dissenting.
Because I believe that a judgment as a matter of law should have been entered for the defendants pursuant to Rule 50(b) of the Federal Rules of Civil Procedure, I must respectfully dissent from the majority opinion. In my view, the combination of several evidentiary errors and plaintiffs’ failure to carry their burden in establishing by a preponderance of the evidence that the motorbike was defective warrants judgment in favor of the defendants or at the very least a new trial.
I.
As noted in the majority opinion, plaintiffs alleged at trial three theories to support their claim that Honda’s motorbike was defectively designed: (1) lack of a safety flag; (2) lack of a key-lock ignition; and (3) the motorbike’s inappropriateness for children under twelve years of age. I agree with the majority opinion’s conclusion that plaintiffs failed to establish that the lack of a key-lock ignition was a proximate cause of the accident. However, I disagree with the opinion’s conclusion that the first and third theories support the jury’s verdict for the plaintiffs on their claim that the motorbike was defectively designed.
a. Lack of a Safety Flag
At trial, plaintiffs contended that the motorbike was defectively designed because it lacked a safety flag. The majority concludes that, due to conflicting expert opinions, reasonable minds could differ as to whether the *520lack of a safety flag was a substantial factor in causing the accident. However, because I believe that the District Court abused its discretion in allowing the plaintiffs’ witness to testify as an expert regarding whether the lack of a safety flag constituted a defective design, the evidence that remains warrants judgment as a matter of law for the defendants.
Plaintiffs’ expert witness, William Kitzes, suggested that Honda should have attached a safety flag to the bike in the following testimony:
since the early ’70’s, Honda has marketed and sold a safety flag for use' on children’s motorized vehicles. And that flag can be as tall as eight feel tall, and it could give visibility to a small rider, especially on a small bike, to other riders and to traffic or whoever else it might be so that they’re more easily spotted when they’re riding the vehicle.1
J.A. 286. Kitzes testimony should not have been admitted. Mr. Kitzes had no special training or experience which would qualify him to testify as an expert regarding the design of the motorbike or any alternative designs for the vehicle.
Rule 702 of the Federal Rules of Evidence governs the admission of expert testimony. Rule 702 provides as follows:
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.
Fed.R.Evid. 702. A district court’s admission of expert testimony under Rule 702 is reviewed for an abuse of discretion. See General Elec. Co. v. Joiner, - U.S. -, -, 118 S.Ct. 512, 515, 139 L.Ed.2d 508 (1997).2 Whether a proposed witness is qualified to testify as an expert is determined, in part, by examining “whether the witness’s ‘knowledge, skill, experience, training, or education,’ are such as to qualify him or her to testify as an expert at all, and it may include a determination of the tests or experiments that the proffered expert conducted, if any.” Cook v. American S.S. Co., 53 F.3d 733, 738 (6th Cir.1995) (quoting Fed.R.Evid. 702).
Kitzes holds a Bachelor’s Degree in history from the University of Wisconsin and a Juris Doctorate from American University. He additionally holds a certificate in safety management from the American Society of Safety Engineers. He participated in.fifty hours of course work in human factors engineering at the University of Michigan and is board certified in Products Safety Management and Hazard Control Management. He was employed with the Consumer Products Safety Commissipn (“CPSC”) for several years3 and, at the time of trial, he owned a company called Consumer Safety Associates operated by Kitzes, his wife, and one assistant. However, Kitzes does not have a degree in engineering nor is he an accident reconstructionist. He has never ridden a Z50R, he has never participated in any testing involving children operating the Z50R, and he never observed Gary Thompson ride the Z50R. Moreover, he has never seen any testing conducted by Honda on the vehicle. During Honda’s cross-examination of Kitzes, he was asked whether he had performed any tests to determine “if you attach a six-foot wind flag to this vehicle whether, when this vehicle was traveling behind a three-foot embankment with a six-foot rolled bale of hay on top of it, you’d be able to see the wind flag?” J.A. 294-95. Kitzes responded that he had not performed any such tests.
