Michael SURACE; Alice Surace, H/w, Appellants, v. CATERPILLAR, INC.; CMI Corporation, Appellees

OPINION OF THE COURT BECKER, Circuit Judge. This is a products liability case, Restatement of Torts 2d § 402A, arising out of a construction accident in which the treads of a huge road profiler machine ran over the foot of plaintiff Michael Surace. Surace brought suit against CMI Corporation (“CMI”), the manufacturer of the machine, in the district court for the Eastern District of Pennsylvania. 1 The district court, concluding that the evidence contained in the summary judgment record failed to demonstrate that the profi-ler’s risks outweighed its utility, and also *1042 that the profiler presented an obvious risk which could have been avoided had Surace exercised reasonable care, granted summary judgment in favor of CMI. Surace appealed. Resolution of the appeal requires us to explore the contours of the Pennsylvania Supreme Court’s decision in Azzarello v. Black Bros. Co., 480 Pa. 547, 391 A.2d 1020 (1978), which established that, for purposes of strict liability, whether a product’s condition justifies placing the risk of loss on the supplier is a threshold question of law for the court to determine. 2 That Court has also made clear that the threshold question turns on a social policy determination to be made by the trial judge. In post-Azzarello defect cases, the Pennsylvania Superior Court has determined that this requirement may be fulfilled by performing a risk-utility analysis, and that the multi-factor list developed by Dean John Wade may be employed in doing so. See John Wade, On the Nature of Strict Tort Liability for Products, 44 Miss.L.J. 825, 837-38 (1973). Though with some diffidence, we predict that the Pennsylvania Supreme Court would adopt that approach. The appeal then requires that we apply the risk-utility factors to our plenary review of the district court’s judgment. When we do so, we find that the risk-utility balance weighs in favor of the plaintiff. In particular, we conclude that the district court erred: (1) in determining that the gravity of the risk of harm and the ability to eliminate it through use of a lockout/tagout device, identified by Surace’s expert as the design solution to the defect, were factors weighing in favor of CMI; (2) in relying on Surace’s own conduct to determine. that the profiler was not unreasonably dangerous; and (3) in weighing the issue of causation as a factor in resolving that question. We also conclude that putative alternative grounds for upholding the summary judgment for CMI do not pass muster. Accordingly, we will reverse the grant of summary judgment and remand for further proceedings consistent with this opinion. However, we will affirm the district court’s judgment insofar as it excluded Suraee’s expert witness Harold Brink from testifying under Fed.R.Evid. 702. We agree that Brink lacks the expertise required to testify regarding the central issue of design defect in the case — habituation. Accordingly, the district court properly excluded Brink’s testimony. I. FACTS AND PROCEDURAL HISTORY On the night of September 16, 1992, Su-raee, an employee of SJA Construction Company, was working on the New Jersey side of the Betsy Ross Bridge. The work crew was using a PR-450 pavement profiler, which had been manufactured by CMI for Caterpillar, Inc., to mill rumble strips at the base of the bridge. The profiler had been equipped with a conveyor assembly which picked up and carried debris generated by the profiler to a waiting receptacle. However, due to space constraints, the crew was operating the profi-ler without the conveyor assembly. Consequently, the crew was required to level manually the piles of debris left behind by the profiler. On the night of the accident, Surace was working as a left-side sensor man. In this position, he was responsible for signaling the profiler’s operator, William Snyder, when to start and stop the profiler, and in which direction to move it. Although the profiler was equipped with horns on the side specifically designed for signaling the operator, Su-race was using hand signals to signal Snyder. The profiler’s design contained a “blind spot,” i.e., the operator’s view of the area directly behind the machine was obstructed. The profiler was equipped with a number of warning devices, including a sign prominently posted on its rear alerting the crew to stay at least 25 feet clear of the machine, an automatic back-up alarm, flashing back-up lights, and a rotating overhead beacon light which signaled when the profiler was in operation. These warning or signaling devices were all in working condition on the night of the accident. Surace was wearing earplugs *1043 to protect his ears from the considerable noise created by the machine. After the first pass of the profiler, Surace signaled Snyder to stop. After moving the profiler forward, Snyder did so. Surace then noticed a pile of debris in the reverse pathway of the profiler which the machiné had generated. Surace picked up a broom or shovel, and, with his back to the machine, began to level the debris. While Surace was behind the machine, and without any signal from Surace, Snyder put the profiler into reverse. Athough the back-up alarms and signals were activated, Surace neither heard nor saw them, nor did he hear the shouts from his crew workers to move out of the way. The profiler backed into Surace and snared his right foot under the treads. As a result of the accident, Surace sustained serious injuries, necessitating the amputation of part of his right foot. Surace and his wife Alice (Surace) filed suit against Caterpillar and CMI alleging negligence and strict liability for defective design. 3 Surace subsequently dropped the negligence claim and, by stipulation of the parties, Caterpillar was dismissed from the action. The complaint alleged that the profi-ler was defectively designed because its warning devices were inadequate. Specifically, Surace alleged that the back-up alarms were prone to “habituation,” a phenomenon by which a person becomes immune to a particular stimulus through constant repetition and exposure, and that without a “lockout/tagout” device, which would prevent the machine from reversing unless activated by the ground crew, the profiler was unreasonably dangerous. Following a period of discovery, CMI moved in limine to exclude the testimony of Surace’s liability experts, Joseph Lambert, Harold Brink and Paul Stephens, pursuant to Fed.R.Evid. 702, 703, and 403. Both Dr. Lambert, a psychologist and specialist in human factors analysis, and Brink, an electromechanical engineer, were to testify that the profiler’s warning devices were defective because they were prone to habituation. Stephens, a mechanical and safety engineer, was to testify that the warning devices were inadequate, and that the failure to equip the machine with additional safety devices caused the accident. Ater conducting an in limine hearing, the district court denied the motion with respect to Lambert and Stephens, but granted it with respect to Brink. Surace v. Caterpillar, Inc., No. CIV.A. 94-1422, 1995 WL 303895 (E.D.Pa. May 16, 1995). The court’s ruling was predicated largely on the fact that Brink’s opinion hinged on habituation, a field, the court found, in which Brink lacked experience and for which he relied exclusively on Dr. Lambert for support. CMI then moved for summary judgment, arguing, inter alia, that the profiler was safe for its intended use. As noted above, the district court granted summary judgment for CMI on the grounds that the. evidence failed to demonstrate that the profiler’s risks outweighed its utility, and that the profiler presented an obvious risk which could have been avoided if Surace had exercised reasonable care. Surace v. Caterpillar, Inc., No. CIV.A. 94-1422, 1995 WL 495123 (E.D.Pa. Aug.18, 1995). We exercise plenary review in determining the propriety of summary judgment. Childers v. Joseph, 842 F.2d 689, 693 (3d Cir.1988). Summary judgment is proper only if there is no genuine issue of material fact. Id. “An