I recently received a copy of Numbersense by Kaiser Fung for review. Fung is the author of a blog I have a lot of respect for : Junk Charts. The current post at the top of Junk Charts is about spider charts, and I knew before reading it that Kaiser would be explaining, kindly, why these are nonsense (I would add, on the topic, that one of the real problems with these graphics is that there is an implied meaning for the area of the chart (in terms of the polygon) and that area is not independent of the order of dimensions around the centre of the diagram).
I really like the idea of the concept of NUMBERSENSE. It is what it sounds like - a set of intuitive behaviours, practices, etc. that amount to common sense for data analysts.
I also really like one of Fung's core opinions regarding 'big data' - that more data is an invitation for more and more analyses which result in an explosion of interpretations and a multiplicity of incorrect conclusions.
The book is structured essentially around a small number of case studies, or studies of characteristic areas where numerical, statistical and probabilistic analytics are employed (often in areas that we unknowingly interact with on a daily basis in our regular lives). This structure is, in my mind, part of the weakness of the book. While reference is made to NUMBERSENSE at various stages, the examples highlight more the general theme of 'don't trust everything you read in the papers' rather than a specific structured thesis. For example, the book could have enumerated, illustrated and re-enforced a small set of common misconceptions regarding statistical inference; or it could have catalogued a small number of misleading sampling scenarios. Rather, it stays more at the case study story telling level.
It concludes with a rather odd chapter detailing the trials and tribulations of Kaiser as he works through some technical details while analysing some data. This fails to provide a clean summary or visualization of the big picture.
Given the quality of communication on his blog, my expectations were high for this book, but I came away feeling that it was perhaps poorly edited and not as impactful as the topic could have been. I already do, and will continue to, evangelize NUMBERSENSE to my team, but would like to do so with a more potent reference.
And on an unrelated note - the title of the book strongly reminded me of that most challenging game show 'Numberwang' which I include here for your enjoyment: