Friday, August 20, 2010

Bringing science to law

For more on DNA, see the second part of the article mentioned last time: New Scientist, 18 August 2010.

It confirms that the likelihood ratio is the best way of presenting conditional probability evidence. This will be familiar to anyone who uses Bayes' Theorem.

The simplicity of this approach is so stark that it can be mistaken for complexity. It asks, how much more is the evidence consistent with guilt than with innocence?

Difficulties with DNA evidence arise when scientists try to assign probabilities to each of these consistencies.

In my own humble little way I have offered lawyers
an easy introduction to Bayesian reasoning.

It is a paper that has long been in draft form, so I can revise and update it as developments occur.