Statistics

[huy.duong@ctsu.ox.ac.uk]

Dear anh Tuan and friends,

I think the medical world has been very slow in coming to term
with statistics. 

I suppose one reason is often statistics is not
something an individual can experience: no one ever has the mean
number of children; of thousands of gamblers who are either lucky
and win or unlucky and loose, there are only a few casino owners
who can see statistic working; if you are a smoker, you either die 
early or you don't, just like non-smokers either die early or don't.

Another reason is that people are used to comparing magnitudes of
something physical, but they are not used to comparing magnitudes
of probabilities; they tend to think of things as either can or can't
happen, but if both A and B can happen they don't often think
which is more likely to happen than which. 

These are general in the population.

For doctors, there is another reason: I think doctors 
don't like statistics becauses it challenges their authority. "I have
always treated my patients this way, and it works for me". "For me" 
is put in deliberately :-). Actually, I'm being unkind. But doctors
are a bit like the gods. And life is so important they can't trust it
to chance, they have to put it in the hands of the gods. But they go
wrong when they face something they *really* don't know about. At
these times if they follow their own judgement, they are relying on
the statistics from a sample of one doctor. What they should do is
admit that they don't know and rely on the statistics from the sample
of many doctors.

Huy