a submarining problem and that the company received no complaints after the cars were placed on the market to suggest that submarining was in fact a problem for small occupants. It argued that contrary to revealing its disregard for the safety of consumers, the fact that its engineers conducted more expensive full car crash tests proved its commitment to producing safe ears.
C.
The Verdict
The jury returned a verdict for Farns-worth, awarding her $2,138,973 in compensatory damages and $5,600,000 in punitive damages. It found that the seat restraint system was defective and that the defect was a legal cause of her injuries. Although the jury concluded that Fennie was negligent, and the superior court instructed that Walters was negligent as a matter of law, the jury found that neither Fennie nor Walters was a legal cause of Farnsworth’s injuries. It then allocated 100% fault for her injuries to GM.
III.
DISCUSSION
A.
Alleged Errors in the Jury Instructions
1.
Standard of review
In reviewing the superior court’s rulings on jury instructions, we determine whether the challenged or refused instruction is a correct statement of the law and then decide whether any error was prejudicial.
See Loof v. Sanders,
686 P.2d 1205, 1211 (Alaska 1984). In evaluating whether the instruction correctly stated the law, we apply our independent judgment.
See Cummings v. Sea Lion Corp.,
924 P.2d 1011, 1019 n. 11 (Alaska 1996). If we decide that an error occurred, we evaluate whether the er ror was prejudicial by putting ourselves “in the position of the jurors and determining] whether the error probably affected their judgment.”
Vincent By Staton v. Fairbanks Mem’l Hosp.,
862 P.2d 847, 851 (Alaska 1993). The appellant bears the burden of establishing prejudicial error.
See Myers v. Robertson,
891 P.2d 199, 208 (Alaska 1995).
2.
Did the superior court err in refusing to instruct the jury on Farnsworth’s comparative negligence ?
GM argues that the superior court erred when it refused GM’s request for a jury instruction on Farnsworth’s comparative negligence. GM contends that this ruling significantly affected the verdict because its central defense theory was that Farnsworth’s misuse of her belt, coupled with the force of the accident itself, caused her injuries.
Farnsworth claims that according to our prior decisions, comparative negligence is a defense in strict liability cases only where the plaintiff knowingly and unreasonably uses a defective product,
not
in cases of
product misuse. We disagree with her interpretation of the case law.
a.
Comparative negligence in strict liability cases
The superior court rejected GM’s proposed instruction because it concluded that
Dura Corp. v. Homed,
703 P.2d 896 (Alaska 1985) permits the defense
of comparative
negligence only in those strict liability cases where the plaintiff knowingly uses a defective product. In
Dura,
the defendant contended that the plaintiff used a product even though he was aware that it lacked a safety device.
See Dura,
703 P.2d at 403-04. We therefore applied a specific set of rules for deciding what kind of plaintiff negligence is relevant in a case where the alleged defect was the absence of a safety device.
See id.
at 404. We did not suggest, however, that comparative negligence was a defense to a strict liability claim only when a plaintiff knowingly used an unsafe product. In fact, we explicitly stated that the comparative negligence defense also extends to cases where the plaintiff misused the product.
See id.
at 403.
Dura
⅛ language represents the law of this state as it was established in
Butaud v. Suburban Marine & Sporting Goods, Inc.,
555 P.2d 42, 46 (Alaska 1976).
In
Butaud,
the plaintiff claimed that he had been injured by a defective snow machine.
See Butaud,
555 P.2d at 46-47. Addressing the manufacturer’s claims on appeal, we ruled that comparative negligence is a defense to strict liability in tort.
See id.
at 45. Further, we stated that the defense of comparative negligence in strict liability cases is not limited to situations where “the plaintiff uses the product with knowledge of the defective condition, but
also extends to those cases where the plaintiff misuses the product and that misuse is a proximate cause of his injuries.’’ Id.
at 46 (emphasis added).
