Having received this book yesterday, I have only had the opportunity to give it a very quick skim through and to cherry-pick a few items.
The first thing that struck me is that it is quite readable. I realise that that probably comes over as damning with faint praise, but for a book that touches on such arcane topics as probability and statistics it is actually an accolade of the highest order.
The second thing is that, I can now actually understand Bayesian statistics — sort of. I shall have to read it again to make sure it goes in and stays in. Don’t forget: I’m one of those people who left school with a brilliant facility for logarithms, a skill that is totally useless these days as far as I can tell, and the ability to recognise an equilateral triangle.
The third thing is that there is a great chapter on communicating accurately using numbers rather than expressions that can be interpreted subjectively. The example the author gives is that of the Bay of Pigs fiasco. Kennedy asked his generals what the success of an invasion was likely to be. They said there was a fair chance. He interpreted that as “reasonably good”. They meant it as almost certain to fail — they’d estimated a 30% likelihood of failure. Because an expression like “fair chance” has no objective meaning, it was open to being interpreted subjectively.
This reminded me of the time when a colleague took me to task for always wanting to measure students’ success in hard numbers (such a 25% improvement in someone’s test score. or a test score of 89%). He said I ought to be using warm fuzzies instead. Well, I’m not averse to such measures, but feeling in the mood for a bit of devilry I answered that I would take on board what he’d suggested and would aim for a 43% increase in my use of warm fuzzies.
Anyway, do check out this book. It’s in the Pelican range, which I’ve always associated with quality. I intend to bring you a fuller review in due course. Here is the Table of Contents, courtesy of Amazon:
