GUY, J.

A 15-month-old child suffered irreparable brain damage after inhaling baby oil. The child's parents and guardian ad litem sued the manufacturer of the oil, Johnson Johnson Baby Products Co., alleging that the injury was proximately caused by the manufacturer's negligence in failing to put a warning on the container. The jury returned a verdict for the child and his parents. The trial judge granted Johnson Johnson a judgment notwithstanding the verdict, or, alternatively, a new trial. The Court of Appeals reinstated the jury verdict. We affirm the Court of Appeals.

FACTS

In the early evening of April 23, 1985, 15-month-old David Ayers and his twin brother were playing in their parents' house. Mrs. Ayers was getting ready to go to work, and Mr. Ayers was watching television. Two of David's sisters were also at home. David entered a bedroom, where he found a purse belonging to his 13-year-old sister, Laurie, on the floor and open. Inside the purse was an unmarked container that Laurie had filled with Johnson Johnson's baby oil for use at school after her gymnastics class. Mrs. Ayers unexpectedly came across David just as he began to drink the oil. In immediate and concerned reaction, she yelled at him to stop, causing him to gasp and inhale some of the oil into his lungs. Mrs. Ayers immediately took the container away from David. She was relieved to discover that it held baby oil transferred from its original container, as she believed that the only effect would be diarrhea. Mrs. Ayers inspected the original container to verify her understanding and found no warning. Interpreting the absence of any warning as an indication that there was no need for concern, she told two of her older children, who were to baby-sit the twins while she was at work, to call her if any problems developed. Expert testimony established that even if Mrs. Ayers had at this time realized the danger, immediate medical attention would not have prevented the oil from diffusing throughout David's lungs, reducing their ability to deliver oxygen to his blood.

Later that evening, David began to have difficulty breathing. His family took him to a hospital, where doctors placed him on a respirator. His progress was not good, and several days later he was sent to St. Louis for extracorporeal membrane oxygenation therapy, a special procedure involving pumping the blood outside the body and enriching it with oxygen. David improved slightly, but shortly after returning to Washington he suffered a cardiac arrest, which in turn led to brain damage.

Today David cannot move his arms or legs, which are stiff and spastic. He has limited control of his head movements. He cannot speak, is mentally retarded, and is subject to seizures.

David's parents and his guardian ad litem (hereafter the Ayerses) brought a product liability action against Johnson Johnson based on failure to warn of the danger of aspirating baby oil. After a 5-week trial, the jury found in favor of the Ayerses, awarding $2 million to David and $500,000 to Mr. and Mrs. Ayers. The trial court granted Johnson Johnson's motion for judgment notwithstanding the verdict on the grounds that the evidence failed to establish proximate causation or foreseeability of harm. The court alternatively ordered a new trial based on jury misconduct. The Court of Appeals reinstated the jury verdict. Ayers v. Johnson Johnson Baby Prods. Co., 59 Wn. App. 287, 797 P.2d 527 (1990). We affirm the Court of Appeals.

ISSUES

This case presents three issues for review. First, did the Ayerses present sufficient evidence to establish that the absence of a warning on the baby oil container proximately caused David's injury? We hold that the Ayerses' evidence was sufficient to sustain the jury's verdict in their favor.

Second, did the Ayerses establish that Johnson Johnson's baby oil was not reasonably safe because adequate warnings were not provided at the time of manufacture, as required under RCW 7.72.030(1)(b)? Resolution of this issue requires that we discuss the nature of a product liability claim based upon inadequate warnings at the time of manufacture, and, in particular, that we address whether foreseeability is an element of such a claim. We hold that the test for inadequate warnings under RCW 7.72.030(1)(b) is based on a strict liability standard, that the test requires no showing of foreseeability, and that it entitled the jury to find that the product was not reasonably safe because adequate warnings were not provided.

Third, did the trial court err in ruling that a new trial was required because of juror misconduct? We hold there was no cognizable juror misconduct, and that in any case the verdict was validated when the trial court polled the jury in open court.

ANALYSIS I

In a product liability suit alleging inadequate warnings, the plaintiff must show that his or her injury was proximately caused by a product that was "not reasonably safe because adequate warnings or instructions were not provided". RCW 7.72.030(1). The trial court granted Johnson Johnson's motion for judgment nothwithstanding the verdict on the grounds that the Ayerses failed to prove that the failure to warn of the dangers of aspirating baby oil proximately caused David's injuries. The Court of Appeals disagreed, holding that the Ayerses had sufficiently established proximate causation to support the jury's verdict. Ayers, at 294. We concur with the Court of Appeals.

