Tags

Charles Seife’s Proofiness is an accessible and entertaining look at the many ways numbers can be used (more to the point, abused) in order to win an argument.  Seife spends the early part of the book outlining his typology for numerical abuses.  For instance, “disestimation” is the act of taking a number too literally, understating or ignoring the uncertainty that surrounds it.  This is often done when some kind of data is presented without taking into account that its calculation contains a great deal of measurement error (think of polling or the US Census).  Seife also shows how visualization can be used to manipulate the meaning of data–what he terms “apple-polishing”.  A classic example is portraying longitudinal data in a graph where the y-axis is truncated instead of starting at zero.  Even a small change over time will be magnified by such a presentation, as you can see from the two graphs below.

Apple-polished Y-axis

Correct Y-axis

The book is packed with great examples.  However, Seife spends a bit too much time on some cases.  More variety would have made the book better.  Seife also tends to focus on the intentional manipulation of data while ignoring the unintentional instances.  There is no doubt that people use many of the tricks he describes to bend data to their advantage, but often times misleading data is the result of people simply making bad calculations rather than purposeful manipulation.  Additionally, Seife’s suggestion as to how to combat proofiness, mathematical sophistication, doesn’t seem capable of solving the problem on its own.  While I agree that the public could benefit from a more robust understanding of numbers and their manipulation, Seife basically ignores the issue of perceptual bias.  Even the most sophisticated consumers of data are subject to fundamental perceptual biases.  Given that we are “predictably irrational”, to quote Dan Ariely, any solution must also take into account that we are hardwired in many ways to be manipulated by proofiness.

The book is not for deep subject matter experts in mathematics or statistics, but it is a fantastic primer for the lay person.