Subject: ,,EMF,,(S) Industry-Sponsored Statistical Nonsense
Date: Fri, 4 Jul 1997 113436 -0500 (CDT)
From: creuss@hitline.ch (Christoph Reuss)
To: Multiple recipients of list <emf-l@mail.llion.org>
--------------------------------------------------
Associated Press reported on the NCI study:
> The study was based on 638 children under age 15 with acute lymphoblastic
> leukemia. The researchers checked the youngsters' exposure to magnetic
fields
> generated by nearby power lines and compared it with that of 620 healthy
> children.
I didn't read the original study, but as I understand the above description,
the statistical design of the NCI study is wrong ("inverted"), and
*DESIGNED*
TO FIND *NO* LINK between EMF exposure and leukemia !!! Here's why:
(1)
First, let's assume that all U.S. children have the *same* MF exposure, and
that this exposure *causes* an individual leukemia probability of p=0.02,
i.e. 2% will get leukemia under age 15. Now we can randomly pick 638
children
with leukemia from those 2%, and randomly pick 620 healthy children from the
other 98%, and compare the MF exposure of the two samples. And
--SURPRISE!!--
both samples will have the SAME MF exposure !!!
Now, you'll say: Stop, the assumption that all children have the same
exposure
was wrong. Okay okay, this was only to lead the NCI approach ad
absurdum....
(2)
So now, let's assume that 10% of all kids live under powerlines, having an
individual leukemia probability of p=0.03, and the other 90% don't live
under
powerlines, having p=0.01 (from other influences). For a population of 100
million kids under 15, this gives the following numbers of leukemia cases:
Living under Gets # of kids
Powerlines Leukemia
--------------------------------------------------
yes yes 3.0 million
yes no 7.0 million
no yes 0.9 million
no no 89.1 million
Thus, from the 3.9M kids with leukemia, MORE kids DO NOT live under
powerlines
(3 million) than do (0.9 million) !!!! So it will be pretty easy to select
638 children with leukemia who have a medium powerline exposure EQUALLY LOW
as
the exposure of 620 healthy kids !! -- ALTHOUGH the individual leukemia is
THREE TIMES HIGHER under powerlines in this model !!
Of course, the basic idea of the study's model is that among the leukemia
kids, a higher *percentage* lives under powerlines than among the healthy
kids. But if they simply pick ill and healthy kids, their selection is all
too prone to be biased, especially as they consider MFs instead of EFs
(powerlines!!) and disregard other leukemia factors such as nuclear
radiation,
heavy metals, smoking etc.!! A big problem is that the *researchers* are
selecting those 638+620 kids, not the 'fate' !!!
(3)
The same applies for the more probable Gaussian distribution of powerlines
exposure and thus p values.
Bottom line: The NCI study WAS *DESIGNED* TO FIND *NO* LINK !!!
================================================================
It's too easy to pick a leukemia sample and a healthy sample such that both
groups have roughly EQUAL distributions of powerlines exposure !!!
The correct design would be to have various groups of kids living under
otherwise equal conditions and expose them to different levels of EMF
exposure, and then correlate the number of leukemia cases in each group with
the EMF exposure of that group -- and this for various types of EMFs.
You'll
say "Crazy or what?! You won't find volunteers for that!!" But after all,
that's exactly what happens in the real world -- JUST NOT UNDER CONTROLLED
CONDITIONS..... because there is no political will for a serious study....
BTW, it's not unusual that industry-sponsored studies have a wrong design
suitable to find what they want to find. For instance, the ADA's position
that Amalgam is *not* associated with brain disorders (like Alzheimer's
Disease) is MAINLY BASED ON the ridiculous "Nun Study", which examined 129
old and largely teethless nuns in a Californian convent and found no link
between Amalgam and brain disorders, because it was STATISTICAL NONSENSE:
- The sample size (N=129) was way too small to find a significant link in
the
range of a few %.
- Nuns *hardly* chew gums, drink Coke/soda drinks/alcohol, smoke and use
VDTs
-- all important factors for mercury leakage from the fillings.
- At the time of the study, most teeth once filled with amalgam were no
longer
in the nun's mouths (especially since mercury causes tooth loss), i.e.
nuns
with *higher* amalgam exposure had *less* fillings at the time of the
study!
Regards,
Chris
creuss@hitline.ch
_______________________________________
NCI = Nonsense Co-funded by Industry ??
Archive provided courtesy of WaveGuide, http://www.wave-guide.org
Reprinted with permission of Roy Beavers, http://www.feb.se/EMF-L/EMF-L.html