[1] The trial court should grant a motion for judgment notwithstanding the verdict only if the court can say, as a matter of law, "that there is neither evidence nor reasonable inference therefrom to sustain the verdict." Hojem v. Kelly, 93 Wn.2d 143, 145, 606 P.2d 275 (1980). The court must view all evidence in the light most favorable to the nonmoving party. Hojem, at 145. The motion should be granted only if the court can say, as a matter of law, that no reasonable person could have found in favor of the nonmoving party. At the same time, the trial court must evaluate the evidence supporting the verdict in order to ensure that the verdict was not founded on mere theory or speculation and that the evidence supporting the verdict was substantial. Hojem, at 145.

In the present case, the jury's conclusion that the absence of a warning proximately caused David's injuries was not so unsupported or unreasonable that a judgment notwithstanding the verdict was appropriate.

[2, 3] To show proximate causation, the plaintiff must show both cause in fact and legal causation. Baughn v. Honda Motor Co., 107 Wn.2d 127, 142, 727 P.2d 655 (1986). "Cause in fact refers to the `but for' consequences of an act — the physical connection between an act and an injury." Hartley v. State, 103 Wn.2d 768, 778, 698 P.2d 77 (1985). In the present case, when the evidence and all inferences therefrom are viewed in a light most favorable to the Ayerses, the evidence adequately supports the jury's verdict on the question of cause in fact.

Expert testimony established that once David inhaled the oil there was no way to get it out, and medical attention would not have prevented injury to his lungs. Therefore, the issue as regards cause in fact does not concern whether had there been an adequate warning, Mrs. Ayers would have been alerted to the danger when she inspected the label on the bottle and so sought immediate medical help. Rather, the issue concerns whether had there been an adequate warning, David never would have inhaled the oil because the Ayerses would have kept it out of his reach. This was what the parties contested at trial and is the issue we examine here.

As the Court of Appeals noted, members of the Ayers family testified that they kept items they knew to be dangerous out of the reach of the twin baby boys. Ayers, 59 Wn. App. at 291. Mrs. Ayers testified that she made a practice of reading labels on products, and that she shelved them at home according to what she read on the labels. Items she knew to be particularly dangerous, such as cleaning waxes or bathroom cleansers, were shelved up high in a cupboard above the kitchen stove or in a box on the top shelf of the bathroom closet. Items she perceived as less dangerous were treated with less care. Before David's accident, she regarded baby oil as one of these less dangerous items, thinking that the only danger it presents is diarrhea if ingested. Accordingly, she usually kept it on the dresser in the babies' room, where she could reach it easily. This location was not, she testified, out of the babies' reach. She testified that if she had been aware of the dangers of aspiration, the baby oil would have been kept up high in the medicine box. She also said that if she had been aware of the dangers, she would have alerted other members of the family. She had specifically told her teenage daughters that if they were carrying in their purses anything that could be dangerous to the twins, such as personal toiletries like perfume, they were to keep the purses out of the twins' reach. David's sister, Laurie, in whose purse David found the baby oil, testified that she thought baby oil might cause an upset stomach if digested, but that she too had no idea of the dangers of aspiration.

On the basis of this evidence, the jury was entitled to infer that if the Ayerses had known of the dangers of aspiration, they would have treated baby oil with greater care; that they would have treated it with the caution they used in relation to items they recognized as highly dangerous, like cleaning products; and that had they done so, the accident never would have occurred. We conclude that the evidence of causation presented to the jury was sufficient to sustain the jury's verdict.

Johnson Johnson argues that the absence of any warning on the baby oil container was not a cause in fact of David's injury. First, Johnson Johnson asserts that the Ayerses knew before the accident occurred that baby oil is for external use only and should be kept out of the reach of children. Johnson Johnson reasons that since the Ayerses already knew this, it is wholly speculative that a further warning on the container would have caused them to modify their behavior in a way that would have prevented the accident. Johnson Johnson also emphasizes that to reach the conclusion that the absence of a warning caused David's injury, one must assume that had there been a warning, it would have been heeded by Mrs. Ayers, that she would have communicated the need for caution to the other members of her family, and that Laurie would not have left her purse on the bedroom floor. Johnson Johnson asserts that under these circumstances it is "rank speculation" to suppose a warning would have prevented the injury.

We reject this argument. All the Ayerses apparently knew was that baby oil could cause diarrhea if swallowed. They did not know of the risks of aspiration, and the evidence they presented, as described above, is sufficient to support the jury's conclusion that if they had been alert to those risks, they would have treated the product more carefully. At most, Johnson Johnson's argument suggests that reasonable persons might disagree as to whether a warning would have made any difference. For this court to uphold the trial court's judgment notwithstanding the verdict, however, more is required. This court must be prepared to conclude that no reasonable person could infer, as did the jury, that a warning would have altered the Ayerses' behavior. The evidence presented at trial was not so weak as to permit such a conclusion.

