MYSE, J.

Neil R. Huss appeals a judgment dismissing his complaint against Yale Materials Handling Corporation based upon Huss's claims of *524 products liability and negligence against Yale. Huss contends that the trial court erred in three respects: (1) by excluding evidence of Yale's and other manufacturers' subsequent design of forklifts with seat belts and subsequent retrofitting of forklifts that had been originally designed without them; (2) by refusing to permit cross-examination of two of Yale's witnesses, who testified as to the appropriateness of manufacturing the forklift without seat belts in 1972, regarding their subsequent recommendations that seat belts be installed for reasons of safety; and (3) by granting partial summary judgment to Yale after concluding that Huss was more negligent than Yale as a matter of law on Huss's claim of negligent design in manufacturing a forklift with a removable overhead guard.

We conclude that the trial court did not err by excluding evidence of subsequent remedial measures by Yale and other manufacturers involving the installation of seat belts on forklifts manufactured after 1972. We further conclude, however, that the trial court erred by excluding evidence on cross-examination of Yale's witnesses that they had recommended at some time after the manufacture of the forklift in question that seat belts be affixed to forklifts for reasons of safety. We further conclude that the trial court erred by granting partial summary judgment to Yale on Huss's claim that Yale was negligent in designing a forklift with a removable overhead guard. We therefore reverse and remand for a new trial.

Huss was operating a forklift, manufactured by Yale in 1972, at a Stokely U.S.A. canning operation. At the time of his injury, Huss was attempting to move empty pallets from the outside area where they were stored to the inside of the plant for use on the canning line. Huss approached a stack of twenty-three wooden *525 pallets each weighing between sixty-fíve and 100 pounds and having dimensions of forty-inches by forty-four inches by a height of five inches. Immediately behind the stack from which Huss was attempting to remove pallets was a stack of thirty-six pallets. Huss inserted the tines of the forklift into the tenth pallet from the ground in the shorter pallet stack, raised the forks and tipped them back to stabilize the load. As he began to back up his forklift, the top five pallets from the adjacent stack fell over the top of the loaded pallets. The forklift Huss was operating had a removable overhead guard which when removed permitted the forklift to be operated in low clearance areas. Although the forklift was not being operated in a low clearance area, the overhead guard designed to protect the operator from falling objects was not in place. It is disputed whether the injuries that rendered Huss a paraplegic resulted from being struck by a pallet or from being knocked from the seat of the forklift by a falling pallet and striking his back upon the pavement.

Huss sued Yale on the theories of products liability, negligence and failure to warn. Huss contended that it was negligent for Yale to manufacture a forklift with a removable overhead guard because of the danger that the guard would be removed and the driver would be subject to injuries by objects falling over the load backrest or mast. Huss further contended that the forklift should have been equipped with a seat belt that would have prevented him from falling from the forklift, which he alleged caused his devastating spinal injuries.

The trial court determined that Huss's negligence exceeded Yale's as a matter of law on the claim of negligent design for permitting a removable overhead guard and the claimed inadequate warning as to the danger *526 of operating without the overhead guard in place. The warning contained on the forklift read as follows: "Truck should be equipped with an overhead guard and low backrest. Use extreme care if operating conditions prevent use of the overhead guard and low backrest. If truck is not equipped with an overhead guard, do not handle loads which are higher than the load backrest or mast."

The trial court however permitted questions of products liability and negligent design for failing to equip the forklift with seat belts to be submitted to the jury for its determination as an enhanced injury claim. The trial court excluded evidence of subsequent remedial measures taken by Yale and other forklift manufacturers to equip all forklifts they manufactured with seat belts. In addition, the trial court found that the subsequent remedial measure doctrine precluded evidence on cross-examination of two of Yale's witnesses who had testified in regard to the appropriateness of manufacturing a forklift without seat belts in 1972. The trial court precluded Huss from demonstrating that the two witnesses had in subsequent years recommended that seat belts be installed in all forklifts for reasons of safety.

Following the jury trial on the issues of products liability and negligence for not installing seat belts, the jury determined that the forklift was not defective and unreasonably dangerous for lack of a seat belt, that Yale was not negligent in the manufacture of the forklift, and that the amount of damages was $1,953,000.