issue is ‘genuine’ only if a reasonable jury, considering the evidence presented, could find for the non-moving party.” Id. at 693-94. II. THE AZZARÉLLO THRESHOLD ANALYSIS A. Introduction Pennsylvania early on adopted the Restatement (Second) of Torts as the law of strict products liability in Pennsylvania. Webb v. Zern, 422 Pa. 424, 427, 220 A.2d 853, 854 (Pa.1966). Section 402A of the Restatement provides in relevant part: (1) One who sells any product in a defective condition unreasonably dangerous to the user or consumer ... is subject to liability for physical harm thereby caused *1044 to the ultimate user or consumer, or to his property, if (a) the seller is engaged in the business of selling such a product, and (b) it is expected to and does reach the user or consumer without substantial change in the condition in which it is sold. Restatement (Second) of Torts (1965). To establish a case under the strict liability doctrine, a plaintiff must prove that the product was defective, and that the defect proximate-iy caused the plaintiffs injuries. Berkebile v. Brantly Helicopter Corp., 462 Pa. 83, 93-94, 337 A.2d 893, 898(1975). In Azzarello v. Black Bros. Co., 480 Pa. 547, 558, 391 A.2d 1020, 1026 (1978), the Pennsylvania Supreme Court held that “the phrases ‘defective condition’ and ‘unreasonably dangerous’ as used in the Restatement formulation are terms of art invoked when strict liability is appropriate.” The Court also announced that the threshold determination as to whether the product’s condition justifies placing the risk of loss on the manufacturer or supplier is a question of law for the court to resolve. Id., 391 A.2d at 1026. If the court determines that the product is defective under the facts as alleged, then the case is submitted to the jury to determine whether the facts indicate that when the product left the manufacturer’s control it “lack[ed] any element necessary to make it safe for its intended use or possess[ed] any feature that renders it unsafe for the intended use.” Id. at 559, 391 A.2d at 1027. The court in Azzarello, however, did not articulate the standard for determining whether the risk of loss should be placed on the manufacturer, except to note that it was a matter of social policy: Should an ill-conceived design which exposes the user to the risk of harm entitle one injured by the product to recover? Should adequate warnings of the dangerous propensities of an article insulate one who suffers injuries from those propensities? When does the utility of a product outweigh the unavoidable danger it may pose? These are questions of law and their resolution depends upon social policy. Id. at 558, 391 A.2d at 1026; see also Ellen Wertheimer, Azzarello Agonistes: Bucking the Strict Products Liability Tide, 66 Temp. L.Rev. 419, 424 (1993) (“Azzarello indisputably failed to provide courts with guidelines for determining precisely when and why strict liability should attach.”); David G. Owen, Rethinking the Policies of Strict Products Liability, 33 Vand.L.Rev. 681, 686-87 (1980) (Azzarello “did very little to help clarify the meaning of defectiveness and its proper standards of measure”) (“the [Azza-rello ] court nowhere explicitly connects the test of liability chosen — ‘unsafe for the intended use’ — to even the weak policies that it does set forth”). B. Risk-Utility Analysis Because we are sitting in diversity, we are, of course, required to the extent necessary to our decision to predict how the Pennsylvania Supreme Court would apply the Azzarello standard. In doing so, we give “ ‘due regard’ to the decisions of Pennsylvania’s intermediate appellate courts as ‘indicia of how the state’s highest court would decidé a matter.’ ” Nowak, 32 F.3d at 758 (quoting Ciccarelli v. Carey Canadian Mines, Ltd., 757 F.2d 548, 553 n. 3 (3d Cir.1985)). Absent further guidance from the Supreme Court, the Pennsylvania Superior Court has determined that in performing the social policy analysis, a court must play a dual role, acting as both a “social philosopher” and a “risk-utility economic analyst.” Fitzpatrick v. Madonna, 424 Pa.Super. 473, 476, 623 A.2d 322, 324 (1993); Carrecter v. Colson Equip. Co., 346 Pa.Super. 95, 101 n. 7, 499 A.2d 326, 330 n. 7 (1985). In doing so, courts, including the district court in the case sub judice, engage in a risk-utility analysis, weighing a product’s harms against its social utility. Smialek v. Chrysler Motors Corp., 290 Pa.Super. 496, 502, 434 A.2d 1253, 1256 (Pa.Super.Ct.1981) (“the question of whether a product is défective reaches the jury only after the court has weighed the relative risks and utility of the product”); see also Burch v. Sears, Roebuck & Co., 320 Pa.Super. 444, 450-51, 467 A.2d 615, 618 (1983). The Superior Court’s approach in this respect seems consistent with the tenor of the Pennsylvania Supreme Court’s § 402A jurisprudence. Indeed, in Azzarello, the Supreme Court indicated that a risk-utility in *1045 quiry may be appropriate in performing the social policy analysis. Azzarello, 480 Pa. at 558, 391 A.2d at 1026 (suggesting that a court inquire as to whether “the utility of a product outweigh[s] the unavoidable danger it may pose”). Furthermore, Dean John Wade’s article setting forth the risk-utility analysis was cited favorably throughout the Azzarello decision. Id. at 556 n. 8, 10, 557-58, 391 A.2d at 1025 n. 8, 10, 1026. That fact, coupled with its long hegemony in Pennsylvania-— risk-utility analysis has been used by state and federal trial courts since at least 1985 without comment by the Pennsylvania Supreme Court — satisfies us that the Supreme Court would adopt it. To be sure, the Pennsylvania Supreme Court has rejected the risk-utility approach to defining design defect in favor of the “intended use” approach. Lewis v. Coffing Hoist Div., Duff-Norton Co., 515 Pa. 334, 340, 528 A.2d 590, 593 (1987), cited in Habecker v. Clark Equip. Co., 942 F.2d 210, 213 n. 2 (3d Cir.1991). 4 And, the Pennsylvania Superior Court has relied on Lewis in rejecting actions which are based on a risk-utility theory of liability. 5 But the courts, including this one, have not interpreted Lewis as supplanting application of the risk-utility analysis as part of the threshold social policy inquiry. See Matter v. Everest & Jennings, Inc., 883 F.2d 1223, 1227 (3d Cir.1989); Shetterly v. Crown Controls Corp., 719 F.Supp. 385, 399 (W.D.Pa.1989), aff'd, 898 F.2d 142 (3d Cir.1990); Marshall v. Philadelphia Tramrail Co., 426 Pa.Super. 156, 165, 626 A.2d 620, 625 (1993). 6 That result is eonsis- *1046 tent with the Pennsylvania Supreme Court’s discussion in Lewis, which confirms, throughout, its “harmony” with Azzarello. At all events, Lewis does not purport to cut back on Azzarello, and the discussion in Lewis that seems to have caused some confusion is background and arguably dicta; the question for decision in Lewis was whether evidence of industry standards was admissible in a design case. We do not minimize the background discussion, and observe that it seems quite correct (as well as consistent with Azzarello, for it establishes no more than that the known hazards of products such as cigarettes or alcohol, see supra n. 6 or, presumably, cigarette lighters, see supra n. 7, do not automatically render their manufacturers or sellers liable on the theory that their utility is outweighed by the risks of their usage). Rather, Lewis established only that, after the threshold Azzarello determination by the court, the jury must determine whether, under the facts, the product, at the time it left the defendant’s control, lacked any element necessary to make it safe for its intended use or contained any condition that made it unsafe for use. See Pennsylvania Suggested Standard Civil Jury Instructions § 8.02 (Definition of “Defect”). In sum, our prediction that Pennsylvania would employ a risk-utility analysis in making the threshold Azzarello determination is not inconsistent with Lewis, and we affirm the district court’s use of a risk-utility analysis in determining whether the risk of loss should be placed on CMI. We regret that