Last week-end, while visiting friends nearby, a copy of the book “bad science” by Ben Goldacre was dropped in my lap. Having read the occasional post on his weblog (http://www.badscience.net/), I had already planned to get it. So I started reading the book with rather high expectations. (This copy, if I did not misunderstand my friend, happens to be an unofficial “homeless” book. The idea is that they are passed along after you have read them. An idea I like!)
The premise of the book is that science (in particular medical science) is not difficult, but that there are certain well-explained and illustrated pitfalls to avoid when testing hypotheses. These range from the straightforward (placebo effects, prior expectation bias) to the slightly more complicated (statistical errors when interpreting data with many measured parameters). The book explains that good science is that in which the methods are clearly explained, so they may be tested and criticized. The very best (in medical science) are the Cochrane reviews; large meta-studies combining the results of many smaller research papers to identify less obvious aspects of medication and medical treatments (http://www.cochrane.org/).
The book starts out with simple examples of groups of people who set out to muddy the public understanding of such science by introducing and advocating pseudoscience, such as homeopaths, nutritionists and some alternative healers. The journalists also get their fair share of criticism, but in my opinion, much too late in the book. The methods of these groups are detailed by examining some examples of fake medicine they have been trying to push.
Thus, the book discusses simple but well-founded examples of medication and nutritionist claims that have not been shown to work (not for lack of effort) or have been shown not to work (and even pose a health risk in some cases). Starting out from “detox” sessions, to “ear candles”, “fish oil”, “vitamin supplements” and “homeopathic medicine”, he then moves on to debunking the claims of nutritionists (people making very specific nutritional claims beyond eating plenty of fruit and vegetables, such as that fish oils improve concentration and IQ, often wildly extrapolated from petri-dish lab tests). Afterwards, he moves on to indicate problems in the pharmaceutical research methods, before moving on to the problems of the media (and their cherry-picking of controversial but inaccurate papers out of a plethora of non-controversial results). In particular he laments the role of the media in the recent vaccination scare, causing a host of parents to refuse their children to be vaccinated and thus a dramatic resurgence (in the UK) of mumps, measles and rubella cases.
In terms of my layman criticism, the book does focus (understandably) on the UK, which is a minor (but perhaps unavoidable) distraction. The book offers good references to its claims, and plenty of examples to illustrate the failings of the research which did show a positive correlation. However, I would like to see the references more closely associated with particular statements, so that these are more easily checked (for example, one could refer to references in the same way as is done in “the god delusion” by Richard Dawkins). Ben Goldacre also has bones to pick with some “experts”, in particular Gillian McKeith and Patrick Holford, but many others are discussed as well. This is funny most of the time, but sometimes comes over as a little petty (which, I suspect, is hard to avoid when talking about your own experiences). Also, a summarizing chapter at the end would not go amiss.
In all, he makes a valid point in his book for the scientist and non-scientist alike, and I wish that many people will read at least the first few chapters. For me, it was striking that the fish oils do not seem to work, and that vitamin supplements (besides multivitamins in some cases) likely do more harm than good. It remains to be seen if the book is as eagerly read by my wife as by me. I may have let slip a few too many negative quotes from the book, causing her to believe the book is only about negative correlations. Sorry about that!