Wednesday, April 19, 2017

Review - Painting by Numbers by Jason Makansi


Short review: Numbers are used for all kinds of purposes, and this book seeks to make it possible for readers to separate the numerical wheat from the chaff.

Haiku
People can be fooled
By numerical models
Think skeptically

Disclosure: I received this book as a Review Copy. Some people think this may bias a reviewer so I am making sure to put this information up front. I don't think it biases my reviews, but I'll let others be the judge of that.

Full review: Subtitled How to sharpen your BS detector and smoke out all the "experts", Jason Makansi's book Painting by Numbers isn't so much about numbers, but rather about how to understand and evaluate the use of numbers in the media, in business, and everywhere else models, charts, graphs, and other statistical representations are used in support of a particular position or projection. Despite being a book about "numbers", there are very few actual numbers in it, because the point of the book is not how to conduct analysis and create mathematical models, but it instead directed at providing the tools that an ordinary non-technically proficient person needs to be able to assess, at least to some extent, the reliability of the myriad of statistical models they are presented with in everyday life.

Painting by Numbers is divided into two broad sections. In the first, Makansi lays out the "Twelve Commandments of a Numerical Skeptic", taking twelve chapters to discuss each commandment in turn. Each chapter outlines one of these "commandments", seeking to arm the reader with a tool capable of picking apart models and presentations that rely upon the use of numerical information. Makansi presents these in an approachable manner, allowing those who are not versed in the intricacies of mathematical analysis to gain an appreciation for the limitations of such methods, and identify when such limitations are present. The commandments also, to a certain extent, allow a skeptic to assess just how critical those limitations are, and how severely they impact the reliability of the model being used. This is not to say that reading this section will make someone capable of doing a complete assessment themselves - that would be well beyond the scope and intent of the book - but rather that they can make a rough evaluation that will serve to separate the wheat from the chaff of numerical models.

The second section of the book is titled "Putting Your BS Detector to Work" and contains ten examples of numerical models to which the author applies the commandments outlined in the first section. Each of the examples is drawn from the real world, and the author states that he picked them and then evaluated them, in order to show how the "twelve commandment" system is applied rather than picking examples that were specifically chosen to make a particular point. This section is interesting, but not quite as effective as the first section. This is, in part, because the examples provided are only sketched out to the extent needed to show how the twelve commandments apply, which means that they feel a little bit disjointed and lacking in context. I would have preferred to see the entire item being discussed (although, to be fair, that would have made the book considerably longer), and then seen an exploration of how the twelve commandments applied to it, as this would have fleshed them out more completely. In a book dedicated to providing readers with the tools to evaluate the context of numerical models, the lack of context for many of the examples given seems odd. The other thing that sticks out about this section is that every example provided is riddled with problems. As I said before, the author chose these examples blind, in an effort to essentially "play fair", but one could come away with the impression that there are no numerical models that are worthwhile. This may be an impression that the author intended, but if so, he doesn't say so.

Despite those minor quibbles, Painting by Numbers remains a useful and informative work, aimed squarely at arming the non-numerically inclined with the tools necessary for them to deal with information provided via numerical models and their close companions, graphs and diagrams. This book is not a technical guide to numerical analysis, and it is not intended to be. The collection of "commandments" cover a wide range of issues that crop up in numerical models, and are presented in a manner that those who are not particularly comfortable with numerical analysis can both understand and apply them. The examples are, in general, easy to follow, and provide a broad spectrum of illustrations of the "commandments" in action. For those looking for a more in depth exploration of numerical analysis, Painting by Numbers comes equipped with a list of fifteen recommended titles on the subject.

At just over 150 pages, Painting by Numbers is exactly what it sets out to be: A clear and concise primer on how to evaluate the numerical data that bombards us on a daily basis. Although the tools provided in its pages are very basic, that should be considered a feature and not a bug. This book is not aimed at those who are already skilled at understanding numerical analysis, but instead seeks to equip those who do not have such a background to they can look at what they find in newspapers, on television, and in the board room with a properly skeptical, and hopefully, reasonably informed eye. For anyone looking for a guide to becoming a skeptic concerning numbers and their meaning, then this book would be an excellent resource.

Jason Makansi     Book Reviews A-Z     Home

No comments:

Post a Comment