In a related argument, Johnson Johnson contends that the Ayerses were required to prove the exact wording of a warning they allege would have been adequate to prevent the injury. This argument is based upon RCW 7.72.030(1)(b), which requires the trier of fact in a failure to warn case to consider whether "the manufacturer could have provided the warnings or instructions which the claimant alleges would have been adequate." Johnson Johnson argues that this statute requires a claimant in warnings cases to put before the jury "warnings or instructions which the claimant alleges would have been adequate." Johnson Johnson reasons that a jury cannot decide that a warning would have prevented an accident without knowing exactly what that warning is. The dissenting opinion in the Court of Appeals' decision agreed with this position. Ayers, at 301 (Reed, J., dissenting).

[4] We reject Johnson Johnson's argument and hold that the language of RCW 7.72.030(1)(b) does not require a claimant to establish the exact wording of the alternative warning. The statute's requirement that "the manufacturer could have provided the warnings or instructions which the claimant alleges would have been adequate" is satisfied if the claimant specifies the substance of the warning. Here, the Ayerses contend that had they been warned of the dangers of aspirating baby oil, the accident would have been avoided. This suffices to indicate the nature of the warning the Ayerses allege Johnson Johnson should have provided and therefore satisfies the requirement of RCW 7.72.030(1)(b). Moreover, requiring claimants in failure to warn cases to establish the exact wording of an alternative warning would impose too onerous a burden. The members of the jury might agree that a certain type of warning should have been provided, but they might not agree among themselves as to exactly how that warning should have been worded.

[5] In addition to arguing that the absence of a warning was not a cause in fact of David's injury, Johnson Johnson further contends that the legal cause requirement is not met. Legal causation depends on considerations of "logic, common sense, justice, policy, and precedent." King v. Seattle, 84 Wn.2d 239, 250, 525 P.2d 228 (1974). It involves the "determination of whether liability should attach as a matter of law given the existence of cause in fact." Hartley v. State, 103 Wn.2d 768, 779, 698 P.2d 77 (1985).

Johnson Johnson asserts that the likelihood of a child aspirating baby oil is extremely low, and that requiring Johnson Johnson to warn against such a remote risk would require it and other manufacturers to deluge consumers with warnings about a vast assortment of other products, including liquids and solids sold for human ingestion such as milk, carrots, or candy. Such a deluge of indiscriminate warnings, argues Johnson Johnson, would lead to warnings being less effective — the "boy-who-cried-wolf" syndrome. Various amici, in briefs submitted in support of Johnson Johnson's petition for review of the Court of Appeals' decision, also make this argument. Thus, the Product Liability Advisory Council, Inc., and the Cosmetic, Toiletry and Fragrance Association argue that overwarning increases the likelihood that important warnings will not be read and heeded, and that it causes some consumers to overreact. The American Tort Reform Association makes the same argument, and also contends that the result of upholding the Court of Appeals' decision would lead to tremendous potential liability and uncertainty for manufacturers. Such a situation, the American Tort Reform Association argues, would impede the development of new products and cause manufacturers to withdraw existing products.

We are unpersuaded by this argument. The evidence presented at trial indicates that baby oil is distinguishable from other products. Baby oil, which is comprised of 99 percent mineral oil, poses special risks if inhaled, and the manner of its use — on babies and around water — creates a significant risk of aspiration. Dr. James Cunningham, a pediatric lung disease expert, testified that mineral oil is a bland substance that does not elicit as good a gag or cough reflex as many other substances, and that its viscosity causes it to render ineffectual the lungs' mucous ciliary clearance mechanism for clearing foreign substances. Dr. Cunningham also testified that macrophages, which are cells in the lungs that surround foreign particles and remove them, cannot function against mineral oil. In a 1985 medical journal article Dr. Cunningham coauthored that discusses the unique dangers of aspirating baby oil, the authors conclude:

The latent danger of this substance needs to be publicized, along with the potentially grave prognosis accompanying aspiration of large amounts of this product. The inadequacy of present therapeutic modalities makes the prevention of baby oil aspiration even more important.