Huss contends that the trial court erred by excluding evidence of the subsequent remedial measures adopted by Yale when, in 1990, it began equipping all forklifts with seat belts and started a retrofit campaign *527 making seat belts available at no cost to the owner except for cost of installation. The admissibility of evidence of subsequent design changes is addressed to the trial court's discretion. Ollhoff v. Peck, 177 Wis. 2d 719, 724, 503 N.W.2d 323, 325 (Ct. App. 1993). Accordingly, we are required to affirm the trial court's exclusion of evidence as long as the court properly applied the law and had a reasonable basis for its conclusion. Id. at 724-25, 503 N.W.2d at 325. Section 904.07, Stats., addresses the issue of admissibility of subsequent remedial measures taken by a manufacturer. It provides:

Subsequent remedial measures. When, after an event, measures are taken which, if taken previously, would have made the event less likely to occur, evidence of the subsequent measures is not admissible to prove negligence or culpable conduct in connection with the event. This section does not require the exclusion of evidence of subsequent measures when offered for another purpose such as proving ownership, control, or feasibility or precautionary measures, if controverted, or impeachment or proving a violation of s. 101.11.

While this provision with some exceptions excludes subsequent remedial measures to prove claims of negligence against a manufacturer, evidence of subsequent remedial measures may be admissible to prove allegations of strict liability. Chart v. GMC, 80 Wis. 2d 91, 258 N.W.2d 680 (1977). Where both strict liability and negligence are alleged, it is up to the trial court to exercise its discretion regarding the admissibility of subsequent remedial measures. D.L. v. Huebner, 110 Wis. 2d 581, 610-11, 329 N.W.2d 890, 903-04 (1983). Because of the similarity between the theories of strict liability and negligence, the manufacturer is at least *528 entitled to a limiting instruction directing the jury to consider the subsequent remedial measures only as to the strict liability claim. Id.

However, where the admissibility of evidence is likely to cause confusion or to be unfairly prejudicial, the trial court in its discretion under § 904.03, Stats., may exclude evidence even though relevant. Section 904.03 provides:

Exclusion of relevant evidence on grounds of prejudice, confusion, or waste of time. Although relevant, evidence may be excluded if its probative value is substantially outweighed by the danger of unfair prejudice, confusion of the issues or misleading the jury....

In this case, the trial court concluded that evidence of the subsequent remedial measures would lead to jury confusion and would unfairly prejudice Yale by permitting the introduction of evidence as to a theory the jury could not properly consider when evaluating Huss's claim of negligence against Yale. The court was concerned that a limiting instruction asking the jury to apply the evidence to one of two theories of liability raises legitimate concerns as to the jury's ability to properly apply the law. Because the court's concern as to unfair prejudice and jury confusion is reasonable, we conclude that the trial court did not err by determining that evidence of the subsequent remedial measures could not be introduced at trial.

Huss further sought to demonstrate that in the mid to late 1980's most of the forklift industry adopted a policy of installing seat belts as standard equipment. *529 Whether evidence of post-manufacture industry custom is admissible on the issue of a manufacturer's negligence in products liability actions is not governed by § 904.07, Stats., or any other specific evidentiary rule. D.L., 110 Wis. 2d at 617, 329 N.W.2d at 906-07. Section 904.07 does not apply to evidence of post-manufacture remedial measures undertaken by entities not charged with liability in the litigation before the court. Id. Because no specific evidentiary rule covers the admissibility of post-manufacture industry custom, the court must determine if the evidence is relevant. Id. at 617-18, 329 N.W.2d at 907. Here, after properly concluding the evidence was relevant, the trial court, under § 904.03, Stats., excluded the evidence of post-manufacture remedial measures. The trial court concluded that the evidence would unfairly prejudice Yale and confuse the jury similar to the evidence of remedial measures taken by Yale. We conclude the trial court did not err in exercising its discretion by excluding evidence of remedial measures taken by other forklift manufacturers since the law was properly applied and the court's concern was reasonable.

