I. Background
PER CURIAM.
District Court granted summary judgment in favor of the defendants on the grounds that plaintiffs expert witness was not qualified to testify and that plaintiff presented no evidence of a design defect. This ruling is affirmed with respect to the findings that Michigan law applied to the issue of punitive damages, and that no defect had been shown with respect to the seatbelt warning light. However, the District Court’s Opinion will be reversed with respect to the findings that the expert witness was not qualified to testify, and that insufficient evidence had been produced to place the question of the seat-belt’s defective design before the jury.
Dr. Nemir testified at deposition that he buckled into his car on the night of December 14, 1993, and drove his Dodge Stealth a short distance before the car left the roadway and impacted a fence post or small tree at 18 miles per hour. The Dodge Stealth used Takata 52 seatbelts. Dr. Nemir was found lying in the back seat with the driver’s side seatbelt unlatched. The shoulder and back of the jacket he wore showed rub marks which indicate that the seatbelt came in contact with his jacket. Dr. Nemir’s head had struck the back pillar support of the vehicle, and he suffered major brain damage. He is wheel-chair bound and is no longer able to practice as a pediatric physician.
During the preparation of his case, Dr. Nemir employed Thomas Horton as his expert witness. Horton is an engineer [269]*269with extensive training in seatbelt design, an individual who had once been Director of Engineering for the company that manufactured the seatbelt in question. Horton surmised that Dr. Nemir’s seatbelt was “partially latched” at the time of impact, and came unlatched during the accident. Horton testifies that this is a design defect. To arrive at this conclusion. Horton relies on his background work and education, his observation and analysis of the seatbelt in question, and some testing he performed on exemplar buckles which were Takata 52 model buckles, but were not taken from the car at issue. The district court did an extensive analysis of Horton’s testing on the exemplar buckles, but did not analyze Horton’s testimony regarding any other issue. The district court struck Horton as an expert witness and granted summary judgment to the defendants.
The district court exercised subject matter jurisdiction in this matter pursuant to 28 U.S.C. § 1332, as the parties are of diverse citizenship and the amount in dispute is over $75,000. This is a direct appeal from the district court’s grant of summary judgment and jurisdiction lies with this Court under 28 U.S.C. § 1291. Summary judgment was entered on July 30, 1999, and a timely appeal was filed on August 9,1999.
II. Standard of Review
The district court’s grant of summary judgment is reviewed de novo. E.E.O.C. v. Prevo’s Family Market, 135 F.3d 1089, 1093 (6th Cir.1998). The Federal Rules of Civil Procedure require that summary judgment shall be granted if “there is no genuine issue as to any material fact and that the moving party is entitled to a judgment as a matter of law.” Fed. R.Civ.P. 56(c).
There exists a genuine issue when “the evidence is such that a reasonable jury could return a verdict for the nonmoving party.” Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 248, 106 S.Ct. 2505, 91 L.Ed.2d 202 (1986). All factual inferences are “viewed in the light most favorable to the party opposing the motion.” Matsushita Elec. Indus. Co. v. Zenith Radio, 475 U.S. 574, 587, 106 S.Ct. 1348, 89 L.Ed.2d 538 (1986).
With regards to the issue of whether or not a product is “inherently dangerous”, Maryland law determines that this issue must be resolved by the court as a matter of law. Lundgren v. Ferna-Washington, 80 Md.App. 522, 565 A.2d 335 (Md.App. 1989). Specifically, the Court must determine whether the plaintiffs claims may reasonably lead a jury to conclude that the product is “inherently dangerous.” Summary judgment cannot be issued if there is a genuine issue of material fact that would, if proven, allow the plaintiff to shoulder their burden of proof. Binakonsky v. Ford Motor Co., 133 F.3d 281, 284-285 (4th Cir.l998)(applying Maryland law).
The exclusion of expert testimony is reviewed for abuse of discretion. Greenwell v. Boatright, 184 F.3d 492 (6th Cir.1999).
