dissenting.
Because I believe that the district court abused its discretion in admitting Dr. Richardson’s testimony, I respectfully dissent.
The majority describes the admissibility of Dr. Richardson’s testimony as a “close case.” I disagree. In my view, this case presents a clear example of a court improperly abdicating its gatekeeping function under Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993), and Kumho Tire Co. v. Carmichael, 526 U.S. 137, 119 S.Ct. 1167, 143 L.Ed.2d 238 (1999).
I.
Expert testimony is admissible under Fed.R.Evid. 702 only if the trial court finds that the testimony is both relevant and reliable. Daubert, 509 U.S. at 589, 594-95, 597, 113 S.Ct. 2786. While Daubert focuses on “scientific” experts, the Supreme Court recently made clear that the Dau-bert principles apply to the opinions of “non-scientific” experts such as engineers. Kumho, 526 U.S. at 141, 119 S.Ct. 1167. Here, there is no question that Dr. Richardson’s testimony was relevant to the material issues presented in this product liability lawsuit. The critical question, then, is whether the testimony was reliable.
II.
The procedural history preceding the admission of Dr. Richardson’s testimony provides the first unmistakable sign of trouble. As the majority acknowledges, the district court refused to hold a hearing on Ford’s pretrial motion in limine, “reserving] to itself the possibility of engaging in a brief voir dire of Dr. Richardson before his testimony is offered.” (Emphasis added.) At trial, after Dr. Richardson summarized his training and experience, plaintiffs’ counsel moved to qualify him as an expert in several engineering-related fields. Defense counsel declined the opportunity to conduct voir dire at that time because the questioning would appear disjointed from the jury’s perspective. The court suggested that, given the breadth of Dr. Richardson’s areas of expertise, defense counsel reserve his questioning of Dr. Richardson for cross-examination. The court then ruled that Dr. Richardson was qualified as an expert in several fields, with the understanding that the defense would make continuing foundational objections as the testimony proceeded. At the conclusion of the plaintiffs’ case, Ford moved to strike Dr. Richardson’s testimony pursuant to Daubert. The court overruled the motion without providing any reasoning on the record.
Although we have stated that a trial court need not hold an evidentiary hearing in every case to comply with Daubert, we have made it clear that the court must make an “initial assessment of the relevance and reliability of the expert testimony.” Greenwell v. Boatwright, 184 F.3d 492, 498 (6th Cir.1999) (emphasis added). Here, the district court did not make such an initial assessment, and indeed made no express assessment at any time as to the reliability of Dr. Richardson’s testimony. The absence of such determination signals the trial court’s failure to perform its gatekeeping function as mandated by Dau-bert and Kumho. In my view, this failure alone constitutes an abuse of discretion. The majority, however, is willing to overlook this failure, in essence substituting its own determination of reliability for that of the district court. Even if I agreed with the majority’s approach, I cannot agree with the majority’s assessment of reliability-
ill.
Daubert sets forth a non-exclusive list of factors for trial courts to consider in determining the reliability of expert testimony: (1) whether a theory or technique can be *675(and has been) tested; (2) whether the theory or technique has been subjected to peer review and publication; (3) the known or potential rate of error in using a particular scientific technique and the existence and maintenance of standards controlling the technique’s operation; and (4) whether the theory or technique has been generally accepted in the particular scientific field. Daubert, 509 U.S. at 593-94, 113 S.Ct. 2786. A trial court may consider one or more of these factors, but the Supreme Court has cautioned that this “list of specific factors neither necessarily nor exclusively applies to all experts or in every case” and that a trial court has “broad latitude” to determine whether these factors are “reasonable measures of reliability in a particular case.” Kumho, 526 U.S. at 141, 153, 119 S.Ct. 1167. Despite the broad flexibility afforded to trial courts, I do not read Daubert or Kumho to mean that the complete disregard of one or more of the Daubert factors in a given case can never be definitive.