While Kitzes was qualified as an expert to testify regarding the manner in which the CPSC collects statistics on consumer product safety, which he did, Kitzes was not qualified to testify as an expert regarding the design *521of the motorbike at issue in this case.4 He has no knowledge, skill, experience, training, or education regarding the design of this motorbike or the designs of other motorbikes. His testimony regarding the lack of a safety flag, therefore, went beyond his area of expertise. “When making a preliminary finding regarding an expert’s qualifications under Fed.R.Evid. 104(a), the court is to examine ‘not the qualifications of the witness in the abstract, but whether those qualifications provide a foundation for a witness to answer a specific question.’ ” Smelser v. Norfolk Southern R.R. Co., 105 F.3d 299, 303 (6th Cir.)(quoting Berry v. City of Detroit, 25 F.3d 1342, 1351 (6th Cir.1994)), cert. denied, - U.S. -, 118 S.Ct. 67, 139 L.Ed.2d 29 (1997). In the case of Mr. Kitzes, his qualifications did not provide a foundation for him to answer the questions he did. Under these circumstances, the District Court abused its discretion in permitting Kitzes to testify as an expert witness.
Furthermore, even if we include Kitzes’s testimony in determining whether there was sufficient evidence to conclude that the motorbike was defective for failing to include a safety flag, Kitzes did not testify as to one critical element of the plaintiffs’ case: proximate cause. The burden is placed on the plaintiffs to prove that the lack of a safety flag was a substantial factor in bringing about Gary’s accident. Kitzes testified only that a flag “could give visibility to a small rider, especially on a small bike, to other riders and to traffic or whoever else it might be so that they’re more easily spotted when they’re riding the vehicle.” (emphasis added). “Could” is not the legal standard utilized to prove causation. See Calhoun v. Honda Motor Co. Ltd., 738 F.2d 126, 130 (6th Cir.1984)(“where an accident could have resulted for any number or reasons, the plaintiff has failed to establish the necessary proximate cause”)(emphasis added). Rather, to establish proximate cause by circumstantial evidence, “ ‘the evidence must be sufficient to tilt the balance from possibility to probability.’ ” Morales r. American Honda Motor Co., 71 F.3d 531, 537 (6th Cir.1995)(quoting Calhoun, 738 F.2d at 130). When asked whether a vehicle on the main road would see the flag had one been attached to the motorbike as it was travelling behind a three-foot embankment topped with a six-foot rolled bale of hay, the crucial question addressing proximate cause, Kitzes could not answer because he had not performed any such tests to determine the answer.
In summary, Kitzes’s opinion that a safety flag could have prevented the accident was based on nothing more than conjecture and supposition and, therefore, should not have been admitted. Furthermore, even if Kitzes’s testimony was properly admitted, Kitzes did not testify that the lack of a safety flag was a probable cause of this accident; consequently, plaintiffs introduced no evidence of proximate causation.
The evidence that remains once Kitzes’s testimony is excluded or once it is understood that Kitzes did not render a sufficiently conclusive opinion as to the proximate cause of the accident, is the testimony of Dr. Graeme Fowler, Honda’s expert. Dr. Fowler did perform tests to detei’mine whether a safety flag would have prevented the accident. Fowler described his testing as follows:
... I went in and actually pulled a flag from one of my lockers that I have at work. And what I’m going to do is assume that a flag is, like, six foot high. Okay? What I did then was to look at the survey to see whether or not — if Gary Thompson had a flag on his vehicle, where would that be visible above the hay bales that lined that entrance road.5 And don’t forget that that entrance road is somewhat depressed from the grassy border on which the hay bales were stacked.
