Koehler, J. J. (1996). On Conveying the Probative Value of DNA Evidence: Frequencies, Likelihood Ratios and Error Rates. University of Colorado Law Review. 67, 859-886.

ABSTRACT

In trials involving probabilistic DNA evidence, there continues to be confusion about the meaning of the frequency statistics that accompany reports of DNA matches between a crime sample and a suspect. Recently, there has been a movement within the DNA statistics community to resolve this confusion by presenting jurors with Bayesian likelihood ratios to convey the significance of the DNA evidence. Although likelihood ratios are, in principle, an excellent way to describe the diagnostic value of DNA evidence, tricky normative and psychological questions remain.

On the normative side, questions remain about how the likelihood ratios should be constructed. What are the hypotheses that should be tested, and what exactly is the evidence that should be examined against those hypotheses? It will be argued that the evidence is best described as a "reported match" rather than as a "certain match," and that this has implications for the value of the resultant likelihood ratios.

On the psychological side, there is little reason to believe that jurors will understand likelihood ratios, and some reason to suspect that they will misinterpret likelihood ratios as posterior odds ratios. Preliminary experimental data that addresses this point will be presented. Furthermore, if jurors tend to treat likelihood ratios as posterior odds ratios, then it is not clear that they will pay sufficient attention to prior odds ratios in cases where these may be important. Preliminary data related to how mock jurors use prior odds evidence in DNA data bank match cases will be presented to address this issue.

OUTLINE

1. Frequency Statistics
   1. Eyewitness characteristics: People v. Collins (1968)
   2. DNA characteristics: People v. Simpson (1995)
   3. What do the frequency statistics mean?
2. The relevance of Bayes's theorem to DNA statistics
   1. What are the priors?
   2. What is the likelihood ratio?
      1. What is the "hypothesis"?
      2. What is the "evidence"?
      3. Should error rates be incorporated into the likelihood ratio?
         1. Counter-arguments
         2. Response to counter-arguments
3. Do people understand likelihood ratios?
   1. Experts don't
   2. Jurors don't
      1. Literature on probabilistic reasoning
      2. Experiment #1 with mock jurors
4. Will jurors understand the role played by prior odds ratios?
   1. Fallacious reasoning: The selection fallacy
   2. DNA data bank matches
      1. Experiment #2 with mock jurors
5. Conclusion