----- Original Message -----
From: "J. van Baardwijk" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Friday, July 06, 2001 6:23 AM
Subject: Re: Earth on Edge
> >The pro side has offered a reply to that - it's in the data. The
"cluster"
> >of cases you're reporting is not really there. The cancer rates are not
> >different from the rates elsewhere.
>
> If you re-read the above quotes, you'll notice something strange: a
> mysterious disappearance.
>
No, it is not a mysterious disappearance. Its an understanding of random
chance. Let me give you a parallel example. People reported a rate of 3x
the national average for certain type of cancers in children living near big
power lines. Various spots around the nation reported this phenomenon.
This was considered, by some, as proof that the power lines cause cancer.
Indeed, there was one paper out of Duke (thanks for the reference Zimmy)
that supported this conclusion.
So what happened, and how did the scientists "remove" the cancers "caused"
by the power lines? It turns out, they didn't have to remove them, they
just used the proper statistical techniques in analyzing them.
Let me try to explain how. Let us say that one is looking at 200 different
types of rare cancer. On average, one expects only 2 cancers per year for a
sample size of 100,000. But, looking through these, 3x the national
average is found for one cancer. Another study, with a different 100,000
finds a similar result, but only for a different rare cancer. After 20 such
studies, the answer seems to be conclusive: there is a correlation. Those
power lines caused cancer.
But, if you look, there is a hole in the logic. In each study, its a
different rare cancer that is observed with a higher rate. The cancer that
has 6 observed incidents in study 1 has zero in study 2. The average for
the 10 studies shows percentile variations that are a bit less than a third
in any given study.
Indeed, on average, one expects to one of these 200 cancers to exhibit 3x
the normal rate, simply from random chance. One also expects to see 0
cancers instead of 2 in about 14% of these cases. When the APS did a
rigorous study, they found that the same techniques that were use to show
that the power lines caused some rare cancers could be used to show that
they prevented others. And, the cancers that were caused in study 1, were
prevented in study 2. The logical conclusion is that neither happens, and
it is just random chance.
Unfortunately, reality is often counter-intuitive. Further, chance unnerves
many people: they need to seek a cause. That's true for a lot of educated
as well as uneducated folks. So, people often jump to conclusions that are
not really justified by the data.
> If medical records show a significant increase (say, three times the
> national average), how can those cancer cases suddenly have disappeared
> when a study is done to determine the cause? Magic?
>
No, then cancers don't go away. When the study is expanded to include other
people, the rate for some subsets are below average and the rate for some
are above. Overall, the rate is within statistical uncertainty of the true
value.
> NOTE: this scenario is not linked to any specific nuclear plant; it's a
> scenario, a hypothetical situation where medical records show a large
> increase in cancer cases. The cause is unknown, but some people consider a
> nearby nuclear plant to be a likely suspect.
>
Part of the problem with that, is some people always consider nuclear power
the first suspect. Even after being shown that the increase in exposure is
only from 200 to 200.01 mrem per year, and that up in Scotland (where the
background rate is 500 mrem/year) no such increase in cancer is noted.
Let me give a parallel. Let us assume that a windmill farm produces an
average noise in a nearby city of 5 dB. It is known that sound can cause
deafness: exposure to 200 dB can quickly damage the hearing. There is a
higher than average incidence of deafness in this city. The obvious
conclusion: blame the windmills.
Would you consider this valid, and if not, why not?
Dan M.
>
> Jeroen
>
> _________________________________________________________________________
> Wonderful World of Brin-L Website: http://go.to/brin-l
>
>