However, based on that survey and looking at the height of the vehicle and I assumed in my analysis that the ... the *522flag ... would be at approximately two feet high ... '
... So you got two feet plus six feet. So you’ve got approximately an- eight-foot-high flag, eight feet above the ground. Well, looking at the survey, I found that certainly as — up on the entrance — if the6 vehicle was traveling along that entrance road with a flag with a maximum height of about eight feet, it would be seen above the bales, and anywhere from two feet to one feet above the bales ... you could probably see a flag bobbing along the road.
One of the things that you’ve got to remember is these — once the vehicles start moving, these flags bend back slightly. But I’m assuming that you could still see the flag.
... the motorcycle came down this road, yes, you could see the flag above the hay bales. Okay? Although it would be obscured by these two trees as the vehicle passed behind the trees.
The next thing I wanted to determine was, okay, given this vehicle had a flag, what was the last point at which Ms. Graham would have had to have seen the flag, okay, before it was too late.
At this point in his testimony, Fowler concludes that, if Graham had spotted the flag by 143 feet from the farm entrance and if Graham had taken 1.25 seconds to react and hit the brakes, she would have just missed him. Fowler termed this situation the “near-miss scenario.” However, Fowler testified that in the actual accident it took Graham 1.27 seconds to hit her brakes. Furthermore, because of the angle at which the flag would be positioned (at 17 degrees to the right of the driver) and the driver’s attention to the road straight ahead, Fowler-concluded it would be very difficult for the driver to see a 12-inch flag in the driver’s peripheral vision. Fowler additionally noted that it would have been difficult for Graham to spot the flag due to “cluttered visual background;” the flag would have been competing with trees and branches that were also in the visual area. Fowler further discussed a study which concluded that drivers look to the right-hand-side of the roadway only 10 percent of the time they are driving and that only 1 to 2 percent of the time will a driver look past 12 degrees to the right side of the roadway. Fowler then concluded that Graham would have likely-been fixated straight ahead more than usual because she was approaching the crest of a hill; watching for any hazards that could be approaching from the other side. Because of all these factors, Fowler concluded that it was unlikely Graham would have detected the flag before it was too late. Fowler’s testimony continues in the joint appendix as follows:
... Suppose she does see the flag. Well, the next thing she’s got to do is then figure out whether that flag is actually going to — is on a converging course and that a collision is likely to occur ... She has no information relative to what type of vehicle the operator has. All you can see is potentially a flag moving along.
Well, it’s absolutely reasonable for her to believe that whatever is coming along that road is going to stop prior to going on the main road.
And, also, at 14 miles an hour, the flag is not moving that fast. So it’s not at such an excessive speed that you know that a person or the thing that’s associated with that flag is going to not be able to stop.
So, for all of these different reasons, I don’t believe that she would have — she would have first detected the flag, perceived there was going to be a collision, decided it was necessary to actually react and hit the brakes before it was too late, before we were into the red zone.
J.A. 328-29. Based on his tests, Fowler concluded that “a flag would not have pre*523vented this accident.” J.A. 327.7 Thus, the only expert who did perform a study of whether the driver would see a flag, with the conditions as they were when Ms. Graham hit Gary Thompson, concluded that a six-foot flag would not have been seen by Ms. Graham in sufficient time to allow her to come to a stop before colliding with Gary Thompson. Fowler’s testimony was the only evidence presented regarding whether the lack of a safety flag was a proximate cause of the collision. Because Fowler testified that the lack of a safety flag could not have caused the accident, there is no evidence supporting proximate cause, an essential element of the plaintiffs’ 'claim. As acknowledged by the majority opinion, the plaintiff has the burden of establishing causation in product liability actions and causation exists where the defendant’s conduct is a substantial factor in bringing about the injured party’s harm. See Morales v. American Honda Motor Co., 71 F.3d 531, 537 (6th Cir.1995). Because Fowler testified that the lack of a safety flag was not a factor in the collision and no evidence exists to contradict his testimony, a judgment against Honda based on this theory is not supported by the evidence.