the Supreme Court has not yet spoken definitively on the matter of risk-utility analysis or its component factors. Since it is almost twenty years since Azzarel-lo, we hope that the Court will speak definitively soon. Unfortunately, we do not have a certification procedure, see generally Hakimoglu v. Trump Taj Mahal Assoc., 70 F.3d 291, 302-04 (3d Cir.1995) (Becker, J., dissenting), through which we could ask that court (in an appropriate case) for an early resolution of the question that is so critically important in many of the large number of diversity cases that are brought in the federal district courts of the Third Circuit. We do not suggest that this is thaiTcaser-because the outcome does not depend on the answer. If the Pennsylvania Court should adopt a certification procedure, we must be careful to be judicious, indeed sparing, in our use of it even though any such procedure would necessarily give that court the absolute right to decline certification in any case. To act otherwise would be to strain the delicate federal-state relationship which needs to be nurtured, not impaired. C. The Wade Factors Absent guidance from the state Supreme Court, the Superior Court has also identified factors that may be considered in making the threshold risk-utility analysis, including the following list developed by Dean John Wade: (1) The usefulness and desirability of the product — its utility to the user and to the public as a whole; (2) The safety aspects of the product — the likelihood that it will cause injury, and the probable seriousness of the injury; (3) The availability of a substitute product which would meet the same need and not be as unsafe; (4) The manufacturer’s ability to eliminate the unsafe character of the product without impairing its usefulness or making it too expensive to maintain its utility; (5) The user’s ability to avoid danger by the exercise of care in the use of the product; (6) The user’s anticipated awareness of the dangers inherent in the product and their avoidability, because of general public knowledge of the obvious condition of the product, or of the existence of suitable warnings or instruction; and (7) The feasibility, on the part of the manufacturer, of spreading the loss of [sic] setting the price of the product or carrying liability insurance. Dambacher v. Mallis, 336 Pa.Super. 22, 50 n. 5, 485 A.2d 408, 423 n. 5 (1984) (citing Wade, supra, 44 Miss.L.J. at 837-38); see also Fitzpatrick, 424 Pa.Super. at 476-77, 623 A.2d at *1047 324 (citing Wade factors); Phillips v. A.P. Green Refractories Co., 428 Pa.Super. 167, 180, 630 A.2d 874, 881 (1993); Riley v. War ren Mfg., Inc., 455 Pa.Super. 384, -, 688 A.2d 221, 225 (Pa.Super.Ct.1997). The district court applied these factors and determined that the profiler was not defective and, therefore, that the risk of loss should not be placed on CMI. We endorse the district court’s methodology. Application of the Wade factors also seems consistent with the tenor of the Pennsylvania Supreme Court’s jurisprudence, and we believe that the Supreme Court would find it acceptable and would probably follow' it, though there are problems of construction, particularly with respect to factor 5, discussed infra. 7 On appeal Suraee submits that CMI should have altered the design of the profiler by equipping it with additional safeguards such as a transmission lock (a “loekout/tagout device”). 8 We turn our attention to the various Wade factors. 1. Gravity of the Risk of Harm and Ability to Eliminate It Through Use of the Lockout/Tagout Device (Wade Factors Two and Four) The gravamen of Surace’s argument is that the profiler was defective and unreasonably dangerous because of the combination of the “blind spot” and the phenomenon of habituation, and that this defect could have been eliminated through the use of a lockout/tag-out device, which would prevent the operator from engaging the profiler in reverse until a switch is activated by ground personnel. The device would incorporate the safety engineering technique of lockouVtagout, a concept .which, according to Surace’s expert, although not currently in use for this specific purpose, has been proven and tested, particularly in the area of machine maintenance. The Occupational Safety and Health Administration has defined a lockout device as one that “utilizes a positive means such as a lock ... to hold an energy isolating device in a safe position and prevent the energizing of a machine or equipment.” See 29 C.F.R. § 1910.147(b)(1996). In this ease, it would prevent the operator “from inadvertently releasing the energy” and reversing the profi-ler without affirmative action by the ground crew. Suraee contends that, in rejecting this theory, the district court failed to view the evidence in the light most favorable to him. Barker v. Deere Co., 60 F.3d 158, 166 (3d Cir.1995) (when performing Azzarello analysis, a court must view the evidence in the light most favorable to the plaintiff) (citing Burch, 320 Pa.Super. at 450-51, 467 A.2d at 618-19). 9 In determining that the profiler *1048 did not pose a grave risk of harm as currently designed (we view this as an application of Wade factor two), the district court primarily relied on its conclusion that Dr. Lambert had not stated in his report that Surace had in fact become habituated to the alarm, but had merely opined that the alann was prone to habituation. The court also based its determination on its conclusion that Lambert had neither tested Surace for habituation nor explained in his report how, when Surace was injured shortly after the first pass of the profiler, he could have become habituated to the alarm, since habituation requires constant or repeated exposure. Lambert’s report, which provided a human factors analysis of the accident, was based on collected accident reports, the results of noise measurements taken of the profiler, and human factors literature. In his deposition testimony, Lambert admitted that, although it was feasible, he had not tested Surace to determine whether or not he was habituated. He further opined that, “because Mr. Surace had been around this piece of equipment for such a long period of time, for months, that he habituated to this alert. And that habituation became a long-term habituation that could carry from day to day.” Thus, although he did not note it in his report, Lambert did conclude that Surace had become habituated, and he further explained how Surace could be habituated on the profiler’s first pass of the evening. In view of the fact that Lambert’s testimony was before the court as part of the summary judgment record, the district court was not at liberty to ignore it. Moreover, because the habituation issue will go to the jury in its determination as to whether the profiler was unsafe for its intended use, the district court could not resolve any dispute over the issue at that stage. We underscore that, in the Azzarello context, the case would not become one for the jury if the district court were able to hold as a matter of law that the risk-utility balance so favored the manufacturer that the profiler could not be deemed unreasonably dangerous. See Barker, 60 F.3d at 161. Given the considerations we have just articulated, and the fact that the profiler will from time to time cause injury and, if so, the injury will be serious given the immensity and huge weight of the machine, we do not believe that the court could properly hold, on account of disputed habituation evidence, that there was not a sufficiently grave risk of harm from the profiler to weigh in favor of Surace on the risk-utility analysis (Wade factor two). Applying the fourth Wade factor, the district court rejected Surace’s contention that, because of the phenomenon of habituation, the profiler should have been equipped with a loekoui/tagout device. As proffered, the lockout/tagout device would prevent the operator from engaging the profiler in reverse until a ground worker activated a switch. As currently designed, the profiler relies on horns located on the sides of the machine which must be activated