III. Analysis
In his Opinion and Order dated July 30, 1999, Judge Feikens addressed three issues. [Record at 39]. First, Judge Feikens found that the plaintiffs proffered expert testimony was inadmissable under FRE 702 and Daubert. Second, the District Court found that Maryland law applied to the design defect issue.1 Finally, it was determined that the plaintiff could [270]*270not produce admissible evidence to create a question for the jury regarding their design defect claim.
Admissibility of Expert Testimony from Thomas Horton
Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 113 S.Ct. 2786, 125 L.Ed.2d 469, (1986), sets forth four factors which may be taken under consideration when determining whether or not to admit an expert’s testimony. Kumho Tire Co., Ltd. v. Carmichael, 526 U.S. 137, 119 S.Ct. 1167, 143 L.Ed.2d 238, (1999), clarified previous case law to indicate that technical testimony falls within the same category as scientific testimony. Therefore, the “factors identified in Daubert may or may not be pertinent in assessing reliability, depending on the nature of the issue, the expert’s particular expertise, and the subject of his testimony.” Id. at 150, 119 S.Ct. 1167. However, the Supreme Court did not intend to establish a definitive set of standards under which district courts must analyze the proffered testimony of expert witnesses. The Supreme Court would “neither rule out, nor rule in, for all cases and for all time[,] the applicability of the factors mentioned in Daubert.” Id. The Kumho decision refused to create a structural method for analysis of “subsets of cases categorized by category of expert or by kind of evidence [as][t]oo much depends upon the particular circumstances of the particular case at issue.” Id.
Federal Rules of Evidence, and the decisions of the Supreme Court, indicate that the district court must carefully review the qualifications and methodology employed by the proposed expert witness. While the Supreme Court has suggested certain indicators which could assist the district court in its determination, these factors are flexible, particularly when reviewing the proffered testimony of an engineer. As the Supreme Court specifically noted, “the test of reliability is ‘flexible,’ and Daubert’s list of specific factors neither necessarily nor exclusively applies to all experts or in every case.” Kumho Tire Co., Ltd. v. Carmichael, 526 U.S. 137, 141—142, 119 S.Ct. 1167, 143 L.Ed.2d 238 (1999). In fact, “the law grants a district court the same broad latitude when it decides how to determine reliability as it enjoys in respect to its ultimate reliability determination.” Id. Accordingly, a district court judge’s exclusion of expert testimony is given strong deference.
However, the court must make a “determination as to the proposed expert’s qualifications.” Morales v. American Honda Motor Co. Inc., 151 F.3d 500, 516 (6th Cir.1998). In order to do this, the Court must “investigate the competence a particular proffered witness would bring to bear on the issue without relying upon any particular label or fact, and must determine whether it would aid the trier of fact in reaching a decision.” Id. referring to Mannino v. International Mfg. Co., 650 F.2d 846, 850 (6th Cir.1981). Although “the decision of whether to admit a witness’s testimony is left to the sound discretion of the trial court, a trial court cannot make an arbitrary decision.” U.S. v. Smithers, 212 F.3d 306, 315 (6th Cir.2000).
In the ease at bar, the district court held numerous oral arguments regarding the admissibility of Thomas Horton’s testimony. While the trial court’s investigation and issued opinion appear to exhaust this subject, it is clear that the district court erred in its analysis by excluding Horton’s entire testimony. The district court opinion focused solely on the unreliability and irrelevance of Horton’s tests on exemplar buckles, while ignoring the fact that much of Horton’s deposition testimony and proposed testimony focused on much more [271]*271than the simple testing of the exemplars at issue. It was arbitrary to decide that analysis would focus on a single aspect of Horton’s testimony.
