[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Cause and Effect, a Response to Tony Cox
A response to Tony Cox's Comment: by
Joe Alvarez and Fritz Seiler
We repeat Tony's statement here as it was a
quite shredded by our line width counters.
He wrote:
Fritz Seiler and Joe Alvarez wrote:
> In order to claim a viable cause-effect relationship,
at least two conditions must be met: One, a causative link
must be established between the agent, its biological action,
and the health effect; and Two, an experimental correlation
must exist between agent exposure and health effect. <
Well, it depends what is being claimed, surely. If someone
is claiming the discovery of a scientific finding, to be added
to the body of accepted facts, and worthy of publication in
scientific journal of record, then I agree.
If, however, the claim is only to have discovered the
existence of a "viable" risk, worthy of further analysis and
possibly worthy of immediate avoiding actions, I would not
agree. The conditions for that are far less restrictive.
Suppose for example that we have identified a possible
mechanism but still await experimental confirmation. We
would then be in a state of uncertainty, i.e. at risk. The
causal mechanism is still "viable" and, absent any
experimental rebuttal, we would remain at risk. If the
effect is a very severe one, we might rationally modify our
behaviour to avoid the risk, in case the experimental results
turn out bad. We might also do this if the experimental results
showed some effect, but with lower confidence than the
conventional scientific yardstick of 95%.
Alternatively, suppose we have found an empirical correlation,
but we have not yet found (and obtained general acceptance
of) a theory to explain it. The same thing applies. The
correlation may be a spurious fluke and our worries
unnecessary (at 95% confidence, one in twenty is like that).
Or, it may have a real basis and we could reduce our risk
by exercising control over one of the correlated variables.
I would accept that our responses to such situations
should be proportionate to the risk that they present.
Thus, the severity of the effect and the degree of
uncertainty are both factors we should weigh up
before deciding to divert resources or make sacrifices
to avoid the risk. But there is nothing to say that our
response should reduce to zero at precisely the same
threshold as that which is used conventionally for an
entirely different purpose in scientific practice.
BTW, none of the above is addressed to the
EMF/cancer issue, which was Seiler & Alvarez's
main concern (on which I happen to agree with them),
but to the wider point of principle.
Tony Cox
Dear Tony,
Quite in general, we are in agreement with your wider point of
principle. We may not fully agree on the degree of response to
a particular suspicion, but that lies the realm of policy and risk
management.
It is obvious from the context of the EMF discussion that it
addresses an issue where suspicion is no longer the issue.
(Additional reference, pointed out to us by Dick Wilson is:
National Academy of Sciences, "Possible Health Effects of
Exposure to Residential Electric and Magnetic Fields".
Washington D.C. National Academy Press, 1996, $39.95)
The culprit EMF was tried and found innocent. ' Innocent',
mind you, rather than 'not guilty'.
Normally, the finding of 'not guilty' is not sufficient to allay
suspicion. Statistically, the 'not guilty' verdict means that
below a given confidence level, 1 - alpha, the null
hypothesis 'not guilty' cannot be rejected. It does,
however, also mean that, at a significance level alpha,
the null hypothesis may not be valid, and thus a residual
suspicion remains.
As we have pointed out before, risk management needs
to be flexible, in particular with respect to the confidence
levels. In several cases , we have advocated levels as low
as 60% or even 50%, if costs and attendant risks are low.
It all depends on the interpretation of the perceived
threat and the action contemplated.
For a general case, as you correctly point out, if one of the
two conditions for a cause-effect relationship is not met, the
situation is somewhat fluid, and different actions may be
indicated. In our opinion the action should be based on a
cost-risk-benefit analysis of all relevant alternatives.
As you agree, in the case of the EMF/cancer link, the
situation is different insofar as both conditions are not met.
Most importantly, any evidence for a physical-chemical/
biological mechanism of causation is lacking. To our way
of thinking, this exonerates EMF, i.e., it declares it to be
"innocent" which is quite a different situation from being
found "not guilty".
Nevertheless, EMF is subjected to continued suspicion on
the part of certain persons and thus the press. In effect
EMF is subjected to the media asking questions of the
misleading type "When are you going to stop causing
cancer?" ("Have you stopped beating your wife yet?")
We appreciate and and support the position that you
advocate, Tony, but would like to see the proviso added
that corresponding actions have to be selected and
justified by cost-risk-benefit analyses.
Joe Alvarez
Auxier & Associates, Inc.
10317 Technology Dr., Suite 1
Knoxville, TN 37932
jalvarez@auxier.com
Tel: 423-675-3669
FAX: 423-675-3677
************************
Fritz A. Seiler, Ph.D.
Principal
Sigma Five Associates
P.O. Box 14006
Albuquerque, NM 87191-4006
Tel. 505-323-7848
Fax. 505-293-3911
e-mail: faseiler@nmia.com
**************************