At 10:03 10/11/04 +0000, Paul Spicker wrote (in part):
>a) Ray Thomas writes: "It is well established in Law that Statistics is
>unreliable, deceptive and misleading." Of course it is, and it should be.
>In criminal law the focus falls on the possibility of convicting an
>individual, and probabilistic arguments are intrinsically unjust. Do we
>really want people to be convicted on the basis that someone of their age,
>nationality and social class probably did it?
No, we obviously would not want that, but it represents an extreme and
(hopefully) unrealistic possibility. However, even though it can be said
that "probabilistic arguments are intrinsically unjust", I think it is
naive to believe that the great majority of convictions are anything BUT
probabilistic. Undisputed forensic evidence that an individual was present
at a murder scene, that his fingerprints were on the murder weapon and that
the victims blood was on his person, together with a suggested motive,
would, I imagine, generally be more than enough to secure a conviction, yet
it is apparent that none of that evidence represents 'certain proof' of
guilt; I am sure that any of us could postulate a scenario in which this
individual was innocent, despite all that evidence. We live in an
essentially probabilistic world, in which there are surprising few true
'certainties'. How can we even think of the concept of 'certainty' in
court cases when a jury takes time, sometimes a lot of time, to reach a
verdict, which sometimes is not even unanimous?
The fact surely is that, whether in a court, the political arena or
wherever, the vast majority of decisions are in one sense or another
probabilistic - perhaps 'intrinsically unjust', but nevertheless
true. Indeed, if there were certainty, then the question of 'a decision'
would not really arise, since the 'answer' would be self-apparent. The
distinction we are discussing really seems to relate to different attitudes
to probabilistic matters when they arise from 'formal' statistics and
Statisticians, as compared with the far more nebulous probabilistic
arguments/decisions made by juries, politicians, doctors or whoever every
single day. At least the formal approach has the merit that the extent of
the lack of certainty can be quantified - yet the public seem so often to
be more critical of this than of the 'gut reaction'/ 'experience'/ whatever
approach to probabilistic issues.
>What the three issues seem to me to have in common is that the writers are
>claiming a much greater degree of certainty than social statistics can
>possibly support.
That is undoubtedly often the case. However, those statistics will very
often be the 'best available estimates' and, as such, probably a better
basis for handling probabilistic situations than 'gut feelings'
etc. Anyone who challenges that view has surely got to offer a more
reliable sort of (imperfect) information than is available from those
social statistics - otherwise their criticism is non-constructive.
>The argument that only experts can debate statistics is particularly
>dangerous - I've heard this, for example, from Eysenck on IQ. If we start
>to claim that social statistics are solid, positive proof of issues, it
>seems to me that we remove the possibility of criticising the use of
>statistics by people with opposing political views, and undermine much of
>the purpose of Radical Statistics as a group.
I cannot disagree with any of that. However, I also think it is equally
potentially dangerous for people to dismiss statistics simply because they
are statistics, and to substitute arguments based upon much more nebulous,
unquantified and untestable approaches. For example, goodness knows how
long it took for the (statistically established) health dangers of smoking
to become universally accepted.
... just my thoughts on the matter!
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]
*******************************************************
|