Butaud
itself concerned the second type
of
comparative negligence, product misuse, which GM contends occurred in this case. As a result,
Butaud
and
Dura
clearly support GM’s interpretation of the law.
b.
Harmless error analysis
Because we conclude that the superi- or court erred in refusing GM’s proposed comparative negligence instruction, we must now examine if this error was harmless. An error is harmless “if the verdict necessarily
shows that the jury would have rejected the defense contained in a requested instruction had the instruction been given.”
Weston v. State,
682 P.2d 1119, 1122 (Alaska 1984).
Farnsworth argues that the superior court’s failure to give the comparative negligence instruction was harmless because of the “all or nothing” strategy that she pursued at trial. According to Farnsworth, the theory of the ease that she presented to the jury was premised on the view that it should rule in her favor only if it found that she had used the belt properly. Thus, although Alaska law may not have barred Farnsworth from recovering against GM even if she had misused the belt,
see
AS 09.17.060, AS 09.17.900, Farnsworth claims that she asked the jury to find GM liable only if it concluded that no misuse had occurred. Because the jury in fact ruled in her favor, Farnsworth contends that it must have rejected GM’s belt misuse theory, thus rendering harmless the failure to instruct on comparative negligence.
Farnsworth’s argument, however, is premised on the assumption that the jurors clearly understood her theory of the case: that GM should be held liable only if they found that Farnsworth had used the belt properly. But we are not convinced that the jurors comprehended Farnsworth’s trial strategy. First, we note that the testimony presented on this issue at trial is equivocal. Although Farns-worth and her biomechanie expert, Dr. Martha Bidez, suggested that GM should be absolved of liability if Farnsworth had worn the belt under the arm, Farnsworth’s design expert, Steve Syson, testified merely that such misuse would “degrade” the belt’s performance; he did not claim that GM should be free of all responsibility if the belt had been misused.
Similarly, in closing argument, Farnsworth’s lawyers failed to mention her “concession” that belt misuse was a complete defense for GM.
Second, and more importantly, none of the
instructions
informed the jury of Farns-worth’s theory of the ease.
Cf. Cummings v. Sea Lion Corp.,
924 P.2d 1011, 1019 n. 11 (Alaska 1996) (stating that the “superior court’s denial of a proposed jury instruction will be upheld if the instructions given, when read as a whole, adequately inform the jury of the relevant law”) (citations omitted). In fact, Instruction 19 may have suggested to the jury that GM could be held responsible for Farnsworth’s negligence.
Instruction 19 stated: “[i]n determining whether a product is used in a reasonably foreseeable manner, the focus is on what is foreseeable to manufacturers. Foreseeable uses include not only careful or intended product use, but also
product misuse
that is known or can be anticipated to occur.” (Emphasis added.) Farnsworth advocated for this instruction because she claimed that “reasonably foreseeable use in this ease [was] ... the use of the belt under the arm_” The superior court included the instruction based on this argument.
Absent further explanation, this instruction was inappropriate in the context of this case. Farnsworth was not alleging, nor could she have, that the belt was defective because it failed to protect her when worn under the arm. Instead, she claimed that her injuries
were caused by submarining. Product misuse entered this case only as a defense for GM. Instruction 19 failed to make this clear to the jurors and instead may have suggested to them that GM could be held fully liable for Farnsworth’s injuries even if she had worn the belt under her arm.
Reviewing the events at trial and the instructions as a whole, we reject Farns-worth’s claim that the jury’s verdict necessarily shows that it rejected GM’s belt misuse theory. Instead, we conclude that GM has carried its burden of showing that the error of denying the comparative negligence instruction probably affected the jurors’ judgment. The instructions provided no opportunity for the jury to allocate fault to Farnsworth if it concluded that she had worn the belt under the arm. As a result, we conclude that the issue of Farnsworth’s comparative negligence must be submitted to the jury on remand.
3.
Did the superior court err in refusing to instruct that Walters caused Farns-worth’s injuries as a matter of law?
