Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!asuvax!anasaz!edluucp
From: John Moore
Newsgroups: sci.electronics
Subject: Re:High Tension Wire Hazards
Message-ID: <4045@anasaz.UUCP>
Date: 19 Apr 91 10:10:58 GMT
Sender: root@anasaz.UUCP
Organization: Anasazi, Inc - Paris, France
Lines: 71

In article Message-ID: <112256.27482@timbuk.cray.com> dadams@cherry10.cray.com (David Adams) writes:

>Yesterday I heard a report on the radio about a study on the 
>Hazards of living near high tension wires.
>
>The study involved, the report pointed out, was not scientific!
>In order to perform a valid scientific study on the subject
>we would have to obtain a random sample of children, randomly
>assign half of them to live near high tension wires and the 
>other half away from high tension wires.  Then we follow the
>progress of these children over the course of many years and
>measure the proportions which develope cancer, etc.

This is not true. In order to perform an OPTIMAL experiment on the
subject, we might have to do as you say, but valid science is often
done with skimpy data.

>
>We cannot do this in a democratic society.  So what the researchers
>did was to find a group of people who had developed some disease.
>(I don't remember what disease was involved.)  Then they tried to

Childhood Leukemia - if you are referring to the Denver study.

>find a *similar* group of people who did not develope the disease.
>Then they looked back to the childhood histories of these people
>to see how many of each group lived near high tension wires.

Not high tension wires - simple backyard distribution wires - with
different levels of ELF magnetic fields.

>This is not a good scientific method.  It was pointed out, for example,
>that it may well be that more affluent people were willing to 
>volunteer for the study, and that fewer of these were likely to have
>lived near high tension wires.  When you set out to find a *similar*
>group, it is entirely impossible to consider all of the variables
>that might be involved.  This is one of the reasons that randomization
>is so critical.
 
>When I see so many people begin to believe conclusions that come 
>from bogus studies, it really makes me stop and think.  We need to 
>learn to be critical when we read, or evaluate a study like this one.
>This doesn't mean that High Tension Wire Hazards do not exist, but
>we cannot use the conclusions of experiments like these as evidence
>that they do.  

Self selection is always a risk in epidemiological studies. It does not
mean that the study is incorrect - it does mean that the results may
not be as conclusive as would be with the sort of experiment that you
admit cannot be run. It does not mean we should simply ignore the
studies or not try to run them. 

Modern medical science would be seriously crippled -where without these 
studies. I would point out that your criticism also applies to most 
smoking-->cancer/smoking-->heart_disease studies. Try to eliminate self
selection from a risk of smoking epidemiological study! Does that mean that 
we are foolish to believe them, or that we should not act upon them?

While the best statistical results come from a pre-designed, randomized,
multiple-blind study, we are not always in a position to do such a study.
Sometimes we have to use what data is available, with recognized corresponding
degradation in experimental accuracy. One of the most powerful challenges
in science is designing statistically valid experiments under these
conditions - and some people are amazingly good at these experimental
designs.

None of what I say above means that I take these studies uncritically. The
Denver study in particular was just barely statistically significant (like
a lot of cold-"fusion" studies were!). It may be wrong. It may also be
right. It is backed up by some other epidemiological studies (also 
perhaps flawed) and some experimental lab work (that is a lot more solid).