Damned Lies and Statistics
Untangling Numbers from the Media, Politicians, and Activistsby
reviewed by Mira de Vries
The indelicate title of this book is taken from the well-known aphorism “There are lies, damned lies, and statistics” attributed to either Mark Twain or Benjamin Disraeli. It nearly caused me to not read it, unjustly, because the language between the covers is perfectly gentlemanly.
Most people, Best tells us, are innumerate. This means that they don’t readily spot implausible numbers, and all big numbers are more or less the same to them, whether a million or a billion. Not only the people who hear statistics are often innumerate, but also the people who report them, and not infrequently even the people who generate them.
Statistics can be wildly off course for many reasons. They may originate in a guess. But even when they originate in research, there are many factors that can influence their accuracy. The researcher may have interviewed people using leading questions. The subject of the research may be poorly defined. The method of measuring it may be flawed. The sample on which the research is based may not be representative. Or perhaps a comparison was made between two entities that aren’t comparable. Numbers may have been mangled by someone who quoted them, such as a reporter. Or the condition described mutates into something else during the retelling. The finesses of complex statistics may be overlooked. Basing new statistics on older ones may result in a chain of bad statistics. And, unavoidably, they are influenced by the interests of the party who compiles them.
The author does not mention the statistics used in modern medical mega-trials, but everything he says about statistics in general applies to those as well.
He warns us not to be naïve or awestruck by statistics, but not to be cynical either. Statistics, he says, are a valid and useful tool. Not all statistics are bad statistics. So we shouldn’t reject all statistics off the bat, just be critical. Of course we should always be critical about everything anyway, he concedes, not just about statistics. I’m afraid I’m not going to take Best’s advice. He provides fine questions to ask when examining statistics, but who will answer them? So I am going to join his cynical group, and remain suspicious of all statistics.
Best’s writing style is refreshingly uncomplicated. This little gem of a book is suitable for a broad audience, including the less sophisticated reader, readers for whom English is a second language, and people who, like me, are innumerate.
Best has since (2004) published a sequel to this book, called More Damned Lies and Statistics.