Panic Nation - Chapter 2: The misuse of numbers - Part 3
Continued from 21.6.2006
Stanley Feldman and Vincent Marks
Leaving aside the fact that this ideal weight is entirely subjective,
it is self-evident that those who are only a few ounces, or even a few
pounds, over the 'ideal' weight run a relatively small risk of any
adverse effects on their health; it is those who are much heavier who
are at risk of their lives being shortened by obesity. But the statistic
gives us no idea of how many children are slightly overweight and how
many are seriously overweight. What is does do, on the other hand, is
serve to scare us and to sell newspapers.
Circumstantial evidence
Many of the scare stories that appear in the media start with an
epidemiological study. These compare the rate, occurrence or coincidence
of two events. Some of these studies are very good, with great care
being taken to prevent bias and to limit the effect of outside factors.
Nevertheless, they all end up trying to demonstrate an association
between a potentially causative factor and an effect. The problem with
these sorts of studies, however, is that by itself the association
between the events is circumstantial and says nothing about their causal
relationship to one another.
Take the example given by Darrel Huff in his book How to live with
statistics the greatest number of suicides in the UK occur in June; June
is also the most popular month for marriages. Does this mean that these
two events are related?
Bias
Many studies depend upon the reports of individuals. The results of
such studies are often subject to bias, particularly when they are based
on telephone or postal questionnaires. If a study trying to find out if
a particular detergent caused skin irritation was carried out by post,
it is likely that someone not troubled by skin irritation would throw
the letter in the bin, whereas someone with skin irritation would reply.
Unless there was a very high return rate, the results would be
subject to bias and be meaningless. A recent postal questionnaire sent
out by the Home Office reported that 1 in 20 women had been raped. This
result in intuitively unlikely, and the nature of the questionnaire was
bound to give the result it did. The chances of happily married women
who have never been raped replying to the questionnaire were very small.
In addition, if the definition of rape was not well defined, it is
not beyond the bounds of possibility that some women, perhaps those in
an unhappy relationship, will have answered the question 'Do you believe
you have been raped?' rather than 'Have you been raped?' which is very
different. The disturbing aspect of this study was not just the waste of
taxpayers' money, but the fact that the results are likely to be used as
the basis for legislation.
Telephone studies are just as likely to be subject to bias, as the
respondent invariably tries to give the questioner answer that they
think he or she wants to hear.
Unbalancing the equation
Benjamin, age five, casually remarked, 'God eats a lot of fish.' The
background to his improbable assertion was his mother's remark that fish
is good for your brain. His tendentious reasoning was that 'God must be
very clever because he can hear what we are saying no matter what
language we use', it follows therefore that 'he must eat a lot of fish'.
A byline in the Sunday Times (August 2004) indicated that salt causes
intellectual failure.
It appears an investigation in Boston showed, in a trail of 2,500
people, that high blood pressure was correlated with minor strokes, and
minor strokes could cause intellectual degeneration, which is not
unreasonable. The author, having believing that salt can cause high
blood pressure wrote up the story to implicate salt as the cause of
intellectual degeneration.
This piece of tendentious reporting ignores the evidence that the
Japanese, who eat twice as much salt as an American on average, are no
less clever and live longer lives.
Then there are numbers apparently picked out of the air but repeated
so many times that they develop an authority of their own.
A letter from Transport for London says that 20,000 people die each
year from atmospheric pollution in London. In fact, pollution in London
is now lower than it has been for centuries. It is possible that a few
with severe respiratory diseases are made worse by diesel fumes from
lorries and buses but these are patients on the edge of breathing
failure who would not be made better even if they lived in a plastic
bubble fed with filtered air.
These figures are frankly false but because they go unchallenged
become accepted as fact. With all the numbers that are bandied about to
show this disease or that food is killing us we should remembers that we
are living longer, healthier lives and that 'life itself is a fatal
disease' if we do not die of one thing it will increase the death rate
from an alternative.
Stanley Feldman is Editor Journal Anaesthetic Pharmacology Review,
contributor Encyclopedia Britannica, Vincent Marks is a former president
of the Association of Clinical Biochemists and erstwhile Vice President
of the Royal College of Pathologists. He is one of England's best-known
nutritionists. |