The Tiger That Isn’t: Seeing Through a World of Numbers
by Michael Blastland and Andrew Dilnot
Reviewed by Michael Hutton
Our lives are more and more dominated by an endless flood of numbers: the average cost of this, the increased risk of that, the number of people who have stopped, or started, doing something else. Yet this increase in data rarely seems to go hand in hand with an increase in understanding and it just seems so difficult to make sense of all the numbers. Does tagging reduce re-offending rates? Should men stop eating cured meats or women stick to one glass of wine a day? What, if any, is the long-term effect of speed cameras on accident rates?
The answers to questions like these are important if we are to make sensible private or public decisions. Michael Blastland and Andrew Dilnot, the creator and presenter respectively of that excellent Radio 4 programme, More or Less, have done something quite unusual: they have written a book about numbers that is lively, readable and entirely practical.
They begin with the assurance that we all know more than we think we do when it comes to making sense of numbers, simply as a result of our own experience. We don’t need to retreat into the “Lies, damned lies and statistics” position, nor see numbers as the deciding factor in any argument. A reasonably sceptical attitude and the habit of asking a few simple questions will usually lead us to the ‘take away’ message buried (often deliberately) in a set of figures. This is certainly not a text book but a book for the consumer of statistics, which is all of us. It is divided into a dozen clearly defined chapters, each dealing with a single topic such as chance, averages, targets, risk, comparisons and correlation, and illustrated with neat everyday examples and reminders of why we should approach even an apparently simple thing like an average with caution. Just remember that an average rainbow would be pure white and most people have more than the average number of feet – which makes them typical.
Perhaps the most valuable chapter is the one on risk, since it is the figures relating to this that cause the most uncertainty and anxiety. A statement like “risk up by 42 per cent” sounds scientific and authoritative but tells us nothing useful (42 per cent of what?) or helps us answer the questions we want to ask: “Does that mean me?” and “What should I do?” The straight- forward way to bring this back into line with personal experience is to use natural frequencies – so many people per 100 or 10,000 or one million. The underlying message of the book, one that every politician and journalist should have drilled into them, is that life is messy and complicated, and that looking for certainty in the numbers that are produced is a waste of time. Truly it has been said: “If you ask a question, statistics will tell you the answer. What they won’t tell you is whether you asked the right question.”
Snake Oil Science: the Truth about Complementary and Alternative Medicine
By R. Barker Bausell
Reviewed by Ray Ward
Bausell covers the rise of CAM, the history of placebos, impediments to valid inferences, why randomised placebo controls are necessary in CAM research, judging scientific evidence, personal research on acupuncture, how we know the placebo effect exists, a biochemical explanation for the effect, what trials and reviews reveal about CAM, and how CAM is hypothesised to work.
He begins by quoting Robert Park’s experience of seeing CAM advocates nodding in agreement despite giving differing views on the field’s most important issues: in a body which regards itself as embattled and besieged there can be no internal dissent. In a new area, poorly conducted research is the norm, and almost invariably produces false positive results. Patterns are often very difficult to distinguish from coincidence. CAM therapists do not value (and most, in Bausell’s experience, do not understand) the scientific process. There is the “file drawer” problem, familiar from other areas of paranormal research: negative results are less likely to be published, and the problem of attrition: subjects who feel they are not being helped tend to drop out.
The nature and quality of the publishing journal is particularly important in CAM research; a homeopathy or acupuncture journal is unlikely to publish a trial that suggests homeopathy or acupuncture doesn’t work. Small studies tend to produce distorted results. Good news is always preferred to bad, and there are always people whose beliefs are more important to them than whether or not they are correct.
Bausell relates how his mother-in-law’s experience with CAM treatments for pain demonstrates the post hoc ergo propter hoc fallacy, the mother of all superstitions: the fact that one thing follows another is not evidence of cause and effect. His mother-in-law attributed pain relief to the treatment, when in fact her pain pattern shows that it would have diminished anyway. Our old friend the principle of parsimony appears, and Occam’s Razor is vigorously plied.
Bausell summarises well over a hundred CAM trials and discusses all the well-known CAM therapies, and his conclusion is unequivocal: “CAM therapies are nothing more than cleverly packaged placebos.”
Submit content / news
For the printed magazine:
For the website or news columns:
Read the submission guidelines.