by ground personnel, who, after visual inspection, signal the operator that it is clear to reverse. It is undisputed that on the night of the accident, Surace was using hand signals rather than these horns to signal the operator and that the operator put the profiler in reverse without waiting for a signal from Surace. The district court concluded that there was no evidence that the profiler was defective without a loekoui/tagout device. It noted that Surace’s expert, Stephens, was unaware of any profiler with this device, and concluded that he could not therefore attest to the technological or economic feasibility of such a device. In his report, Stephens explained that the machine was inherently dangerous without a lockout/tagout device because of its “blind spot,” coupled with the fact that crew members were required to work in close proximity to it. Stephen’s Report at 3 (“severity of hazard and frequency of laborer exposure to the hazard dictated that [such a device] be provided on the machine”). Stephens further stated that the lockout/tagout device was both technically and economically feasible. In concluding that he had not shown *1049 that the device was mechanically feasible, the district court emphasized that Stephens did not know of any currently designed construction machinery that uses the device. That conclusion, however, runs afoul of our decision in Barker, where we held that “a district court, during its threshold determination, may [not] consider the nonexistence of a safety device as evidence of its nonfeasibility.” Barker, 60 F.3d at 166-67; see also Habecker v. Clark Equipment Co., 36 F.3d 278, 286 (3d Cir.1994) (“The fact that the [safety device] did not exist ... does not mean that it was incapable of being placed on the [profiler] if it did in fact exist.”). Although Stephens admitted that the device was not currently employed by construction machinery for this purpose, he repeatedly testified that such a device was used in other applications, that the concept had been tested and proven on machines comparable to the profiler, and that it could be applied to work for this specific purpose. We have previously held that expert testimony alone may be sufficient, for purposes of summary judgment, to demonstrate feasibility, see Hollinger v. Wagner Mining Equip. Co., 667 F.2d 402, 409-10 (3d Cir.1981), and further held that, while “a clear' and concise diagram or verbal picture of the type of device” would be helpful, it is not required to defeat a motion for summary judgment. 10 Id. at 410. While it would have been preferable for Stephens to have proffered design drawings or developed a prototype of the device, his testimony was sufficient for purposes of the threshold risk-utility calculus to make a showing of the feasibility of a lockout/tagout device, at least in the absence of a countervailing showing by CMI. We note that, in accordance with Azzarello, the technical feasibility issue will go to the jury in determining whether the loekout/tagout device was an element necessary to make the profiler safe for its intended use. Azzarello, 480 Pa. at 557, 391 A.2d at 1026. 11 Finally, the district court concluded that there was insufficient evidence from which to determine whether the profiler would be safer if equipped with a lockout/tagout device. Emphasizing that, as currently designed, the profiler relies on the crew to signal the operator that it is safe to reverse, the court concluded that the lockout/tagout device performs the same function and, because it relies on crew members to activate the device, “provides no assurance that human error as occurred here” would not cause another accident. Surace, 1995 WL 495123, at *7. However, unlike the current design, the lock-ouVtagout device would prevent the machine from going in reverse without affirmative action by the ground crew which, unlike the operator, have an unobstructed view of the reverse path of the machine. Stephens acknowledged that he could not state with a reasonable degree of scientific certainty that the device could prevent the type of accident from occurring; however, he did attest that it would “cut the risk significantly.” In sum, while such a design could not entirely eliminate the possibility that after the machine was put into motion a crew worker could walk into its pathway, it would obviously render the machine safer. This conclusion is “in sync” with the Pennsylvania courts’ approach to determining whether the risk of loss should be placed on the manufacturer. The court must balance “the utility of the product against the seriousness and likelihood of the injury and the availability of *1050 precautions that, though not foolproof, might prevent the injury.” Burch, 320 Pa.Super. at 450, 467 A.2d at 618 (emphasis added). Although a conclusion that the profiler would be made safer if equipped with a lockout/tag-out device does not require a finding by the court, or even the jury, that the profiler is defective, viewing the evidence in the light most favorable to Surace we conclude that the district court could not, on this basis, decide that the profiler was not unreasonably dangerous. Couching this discussion in terms of the Wade factors, since it appears that the lockout/tagout device could eliminate the unsafe character of the product and since it does not appear that the lockout/tagout device would be expensive or would otherwise impair the utility of the profiler, the second and fourth Wade factors weigh in favor of Surace. Therefore, unless other factors control the balance, the case must be submitted to the jury2. Consideration of Surace’s Conduct and the Fifth Wade Factor In finding that the profiler’s risks were not outweighed by its utility, the court concluded that the accident could likely have been avoided had Surace exercised due care. Specifically; the court found that because Surace was an experienced construction worker, fully aware of the dangers posed by the profi-ler, his conduct in wearing earplugs and turning his back to the machine while standing in its pathway was both careless and a cause of the accident. The court observed that “[i]t would be unjust to burden CMI with liability in a situation where there is clear evidence that Mr. Surace’s own lack of care played a role in bringing about the accident.” Surace, 1995 WL 495123, at *9. On appeal, Surace argues that the district court erred in considering his conduct as part of the Azzarello threshold analysis. He submits that such a consideration impermissibly interjects concepts of negligence into a strict liability case. Pennsylvania courts generally bar consideration of contributory negligence, in strict liability actions. See Kimco Dev. Corp. v. Michael D’s Carpet, 536 Pa. 1, 8, 637 A.2d 603, 606 (1993) (rejecting comparative negligence as a defense in a strict liability case); see also Dillinger v. Caterpillar, Inc., 959 F.2d 430 (3d Cir.1992). 12 Indeed, the Pennsylvania Supreme Court eschews the use of negligence concepts in a strict liability case. See Lewis, 515 Pa. at 341, 528 A.2d at 598 (“negligence concepts have no place in a case based on strict liability”). The theoretical basis for this approach is that strict liability focuses on the condition of the product; it is irrelevant that the injury was the result of the manufacturer’s or consumer’s negligence. Kimco, 536 Pa. at 7-8, 637 A.2d at 605-06; Lewis, 515 Pa. at 341, 528 A.2d at 593. An individual plaintiffs failure to exercise care in the use of a product is not relevant to whether the product is unreasonably dangerous in the first place. See Fleck, v. KDI Sylvan Pools Inc., 981 F.2d 107, 119 (3d Cir.1992) (“product liability laws ... encourage manufacturers to make safe products even for the careless and unreasonable consumer”); Berkebile, 462 Pa. at 95 n. 6, 100, 337 A.2d at 899 n. 6, 902 (1975) (rejecting even the “reasonable” consumer standard for the “ordinary” consumer); see also William J. McNichols, The Relevance of the Plaintiffs Misconduct in Strict Tort Products Liability, the Advent of Comparative Responsibility, and the Proposed Restatement (Third) of Torts, 47 Okla.L.Rev. 201, *1051 207 (1994) (“unreasonably dangerous” determination requires objective inquiry into the “class of ordinary purchasers”). Therefore, the district court’s consideration of Surace’s conduct runs afoul of Pennsylvania § 402A jurisprudence. 