Plaintiff was injured in a rollover car accident, and sued the car manufacturer alleging that a defect in the seat belt buckle caused it to release during the accident and allow her to be ejected from the vehicle. In support of this theory, plaintiff presented the opinion of Dr. Good, a mechanical engineer, who theorized that the buckle most probably inertially unlatched during the accident due to an alleged design defect. He ran a series of tests on buckles allegedly similar to the one in the accident, but ran into issues when he needed to make a comparison of the data from his lab tests to data from crash rollover tests to determine if the situation measured in the lab could actually occur in the real world. Specifically, there was an absence of available data from relevant rollover crash tests (which present dynamic, multi-dimensional forces working on the component), and so he compared his results to data from planar crash data -- ones focused on only the horizontal plane (for example, a frontal car crash).
Ford moved to exclude the opinions as unreliable under Daubert, but the district court (without a hearing) denied the motion, concluding Ford had failed to prove that the differences between the lab test results and the real world rollover accidents were significant. Defendant appealed. (Note, whether she was even wearing her seat belt at all was hotly contested at trial. For purposes of the Daubert issue, the court assumed she was.)
The court of appeals concluded that in permitting the testimony, the district court had not been "a sufficiently exacting gatekeeper; Daubert requires more precision." Plaintiff failed to present a sufficient scientific connection between the accelerations and forces the expert found necessary to unlatch the buckles in the lab, and the acceleration and forces that would have occurred in the actual accident on the street.
Specifically, the court of appeals held that the trial court should NOT have chastised the defendant for failing to show how the deficiency mattered, the failure to use rollover crash data. And the trial court should not have deemed it "unfair" for Ford to criticize the plaintiff because of the limited amount of rollover crash data available to the expert. The state of the science is what it is. And Ford did more than point out a deficiency in the method; it also explained why the deficiency rendered the testing and comparison suspect. More importantly for our readers, "it was not Ford's burden to show Good's inertial unlatch opinion was unreliable and irrelevant. Rather, it was plaintiff's burden to show reliability and relevancy."
It was undisputed engineering science that once a component is tested, the results must be applied to the whole vehicle setting; the lab results must be compared to data from the real world. Merely showing that similar buckles can be made to unlatch under certain lab conditions is irrelevant to whether the buckle at issue unlatched in the accident absent proof that the lab conditions were present and can be adequately and accurately related to the actual rollover-type accident. Plaintiff's expert failed to explain adequately how the acceleration and forces present in the planar crash tests were similar enough to those present in a rollover accident. Nowhere did he show how his comparison was scientifically valid. Thus, his opinion was based on mere speculation, or on the assumption, that the levels of forces he found necessary to unlatch buckles in the lab were substantially similar to those that occurred in the subject accident.
Absent such evidence, plaintiff could not meet her burden. Since plaintiff had a full and fair opportunity to present the case, and made no attempt to add or substitute other evidence, the court of appeals remanded with instruction for the district court to enter judgment as a matter of law for defendant.