Bayesian Statistics

[Dan Perkins]

Hi,

I was wondering if anyone could point me in the direction of somewhere I could find some basic information regarding Bayesian statistics and how they relate to fingerprints?

thanks

Dan

[Gerald Clough]

Here are some I found when I was looking for information. The common thread has to do with combined probabilities in situations in which there is a probability of less than 1 of correlation between an evidence and guilt and other probabilities, such as the likelihood that the real actor is in a given population or some other measure of how likely it is that the person would or could commit the crime. The first article proposes the importance of populations when the actor can reasonably be placed in a limited population. A separate issue from the t-model stuff. It was my jumping off point.

This work, cited in a recent thread, illustrates some issues where Bayes Theorem can be applied, where he gets into error rates and limited populations.

http://www.henrytempleman.com/

I found the Wikipedia article to be a useful short presentation.

http://en.wikipedia.org/wiki/Bayes%27_theorem

And Bayes applied to reasoning error in courtroom evidence.

http://www.agenarisk.com/resources/prob ... ence.shtml

The parent link for the one above.

http://www.agenarisk.com/resources/prob ... ility.html

And

http://en.wikipedia.org/wiki/Prosecutor%27s_fallacy

Another on that topic.

http://www.colchsfc.ac.uk/maths/dna/discuss.htm

A paper on Bayesian Networks and legal arguments.

http://www.dcs.qmul.ac.uk/researchgp/sp ... allacy.pdf