13 The district court believed that by endorsing the Wade factors, specifically, the fifth factor, (the user’s ability to avoid danger by the exercise of care in the use of the product), the Pennsylvania Superior Court has expressly sanctioned consideration of a product user’s conduct in failing to exercise care as an appropriate factor in performing the threshold analysis. Although the Pennsylvania Supreme Court has not sanctioned use of the Wade factors, we have predicted that it will. See supra at 1046. That prediction nonetheless admits of the possibility that the court will adopt some but not all of the Wade factors. We note that at least one justice has commented on the limitations of these factors in one respect. See Sherk v. Daisy-Heddon, 498 Pa. 594, 625-26, 450 A.2d 615, 631-32 (1982) (Larsen, J., dissenting) (noting that use of these factors as the standard for determining whether a product is defective fails in its “attempt to maintain a distance from negligence concepts”). In applying the fifth Wade factor, the district court looked to Surace’s conduct, rather than to an ordinary product user’s conduct. We believe that it erred in so doing. The Wade factors set forth an objective test to determine whether a product is defective; the “user” referred to in the factors is the ordinary consumer who purchases or uses the product. Williams v. Briggs Co., 62 F.3d 703, 707 (5th Cir.1995) (applying Mississippi law and noting that the fifth Wade factor focuses on “an ordinary person’s ability to avoid the danger by exercising care”); Riley v. Beaton Dickinson Vascular Access, Inc., 913 F.Supp. 879, 889-90 (E.D.Pa.1995) (“we are concerned with the ability of the [product’s users], in general, to avoid the risks inherent in the product, not with the particular circumstances of [a] plaintiffs accident” (emphasis added)); Johansen v. Makita U.S.A., Inc., 128 N.J. 86, 100-01, 607 A.2d 637, 645 (1992) (“risk-utility analysis is an objective test that focuses on the product” and the fifth Wade factor requires consideration of “the extent to which the hypothetical ‘average user’ of the product — not the plaintiff — could avoid injury through the use of due care.”); see generally Wade, supra, 44 Miss.L.J. at 847 (“strict liability ... is imposed on an objective basis”). The proper focus in applying the fifth Wade factor then is an objective inquiry into whether the class of ordinary purchasers of the product could avoid injury through the exercise of care in use of the product, not whether this particular plaintiff could have avoided this particular injury. Put differently, the user’s ability to avoid injury by the exercise of care in the use of the product appears to be a design factor that may justify a more or less exacting design depending on the facts, but it is, in any case, not a vehicle for injecting a plaintiffs (alleged) failure to exercise due care into the case. Thus, the district court misapplied this factor. We acknowledge that, notwithstanding the foregoing discussion, it is unclear whether the Pennsylvania Supreme Court would endorse even an objective application of the fifth Wade factor in performing the Azzarello threshold analysis. The court has held that the existence of due care in strict liability cases is irrelevant,' both with respect to the supplier and the consumer. Berkebile, 462 Pa. at 94, 337 A.2d at 899. Although it may appear that in doing so, that court has implicitly rejected the fifth Wade factor, its concern is with divorcing negligence concepts *1052 from strict liability proceedings; 14 we do not believe that the inquiry suggested by the fifth Wade factor injects negligence into the action or diverts the focus away from the condition of the product, but rather it informs. the decision as to whether the product, as designed, is not reasonably safe when used as intended. As Dean Wade explained in his seminal article enunciating the factors, the focus of the inquiry is on the product: Suppose that a consumer buys and wears shoes that are too little or tires that are too large for his automobile, or that he uses the product without following instructions. If he is injured as a result and brings suit, the problem may be posed in terms of whether he was at fault and whether his fault should bar recovery in an action based on strict liability. The initial, and really significant, problem is whether the product was duly safe or not. A good pair of shoes size 5 is not unduly unsafe because it may be worn by a woman with feet size 7____ A product with adequate instruction for its safe use may as a result be duly safe, and it is not rendered unsafe by the fact that the consumer did not follow the instructions.... Further illustrations easily present themselves. There is no drug, and perhaps no food, that is not dangerous if too much of it is consumed. It is missing the real point to pose the issue in terms of whether the plaintiff was eontributorily negligent in taking too many pills or too much food. Wade, supra, at 846 (footnotes omitted). The analysis does not center on the due care vel non of the consumer but rather highlights whether a product is duly safe for its intended use. This is true of all of the Wade factors. For example, the focus of the sixth factor, which considers “the user’s anticipated awareness of the dangers inherent in the product and their avoidability, because of general public knowledge of the obvious condition of the product, or of the existence of suitable warnings or instruction,” is on the product: [T]he dangers of a hoe or an axe are both matters of common knowledge and fully apparent to the user. But it is not necessarily sufficient to render a product duly safe that its dangers are obvious, especially if the dangerous condition could have been eliminated. A rotary lawn mower, for example, which had no housing to protect a user from the whirling blade would not be treated as duly safe, despite the obvious character of the danger. Note that the question here is whether the product possesses the quality of due safety, not whether the plaintiff assumed the risk or was eontributorily negligent. Id. at 842-43. Thus, insofar as the fifth Wade factor inquires into the (objective) conduct of the average product user as a factor that may justify a more or less exacting design depending on the facts, it seems to be an appropriate ingredient in the Wade risk-utility balance, which itself seems a useful approach to performing the Azzarello threshold analysis. We thus predict that the Pennsylvania Supreme Court would follow that approach. But only that Court can tell us, see Hakimoglu, 70 F.3d at 302-04 (Becker, J., dissenting), and we hope that it will do so soon. Properly applying the fifth Wade factor to determine the objective user’s ability to avoid danger by the exercise of care in the use of the profiler, we find that the factor weighs slightly in Suraee’s favor. Although an individual working on the ground behind the profiler could, in theory, avoid danger by exercising care to always remain out of the machine’s blind spot, it seems likely that ordinary workers at a highway construction site will occasionally find it necessary to step *1053 behind the machine, and that such workers may, like Surace, be habituated to the profi-ler’s alarm and thus unable to avoid danger if the profiler’s operator backs up without signaling. 