In its application of Daubert, the majority first devotes significant attention to Dr. Richardson’s general education and experience. There is no question that a witness’s education, personal knowledge, and experience are important components of the witness’s qualification to render an expert opinion, viz, whether he or she is an “expert.” But in cases involving scientific opinion {Daubert cases) or applied scientific opinion as in matters of engineering (Kumho cases), it is the methodology employed by the expert, not the expert’s general educational qualifications, that is in issue. Dr. Richardson’s impressive academic and experiential history tells us nothing about how he did what he did to reach his conclusions in this case. And I do not read the majority’s opinion to suggest otherwise.
To be sure, the majority opinion summarizes Dr. Richardson’s testimony describing the methods he employed in reaching his conclusions. For example, before he opined on accident reconstruction, Dr. Richardson reviewed the process he employed to reach his conclusions. He personally reviewed several discovery depositions in the litigation, reviewed police reports, and prepared a scale drawing reflecting points identified in the police reports. He also visited the scene of the accident, but this visit occurred the day before the trial, well after he had formed his opinion. Finally, Dr. Richardson reviewed photographs of the vehicle following the accident. He admitted, however, that he did not personally inspect the vehicle, but rather sent his son to take the photographs. While this evidence indirectly suggests that Dr. Richardson thought his methodology was reliable, it hardly suffices as evidence of reliability under Daubert. See id. at 157, 119 S.Ct. 1167. Nothing in this testimony touches on any of the Daubert factors, or any other measures of reliability, for that matter. The record is absolutely devoid of any indication that the process or methodology Dr. Richardson employed in reaching his accident reconstruction opinion was “good science” or “good engineering.”
The majority minimizes this deficiency by noting that Ford’s accident reconstruction expert, Dr. Carr, performed some of the same procedures in reaching his conclusion about how the accident occurred, which, of course, was very different than Dr. Richardson’s conclusion. It is not clear to me how this observation is useful. Similarity in the methods employed by two groups of experts tells us nothing about the legal sufficiency {i.e., the reliability) of either. Moreover, plaintiffs’ counsel did not object to the admission of Dr. Carr’s testimony. The majority also overlooks potentially key distinctions between the experts’ methods. Unlike Dr. Richardson, Dr. Carr personally inspected both the site of the accident and the vehicle following the accident. He also testified that his opinion on the accident reconstruction was based, in part, on his extensive personal experience testing the design and handling of the vehicle model involved in this case. *676Dr. Richardson’s experience testing any vehicle was much more limited, and he did not rely on this experience in reaching his accident reconstruction opinion. More importantly, he conducted no tests whatsoever on the model of vehicle involved in this case.
The record is also devoid of any evidence showing that Dr. Richardson employed recognized, reliable scientific or engineering methodology in reaching his conclusion that the Bronco II has handling and stability design defects. What’s more: the district court required no such evidence in 'discharging its gatekeeping function. In general, a district court should require compliance with at least some of the Daubert factors in determining whether an engineer’s opinion on a design defect is reliable. Engineering is, after all, nothing more than applied science that “rests upon scientific foundations.” Id. at 150, 119 S.Ct. 1167. At a bare minimum, Dau-bert ’s first factor — “testing” or “falsifiability” — will generally require an engineer to test his hypothesis of design defect. As one of Ford’s experts explained, an engineer’s evaluation of a vehicle’s design involves not only following generally accepted design procedures, but also performing a variety of actual tests on the design during the development process. This is precisely what Ford did in developing the Bronco II.
. Dr. Richardson, however, based his opinion that the Bronco II was defectively designed solely on his review of Ford’s internal memoranda, industry comparisons of vehicle rollover rates, and Ford’s vehicle testing. While there is a certain logical appeal to the notion that Dr. Richardson’s opinion must be reliable if it rests upon data produced by the defendant, the notion does not withstand close consideration. What Daubert and Kumho require of the proponent of expert opinion is evidence that the methodology underlying the expert’s conclusion is “good science” or “good engineering.” That means that the plaintiffs were obligated to introduce at least some evidence that Dr. Richardson’s method — that is, examining depositions, police reports, photographs of the vehicle, Ford’s internal memoranda, industry comparisons of vehicle rollover risks, and Ford’s vehicle testing data, but never personally examining the accident vehicle or testing any other vehicle of the same model- — -is a sound engineering methodology for evaluating vehicle design.
