Book Reviews

David Spiegelhalter. The Art of Statistics: Learning from Data. London, UK: Pelican, 2019. 426p. 
Paper, $16.99 (ISBN 978-1-5416-7570-4).

We are presented with statistical claims in the news, on social media, and 
in the workplace. How can we determine if these claims are trustwor-
thy? University of Cambridge professor David Spiegelhalter describes 
how real-life problems are solved with statistical analysis as a way of 
introducing concepts necessary for data literacy. He defines data literacy 
as “the ability to not only carry out statistical analysis on real-world 
problems, but also to understand and critique any conclusions drawn 
by others on the basis of statistics.” This book will help readers acquire 
these abilities.

In addition to the statistical concepts discussed, Professor Spiegel-
halter presents 10 questions for readers to ask themselves when faced 

with a statistical claim. For example, “What is the statistical uncertainty/confidence in the 
findings?” Readers are advised to consider margins of error, statistical significance, sample 
size, and bias. He also applies a practical framework to approach problems that readers can 
use. Problems discussed in the book are approached through the PPDAC cycle, which stands 
for explaining the Problem, creating a data collection Plan, collecting the Data, performing the 
Analysis of results, and interpreting the results to form a Conclusion that is communicated 
to interested parties. The first half of the book covers summarizing and communicating data 
effectively with tables and graphs, the process of inductive inference (in other words, going 
from examining collected data to making observations that are generalized to the population), 
the challenges in determining causation, employing regression to investigate relationships, and 
the use of algorithms for making predictions. The second half explains margins of error, prob-
ability, hypothesis testing, and the Bayesian approach to statistical analysis. The final chapters 
focus on stories of incorrect statistical analysis and suggestions for how we can improve. The 
author concludes by summarizing the main issues dealt with in the book, including the fol-
lowing points: statistical analyses should answer specific questions; variability exists in any 
data collected; researchers should understand why they are employing specific formulas and 
procedures to analyze the data; the communication of results should be as simple as possible; 
margins of error should be presented to readers; and statistical analysis should be reported 
clearly, and in enough detail, to make it reproducible.

This book is written for both students and consumers of statistics. The clarity of the author’s 
explanations makes the subject accessible to a general audience. Professor Spiegelhalter also 
demonstrates how to communicate concepts that are often reported incorrectly, such as a 95 
percent confidence interval. The different problems discussed with their accompanying tables 
and/or graphs provide a rich source of examples to help librarians in their work, whether it is 
to help them design a study, analyze a data set, communicate results, or teach students how 
to read the scientific literature.—Giovanna Badia, McGill University

923