However, the court also used the subsequent remedial measure exclusion and § 904.03, Stats., to limit Huss's cross-examination of two of Yale's witnesses. Yale's liability expert, R. Kevin Smith, testified that the forklift truck was reasonably safe without a seat belt in 1972. Alvin Kirby, a fact witness for Yale, testified as to the reasons Yale did not install seat belts in 1972. Both witnesses recommended at some time after 1972 that seat belts be installed on forklifts of the type Huss was using. While each witness had recommended the installation of seat belts on forklifts at some time after the manufacture of the forklift in question, we *530 conclude that their recommendations were a proper area for cross-examination. We conclude that Huss was entitled to explore the differences between their opinions expressed as of 1972 and their later recommendations for the installation of seat belts on forklifts of this type. While there may be perfectly legitimate explanations for their subsequent recommendations, the apparent inconsistency of the two opinions is a legitimate avenue for Huss's cross-examination. Eliminating inquiry into the apparent inconsistent opinions of these witnesses unduly restricted Huss's attempts to demonstrate that a forklift manufactured without seat belts in 1972 was unreasonably dangerous.

The trial court apparently based its decision to exclude this area of inquiry from cross-examination upon the subsequent remedial measure rule. We do not agree that the recommendations made by these witnesses are foreclosed by the trial court's determination that the evidence of subsequent remedial measures is not admissible. The recommendation in question need not disclose that the industry as a whole or Yale specifically adopted that recommendation. Nor does the fact that these specific witnesses made the recommendation necessarily open up the question of industry standards and practices after the manufacture of the forklift in question. Sufficient prophylactic safeguards could have been imposed by the court to exclude evidence of the change of industry practice without restricting Huss's cross-examination of the allegedly inconsistent opinions regarding seat belt installation. As long as the trial court can impose adequate safeguards, the existence of the allegedly contradictory opinion is sufficient to allow cross-examination without *531 subjecting Yale to the danger of undue prejudice or confusing the jury.

While questions of admissibility are addressed to the trial court's discretion, the trial court's erroneous conclusion that subsequent remedial measures restrictions preclude inquiry into what Huss alleges to be contradictory opinions expressed by witnesses is an error of law. Discretion exercised based upon an erroneous application of the law is an unreasonable exercise of discretion. Allstate Ins. Co. v. Konicki, 186 Wis. 2d 140, 150, 519 N.W.2d 723, 726 (Ct. App. 1994). An evidentiary error requires reversal or a new trial if the error affects the substantial rights of the party seeking relief on appeal. Johnson v. Kokemoor, 188 Wis. 2d 202, 214, 525 N.W.2d 71, 75 (Ct. App. 1994). Under the harmless error test, we "reverse where there is a reasonable possibility that the error contributed to the final result." Id. Here, expert testimony was the dominate evidence on the issue of whether Yale was negligent in manufacturing the forklift without a seat belt. The improperly excluded evidence could have impacted substantially on the jury's determination as to which of the conflicting experts' opinion to accept and the weight they should attach to it. Accordingly, we conclude that there is a reasonable possibility that the court's erroneous exclusion of what could be perceived as conflicting opinions by the witnesses prejudiced the plaintiffs right to a fair trial. We therefore direct the trial court to grant Huss a new trial on the issues of whether Yale was negligent in manufacturing the forklift without a seat belt and whether Huss suffered an enhanced injury as a result of the lack of a seat belt.

*532 Huss's final contention is that the trial court erred by granting partial summary judgment to Yale and determining that as a matter of law Huss's negligence exceeded Yale's negligence for manufacturing a forklift with a removable overhead guard. When reviewing a grant of summary judgment, we independently apply the same methodology as the trial court. Hertelendy v. Agway Ins. Co., 177 Wis. 2d 329, 333, 501 N.W.2d 903, 905 (Ct. App. 1993). That methodology has been set forth numerous times and need not be repeated here. See Grams v. Boss, 97 Wis. 2d 332, 338, 294 N.W.2d 473, 476 (1980).

The trial court concluded that Huss knowingly confronted an open and obvious danger when he operated the forklift in an area of highly stacked pallets without the overhead guard. However, the fact that this may have been an open and obvious danger is merely an element of the comparison of negligence and is not an absolute defense. Kloes v. Eau Claire Cavalier Baseball Ass'n, 170 Wis. 2d 77, 86-87, 487 N.W.2d 77, 81 (Ct. App. 1992). Accordingly, we must compare the negligence of the parties and determine whether Huss's negligence exceeded Yale's as a matter of law. In the comparison, we must also consider the employer's negligence because a jury must have the opportunity to consider the possible negligence of all parties to the transaction whether they are parties to the lawsuit or not. Connar v. West Shore Equip., 68 Wis. 2d 42, 44-45, 227 N.W.2d 660, 662 (1975).

