On 8 DFómh 2009, at 17:14, Maconochie, Ian K wrote:
This is not necessarily true - see Prof Meadows, paediatrician in Uk
on the probablity of serial cot deaths! There may be predilections to
their occurence which negate the routine estimation of risk.
This is exactly the fallacy of Prof Meadows - he argued that cot
deaths were so rare that the chances of a second one happening to the
same family were utterly remote, so some factor other than chance had
to be at work. His argument is flawed for three reasons:
1. The fact that you have had a cot death in the house does not alter
your chances of having another one. If you have a son, your chances of
your second child being a son are 50%, the same as they were when you
had no children. More formally, the events are independent, so the
probability of one event is uninfluenced by the occurrence of the
other (but see 3 below)
2. Prof meadows argued that two cot deaths are so unlikely that they
cannot simply be a coincidence, but actually one cot death is so
unlikely that, were we to follow his argument, we would declare foul
play each time a cot death happened. The fact that something is
unlikely does not prevent it from happening.
3. He ignored the possibility of a common underlying mechanism. Many
conditions cluster in families due to shared genes and environment.
For this reason, the argument that the events are independent is hard
to sustain. And the effect of shared mechanism is to make multiple
events more likely, not less.
The whole case was a dreadful example of the utter failure of the
legal system to understand probabilistic reasoning.
Ronan Conroy
=================================
[log in to unmask]<mailto:[log in to unmask]>
Royal College of Surgeons in Ireland
Epidemiology Department,
Beaux Lane House, Dublin 2, Ireland
+353 (0)1 402 2431
+353 (0)87 799 97 95
+353 (0)1 402 2764 (Fax - remember them?)
http://rcsi.academia.edu/RonanConroy
P Before printing, think about the environment
|