b. Age Appropriateness
Plaintiffs also alleged at trial that the motorbike was defective because, while Honda recommended the motorbike for children seven and older, the bike was not appropriate for children under age twelve. Plaintiffs specifically argued that children under the age of twelve do not possess the requisite cognitive skills to safely operate a motorbike like the one purchased by Ms. Morales. Plaintiffs presented their theory of inappropriate age recommendations via two avenues: (1) Dr. Kitzes’s use of accident and injury statistics produced by the Consumer Product Safety Commission; and (2) the testimony of Dr. Karen DeMoss, a psychologist who testified that children between the ages of seven and nine (Gary’s age at the time of the accident) do not have sufficient cognitive skills to operate a motorbike.8
i. Admission of Accident Statistics
Over Honda’s objection and through witness Kitzes, plaintiffs admitted a CPSC document entitled “Hazard Analysis of Mini-Bike Related Injuries,” dated November, 1974, as well as other statistics. This publication indicated that, by virtue of selected hospital emergency room entries into a database called the National Electronic Injury Surveillance System, 31,000 accidents involving off-road motorbikes occurred in 1973. Fifty percent of these injuries occurred to children between the ages of 10 and 14 and fifteen percent occurred to those between ages 5 and 9. In addition, Mr. Kitzes testified regarding the statistics for other years. For example, in 1981, 34,542 injuries occurred. Of the 34,542 injuries, 15,000 occurred to children ages 5 to 14. From those statistics, however, Mr. Kitzes could not inform the jury regarding how many injuries occurred to children between the ages of 7 and 9, how many involved the specific model at issue in the instant case, or how many generally involved a Honda vehicle. Furthermore, the statistics did not identify how many accidents occurred due to a child’s failure to stop before entering a public roadway. In fact, only 4 of the 10,000 to 16,000 accidents involving off-road motorbikes even occurred on public thoroughfares, according to Ms. DeMoss.
In a products liability action wherein the plaintiff alleges the product has been defectively designed, plaintiff may introduce evidence of other accidents if the other accidents occurred under substantially similar circumstances. See Brock v. Caterpillar, Inc., 94 F.3d 220, 224 (6th Cir.1996), cert. denied - U.S. -, 117 S.Ct. 1428, 137 L.Ed.2d 537 (1997); see: also Montgomery Elevator Co. v. McCullough, 676 S.W.2d 776, 783 (Ky.1984). Because this is a relevancy determination, we review the District Court’s decision to admit the evidence for an abuse of *524discretion. See Montgomery Elevator, 676 S.W.2d at 783 (citing Fed.R.Evid. 401).
I disagree with the majority’s conclusion that the District Court did not abuse its discretion in admitting these statistics. In my view, the District Court plainly abused its discretion in admitting these general statistics regarding off-road motorbike accidents. The facts underlying these accident statistics are not substantially similar to those underlying the instant case. The statistics do not distinguish this model of Honda, motorbike from other models and manufacturers, how each accident occurred, or how many occurred to children of Gary Thompson’s age. According to Kitzes, he could not inform the jury as to how many of the injuries noted occurred to “children between the ages of 5 and 9, [and] occurred because a child didn’t stop ánd entered a public highway with a minibike.” Because these statistics played such a large role in the plaintiffs’ attempt to establish the age inappropriateness theory, I cannot conclude that this error was harmless under Rule 103(a) of the Federal Rules of Evidence.9
I agree with the majority’s conclusion that, once the statistics were admitted, Honda should have been permitted to introduce the CPSC’s denial of a rule-making petition despite 15 U.S.C. § 2074(b), which prohibits the admission of evidence of the CPSC’s inaction in suits relating to consumer products. However, while the majority holds that the erroneous exclusion of the denial of the rule-making petition did not affect the substantial rights of the defendants, I submit that the exclusion of this rebuttal evidence was not harmless. The plaintiffs admitted the statistics in an attempt to establish a direct correlation between the number of motorbike accidents and the youth of the children involved in the accidents. The correlation, plaintiffs contended, supported their1-theory that the motorbike was inappropriate for children Gary’s age. The denial of the rule-making petition rebutted that theory; the denial constituted direct evidence that the age of the riders was not related to the number of accidents, in direct contradiction to plaintiffs’ theory. As noted by the majority opinion, the petition submitted by two physicians requested regulation of unlicensed, two-wheeled vehicles. Specifically, the petition “requested design and labeling requirements to address the risk to children under age 14 from these off-road vehicles.” J.A. 178. The Commission’s denial was “based on the fact that the vast, majority of injuries associated with these vehicles are. related to the way they are used and not to design characteristics which the Commission could effectively or practically regulate.”, J.A. 178.. I have little doubt that the government statistics impacted the jury’s decision. To permit the introduction of the-underlying statistics, but not the ultimate decision made by -the government regarding those statistics, clearly constituted prejudicial error. At .the very least, the error entitles defendants to, a new trial.