3. Other Wade Factors We do not discuss the other Wade factors except in passing because they are either neutral or favor Surace in the risk-utility balance. The profiler is, of course, useful and desirable. (Wade factor one) There does not appear to be a substitute product that would meet the same need and not be unsafe, (factor three) Finally, it seems feasible for the manufacturers to spread any loss implicated by a safer design in a variety of ways, (factor seven) 4. Summary We have rejected both the factual and legal bases of the district court’s holding that the profiler was not defective as a matter of law, and conclude that the threshold Azzarel-lo test has been met. Specifically, when we evaluate the risk-utility factors in the light most favorable to Surace, we conclude that the profiler may pose a grave risk of harm absent a lockout/tagout device, in view of the phenomenon of habituation. The summary judgment must therefore be set aside and the case must go to the jury, though, of course, the jury may find for the defendant if it determines that the facts do not support a finding of defect. See supra at 1045-46. There is, however, one other problem in the case — that presented by the district court’s conflation of the causation issue into the Azzarello analysis. III. CAUSATION; SEPARABILITY FROM AZZARELLO ANALYSIS In a footnote in its opinion, the district court indicated that, although contributory negligence is irrelevant in a strict liability case, consideration of Surace’s conduct in wearing the earplugs and turning his back to the machine was appropriate as part of the Azzarello threshold analysis to the extent that it bears on causation. Surace, 1995 WL 495123 at *8 n. 10. We hold that it was error for the district court to have weighed the issue of causation as a factor in resolving the legal question of risk allocation. There are two elements to a strict liability claim. The plaintiff must establish that: (1) the product was defective; and (2) the defect was a proximate cause of the injury. See Berkebile, 337 A.2d at 898. It is only the first element that a court must address as part of the Azzarello threshold analysis. See Azzarello, 480 Pa. at 556-58, 391 A.2d at 1025-26; see also Eon v. Stroh Brewery Co., 835 F.2d 510, 512-13 n. 3 (3d Cir.1987) (focusing on whether the product was defective under Azzarello and specifically declining to address proximate cause); Phillips v. A-Best Prods. Co., 542 Pa. 124, 131, 133 n. 7, 665 A.2d 1167, 1171, 1171 n. 7 (1995) (noting that its decision rested not on whether the product was defective under Azzarello, but rather on the lack of causation). This threshold analysis focuses on the condition of the prod: uct at the time it is marketed, and whether that condition justifies placing the risk of loss on the manufacturer. Azzarello, 480 Pa. at 559, 391 A.2d at 1027. If the plaintiff ultimately proves that the product is defective, then the distinct question of whether the defect proximately caused the injury must be resolved. Pacheco v. Coats Co., 26 F.3d 418, 422 (3d Cir.1994); see generally Wade, supra, at 842-43. Should the court determine that the defect was not a legal cause of the injury, then the defendant is entitled to judgment as a matter of law. Although the district court followed the correct methodology in balancing the profiler’s inherent risks against its utility, it erred in factoring the specific circumstances surrounding the cause of the injury into this threshold inquiry. IV. ALTERNATIVE GROUNDS FOR SUMMARY JUDGMENT CMI moved for summary judgment on several alternative grounds that the district court did not reach. In view of the foregoing discussion, we must reach them. See United States v. Taylor, 98 F.3d 768, 774 (3d Cir.1996) (citing Colautti v. Franklin, 439 U.S. 379, 397 n. 16, 99 S.Ct. 675, 686 n. 16, 58 L.Ed.2d 596 (1979) (appellee may assert any *1054 ground in support of the judgment below, whether or not that ground was relied upon or even considered by the district court)), cert. denied, — U.S. -, 117 S.Ct. 1016, 136 L.Ed.2d 892 (1997). Only two of these grounds merit discussion. 15 A. intended Use The district court observed in a footnote in its opinion that the evidence suggested the profiler was not being used as intended at the time of the accident. Under Azzarello, the inquiry is whether the product was safe for its intended use. Azzarello, 480 Pa. at 559, 391 A.2d at 1027; Marshall, 426 Pa.Super. at 162, 626 A.2d at 624. Although it. did not rely on this basis in granting summary judgment, the district court opined that “[i]t is doubtful that Plaintiffs would be able to convince [the district court], or a jury, that using the profiler without the conveyor system was a way in which the profi-ler was intended to be used.” Surace, 1995 WL 495123, at *9 n. 11. On appeal, CMI continues to assert that liability should not be imposed because the profiler was not used as intended. CMI submits that the profiler was not meant to mill rumble strips, and that it was not meant to be operated without the conveyor attached. Because the profiler was being operated without the conveyor attached, the debris generated by it had to be picked up manually. At the time of the accident, Su-race'was sweeping or shoveling debris out of its pathway. Surace testified that this was both the first time he had ever worked with the profiler without the conveyor attached, or that anyone had been injured when the machine was in back-up mode. It is clear that “[u]nless the use giving rise to a strict liability cause of action is a reasonably obvious misuse ... or unless the particular use ... is clearly warned against, the manufacturer is not obviously exonerated.” Metzgar v. Playskool, Inc., 30 F.3d 459, 465 (3d Cir.1994). CMI’s Engineering Manager John Frost Phillips testified that the machine was not intended to be operated without the conveyor attached, and that it was not intended to mill rumble strips. Although Phillips testimony is probative, there was no warning in the manual indicating that the conveyor should not be removed. Moreover, the profiler had an on/off switch to control the conveyor and, obviously, the profiler could be operated without the conveyor attached and used to mill rumble strips. We hold that, on the record before us, it cannot be determined as a matter of law that such use of the profiler was a “reasonably obvious misuse.” Accordingly, this cannot serve as an appropriate basis for summary judgment. B. Assumption of Risk CMI argues that it is entitled to summary judgment because Surace assumed the risk of his injuries. Assumption of the risk is a viable defense to strict liability actions. McCown v. International Harvester Co., 463 Pa. 13, 15, 342 A.2d 381, 382 (1975). To prevail on an assumption of the risk defense, a defendant must show “that the plaintiff knew of the defect and voluntarily and unreasonably proceeded to use the product or encounter a known danger.” Wagner v. Firestone Tire & Rubber Co., 890 F.2d 652, 657 (3d Cir.1989) (citation omitted). “[W]hether the plaintiff knows of the existence of the risk, or whether he understands and appreciates its magnitude ... is a question of fact, usually to be determined by the jury under proper instructions from the court. The court may itself determine the issue only where reasonable men could not differ as to the conclusion.” Mucowski v. Clark, 404 Pa.Super. 197, 202, 590 A.2d 348, 350 (1991) (quoting Staymates v. ITT Holub Indus., 364 Pa.Super. 