Horton testified on numerous issues during deposition. He testified regarding the mechanics of seat belt operation, based on his of experience as an automotive safety engineer. During his career, Horton worked for General Motors, TRW, Inc., and was the Director of Engineering at Takata, Inc., the company which manufactured the Takata 52 seatbelt, Record at 171-172. Horton was qualified to testify regarding the internal latching process of the seatbelt in question. This testimony was clearly relevant to the case at bar, as such testimony would help laymen jurors understand the complex internal structure of the seatbelt at issue. However, the district court fully excluded all of Horton’s testimony without consideration of the relevance of this line of testimony.2
Horton also testified regarding the partial latch phenomenon, the federal regulations governing it, and that an existing alternative seatbelt design circumvents the issue of partial latch. It is undisputed that partial latch is a real phenomenon that is tested and regulated. 49 C.F.R. § 571.209 addresses seatbelt safety standards. This regulation requires all metal-to-metal buckles, like the Takata buckle at issue, to undergo testing to insure that a partial engagement, which requires more than five pounds of force to separate, shall not occur. During his deposition, Horton testified about this regulation, and about the mechanics of a partial latch situation. This testimony was independent of his testing of exemplar buckles, and would help the jury to understand the mechanics of a partial latch situation. In addition, Horton testified that the RNS-3 belt is structurally different from the Takata 52 seatbelt. Record at 353-355. Horton also explained how he had tested the RNS-3 and found that it is incapable of partial latch. Record at 371. Horton explained that the RNS-3 is a viable alternative to the Takata in question because it is not subject to partial latch, could be manufactured on an automated assembly line, contains “far fewer” parts, and is made at a lower cost than the Takata 52. Record at 372. While all of this testimony would assist the trier of fact in reaching a determination on the pivotal issues in this case, and is independent of the excluded testimony on exemplar testing, the district court nevertheless struck such testimony.
Horton also testified that his examination of Dr. Nemir’s latch plate indicated “numerous scratch marks” which are “indicative of frequent use.” Record at 336. This testimony would bolster Dr. Nemir’s testimony that he remembers fastening his seatbelt on the night in question, and that he was in the habit of wearing his seatbelt. While this testimony is based on Horton’s expertise as a automotive safety restraint engineer, and is independent of his testimony regarding the testing of exemplar buckles, the district court still excluded said testimony.
Finally, Horton had examined and demonstrated the partial latch hypothesis on the actual seatbelt at issue. The Court inexplicably ignored this portion of Horton’s research, and instead dismissed Horton as an expert based on the fact that [272]*272Horton’s testing methods involved exemplars taken from other vehicles. In fact, in the district court’s opinion, footnote 13 [Record at 56], the district court notes that Horton “has not even attempted to perform his testing on the buckle from Dr. Nemir’s driver’s seat.” The district court describes this omission as “inexplieabl[e].”3 Id. However, this statement is clearly contrary to the evidence before the district court. During his deposition, Horton testified that he had inspected the Nemir seatbelt, and while he did not “measure any loads”, he did “check it for partial latching.” Record at 330. He testified that he inspected the seatbelt on December 9, 1997. Record at 335. During this inspection, he determined that the “buckle and latch plate will partial latch after full latch plate engagement and partial depress of push button.” Record at 339. Horton also testified in his capacity as a designer and engineer of seatbelts, that a properly designed “buckle shouldn’t get into partial engagement positions.” Record at 344. On March 2, 1999, the plaintiff presented the Court with an affidavit to further clarify Horton’s deposition testimony.4 This document clearly presents evidence which would allow Horton to testify that Nemir’s seatbelt could be put into a position of partial latch, and that as such, the seatbelt was defectively designed. Record at 215-217. This affidavit clarifies Horton’s deposition testimony and sets forth Horton’s opinion that “the ability of a belt buckle to achieve partial latch at all, by virtue of either method [of testing], presents an unreasonable risk to the consumer and is a dangerous and defective design.” Record at 215(emphasis in original). Horton also notes that, on February 11, 1999, Horton further tested Nemir’s seatbelt and determined that partial latch occurred two out of twenty tries, and that partial latch was achieved without manipulation of the push button. Record at 216.
