At 11:16 16/06/03 +0100, Ted Harding wrote in part):
>To some extent, there is evidence in the criminal law system as to
>how "beyond reasonable doubt" may be interpreted probabilistically.
>.... The clearest example I know is that of driving a motor vehicle
>with excess alcohol in the blood. ... When a blood sample is
>analysed, 3 SDs are subtracted from the result, and this is the
>figure which is quoted in court ... Anyway, this establishes that
>according to Home Office procedures a P-value of 0.001 or less is "proof
>beyond reasonable doubt" for this crime.
Interesting. I often wondered what degree of precision was required.
However, as you go on to say, there are very few 'numerical crimes' and I
really don't think that the degree of precision you are talking about is
anywhere near the ballpark of what is realistic for non-numerical crimes ....
.... if 'beyond reasonable doubt' meant that I had to believe that there
was only a 1 in 1000 chance or less that the argument presented in defence
was true, then (unless lawyers and expert witnesses were going way beyond
what is professionally/legally acceptable - i.e. simply, and obviously,
lying!), I'd be very surprised if I could ever find anyone gulity, if they
had pleaded 'not guilty' and had had some sort of defence (i.e. a
suggestion of possible innocence) presented professionally. If that
defence were even remotely 'credible' (no matter how unlikely), I would
think the risk of a 'Type I Error' (in relation to the null hypothesis of
innocence) would be very much (probably at least an an order of magnitude)
greater than 0.001
Kind Regards,
John
----------------------------------------------------------------
Dr John Whittington, Voice: +44 (0) 1296 730225
Mediscience Services Fax: +44 (0) 1296 738893
Twyford Manor, Twyford, E-mail: [log in to unmask]
Buckingham MK18 4EL, UK [log in to unmask]
----------------------------------------------------------------
******************************************************
Please note that if you press the 'Reply' button your
message will go only to the sender of this message.
If you want to reply to the whole list, use your mailer's
'Reply-to-All' button to send your message automatically
to [log in to unmask]
*******************************************************
|