ii. Testimony of Dr. DeMoss
To further support their age inappropriateness theory, plaintiffs offered the testimony of Dr. Karen DeMoss. In addition to testifying regarding the CPSC statistics, Dr. DeMoss, a psychologist for the Fayette County School District, testified generally regarding the psychological development of children and concluded that Honda’s age recommendation of seven-years-old was not appropriate. In 'her opinion, a seven-year-old is “not 'cognitively developed to a degree, to safely operate a machine that is much more sophisticated than the child’s thinking skills.” J.A. 245. More specifically, Dr. DeMoss stated, “[tjhey’re not at a stage cognitively where they’re capable of handling a machine with this sort of skill that’s required with regards to speed, 'depth perception, remembering rules, and operating the motorcycle at the same time; too many factors and variables that they have to take into- account.” J.A. 246. In hei opinion, children between the ages of 7 to 9 are capable of only performing one task at a time.10 Dr. DeMoss *525ultimately concluded that the motorbike at issue was unsafe for children under the age of 12 and therefore should not have been recommended for use by a child under that age.
The majority opinion relies, in part, on Dr. DeMoss’s testimony in concluding that sufficient evidence was presented to permit a reasonable jury to find that Gary Thompson’s inability to concentrate on the multiple tasks necessary to safely operate the motorbike was a substantial factor in causing the accident. Dr. DeMoss’s testimony, however, did not establish the requisite causation element; Dr. DeMoss did not testify that Gary’s age was a substantial factor in the accident.11 Dr. DeMoss’s testimony did not prove that Gary’s immature cognitive skills in any way caused the accident in question. Dr. DeMoss simply never tested, evaluated or reviewed the cognitive skills of Gary Thompson. According to Dr. DeMoss, she never spoke with Ms. Morales regarding Gary’s pre-accident “riding behavior,” and she never observed Gary ride the motorcycle. Furthermore, while she could not specifically recall that Gary rode the motorbike without incident for three years before the accident, she understood that to be the case. Thus, Dr. De-Moss’s testimony does not provide the evidence needed to establish causation between Gary’s age at the time of the accident and the occurrence of the accident.
The only remaining evidence regarding the age appropriateness theory was briefly introduced through Dr. Kitzes:
Q: Do you feel that this 1988 Z50R motorcycle is defective and unreasonably dangerous for its intended users of 7,8,9, and 10 years old?
A: Ido.