37, 49, 527 A.2d 140, 146 (1987)). CMI asserts that Surace was aware of the danger of being in the path of the profiler while it was operating. To infer assumption of the risk from Surace’s conduct, the conduct “must be such as fairly to indicate that [Surace was] willing to take his chances.” Wagner, 890 F.2d at 657 (quoting Restatement (Second) of Torts § 496C, com *1055 ment h). It is undisputed that Surace was responsible for signaling the operator when to reverse and that Snyder reversed the profiler without waiting for a signal from Surace. The evidence further indicates that Snyder had never before reversed the profi-ler without being signaled. Moreover, there is no evidence that Surace knew that, due to habituation, he was unlikely to hear the pro-filer’s backup alarm if Snyder should advance without signaling. Therefore, the evidence is insufficient to establish as a matter of law that Surace knowingly assumed the risk of injury when he stepped into the profiler’s pathway. Accordingly, summary judgment is not warranted on this ground. V. BRINK’S EXPERT TESTIMONY Surace challenges the district court’s exclusion of Harold R. Brink as an expert witness. Surace had proffered Brink, an electromechanical engineer, to testify to the inadequacy of the profiler’s warning devices from an engineering standpoint. Brink’s “expert opinion” was that the profiler was defective because of the obstructed view, and that the back-up alarm was insufficient to alert crew workers because of the phenomenon of habituation. The district court excluded Brink’s testimony, concluding that he did not meet the Supreme Court’s test for the admission of expert witnesses set forth in Daubert v. Merrell Dow Pharmaceuticals, 509 U.S. 579, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993), and interpreted by us in In re Paoli R.R. Yard PCB Litig., 35 F.3d 717 (3d Cir.1994) (“Paoli ”). “A district court’s ruling on admissibility of evidence is reviewed for abuse of discretion, ‘but to the extent the district court’s ruling turns on an interpretation of a Federal Rule of Evidence [its] review is plenary.’” Paoli, 35 F.3d at 749. The district court excluded Brink’s testimony under Fed.R.Evid. 702, which provides: If scientific, technical, or other specialized knowledge will assist the trier of fact to understand the evidence or to determine a fact in issue, a witness qualified as an expert by knowledge, skill, experience, training, or education, may testify thereto in the form of an opinion or otherwise. The first requirement under Rule 702 is that the witness “proffered to testify to specialized knowledge must be an expert.” Paoli, 35 F.3d at 741. Although the Rule mandates a policy of liberal admissibility, both with respect to the substantive as well as the formal qualification of experts, id., we agree with the district court that Brink did not qualify. Brink earned a degree in electrical engineering from Lehigh University and has had an extensive career in electrical and mechanical engineering, including twenty years of employment with Mack Trucks. However, as the district court noted, Brink’s theory of liability “hinged on habituation,” an area in which Brink has no training and no experience. Surace contends that through his work experience, Brink “has become highly familiar with the concept of habituation and the need to avoid this phenomenon when designing construction machinery.” But the record does not support this contention. Regarding habituation, Brink only averred generally that “over the years, I’ve learned that people can tune things out.” Significantly, his testimony focuses on making the machines “user friendly” as opposed to considering the safety implications of a device. 16 Moreover, there is no evidence in the record that Brink had experience in designing equipment from a human safety standpoint. Indeed, in his deposition testimony, Brink testified that he did not design back-up alarms; rather they were purchased by Mack and he would make sure they were mounted and wired properly. Brink further stated that he did not remember ever having tested back-up alarms, and that his knowledge of back-up alarm systems is limited to the extent to which they could be considered *1056 part of the design of an electrieal/mechanical system. Although he was a member in the Human Factors Society, Brink admitted that he had not read any literature on the phenomenon of habituation stemming from the alarms on construction machinery, nor had he ever participated in a habituation testing or study. Most significantly, Brink admitted that he relied on Dr. Lambert as the sole authoritative basis for his conclusions regarding habituation. Indeed, when questioned about the specifics of habituation, Brink replied “I’ll let [Lambert] be the expert on this.” While we have recognized that there is no set litmus test to qualify as an expert, see Hammond v. International Harvester Co., 691 F.2d 646, 653 (3d Cir.1982) (permitting person with sales experience in automotive and agricultural equipment, who had also taught high school automobile repair, to testify as an expert witness in products liability action involving tractors), there is no evidence in the record to suggest that Brink possesses sufficient knowledge of the phenomenon of habituation, either through training or experience, to testify as an expert. Because habituation was the crux of his theory of liability, and indeed the central issue of design defect in the ease, the district court properly concluded that Brink did not qualify as an expert. VI. CONCLUSION For the foregoing reasons, the judgment of the district court will be reversed and the case remanded for further proceedings consistent with this opinion. However, the judgment will be affirmed insofar as it precluded the Brink expert testimony. Footnotes [1] Jurisdiction was based upon diversity of citizenship, 28 U.S.C. § 1332. [2] The parties agree that Pennsylvania substantive law governs this diversity action. Nowak By and Through Nowak v. Faberge, U.S.A., Inc., 32 F.3d 755, 757 (3d Cir.1994). [3] Alice Surace claimed loss of consortium. [4] In Lewis, the Supreme Court acknowledged various approaches to determining whether a product is defectively designed. The court noted that under a "consumer expectations” approach, adopted by the California Supreme Court in Barker v. Lull Engineering Co., 20 Cal.3d 413, 143 Cal.Rptr. 225, 573 P.2d 443 (1978), a product is deemed defective in design “if it failed to perform as safely as an ordinary consumer would expect when used in an intended or reasonably foreseeable manner.” 515 Pa. at 340, 528 A.2d at 593. The Lewis court took cognizance of the risk-utility approach, under which a product design is defective where "on balance, the benefits of the challenged feature outweigh the risk of danger inherent in such design,” id., 528 A.2d at 593, but went on to state that the Azzarello court "sets forth yet another approach” to determining design defects — the intended use approach. Id., 528 A.2d at 593 (citing Azzarello, 480 Pa. at 559, 391 A.2d at 1027). [5] In Hite v. R.J. Reynolds Tobacco Co., 396 Pa.Super. 82, 90-91, 578 A.2d 417 (1990), the plaintiff brought a strict liability suit against a cigarette manufacturer. Rather than allege a specific defect, the plaintiff argued that the product was defective because the risks of cigarettes are outweighed by their social utility. The Superior Court, noting that the Supreme Court in Lewis had declined to embrace the risk-utility approach to defining design defect, rejected the plaintiff's theory. Id. at 91, 578 A.2d at 421; accord Miller v. Brown & Williamson Tobacco Corp., 679 F.Supp. 485, 489 (E.D.Pa.1988), aff'd, 856 F.2d 184 (3d Cir.1988). Likewise, in Dauphin Deposit Bank & Trust Co. v. Toyota Motor Corp., 408 Pa.Super. 256, 265, 596 A.2d 845 (1991), a panel of the Superior Court rejected the plaintiff's theory that strict liability should be imposed against an alcohol manufacturer solely because the risks associated with alcoholic consumption outweigh their utility. The Superior Court panel suggested that the risk-utility theory of liability was not a cognizable approach to defining defect under Pennsylvania strict liability law. Id., 596 A.2d at 849. [6] In Griggs v. BIC Corp., 786 F.Supp. 1203 (M.D.Pa.1992), the plaintiffs sought to have strict liability imposed, not because the product, a disposable butane lighter,' was unsafe for its intended use, but because it was unreasonably dangerous to foreseeable users, i.e., children. The plaintiffs advocated use of the risk-utility approach to design defect, rather than the “intended use” approach, arguing that because it was foreseeable that children would misuse the lighters and it was feasible to design child-proof lighters, on balance, the product was defective. The district court, citing Hite and Dauphin, noted that the Pennsylvania courts have rejected the risk-utility approach to design defect cases. Id. at 1206-07. The court, however, correctly distinguished between the use of risk-utility as an approach to defining defect and as a method for