Clerk's Papers, at 101. This article also explains that the aspiration of mineral oil poses a significant risk of lipoid pneumonia. Dr. Marvin Scotvold, chief of dermatology at Children's Hospital and Medical Center in Seattle, testified that he teaches resident physicians training under him not to use baby oil on children under the age of 3 years because of the danger of aspiration. Dr. Scotvold further testified that he felt strongly that there should be a warning about aspiration on bottles of baby oil. Among the reasons he gave for this view is that the product is sold for use on and around babies, who have a well-known tendency to put things in their mouths, and for use in and around water, where the chances of the baby aspirating some liquid containing the oil are increased.

In short, uncontroverted evidence presented at trial demonstrates the danger of aspirating baby oil. We recognize that other oils may be equally as dangerous, and that other products may have attributes that render them no less dangerous than baby oil. What makes baby oil unique, and what is the sine qua non of our decision, is that baby oil is intended for use on babies. When he drank the oil, David Ayers was acting in conformity with the behavior to be expected of any 15-month-old child. It is this kind of predictable infant behavior which necessitates that consumers and parents be alerted to the dangers of a product promoted for use on babies.

Our holding is limited to the proposition that baby oil, because it is intended for use on and around babies, may be a proximate cause of injuries due to aspiration if sold without an adequate warning of the danger of such injuries. We reject the notion that any product with physical properties similar to, or as dangerous as, those of baby oil may for that reason alone create manufacturer liability for failure to warn. Because our holding is limited in this way, we reject the argument that upholding the jury's verdict in favor of the Ayerses will dramatically encourage overwarning.

We conclude that the Ayerses provided sufficient evidence of causation to support the jury's verdict in their favor. We next turn to the issue whether the Ayerses provided sufficient evidence that Johnson Johnson's baby oil was not reasonably safe because adequate warnings or instructions were not provided at the time of manufacture.

II

Under RCW 7.72.030(1), "[a] product manufacturer is subject to liability to a claimant if the claimant's harm was proximately caused by the negligence of the manufacturer in that the product was not reasonably safe as designed or not reasonably safe because adequate warnings or instructions were not provided." The definition of "not reasonably safe", for purposes of a suit alleging failure to warn of dangers existing at the time of manufacture, is provided by RCW 7.72.030(1)(b) (hereafter subsection (b)):

A product is not reasonably safe because adequate warnings or instructions were not provided with the product, if, at the time of manufacture, the likelihood that the product would cause the claimant's harm or similar harms, and the seriousness of those harms, rendered the warnings or instructions of the manufacturer inadequate and the manufacturer could have provided the warnings or instructions which the claimant alleges would have been adequate.

Also relevant to a determination of "not reasonably safe" is the consumer expectation test of RCW 7.72.030(3), which states:

In determining whether a product was not reasonably safe under this section, the trier of fact shall consider whether the product was unsafe to an extent beyond that which would be contemplated by the ordinary consumer.

The trial court interpreted these statutes as requiring the Ayerses to prove that David's injuries were a foreseeable consequence of Johnson Johnson's failure to place a warning on its baby oil containers. The trial court ruled that the evidence submitted by the Ayerses was insufficient to meet this requirement. The Ayerses contend they were not required to prove foreseeability because it is not an element of a failure to warn claim under subsection (b). They further allege that even if they were required to prove foreseeability, they did so. The Court of Appeals held that foreseeability is not an element of a failure to warn claim under subsection (b), which the court interpreted as providing a strict liability test. Ayers, at 294. Applying that test, the court concluded the jury was justified in finding that baby oil is a dangerous product that should have been accompanied by a warning. Ayers, at 295. The Court of Appeals also applied the consumer expectation test of RCW 7.72.030(3), and held that a jury could find that the product was unsafe to an extent beyond that which would be contemplated by the ordinary consumer. Ayers, at 296. We agree with the Court of Appeals that foreseeability is not an element under subsection (b), and concur with that court's application of both the subsection (b) test and the consumer expectation test.

Johnson Johnson contends that foreseeability is an element of a product liability claim based on inadequate warnings for two reasons. First, Johnson Johnson emphasizes that the word "negligence" is used in RCW 7.72.030(1): "A product manufacturer is subject to liability to a claimant if the claimant's harm was proximately caused by the negligence of the manufacturer". (Italics ours.) Absent evidence to the contrary, Johnson Johnson asserts, the term "negligence" as used in this section should be interpreted to mean common law negligence, which includes the element of foreseeability. Johnson Johnson also quotes from a report prepared by the Senate Select Committee on Tort and Product Liability Reform during passage of the tort reform act of 1981, of which RCW 7.72.030 is a part. That report, argues Johnson Johnson, indicates that the Select Committee viewed RCW 7.72.030 as adopting a negligence standard for Washington.

