RogerBW's Blog

2014; a collection of Ben Goldacre's short writing, mostly for the Guardian.

If like me you've been reading Bad Science since the early days, much of this material will be familiar. It's loosely ordered by subject matter rather than chronologically (sections include "Statistics", "Epidemiology", "Evidence-Based Policy", "Bad Journalism", "Libel", etc.), and even within sections some pieces refer to other earlier ones that are later in the book. This makes reading straight through more effort than it really needs to be, since Goldacre obviously enough tends to say the same sort of thing about the same sort of subject, and they tend to come in clusters.

On the other hand I don't think that's the way this book is best read; one would do better to dip into it every now and then and read a random short piece.

While there's often a sense of frustration in these columns (why, one imagines Goldacre thinking, don't other people complain about obviously dodgy claims, or at least why isn't this reported) there isn't the consistent sense of anger that was in Bad Pharma. This is much more in the tone of Bad Science, which of course was rewritten from the Guardian columns. Those originals aren't repeated here, and some other material is deliberately omitted (it looks as if Goldacre's next book will be on randomised controlled trials in education).

"Bad Science" would have been a better title for this book than the one it got: most of the time there isn't room in a few hundred words to go into the more complicated bit, so instead there's a simple statement of whatever claims are being made, then a quick explanation of why they're blatant rubbish. Some of the saddest material is in the complaints sent about the columns: obviously one wouldn't expect fraudsters to admit that they were being fraudulent, but they appear not even to understand the nature of the objection, and simply repeat their original claims as though shouting louder made something more likely to be accepted as true.

Which all too often it does.

There's other material too: the pieces on care.data (both the original boosting one and the later "OK, still a good idea but not like this" one), some early writing on the treatment of drug addicts, and an introduction to the Romney, Hythe and Dymchurch Railway.

If you want the contents of badscience.net in dead tree form, or you're a Goldacre completist, or you want to see occasional notes from years after the original article confirming that the promised paper was indeed never published, then this book is recommended. If you've read most of it before, it's probably not worth the new price.

Even so, I actually prefer Goldacre's short writing to his longer works on the evidence so far; alas, he's unlikely to return to writing these columns, as he's working on two more books and apparently enjoys developing an argument at greater length.