Horton also testified that a seatbelt’s ability to partially latch utilizing any method available, rather than the actual use standard employed in the district court’s analysis, creates an inherently dangerous product. Horton further testified that he believed that this design defect is causally connected to Dr. Nemir’s resulting injury. Record at 351. While he did not measure the exact amount of force required to pull Dr. Nemir’s belt out of partial engagement, he testified that the buckle took [273]*273“more than a reasonable amount of poundage to pull it out of partial engagement.” Id. From this testimony, the jury could conclude that Horton’s demonstrations and analysis of the actual buckle show that the buckle in Dr. Nemir’s car could be put into a position of partial engagement, and remain in that position despite the natural shifting of positions that occurs when someone is driving a car, and therefore led Dr. Nemir to assume that his belt was adequately fastened, despite the fact that the buckle failed on impact. This could lead a jury to conclude that the design defect caused the injury to Dr. Nemir.
The district court struck Horton as an expert witness based on the finding that his testing of exemplar buckles was inadmissible under Daubert. However, the district court neglected to review the admissibility of the remainder of Horton’s testimony. Horton’s explanation of the mechanical functions of a properly working seatbelt, the occurrence and testing of the partial latch phenomenon, the availability of an alternative design, the physical evidence of Dr. Nemir’s frequent seatbelt usage, and the testing of Dr. Nemir’s actual seatbelt, are all components of Horton’s testimony which are isolated from his testimony regarding the testing of exemplars. Plaintiff Presented Sufficient Proof to Establish a Design Defect
In reaching its determination that the plaintiff failed to produce evidence of a design defect, the district court erred in its analysis by ignoring relevant evidence properly before the court. The district court concludes that, “the principal shortcoming of plaintiffs proofs is the complete lack of evidence as to the existence of inadvertent partial latching in actual use.” Opinion and Order, at footnote 16, Record at 24-25. The district court reiterates this conclusion four more times,5 indicating that this lack of evidence is the case’s “fatal flaw.” Record at 68. The assumption that the plaintiff failed to produce any evidence of partial latching in actual use is flawed in two ways. First, such evidence was produced. Second, the lack of such evidence is not a death knell to the case at bar.
The evidence indicating that partial latching occurs in actual use can be divided into two categories. The first category involves the evidence showing that Dr. Nemir actually experienced a partial latch in the case at bar. The second category presents evidence that partial latching can and has occurred within the general population.
There are five important points of evidence that relate specifically to Dr. Nemir.6 None of these factors appeared in the district court’s analysis of this case. Therefore, it is unclear whether or not the district court considered any of this evi[274]*274dence in reaching the conclusion that the plaintiff could not assert a claim for design defect. First, Dr. Nemir testified that he clearly remembers buckling his seatbelt prior to the accident in question. Hearing Transcript, Record at 486, Dr. Nemir had a shoulder injury and was unable to latch his belt on the first try. Therefore, he got out of the car, took ahold of the seatbelt with his good arm, and buckled up as he entered the vehicle. Id. The testimony that Dr. Nemir buckled his seatbelt before driving on the night in question must be assumed true for the purposes of a summary judgment motion. Matsushita Elec. Indus. Co. v. Zenith Radio, 475 U.S. 574, 587, 106 S.Ct. 1348, 89 L.Ed.2d 538 (1986).
Second, Nemir’s testimony that he buckled his belt is bolstered by Horton’s testimony regarding the markings on Nemir’s latch place, which indicated frequent use. Horton Deposition, Record at 336. The district court did not mention or consider this testimony. Likewise, the opinion and order made no reference to the fact that Michael Klima, the defense’s expert witness, testified that he “can’t make a real strong affirmative statement that he wasn’t belted” into his car on the day of the accident in question. Record at 138.
Third, Horton also testified that Nemir’s belt did not show signs of abuse, or extraordinary wear and tear, or corrosion. Record at 348. He did not find any evidence of dirt or debris in the buckle. Horton only found a little dust near the mouth of the buckle. Record at 349. This testimony works against the presumption that an object or article of clothing became lodged in the buckle, preventing proper latching.
Fourth, as discussed in the previous section, Horton tested Nemir’s belt buckle and determined that it could be placed into a state of partial engagement by slightly depressing the release button, and by improper or incomplete insertion. Record 339, 215-217. He found that it took an unreasonable amount of force to disengage the budde once it was put into a state of partial engagement.
Finally, the district court made no mention of the fact that the load marks along the shoulder and back of the jacket worn by Dr. Nemir on the night of the accident could indicate that his seatbelt was engaged at the time of impact. Hearing Transcript, Record at 476. Dr. Nemir’s jacket shows rub marks, possibly from the seatbelt. This inference is supported by the fact that jacket fibers were found in the seatbelt, indicating that his body was straining at the shoulder harness. The district court makes no comment with regard to this evidence.
When viewed as a whole, the above evidence could be utilized by a jury to conclude that Dr. Nemir had buckled his seatbelt on the night of the accident, and that the seatbelt was clean, and appeared to be in working order. Horton’s examination of the buckle and the load marks on the jacket could lead a reasonable jury to conclude that the seatbelt remained attached until impact, when it released and retracted, causing damage to the jacket.7 The jury could also conclude that the release of the seatbelt caused the injury to Dr. Nemir. This evidence could lead a jury to conclude that the partial latching phenomenon can not only occur in actual [275]*275usage, but did in fact occur in the situation at bar.
The Sixth Circuit has recently held that an expert witness’ conclusion regarding all admissible evidence “need not eliminate all other possible causes of injury” 8 in “order to be admissible on the issue of causation.” John v. Equine Services, PSC, 233 F.3d 382, 390 (6th Cir. 2000). Accordingly, Horton could use this evidence to testify that the partial latch of Dr. Nemir’s seatbelt caused the damage in question. While other possibilities exist with regards to causation, the “fact that several possible causes might remain ‘uneliminated’. .. only goes to the accuracy of the conclusion, not the soundness of the methodology .” Id. In evaluating an expert witness, “Daubert and Rule 702 require only that the expert testimony be derived from inferences based on a scientific method and that those inferences be derived from the facts on the case at hand ... not that they know the answers to all the questions a case presents-even to the most fundamental questions.” Id. The district court, by requiring “specific knowledge of the precise physiological cause” of the accident “held the expert[ ] up to entirely too strict a standard when considering the admissibility of them testimony.” Id. The testimony of Horton, establishing causation based upon the above referenced facts, would adequately shoulder the Michigan burden of proof that plaintiff show “‘a reasonable likelihood of probability’ that his explanation of the injury is correct.” McLean v. 988011 Ontario, Ltd., 224 F.3d 797, 800 (6th Cir.2000) quoting Skinner v. Square D Co., 445 Mich. 153, 516 N.W.2d 475 (1994).
In addition to the above evidence, the district court failed to discuss the evidence that partial latching can occur within the general population’s actual seatbelt usage. First, the defense’s expert witness, Michael Klima, testified at deposition that he could not “rule out the possibility that a partial latch could end up occurring in the real world.” Record at 136. Second, the district court perfunctorily disregarded hundreds of pages of consumer seatbelt complaints. This evidence was released to the plaintiff in the eleventh hour, after years of discovery disputes over whether or not such complaints were discoverable material. The district court only mentioned these documents briefly in a footnote stating, “plaintiff has been unable to present even one case in which partial latching or engagement was identified as the cause of a customer’s complaint or injuries.” Opinion and Order, footnote 19, Record at 63. Accordingly, the district court ruled that these complaints were of no probative value and refused to grant the plaintiff additional time to contact the customers in order to more specifically detail their seatbelt complaints.
[276]*276The customer complaints proffered by the defense all contain short vernacular sentences or phrases to describe the customer’s grievance. Often, the complaints appear to be utilizing the customers own phraseology. It is not surprising the term “partial latch” does not appear in any of the complaints. However, this is not grounds for ruling that the complaints are distinct from the case at bar for two reasons. First, the term “partial latch phenomenon” is a term of art employed by engineers, seatbelt designers, and lawyers. The layman with a concern that his seat-belt will not remain fastened will not use this term. Second, the term itself is a red herring. To describe this problem as a “partial latch phenomenon” does not encompass the breadth of the problem. This is not a case where the seatbelt will be wrongfully engaged each time it is used. If this were the case, the vehicle owner would most likely be aware of the issue as his belt would disengage each time he made a sudden stop or hit a bump in the road. Mechanically and internally, the buckle in question will, at times, partially latch. However, the outside observer does not know that the internal mechanism, at times, partially latches. From the driver’s point of view, the buckle intermittently unlatches: sometimes the belt works, sometimes it does not. The correct term that must be found in customer complaints, in order to render the complaints analogous to the case at bar, would be “intermittent partial latching.5’ However, this term is beyond the ken of the average car owner.