Second, Johnson Johnson maintains that foreseeability is specifically included as an element in subsection (b) inasmuch as the statute uses the expression "the likelihood that the product would cause the claimant's harm or similar harms". Johnson Johnson apparently views the latter phrase as synonymous with "foreseeability".

[6] We reject both of Johnson Johnson's arguments. This court's prior decisions demonstrate that the test in subsection (b) is one of strict liability, and consequently that foreseeability is not an element of a failure to warn claim under subsection (b).

In Couch v. Mine Safety Appliances Co., 107 Wn.2d 232, 728 P.2d 585, 78 A.L.R.4th 139 (1986), this court discussed in dicta the legislative history of the design defect provision of RCW 7.72.030. We explained that this provision, RCW 7.72.030(1)(a) (hereafter subsection (a)), defines a strict liability standard for design defect cases, despite the use of "negligence" in RCW 7.72.030(1). Couch, at 239 n. 5. Subsection (a) provides in pertinent part:

A product is not reasonably safe as designed, if, at the time of manufacture, the likelihood that the product would cause the claimant's harm or similar harms, and the seriousness of those harms, outweighed the burden on the manufacturer to design a product that would have prevented those harms and the adverse effect that an alternative design that was practical and feasible would have on the usefulness of the product[.]

We recently affirmed this view in Falk v. Keene Corp., 113 Wn.2d 645, 782 P.2d 974 (1989). There, we held that under subsection (a), strict liability, not ordinary negligence, is the standard for design defect claims, and we elaborated on the statute's legislative history as earlier sketched in Couch. Falk, at 650-51. We explained that Seattle-First Nat'l Bank v. Tabert, 86 Wn.2d 145, 542 P.2d 774 (1975) established that strict liability principles apply to design defect claims, and that Tabert formulated the test eventually enacted as subsection (a). Falk, at 648-49. We also explained that the Senate Select Committee, in making recommendations to the Legislature prior to the enactment of the tort reform act of 1981, advocated adopting the Tabert test, and that in its final report to the Legislature, the Select Committee stated that the Tabert test essentially reflected a negligence standard. Falk, at 651. The use of "negligence" in RCW 7.72.030(1) thus reflects the Select Committee's perception that the Tabert test, which was actually a strict liability test, was "akin to" a negligence test. Falk, at 653. This, we said in Falk, "tends to explain the use of the term `negligence' in the statute". Falk, at 651. Analyzing RCW 7.72.030, we stated that "it is apparent from the statute itself that the Legislature did not intend to engraft ordinary negligence principles onto the law of design defect product liability claims." Falk, at 650.

Thus, Falk shows that "negligence" as used in RCW 7.72.030(1) does not mean common law negligence, as Johnson Johnson contends. It means "negligence-like". Falk, at 651. Moreover, Falk demonstrates that the test for design defects in subsection (a) is the Tabert test, which is one of strict liability. Foreseeability is not an element of strict liability, which focuses not on the conduct of the manufacturer but upon the product and the consumer's expectation. Davis v. Globe Mach. Mfg. Co., 102 Wn.2d 68, 72, 684 P.2d 692 (1984). Therefore, Falk implies that foreseeability is not an element of a design defect claim under subsection (a). The similarities between the design defect claim in subsection (a) and the failure to warn test in subsection (b) indicate that subsection (b)'s test should also be interpreted as one of strict liability, not common law negligence, and therefore as not including an element of foreseeability.

Both subsection (a) and subsection (b) require a balancing test. On one side of the balance in subsection (a) are the likelihood that the product would cause the claimant's harm or similar harms and the seriousness of those harms. On the other side of subsection (a)'s balance are the burden on the manufacturer to design a product that would have prevented those harms, and the adverse effect that a feasible alternative design would have on the usefulness of the product. Similarly, on one side of the balance in subsection (b) are the likelihood that the product would cause the claimant's harm or similar harms and the seriousness of those harms. On the other side of subsection (b)'s balance are the adequacy of the warnings that were provided and the ability of the manufacturer to have provided an alternative warning that would have prevented the injury. Thus the test in subsection (b) is quite similar to the test in subsection (a). Because of this similarity, and since the test of subsection (a) is one of strict liability, we hold that the test of subsection (b) is also one of strict liability. That is, once the balancing test of subsection (b) has been applied and the product is found to be not reasonably safe because adequate warnings were not provided, the manufacturer is liable for harm proximately caused by the inadequate warnings. Cf. Falk, at 651 (defining nature of strict liability standard for design defect claims). Consequently, inasmuch as foreseeability is not an element of a strict liability claim, neither is it an element of a failure to warn claim arising under subsection (b).

