The Bayesian Analysis of Testimony,
NORTHWESTERN UNIV EVANSTON IL DEPT OF STATISTICS
Pagination or Media Count:
The use of mathematical probability to analyze the probative force of testimony was among its earliest applications. For almost a century, the subject generated a substantial literature and was a common topic in the textbooks of the 19th century. Among its many contributors are such distinguished names as Condorcet, Laplace, Poisson, De Morgan, and Boole. From the very beginning, however, the theory suffered from a nearly universal failure to be grounded in empirical fact. In the end it became largely discredited, and today such attempts are often considered mere curiosties naive, erroneous, or uninformative. Nevertheless, there are several reasons why there is value in reviewing the history of such attempts. First and foremost, the analysis of testimony -- how to combine several items of evidence, and how to combine the testimony of several witnesses -- stands as a genuine and important problem that any theory of uncertainty must grapple with. It is the thesis of this paper that the best such attempts provide useful insights, and that the admitted limitations in such analyses are largely inherent in the questions asked.
- Statistics and Probability