J.A. 285-287.12 Kitzes, however, does not have a degree in child development or in child psychology. He has never ridden a Z50R, he has never participated in any testing involving children operating the Z50R, and he never observed Gary Thompson ride the Z50R. Kitzes was not, therefore, qualified to testify as an expert regarding the ability of children under 12 to ride this motorbike for he has no knowledge, skill, experience, training, or education regarding the ability of children to safely ride this motorbike. Therefore, his testimony concerning the safe age of children to ride the bike went beyond his area of expertise. For this reason, the District Court also abused its discretion in permitting Mr. Kitzes to testify regarding the age at which children can appropriately operate Honda’s motorbike.
c. Conclusion
Because there is insufficient evidence to conclude that either of the two alleged defects enumerated above was the proximate cause of the accident, it is my opinion that the District Court erred in failing to grant a motion for judgment as a matter of law under Rule 50(b) of the Federal Rules of Civil Procedure. Based on the totality of the evidence presented, the jury’s conclusion that Honda was liable for Gary Thompson’s injuries was not supported by sufficient evidence.
II.
Accordingly, in light of the several eviden-tiary errors and plaintiffs’ failure to carry *526their burden in establishing by a preponderance of the evidence that the motorbike was defective and that the defective design was a proximate cause of Gary Thompson’s injuries, I would reverse the order of the District Court denying defendants’ motion for judgment as a matter of law.
. Kitzes added, however, that no Kentucky standard or regulation required an off-road motorbike to have a wind flag.
. Due to the Supreme Court’s clarification of the standard of review in Joiner, I do not believe the majority's opinion needs to struggle with the question of the proper standard of review to apply and whether Cook v. American Steamship Co., 53 F.3d 733 (6th Cir.1995), conflicts with other decisions of our Circuit.
.The testimony in the joint appendix does not indicate how many years Kitzes worked for the CPSC.
. Furthermore, given his experience with the CPSC, Kitzes was likely qualified to testify regarding necessary or appropriate warnings for the motorbike. Notably, however, the jury found that Honda’s warnings were adequate.
. Trooper Ronald Wardrip testified Lhal the hay bales were between 5 1/2 and 6 1/2 feet tall and that they sat on a 3 to 3 1/2 foot embankment.
. His testimony in the joint appendix ends here and picks up again twelve pages later. The next few paragraphs are from the record transcript. While the majority opinion properly emphasizes that it is the responsibility of both parties to submit a complete joint appendix to the court, we need not limit our review to the joint appendix as the majority indicated it has.
. ■ Supporting Fowler's testimony and description of the scene. Trooper Ronald Wardrip testified that the embankment, the bales of hay, a fence, a telephone pool and a large tree obstructed Ms. Graham's view. Wardrip concluded that the combination of the obstructed view, Gary’s failure to yield the right-of-way, and his inattention caused the accident.
. Dr. DeMoss utilized the CPSC statistics as well.
. Rule 103(a) provides, “Error may not be predicated upon a ruling which admits or excludes evidence unless a substantial right of the party is affected.” Fed.R.Evid. 103(a).
. Therefore, Dr. DeMoss testified that children of this age do not have the requisite cognitive skills to even ride a bicycle becausé that activity *525requires the ability to concentrate on and process several tasks at once.
. I am of the further opinion that, while Dr. DeMoss was generally qualified to testify regarding the cognitive skills of children, she was not qualified to testify regarding the cognitive skills required to ride a motorbike. In order to determine whether Dr. DeMoss was qualified to testify regarding the cognitive skills required to ride a motor bike, we examine her knowledge, skill, experience, and training. Dr. DeMoss testified on cross-examination that none of her published papers had anything to do with "child skill development and the operation of motor vehicles” or even "skill development in the use of any type of recreational vehicle.” J.A. 255. Furthermore, Dr. DeMoss did not "physically observe young children operating the Z50R.” J.A. 256. Given this testimony, it is my opinion that the District Court abused its discretion in determining that Dr. DeMoss was qualified to give expert testimony on the cognitive skills required to operate a motorbike. However, defendants have not argued to this Court that Dr. DeMoss’s testimony should not have been permitted.
. Just prior to this portion of his testimony, Mr. Kitzes opined that age 12 was the absolute minimum appropriate age for a child to ride the motorbike. J.A. 285.