In this case, manufacturing a forklift with a removable overhead guard with an affixed warning of the dangers of operating without the guard would seem *533 reasonable and inconsistent with the claim of negligence. Indeed, as to the adequacy of the warning, we agree with the trial court that the warning is legally sufficient and conclude that the trial court did not err by striking this allegation of negligence from Huss's complaint. The issue, however, as to whether the forklift should have been manufactured with a removable overhead guard raises a more difficult question. Huss's experts opined that Yale should have known that the guard would be removed and the forklift would be operated in this dangerous condition. Based on this testimony, a jury could draw the permissible inference that Yale knew that the forklift would be operated without the overhead guard if it made the guard removable and Yale was aware that the operation of the forklift without an overhead guard presented an unreasonably dangerous condition. We conclude it is therefore possible for the jury to find Yale negligent to some degree by manufacturing the forklift with a removable overhead guard.

We must now examine Huss's operation including his determination to operate the vehicle without an overhead guard. The overhead guard was apparently removed because once or twice a day a forklift was required to go into a low clearance area of the plant to correct problems on the canning line. The work undertaken by Huss at the time of the injury, however, did not require the overhead guard to be removed. While Huss's employer, Stokely U.S.A., had a policy against operating the forklift without the overhead guard unless in a low clearance area, Huss's supervisors knew he had been operating the forklift for days without the guard and said nothing. Huss claims that he did not know the policy against operating the forklift without the overhead guard, and, in his four or five *534 weeks operating the forklift at Stokely before his injury, the guard was attached only one day. Someone else removed the guard, and Huss never saw it again. Thus, the jury must determine whether Huss could have affixed the guard prior to his injury and the amount of negligence, if any, attributable to Huss for not insisting that the overhead guard be affixed to the forklift before undertaking the task of replenishing the supply of pallets in the plant.

There is certainly adequate evidence to determine that Huss was negligent in his operation of the forklift. However, the degree of negligence may depend upon the inferences drawn by the jury as to whether the manner of Huss's operation of the forklift rendered the adjacent pile of pallets unstable and whether he reasonably should have foreseen the danger in operating the forklift in such an environment. We conclude that a reasonable jury could find Huss negligent but ascribe a reasonably small amount of negligence to Huss depending on the inferences drawn from the evidence in regard to his ability to obtain the overhead guard and affix it before undertaking this task, the reasons for the adjacent pile falling, and the perceived dangers existing at the time of his injuries.

Because a reasonable jury could find some degree of negligence against both parties, we examine the trial court's conclusion that Huss's negligence exceeded Yale's as a matter of law. The apportionment of negligence is ordinarily a question for the jury. Kloes, 170 Wis. 2d at 86, 487 N.W.2d at 81. The instances in which a court may rule that, as a matter of law, the plaintiffs negligence exceeds that of the defendant are extremely rare. Davis v. Skille, 12 Wis. 2d 482, 489, 107 N.W.2d *535 458, 462 (1961). The apportionment of negligence is a matter that rests within the sound discretion of a jury based upon the inferences it draws from the evidence presented, together with its determination as to the standard of care required of the parties. Accordingly, summary judgment should only be used in the rare case where it is clear and uncontroverted that one party is substantially more negligent than the other and that no reasonable jury could reach a conclusion to the contrary. See Cirillo v. Milwaukee, 34 Wis. 2d 705, 716-17, 150 N.W.2d 460, 465-66 (1967); Bush v. Mahlkuch, 272 Wis. 246, 251, 75 N.W.2d 283, 286 (1956).

We conclude that the issue should have been presented to the jury in this case because there was no clear and compelling evidence that Huss's negligence exceeded Yale's. The jury's determination may range widely in this case depending on the inferences it draws from the evidence. The jury may determine that Yale was not negligent or was casually negligent to a small degree. However, the jury may also decide that Huss was only minimally negligent based upon the work rules, the environment within which he was operating and the rest of the circumstances surrounding the accident. Since we cannot determine what inferences or conclusions a jury would make as to the negligence against either party, we conclude that the trial court erred by removing this issue from the province of the jury. Because the jury could reasonably conclude that both parties were negligent to a small extent with the majority of negligence ascribed to the employer, partial summary judgment was error.

*536 By the Court. — Judgment reversed and cause remanded.