Conquering Statistics: : Numbers without the Crunch

Citation

Carver, R.H. (1999), "Conquering Statistics: : Numbers without the Crunch", Journal of Consumer Marketing, Vol. 16 No. 1, pp. 101-104. https://doi.org/10.1108/jcm.1999.16.1.101.4

I have killed more than one cocktail party conversation with the revelation that I teach statistics for a living. Clearly, the victims of introductory statistics courses are well represented in the suburbs. Some of them might find Weaver’s Conquering Statistics an amiable introduction to this much‐maligned discipline.

The goal of the book is to present “the central concepts of statistics and the ways in which commerce and industry have benefited from the use of these tools” ‐ with a substantial dose of humor and pathos for the reader. Despite some serious flaws, it does basically that. Think of it as a short package tour of the capitals of Europe, with a guide who is a frustrated (but poor) comedian. At the end of the trip, you might have a sense of a place worth revisiting. For the marketing professional lacking a grounding in statistics, this book may be a good read.

Although not a textbook, the book follows the normal flow of a standard introductory statistics text. After an engaging bit of history, Weaver leads us through a clear presentation of the ideas of sampling, elementary probability, variables, normal distributions, and several techniques of statistical inference. At every turn, he mixes a non‐formulaic, intuitive style with light digressions and silly‐sounding but often useful examples. In fact, the balance between statistical content and intended humor is decidedly in favor of the latter.

Nonetheless, the chief virtue of this book is Weaver’s clear‐headed and clearly worded explanations of statistical concepts. Although the reader will not be able to do statistical analysis on the strength of this book alone, a committed reader will be a better consumer of such analysis, able to comprehend statistical reasoning and to critique its misuse. For example, in a discussion of the scope and limitations of probabilistic forecasting, Weaver writes “… its predictive power depends on focusing on a group of events instead of a single event. Indeed … while individual events such as the exact date when a person dies cannot be predicted with any certainty, one can approximate with some certainty the average age at which an individual may perish based on an analysis of the entire population” (p. 17). This concisely presents the essence of a crucial and oft‐misunderstood concept.

Sometimes, the combination of humor and the fanciful nicely conveys a kernel of meaning. Consider a particular type of estimate called a confidence interval, which is a prediction with a precisely built‐in fudge factor. A forecast of “two to four inches of snow” is an example. For many undergraduates, a bothersome feature of these intervals is that we remain uncertain of their accuracy, even after laboring to generate one correctly. Given the task of estimating the average of a large group based on evidence from a small portion of the group, students want to know if their interval truly contains the “right answer”.

Weaver captures both the frustration and its source: “This sort of uncertainty is absolutely maddening to those who like to have everything in its proper place and who go so far as to stack cornflakes neatly in their cereal bowls each morning. The crux of the problem is that we simply cannot know with any certainty whether a particular confidence interval actually contains the population mean” (p. 142).

I find four principal shortcomings in the book, detailed below. First, despite the predominantly clear presentation, there are points at which a statistical novice is apt to be misled. Second, Weaver’s somewhat arbitrary choices regarding the use of algebraic notation sometimes serve to obfuscate rather than illuminate. Third, his efforts at humor fall short more often than not, occasionally adding weight rather than lightening the load. Finally, the humor too often is demeaning, particularly to statistics students and to women.

Though infrequent, sometimes the author uses technical terms inconsistently or inappropriately. The word variable is such a term. A quantitative variable is a number which is (potentially) different each time we measure or observe it. In some places, Weaver uses the word variable in the standard way. At other points (e.g. p. 96) the term refers to the range of possible values of a variable. Elsewhere (p. 113) Weaver refers to the mean and standard deviation (average and dispersion of a variable) themselves as variables. While hardly a capital offense, this could be mightily confusing for the reader who is trying to come to terms with the terms! Worse still, for a reader using this book as a companion in a statistics course, multiple meanings for a single word could be distressing.

The author has decided to keep the use of formulae and graphs to a bare minimum (the first algebraic expression sneaks onto the scene on p. 55). This is a defensible choice, though at times it seems to enslave both author and reader. To be sure, unfamiliar symbols can be intimidating to a reader; on the other hand, algebra is sometimes clearer and less clumsy than English. For example, his discussion of the computation of maximum sampling error (p. 140) labors along in this sentence: “The maximum error of the estimate of the population mean can be found by multiplying 1.96 by the sample standard deviation (7) and then dividing that amount by the square root of the number of persons in our sample group (10). This calculation gives the resulting value for E of 4.34.” These two sentences could be replaced by this one: “E = 1.96(7/{square root of 10}) = 4.34” without losing the reader nearly so soon.

Humor is, of course, subjective so I hesitate to criticize Dr Weaver on this score. Simply put, I found few occasions to chuckle while reading this book. More often than not, the far‐fetched examples and digressions interfered with Weaver’s skillful presentation of obscure and admittedly dense material. The entertainment is from the adolescent male “cars, sports, and chicks” school of American humor writing … though Dave Barry need lose no sleep. Ironically, the fabricated examples are often a poor match for the drama, humor, and (sometimes) absurdity found in authentic statistical studies.

In a related vein, beyond the jokes that simply fall short are the scenarios and comments which demean. The two main foils of Weaver’s word pictures are:

1. 1

   anyone who might find the subject interesting; and

2. 2

   women.

It is not merely that so many examples treat stereotypically male subject matter (fuel efficiency, body building, weight‐lifting); it is the frequent examples dealing with, say, the illicit affairs of Congressmen or college professors. Less offensive ‐ but counter‐productive ‐ are the turns of phrase slamming the subject itself, with the implication that a reader who is “getting into the material” is somehow socially lacking.

In short, readers should be cautioned about what passes for humor and “fanciful examples” in this book. That said, though, Weaver’s approach to the core of the material provides a clear, often inviting introduction to a field feared and avoided by many.