GM argues that the superior court erred in refusing to instruct the jury that Walters caused Farnsworth’s injuries as a matter of law. We agree.
Walters was tried and convicted of criminal assault in the fourth degree for “causing personal injury” to Fennie and Farnsworth. No one disputed that Walters’s negligence in fact caused the accident. Analogizing to other areas of tort law, GM argues that the original tortfeasor is always considered a legal cause of a plaintiffs further injuries unless there is a superseding cause. Because Farnsworth did not allege that the defects in the restraint system were a superseding cause, GM claims that the only question that should have been posed to the jury was the proportion of Walters’s fault.
GM’s position is consistent with related tort principles. An original tortfeasor is considered a proximate cause, as a matter of law, of injuries caused by subsequent negligent medical treatment.
See
Restatement (Second) of Torts, § 457 (1965) (“If the negligent actor is liable for another’s bodily injury, he is also subject to liability for any additional bodily harm resulting from normal efforts of third persons in rendering aid which the other’s injury reasonably requires, irrespective of whether such acts are done in a proper or a negligent manner.”). Just as
the injury caused by the original tortfeasor exposes the injured individual to the potential negligence of doctors, so did Walters’s negligence in causing the accident expose Farnsworth to the potential defects in the seat restraint system.
Farnsworth suggests that Walters should not be considered a legal cause of injuries caused by her seat belt because a seat belt’s sole purpose is to protect someone after a collision occurs. However, a doctor, like a seat belt, is necessary only
because
something goes wrong. Phrasing the issue in terms of foreseeability, it is just as foreseeable to an original tortfeasor that equipment in a car may malfunction as it is that a doctor may act negligently in treating the plaintiffs injuries.
Holding Walters liable as a matter of law for Farnsworth’s enhanced injuries also furthers one of the policy goals of tort law. Tort law seeks to deter future behavior that exposes others to injury. In this ease, Walters drove while he was under the influence of cocaine. His actions caused the collision that exposed Farnsworth to the defect in her seat restraint system. Allowing a jury to return a verdict allocating no fault to Walters for her injuries frustrates the purpose of deterring Walters and others from driving while they are under the influence of drugs.
Finally, ruling that Walters caused Farns-worth’s enhanced injuries as a matter of law does not significantly reduce the jury’s role in the complicated factual determination of allocation of fault. Adopting GM’s position only slightly reduces the jury’s field of choice: instead of allocating fault to Walters from 0% to 100%, the jury would have had to allocate some fault to Walters. As a result, we conclude that the superior court erred in refusing to instruct the jury that Walters caused Farnsworth’s injuries as a matter of law. The court should give GM’s requested instruction on remand.
4.
Did the superior court err in instructing the jury on apportionment of damages?
Instruction 24 provided that if the jury found that a defect in the seat restraint system was a substantial factor in causing Farnsworth’s enhanced injuries, GM could be held liable for all of her injuries unless GM proved that they could be apportioned between enhanced and non-enhanced injuries. GM claims that Instruction 24 did not belong in the ease, and that even if it did, it placed the burden of apportionment on the wrong party.
An apportionment of damages instruction is appropriate only when the plaintiff is alleging that a single harm, such as paraplegia, was caused both by the defect and the severity of the crash itself.
See, e.g., Mitchell v. Volkswagenwerk, AG,
669 F.2d 1199, 1201 (8th Cir.1982). In such cases, the instruction informs the jury that once the plaintiff proves that the defendant’s defective design was a substantial factor in causing the injury, the defendant must be held jointly and severally hable for the entire harm.
See id.
at 1201-02. The instruction is based on the reasoning that it is impossible to divide a single harm: a plaintiff cannot be expected to prove what portion of an injury like paraplegia was caused by the defect and what portion was caused by the collision.
See id.
at 1203 n. 2. The apportionment instruction therefore permits the jury to hold the car manufacturer hable for all of the plaintiffs damages even though the car manufacturer is theoretically responsible only for damages caused by the defect.