In reaching this holding, we are aware that the Model Uniform Product Liability Act (UPLA), promulgated by the United States Department of Commerce, 44 Fed. Reg. 62,714-62,750 (1979), served as a basis for the Washington tort reform act of 1981. Senate Journal, 47th Legislature (1981), at 624. We also recognize that UPLA utilizes a negligence standard for failure to warn claims, requiring the trier of fact to find that the manufacturer "should" have provided adequate warnings. UPLA § 104(C)(1), 44 Fed. Reg. at 62,721. However, unlike UPLA § 104(C)(1), subsection (b) does not contain the term "should", or any other term that suggests a fault basis. Moreover, the commentary regarding UPLA's design defect provision states that the provision places design defect cases on a fault basis. UPLA, 44 Fed. Reg. at 62,722. We explicitly held in Falk that strict liability is the standard for design defect claims under subsection (a) of RCW 7.72.030(1). As explained above, given the similarities between subsection (a) and subsection (b), consistency with Falk requires we hold that strict liability is also the standard under subsection (b).

[7] We also reject Johnson Johnson's argument that foreseeability is specifically included as an element in subsection (b) because of the statute's use of the phrase "the likelihood that the product would cause the claimant's harm or similar harms". This phrase also occurs in subsection (a). Therefore, if foreseeability were entirely a matter of likelihood, then foreseeability would be an element of a design defect claim under subsection (a). As we explained above, Falk clearly established the contrary. Moreover, foreseeability is a matter of what the actor knew or should have known under the circumstances; it turns on what a reasonable person would have anticipated. The likelihood, or probability, that an event would occur, on the other hand, does not depend on what a reasonable person would have anticipated under the circumstances, but on an assessment of all relevant facts, including those available only in hindsight. Thus harm might be likely but unforeseeable, or foreseeable but unlikely. Therefore, foreseeability is not simply a matter of likelihood, and the occurrence of the phrase containing the term "likelihood" in subsection (b) does not show that foreseeability is an element of a claim arising under that statute.

Finally, we note that RCW 7.72.030(1)(c) (hereafter subsection (c)) clearly embraces a negligence standard, requiring that "where a reasonably prudent manufacturer should have learned about a danger connected with the product after it was manufactured", the manufacturer must provide a warning. Subsection (b), on the other hand, which concerns dangers that exist at the time of manufacture, rather than postmanufacture dangers, does not include language incorporating a negligence standard. We explained in Falk that the dissimilarities between the language of subsections (a) and (c) supported our conclusion that negligence is not the standard for design defect claims. Falk, at 653. Just so, the fact that subsection (b) does not include language like that found in subsection (c) supports the conclusion that negligence is not the standard for failure to warn claims.

In sum, the test in subsection (b) is one of strict liability, not ordinary negligence, and so foreseeability is not an element of a failure to warn claim arising under subsection (b). We next consider whether there was sufficient evidence to justify the jury's conclusion that baby oil is a dangerous product requiring an adequate warning.

[8] Under the balancing test of subsection (b), the trier of fact must balance the likelihood that the product would cause the harm complained of, and the seriousness of that harm, against the burden on the manufacturer of providing an adequate warning. Here, the likelihood that baby oil would cause the harm David suffered is low. However, given the seriousness of that harm and the slight burden on the manufacturer of providing a warning, the jury was justified in concluding that baby oil is a dangerous product that should have been accompanied by an adequate warning.

[9] The consumer expectations test in RCW 7.72.030(3) provides an independent basis for liability. See Falk, at 654. Here, we believe the ordinary consumer is unaware of the danger presented by the inhalation of baby oil. This misconception is encouraged by the presence of the words "pure and gentle" on the baby oil container. Accordingly, we agree with the Court of Appeals that because of the gravity of the potential harm, and because the product is promoted for use on and around babies, the jury could have reasonably concluded that the product was unsafe to an extent beyond that contemplated by the ordinary consumer. Ayers, at 296.

We recognize that reasonable persons might differ as to the proper conclusions to be drawn under subsection (b) and the consumer expectations test. Nonetheless, the jury evidently regarded the likelihood of injury to be sufficiently great, when balanced against the gravity of injury and the burden on the manufacturer, as to render the product not reasonably safe without a warning. Alternatively, the jury may have concluded that the product, when sold without a warning, is unsafe to an extent beyond that contemplated by the ordinary consumer. As we have already observed, Dr. Scotvold regards the risk of aspiration as significant, as he teaches his resident physicians not to use baby oil on children under age 3. The authors of the medical journal article the Ayerses presented stated that "the latent danger of [baby oil] needs to be publicized". Clerk's Papers, at 101. In light of such evidence, we are unwilling to declare the jury's verdict unjustified.