evaluating a product's 'unreasonable dangerousness' under the rubric of strict products liability. Id. at 1207 n. 4. On appeal, a panel of this Court relied on the Supreme Court's decision in Lewis in affirming the district court on this issue. Griggs v. BIC Corp., 981 F.2d 1429, 1433 n. 6 (3d Cir.1992). In light of the underlying facts of the case, we read the panel’s decision in Griggs as rejecting the argument that the risk-utility approach to defining defect should be used instead of the "intended use” approach; however, to the extent that the panel's decision can be read as rejecting outright the use of a risk-utility analysis as a part of the threshold determination, it is contrary to our decision in Motter, supra, which, subsequent to the Lewis decision, sanctioned this approach, and, therefore, carries no precedential weight. *1046 See O. Hommel Co. v. Ferro, 659 F.2d 340, 354 (3d Cir.1981) ("[A] panel of this court cannot overrule a prior panel precedent.... To the extent that [the later case] is inconsistent with [the earlier case, the later case] must be deemed without effect.” (citation omitted)). [7] We note that in addition to the Wade factors, the Pennsylvania Superior Court has suggested another set of factors, developed by the California Supreme Court, which may be used in performing the risk-utility analysis: (1) The gravity of the danger posed by the challenged design; (2) the likelihood that such danger would occur; (3) the mechanical feasibility of a safer design; (4) the financial cost of a safer design; and (5) the adverse consequences to the product that would result from a safer design. See Dambacher, 336 Pa.Super. at 50 n. 5, 485 A.2d at 423 n. 5 (citing Barker, 20 Cal.3d at 431, 143 Cal.Rptr. at 237, 573 P.2d at 455). The Barker factors also reflect the Pennsylvania Supreme Court's approach to strict liability. We focus on the Wade factors which are more widely accepted (in Pennsylvania and elsewhere) and are more comprehensive, including the Barker factors within their compass. At all events, it is the Wade factors that the district court applied, albeit incorrectly, and hence, we limit our discussion to those factors. [8] In the district court, Suraee also argued that CMI should have provided the operator an unobstructed view of ground personnel, or added a rear guard, described as a cow catcher. At oral argument, Suraee indicated that he was abandoning the cow catcher design and pursuing the lockout/tagout device as the "main [but not exclu sive] thrust" of his argument. However, his brief is devoid of argument with respect to the district court’s disposition under Azzarello of the alternative proffered designs, including the need for variable alarms. Accordingly, appellate review of these alternative arguments has been waived. United States v. Voigt, 89 F.3d 1050, 1064 n. 4 (3d Cir.) (failure to raise a theory as an issue on appeal constitutes a waiver), cert. denied, - U.S. -, 117 S.Ct. 623, 136 L.Ed.2d 546 (1996). This case does not present extraordinary circumstances that might warrant review of any unpreserved issues. Id. [9] We note that this conclusion is problematic. This is because the risk-utility calculus (or indeed any mode of making the social policy determination required by Azzarello) is a legal determination which should probably not be predicated upon a weighted view of the evidence. The Pennsylvania Supreme Court might want to re *1048 visit this aspect of the matter if and when it definitively comes to grips with the issues we have identified in this opinion. [10] In Hollinger, a panel of this Court noted that the district court had analyzed the question of summary judgment “under the traditional standard of sufficiency of the evidence to present a jury question," and not as threshold matter under Azzarello. Hollinger, 667 F.2d at 410 n. 11. However, in the context in which the case was decided, the panel did not pass on the propriety of the district court’s action. Id. We have previously held that the threshold legal determination under Azzarello regarding the unreasonable dangerousness of the product is properly made in the context of summary judgment or directed verdict. Nowak, 32 F.3d at 758. However, the question for the court to determine is whether the evidence is sufficient, for purposes of the threshold risk-utility analysis, to conclude as a matter of law that the product was not unreasonably dangerous, not whether the evidence creates a genuine issue of fact for the jury. [11] Stephens merely stated, without explanation, that the lockout/tagout device was economically feasible. However, although we conclude that his evidence was marginal with respect to this Wade factor, the totality of the factors relevant here would require the same result. [12] In Dillinger, after a thorough analysis of Pennsylvania strict liability law, we concluded that the “[Pennsylvania] Supreme Court has unequivocally excluded negligence concepts from product liability cases” and, therefore, the district court had erred in ruling that evidence of the plaintiff’s contributory negligence was admissible to rebut causation. 959 F.2d at 443, 444. Our opinion in Dillinger has, however, not put to rest all questions relating to the manner in which a plaintiff's negligence may be considered in a § 402A case. Compare Kramer v. Raymond Corp., 840 F.Supp. 333, 335 (E.D.Pa.1993) (relying on Dillinger for the proposition that evidence of a plaintiff's conduct is admissible only to show that the plaintiff has assumed the risk or misused the product), with Kern v. Nissan Indus. Equip. Co., 801 F.Supp. 1438, 1441 (M.D.Pa.1992) (evidence of negligence is admissible to rebut causation where the plaintiff's conduct triggered events resulting in injury), and Kolesar v. Navistar Int’l Transp. Corp., 815 F.Supp. 818, 822 (M.D.Pa.1992) (same), aff'd, 995 F.2d 217 (3d Cir.1993). [13] We note further that, even assuming arguen-do, that consideration of Surace's conduct had been appropriate under Azzarello, the district court erred in its application of the facts. The district court first suggested that Surace’s use of earplugs was careless. However, SJA required its crew to wear earplugs. Moreover, on appeal, CMI concedes that federal regulations require the use of protective ear gear at the level at which the profiler's alarms sounded (100 db). CMI Br. at 22. Furthermore, although it was undisputed that Surace had turned his back to the machine, it was also clear that the operator moved the machine without being signaled, and that he had never done this in the past. Surace Dep. at 150; Fisher Dep. at 44. Therefore, a reasonable jury could infer that Surace's actions were not careless. [14] See Berkebile, 462 Pa. at 97, 337 A.2d at 900 (holding trial court erred in instructing on manufacturer’s foreseeability, as ”[t]o require foreseeability is to require the manufacturer to use due care in preparing his product. In strict liability, the manufacturer 'is liable even if he has exercised all due care.”); see also Brandimarti v. Caterpillar Tractor Co., 364 Pa.Super. 26, 33, 527 A.2d 134, 138 (1987) (where trial court had instructed jury that plaintiff’s “misuse”, “abuse”, or “abnormal use” of the product was a defense to a strict liability claim, the court cautioned that, on remand, the introduction of the element of due care was not an issue). [15] CMI also sought summary judgment on the grounds that the profiler had been "substantially changed” and that Surace's conduct was the sole cause of the accident, but these claims are patently lacking in merit. [16] For example, Brink described his human factors experience as follows: Everything was aimed at making it as easy as possible for the person to operate the vehicle as far as locations of instruments and controls, location of steps for entrance and egress ... we tried to design our product that was as manufacturable and maintainable as possible. So to that extent, we always had human factors in our mind____ We never called it human factors, but we were interested in how the product related to the customer or user.