It is true that in this ease neither Farnsworth nor GM contended that some of Farnsworth’s injuries were caused by the defect and some were not; this fact suggests that apportionment was not at issue in this case.
See Czarnecki v. Volkswagen of America,
172 Ariz. 408, 837 P.2d 1143, 1148 (Ariz.App.1991) (stating, in a crashworthiness case, that an apportionment instruction would not
be proper where both sides had pursued an all or nothing strategy). But when Judge Shortell offered to eliminate the apportionment instruction if GM stipulated that the injuries were
not
apportionable, GM declined. In these circumstances, it was reasonable for Judge Shortell to include the instruction in order to provide guidance to the jury in case it decided to reject the parties’ all or nothing approach.
Having concluded that the superior court did not err in including Instruction 24, we now turn to the question of whether the instruction accurately stated the law of apportionment. GM argues that the instruction should have placed the burden of apportioning injuries on Farnsworth, and Farnsworth claims that the instruction correctly placed the burden on the defendant, GM. The issue of who has the burden to apportion injuries in a crashworthiness case is hotly contested. See Restatement (Third) of Torts,
supra,
§ 16 cmt. d (describing the debate in the cases and law reviews);
see also
Note,
Apportionment of Damages in the “Second Collision” Case,
63 Va. L.Rev. 475 (1977) (advocating Farns-worth’s position); Levenstam & Lapp,
Plaintiffs Burden of Proving Enhanced Injury in Crashworthiness Cases: A Clash Worthy of Analysis,
38 De Paul L.Rev. 55 (1989) (advocating GM’s position). The rule advocated by Farnsworth is the majority position, dubbed the
Fox-Mitchell
approach.
See Mitchell v. Volkswagenwerk, AG,
669 F.2d 1199 (8th Cir.1982). It is also the position supported by the drafters of the third Restatement of Torts.
See
Restatement (Third) of Torts,
supra,
§ 16. The minority position, advocated by GM, is referred to as the
Huddell
approach, named after
Huddell v. Levin,
537 F.2d 726 (3rd Cir.1976).
Adopting the reasoning of the
Huddell
court, GM contends that the plaintiff must bear the burden of apportioning injury because a crashworthiness claim rests on the idea that a defect enhanced the plaintiffs injuries.
See Huddell,
537 F.2d at 738. Thus, according to GM, a plaintiff who fails to show what injuries he or she would have sustained absent the defect has failed to show enhancement and therefore failed to establish a crashworthiness claim.
Relying on the
Fox-Mitchell
approach, Farnsworth responds that the burden of apportionment belongs to the defendant. In
Mitchell,
the court reasoned that the same rules should apply to a crashworthiness claim as apply to claims of harm caused by multiple tortfeasors: once a plaintiff proves that the tortious conduct of two or more defendants has combined to bring about harm to the plaintiff, the burden is placed on the defendants to limit their liability by apportioning the injury.
See Mitchell,
669 F.2d at 1206 (citing Restatement (Second) of Torts,
supra,
§ 433B).
Farnsworth also argues that three policy reasons justify placing the burden of apportionment on the defendant. First, it would be unjust to allow a proven wrongdoer to escape liability when the harm cannot be apportioned. Second, the manufacturer is in the best position to perform such apportionment given its extensive knowledge of the product and its technical resources. Third, a contrary rule would allow a manufacturer to escape liability even when a plaintiff had shown that the manufacturer had sold a defective product.
Using the concept of “enhanced” injury as the framework for a crashworthiness case does lead one to the conclusion that a plaintiff who fails to show what injuries he or she would have suffered absent the defect has failed to show enhancement at all. However, it is just as reasonable to use the concept of multiple tortfeasors as the framework for an indivisible injury claim. Once this framework is adopted, the rules of section 433B of the Restatement of Torts are triggered and the burden of apportioning injury falls on the defendant. Alaska has explicitly adopted the Restatement’s rules for apportioning injury in multiple tortfeasor cases.