We hold that under the test of RCW 7.72.030(1)(b), as well as under the consumer expectation test of RCW 7.72.030(3), the jury was justified in concluding that the product was not reasonably safe because adequate warnings were not provided. We next consider whether the trial court properly granted a new trial based on juror misconduct.

III

In addition to granting Johnson Johnson's motion for judgment notwithstanding the verdict, the trial court ruled in the alternative that a new trial was required. The trial judge stated that he would grant a new trial for the same reasons he granted the judgment notwithstanding the verdict: insufficient evidence of causation and foreseeability. The trial judge also added another reason: jury misconduct. The trial judge found that the jurors never voted on the issue of liability, and he identified two reasons for this finding. First, the judge appears to have been persuaded by juror declarations and affidavits submitted by Johnson Johnson in which the jurors state that during deliberations, they became deadlocked on the issues of negligence and proximate cause. According to these jurors, the jury foreman then proposed that the jury consider the question as to what would be a fair amount of damages to award the Ayerses if the jury were to find that Johnson Johnson was liable. The jury arrived at the figure of $2.5 million, these jurors allege, and the jury foreman then presented that figure to the court as if it represented a judgment for the Ayerses. The trial judge felt that the allegations in those juror declarations were substantiated by the fact that during the jury's deliberations, the jury informed him that it was deadlocked on the issue of liability, and then only 30 minutes later returned with its verdict of $2.5 million. The judge felt that this short time frame substantiated the juror declarations.

The Ayerses present a different version of the jury's deliberations. They argue that the jury was given a general verdict form and was not required to vote consecutively on liability and damages. The jury was thus permitted to vote simultaneously on the two issues, and this, according to the Ayerses, is what the jury did. In support of their version of the jury deliberations, the Ayerses presented their own set of juror declarations. According to these declarations, the jurors unanimously agreed to the following voting procedure. Voting on liability and damages simultaneously, they started at the highest monetary figure any of them felt represented the amount of damages for which Johnson Johnson should be held liable, and then they worked their way down to successively lower damage figures until they reached one on which 10 of them agreed. In this way, these jurors claim, they arrived at the $2.5 million figure.

The Ayerses also point out that after the jurors returned with the verdict, the jury was polled in open court, and this polling indicated a vote of 10 to 2 in support of the verdict. This polling in open court, the Ayerses argue, cured any alleged voting defects or misconduct.

The Court of Appeals reversed the trial court's order for a new trial. The court held that the jury's voting procedure inhered in the verdict, and that therefore there was no juror misconduct. The court also held that the jury's verdict was validated by the poll in open court. Ayers, at 297. We agree with the Court of Appeals as to both issues.

[10] We recently stated that "[a]n order granting or denying a new trial will not be reversed except for an abuse of discretion; this principle being subject to the limitation that, to the extent that such an order is predicated upon rulings as to the law, no element of discretion is involved". Robinson v. Safeway Stores, Inc., 113 Wn.2d 154, 158, 776 P.2d 676 (1989) (quoting Coleman v. George, 62 Wn.2d 840, 841, 384 P.2d 871 (1963)). Here, the trial court's order for a new trial involves the questions whether the jury's voting procedure inheres in the verdict, and whether the polling of the jury in open court cured whatever irregularities there may have been in that procedure. These questions are questions of law. Therefore, the deference this court ordinarily gives a trial court's granting of a new trial does not apply in this case.

[11, 12] Considering first the question whether the jury's voting procedure inhered in the verdict, we emphasize that juror affidavits may not be used for the purpose of contesting the thought processes involved in reaching a verdict. Robinson, at 160. This court has explained that

[t]he mental processes by which individual jurors reached their respective conclusions, their motives in arriving at their verdicts, the effect the evidence may have had upon the jurors or the weight particular jurors may have given to particular evidence, or the jurors' intentions and beliefs, are all factors inhering in the jury's processes in arriving at its verdict, and, therefore, inhere in the verdict itself, and averments concerning them are inadmissible to impeach the verdict.

Cox v. Charles Wright Academy, Inc., 70 Wn.2d 173, 179-80, 422 P.2d 515 (1967). Thus, a verdict may not be affected by the circumstances that some jurors misunderstood the judge's instructions. Gardner v. Malone, 60 Wn.2d 836, 841, 376 P.2d 651, 379 P.2d 918 (1962). A juror's failure to follow the court's instructions inheres in the verdict, and affidavits relating to such alleged misconduct may not be considered. Rasor v. Retail Credit Co., 87 Wn.2d 516, 531-32, 554 P.2d 1041 (1976); cf. Ralton v. Sherwood Logging Co., 54 Wn. 254, 103 P. 28 (1909) (court refused to grant new trial where jury failed even to consider court's instructions).

