The following point may be somewhat beside the point but ...: It should be noted that lack of veracity -- being a liar -- is only one of several "testimonial defects" that can detract from the probative value of a testimonial report; for example, thoroughly "sincere" witnesses -- people who _want_ to tell the truth -- can give inaccurate reports because their memories are bad, or because their sensory capacities (e.g., eyesight) are poor, or because of matters such as expectations or biases they honestly but mistakenly misinterpret the evidence of their senses, etc.
For a mathematical treatment (Bayesian) of such matters see the work of David Schum. (He is fond of calling "credibility a "multiattribute characteristic." Best wishes, Peter T > > From: Rolf Haenni <[EMAIL PROTECTED]> > Date: 2003/03/17 Mon AM 10:51:18 EST > To: [EMAIL PROTECTED] > Subject: [UAI] Laplace > > Dear UAI readers, > Pierre-Simon Laplace in his book "Théorie analytique des > probabilitiés, Paris, 2nd édition, 1814" mentioned a formula for the > probability p(h) of an event h given confirming reports from n > independent witnesses. The probability that an individual witness > tells the truth is p, whereas (1-p) is the probability the witness is > a liar: > p^n > p(h) = ------------- > p^n + (1-p)^n > Who knows more recent publications containing this formula? > Thanks and best regards. > Rolf Haenni > -- > ------- End of Forwarded Message