See State v. Abbott,
498 P.2d 712, 728 (Alaska 1972);
see also
Restatement (Third) of Torts,
supra,
§ 16 cmt. d (stating that the majority of courts that have addressed the issue as well as the drafters of the Restatement find the analogy to section 433B compelling).
Although both the
Fox-Mitchell
approach and the
Huddell
approach are logically defensible, we find the majority rule more compelling for policy reasons. We agree with Farnsworth that it should be the proven wrongdoer who must bear the burden of limiting its liability; it would be unfair to require a plaintiff who has already proved that a defect was a substantial factor in causing his or her injuries to try another case based upon what
might
have happened absent the defect. Thus, we conclude that Instruction 24 accurately stated the law of apportionment by placing the burden of apportioning injury on the defendant. The superior court committed no error with respect to this instruction.
5.
Did the superior court err in instructing the jury that it could find defect under the consumer expectation test?
GM next argues that the superior court erred in permitting the jury to find the restraint system defective under the consumer expectation test. We reject GM’s argument.
In
Caterpillar Tractor Co. v. Beck,
593 P.2d 871 (Alaska 1979), we adopted a two-prong test for finding defect in design cases. Under this test, the factfinder can find a product defective either
if the plaintiff demonstrates that the product failed to perform as safely as an ordinary consumer would expect when used in an intended or reasonably foreseeable manner [consumer expectation test] ... or if the plaintiff proves that the product’s design proximately caused his injury and the defendant fails to prove ... that on balance the benefits of the challenged design outweighed the risk of danger inherent in such design [risk benefit test].
Id.
at 884 (citing
Barker v. Lull Engineering Co.,
20 Cal.3d 413, 143 Cal.Rptr. 225, 573 P.2d 443, 457-58 (Cal.1978)).
GM argues that the consumer expectation test does not make sense in a complex design case like this one and urges that it should be abandoned for three reasons. First, it contends that consumers may have unreasonable or uninformed expectations. Second, GM claims that the test is standardless and thus leads to haphazard verdicts. Third, it argues that the test is particularly out of place in complex design cases because consumers have no basis for forming expectations about how products like cars should perform in serious accidents. GM buttresses its position by pointing out that the proposed new Restatement of Torts and the majority of states do not allow recovery unuer the consumer expectation test without also requiring that the product’s risks outweigh its benefits.
Farnsworth responds that the consumer expectation test is well established in Alaska’s product liability law and has been applied in numerous cases since it was adopted by this court in
Beck.
She also contends that even if the test were inappropriate in certain situations, this case does not present one of those situations because seat belts are familiar products about which consumers can form reasonable expectations. In addition, Farnsworth challenges GM’s interpretation of the case law and insists that only a minority of states reject the consumer expectation test as an alternative to the risk benefit test.
We agree with the California Supreme Court that consumers can form reasonable and educated expectations about how certain products should perform.
See Soule v. General Motors Corp.,
8 Cal.4th 548, 34 Cal.Rptr.2d 607, 882 P.2d 298, 310 (Cal.1994) (stating “we cannot accept GM’s insinuation that ordinary consumers lack any legitimate expectations about the minimum safety of the products they use. In particular circumstances, a product’s design may perform so unsafely that the defect is apparent to the common reason, experience, and understanding of its ordinary consumers.”).
When a seat belt, designed to be an instrument of protection, becomes an instrument of life-threatening injury, a consumer is justified in concluding that it did not perform as safely as promised. A seat belt is a familiar product whose basic function is well understood by the general population.
In addition, our initial reasons for adopting the consumer expectation test are sound. In
Beck,
we adopted the test as part of Alaska’s product liability law partly because it “incorporates notions of the implied warranty of fitness for reasonable use, a primary concept in the evolution of strict products liability....”
Beck,
593 P.2d at 885. The ‘implied warranty of fitness or merchantability requires that goods ‘be fit for the ordinary purposes for which the goods are used.’”