Here, as the Court of Appeals noted, the jury was given general verdict forms, which it was required to use in rendering its verdict, and separate instructions to decide each element of the case. Ayers, at 297. The juror affidavits and declarations submitted by Johnson Johnson allege that the jury returned a verdict for the Ayerses without ever having voted on liability, and thus that the jury failed to follow the court's instructions. Such alleged misconduct in failing to follow the court's instructions inheres in the verdict, and therefore may not be considered. Moreover, the affidavits and declarations Johnson Johnson submitted regarding the alleged misconduct require the court to inquire as to the jurors' understanding of the procedure they followed in voting and their motives for voting as they did. Such an inquiry is not permitted. Therefore, the trial court erred in considering the juror affidavits and declarations.

[13] Other than the juror declarations, the trial judge stated that he was convinced there was misconduct because of the short time period between when the jury reported it was deadlocked and when it returned a verdict. When the juror declarations are discounted, this short time frame provides no basis for ordering a new trial.

We hold that the jury's voting procedure inhered in the verdict. Therefore, any allegations relating to such alleged misconduct were inadmissible.

Arguendo, even if the jury's voting procedure did not inhere in the verdict, the polling of the jury in open court validated the verdict. In Hamilton v. Snyder, 182 Wn. 688, 48 P.2d 245 (1935), several jurors submitted affidavits stating that the jury foreman had signed the verdict form and presented it to the court despite their protests that the verdict was not theirs. The court refused to consider the affidavits because a poll had been taken and 10 of the 12 jurors had answered that the verdict was their verdict. The court was willing to assume that the procedural irregularities complained of in the affidavits had in fact occurred, but the court held that the poll overcame those irregularities. At the time of the poll, the court explained, "[e]ach juror then stood before the court charged by his oath and his conscience to then and there register his honest convictions, and presumably each juror did so. No juror can now be permitted to say that he did not do so." Hamilton, at 692. Hamilton thus stands for the proposition that polling the jury can cure procedural irregularities that occurred during the jury's deliberations.

The same conclusion was reached in the more recent case of Butler v. State, 34 Wn. App. 835, 663 P.2d 1390, review denied, 100 Wn.2d 1009 (1983). There, the jurors were given special interrogatories to answer and instructions from the court as to the procedure to follow in answering those interrogatories. The jury returned a verdict, which was confirmed in a poll. The plaintiff then moved to vacate the verdict and for a new trial. To support the motions, the plaintiff presented juror affidavits alleging the jury had followed an improper voting procedure. The trial court agreed and ordered a new trial. The Court of Appeals reversed, holding that the poll validated the jury's verdict, regardless of whatever irregularities may have occurred in voting procedure during the jury's deliberations. Butler, at 837-38.

In the present case, Johnson Johnson alleges, on the basis of juror affidavits, that irregularities occurred in the voting procedure the jury followed. Even if the evidence of those irregularities is considered and accepted at face value, Hamilton and Butler demonstrate that the polling of the jury in open court validated the verdict.

Johnson Johnson seeks to distinguish Butler on the grounds that there the jurors answered special interrogatories, whereas in the present case the jurors merely indicated a vote for or against the verdict. This argument is unpersuasive. The distinction Johnson Johnson would draw between the present case and Butler cannot be drawn as regards Hamilton, where the court did not mention whether the polling was in the form of special interrogatories or a general question. This omission is not surprising inasmuch as the distinction in this context is unimportant. The trial court must assume that a juror polled in open court understands the question being asked in the poll. See Gardner v. Malone, 60 Wn.2d at 841 (verdict may not be overturned due to jurors' misunderstanding of instructions). Presumably each juror in the present case was asked whether his or her vote was in favor of the verdict. Since 10 out of 12 jurors said the verdict was theirs, the poll validated the verdict.

CONCLUSION

We conclude that the trial court erred in granting Johnson Johnson's motion for a judgment notwithstanding the verdict and in ruling in the alternative that a new trial was required. The Ayerses presented sufficient evidence of causation and of inadequacy of warning to support the jury's verdict. The new trial order was inappropriate because the jury's voting procedure inhered in the verdict and because the polling in open court validated the verdict. The Court of Appeals is affirmed. We remand the case to the trial court with instructions to reinstate the verdict.

DORE, C.J., and UTTER, BRACHTENBACH, DOLLIVER, ANDERSEN, DURHAM, SMITH, and JOHNSON, JJ., concur.