Forensic Metrology: The New Honesty about the Uncertainty of Measurements in Scientific Analysis
33 Pages Posted: 6 Dec 2012
Date Written: December 6, 2012
Abstract
In many cases, forensic scientists rely on measurements as a basis for their opinions. For instance, toxicologists measure the concentration of the toxic drug in human bodies to determine whether the level of toxin was high enough to have caused the person’s death. Accident reconstruction experts measure the length of skidmarks and yaw marks to determine the velocity of a vehicle. Measurements of blood alcohol concentration have long been critical in drunk driving prosecutions. Metrology is the science that studies measurement. Metrologists have long recognized that there is an unavoidable, inherent element of uncertainty in every measurement. Thus, if at trial the court permits the expert to testify to a single point value for the measurand, the trier of fact may erroneously conclude that the number is the exact value of the measurand. The presentation of a single point value can easily mislead the trier.
It was understandable that in the past, the courts allowed the expert to present such testimony. The early view was that the expert had to vouch for his or her opinion as a scientific or medical certainty. Many courts demanded that the expert use the term “certainty” in the phrasing of his or her opinion.
However, that view has fallen into disrepute. In Daubert, the Supreme Court recognized the limitations of the scientific enterprise. In one passage, Justice Blackmun wrote that “arguably, there are no certainties in science.” In investigational science, researchers typically rely on inductive reasoning. There are limits to inductive logic. It is true that if repeated experiments yield the same result, the outcomes can give the researcher a good deal of confidence in the hypothesis. However, another experimental test is always conceivable; and so long as that possibility remains, subsequent falsification of the hypothesis remains possible. Thus, the expert cannot regard the hypothesis as certainly or conclusively validated.
This new understanding of the limits of the scientific process is impacting the judicial treatment of expert opinions. The initial step was to abandon the insistence that the expert vouch for his or her opinion as a certainty; the courts allowed the expert to testify in terms of probabilities and possibilities. However, the logic of the new understanding led farther. Post-Daubert, now a substantial number of courts forbid the expert from testifying to a “certainty” or “zero error rate.” This article points out that some courts are now taking an additional step. In several drunk driving cases in Washington, the courts have insisted that the prosecution expert provide the trier of fact with both a best estimate (e.g. a bias-adjusted mean) and a quantitative measure of the error or uncertainty of the estimate. The Washington cases have demanded that the expert compute a confidence interval for the estimate. However, as this article explains, in metrology there are other possibilities such as coverage intervals.
This development is hopeful for two reasons. First, this development reduces the risk that the trier of fact will attach undue weight to a point value cited by an expert. The development may produce more accurate factfinding. Second and perhaps even more importantly, the development can promote a more honest, open relationship between law and science. Scientists have long known that there is an inherent element of uncertainty in measurement, and this judicial development permits them to be honest and frank in their courtroom testimony.
Suggested Citation: Suggested Citation