Instead, the car owner is likely to explain the problem using terms such as “sometimes pops out”, or “not staying locked all the time”, or “does not work at times.” The vernacular possibilities are endless. A complete review of the record indicates that phraseology analogous to “intermittent partial latching” occurred 188 times in the customer complaints submitted as evidence. To summarily dismiss this evidence without affording the plaintiff time to investigate and further elicit information regarding the exact nature of these incidents is clear error. It is possible that this investigation could reveal that intermittent partial latching occurs frequently during the general population’s actual seatbelt usage. This issue potentially impacts the safety of millions of American consumers and the plaintiffs should be granted sufficient time to fully investigate their case.
Accordingly, if the plaintiffs “fatal flaw” in his design defect claim is the lack of evidence of partial latch occurring in actual use, the evidence clearly demonstrates that plaintiff has produced enough evidence to submit this claim to the jury. Plaintiffs evidence meets the necessary elements set out in the district court’s opinion and order. Record at 39. Plaintiff demonstrates a triable issue of fact with regard to the design defect condition and the unreasonably dangerous condition of the product. The Sixth Circuit notes that “the facts established by admissions in the pleadings and in the record must be construed most strongly in favor of the party against whom the motion is made.” Clay v. Ford Motor Co., 215 F.3d 663, 669 (6th Cir.2000). Summary judgment was inappropriate as “[rjeasonable minds considering this evidence could reach different conclusions on whether the alleged design defects caused the accident. Construing the evidence most strongly in favor of the appellees, there was substantial competent evidence upon which reasonable minds could reach different conclusions.” Id. at 672. The plaintiffs evidence could lead a reasonable jury to conclude that the product poses an inherently unreasonable risk. The plaintiffs have “presented a plausible theory, supported by what evidence is [277]*277available, which if believed by the jury” could establish the defendants’ liability. McLean v. 988011 Ontario, Ltd., 224 F.3d 797, 805 (6th Cir.2000).
However, it is also important to note that evidence regarding the occurrence of partial latch in actual use is unnecessary for the plaintiff to create a jury issue in this case. Horton’s properly submitted affidavit states that “the ability of a belt buckle to achieve partial latch at all, by virtue of either [testing] method, presents an unreasonable risk to the consumer and is a dangerous and defective design.” Record at 215. This testimony creates a jury issue with regards to whether or not the Takata 52 seatbelt is “unreasonably dangerous” under Maryland’s adopted Restatement of Torts § 102A.
Plaintiffs Failure to Warn Claim
Plaintiff argues that his “fasten seat belt” light did not come on when his seatbelt was in the partially engaged position. However, no evidence has been shown to indicate that this design defect is actionable. While the mechanism in place will not light up when the seatbelt is partially latched, Plaintiffs expert witness testified that it would not be possible to “fix” that problem and that an alternative design is not available. Therefore, summary judgment on this issue is upheld.
Plaintiffs Claim for Punitive Damages
This claim was dismissed when the district court determined that Michigan law applied to the issue of punitive damages. A federal court sitting in diversity must apply the choice of law rules of the state in which it sits. Klaxon v. Stentor Electric Mfg. Co., 313 U.S. 487, 61 S.Ct. 1020, 85 L.Ed. 1477 (1941). In Michigan, courts apply Michigan law unless there is a “rational reason” to justify displacing it. Olmstead v. Anderson, 428 Mich. 1, 400 N.W.2d 292 (Mich.1987). The district court did not find a reason to displace Michigan law, which does not allow punitive damages. This finding is not contrary to case law on the issue, and are upheld.
This case is AFFIRMED IN PART, and REVERSED IN PART with instructions to remand this case to the district court for further proceedings consistent with this opinion.