Id.
at 885 n. 49 (quoting AS 45.05.096(b)(3), renumbered as AS 45.02.314(3)).
In this case, a jury that believed Farns-worth’s version of the facts could intelligently conclude that the Jimmy’s restraint system was unfit for ordinary use. According to Farnsworth, she and Fennie were wearing their seat belts normally in a typical frontal collision of moderate severity. Fennie was protected by his seat belt and walked away from the accident with minor injuries. In contrast, Farnsworth submarined under her lap belt and lingered near death in a coma for five weeks. The parties did not contest that the seat belt itself caused her injuries. Given these facts, there is no reason why a jury would be “haphazard” in deciding that Farnsworth’s seat restraint system was unfit for ordinary use because it failed to protect smaller occupants from submarining. The superior court did not err in permitting the jury to find the restraint system defective under the consumer expectation test.
B.
Other Alleged Errors
1.
Did Farnsworth present sufficient evidence to support the jury’s finding of causation?
GM argues that there was insufficient evidence to support a finding that defects in the restraint system caused Farnsworth’s injuries. Although we believe that the record would support a finding of causation, we do not fully address GM’s argument because we conclude that the failure to give a comparative negligence instruction requires retrial of the causation issue.
“Where two issues are interwoven and not reasonably divisible ... a new trial on both issues is required.”
Caterpillar Tractor Co. v. Beck,
624 P.2d 790, 795 (Alaska 1981). In this case, the issue of comparative negligence was closely tied to the issue of causation. GM argued not only that Farnsworth had been negligent in misusing the seat belt but also that her negligence caused her injuries; according to GM’s theory of the case, Farns-worth was injured because her belt misuse caused her to “jackknife” over the belt. As we concluded above, however, the jury instructions provided no opportunity to attribute fault to Farnsworth for her alleged negligence. In fact, Instruction 19 may have suggested to the jury that GM could be held responsible for any belt misuse and thus may also have suggested that GM could be held accountable for injuries
caused
by this misuse. We therefore conclude that the superi- or court’s error in failing to give a comparative negligence instruction taints the finding of causation. The parties should retry this issue on remand.
2. Challenges to the punitive damages award
GM challenges the punitive damages award on four grounds: (i) that its compliance with applicable federal regulations precluded punitive damages as a matter of law; (ii) that there was insufficient evidence to support a finding of punitive damages; (iii) that the award was excessive; and (iv) that the award was extraterritorial and thus unconstitutional under
BMW of North America, Inc. v. Gore,
517 U.S. 559, 116 S.Ct. 1589, 134 L.Ed.2d 809 (1996).
Addressing other claims raised by this appeal, we concluded above that the superior court erred in rejecting an instruction on Farnsworth’s comparative negligence and in refusing to require the jury to allocate fault to Walters. Because these erroneous rulings may have increased GM’s culpability in the jury’s eyes by diminishing the culpability of other potential wrongdoers, we vacate the punitive damages award. We therefore need not address GM’s arguments on excessiveness, sufficiency of the evidence or extraterritoriality
because the jury may award a different amount of punitive damages or no punitive damages on remand. We also do not address GM’s argument regarding its compliance with relevant federal regulations. Because GM raised this point for the first time on appeal,
the superior court never had the opportunity to address it and thus there is no decision for us to review.
See Nenana City School Dist. v. Coghill,
898 P.2d 929, 934 (Alaska 1995).
IV.
CONCLUSION
We REMAND for a new trial limited to the issues of causation, punitive damages,
Farnsworth’s comparative negligence, and allocation of fault to Walters and Farnsworth. Because no error tainted the jury’s finding that GM’s seat restraint system was defective or its compensatory damage calculation of $2,138,973, these aspects of the verdict should stand on remand.
REVERSED in part, AFFIRMED in part, and REMANDED.
EASTAUGH, J., not participating.