Daubert and Kumho teach that whether such methodology is “good engineering” is determined by testing such data for “falsifiability”; considering whether peer review and engineering publications approve of such an approach; considering the rate of error such data might produce; and considering whether the relevant scientific/engineering communities generally accept consideration of such data as a sound methodology for reaching a conclusion about the design of a vehicle. However, Dr. Richardson conducted no “falsifiability” testing at all on a vehicle of the model involved in this case, and although he conducted very limited testing on a Bronco 4X4, a model not involved in this case, he did not base his design defect opinion on such tests; he never subjected his theories or techniques to peer review or publication; he produced no evidence of an error rate or standards controlling his “technique”; and he did not attempt to show that his “technique” or methodology was “generally accepted” in the scientific community. And the plaintiffs offered no proxies for these indicia of reliability.
The majority is very well aware, of course, of these deficiencies in the plaintiffs’ evidence, but seeks to dismiss them as unimportant. With regard to the first Daubert factor — “testing” or “falsifiability” — the majority acknowledges that Dr. Richardson never tested his hypothesis but concludes that the district court “could have” decided that this failure “went to the weight of [Dr. Richardson’s] testimony regarding defects in the Bronco II, not to its admissibility.” Whether the court did so or not, we do not know, because the court *677made no finding on the matter in the record. I agree that, as a general matter, a hypothesis may satisfy Daubert even -if it is untested by the expert proposing it. However, in such cases the proponent of the testimony must demonstrate other appropriate indicia of reliability. This requirement is all the more critical in a case in which an engineer who conducted no testing is testifying about a design defect, particularly given that actual testing is the principal method employed by the engineering community to prove or disprove an engineering hypothesis. But the plaintiffs submitted no evidence whatsoever that Dr. Richardson employed any acceptable scientific/engineering methodology, whether mentioned in Daubert or otherwise, to support his conclusion that the Bronco II was defectively designed.
I note in passing that even Dr. Richardson’s general qualifications as an expert in vehicle design defect are questionable, given that he has never published a single article on vehicle handling or stability; has never conducted any testing on a Bronco II, the vehicle model involved in this case; and has never designed a vehicle or vehicle components. The only machine Dr. Richardson has designed himself is an okra picker. To be sure, such lack of specific experience will not always disqualify an expert, and Ford does not focus its, challenge to Dr. Richardson’s opinion on his general professional qualifications. But Dr. Richardson’s questionable qualifications to render an opinion concerning the adequacy of Ford’s design of the vehicle in this case should at least have signaled to the trial court that the methodology this witness employed to get to his conclusion would require close gatekeeping scrutiny.
The majority further excuses the plaintiffs’ manifest failure to introduce proof of the engineering soundness of Dr. Richardson’s methodology, and perforce, the reliability of his ultimate opinion in accordance with Daubert and Kumho, by noting that Ford had an opportunity to cross-examine Dr. Richardson. It is not clear to me how this point is relevant, but in any event, I do not understand the majority’s observation to suggest that such cross-examination would properly substitute for a trial court’s gatekeeping function. This is not a duty a court may delegate to the jury under any circumstances. It was the trial court’s gatekeeping duty to require proof that Dr. Richardson’s methodologies were sound as a condition of admitting Dr. Richardson’s opinions; the court did not do so and, therefore, the expert’s opinion was improperly admitted.
In short, because I believe that Dr. Richardson’s testimony should have been excluded, and that the error in admitting it was not harmless, I would reverse the district court’s judgment.