45 pages • 1 hour read
How to Lie with Statistics
A modern alternative to SparkNotes and CliffsNotes, SuperSummary offers high-quality Study Guides with detailed chapter summaries and analysis of major themes, characters, and more.
Themes
The Importance of Critical Thinking
The central theme of How to Lie with Statistics is the importance of critical thinking. The book’s purpose, as Darrell Huff states in Chapter 10, is to provide a general audience with the means to spot manipulative statistics in their everyday lives. Most of these faulty statistics rely on people not paying much attention to what they are seeing and the fact that, for many, “the magic of numbers brings about a suspension of common sense” (140). The issue is not merely the bad statistics themselves but that ordinary people fall for them because numbers and data are “appealing in a fact-minded culture” (10). Logic, data, and demonstrable fact are uplifted in the modern world as having more value than opinions. When people want to lend credibility to opinions or other unprovable matters, they often mimic sound statistical studies. As a result, the untrained reader cannot differentiate good statistics from bad ones.
Each book chapter provides ways to apply a critical lens to many deceptive practices. Huff uses the first few chapters to address how to be aware of problems in collecting and analyzing data samples. The following few chapters cover how to give visuals a second look to see if they are missing or manipulating the results they present. The rest of the body of the work focuses on broader presentation issues and being aware of statistics used to draw unearned conclusions. While Huff provides numerous examples of these tricks and errors of bad statistics, the common thread is that readers should be aware that both tricks and mistakes can happen. He reiterates throughout the book that one should never take anything at face value. Instead, readers should view the world with a critical eye.
The book’s ultimate purpose isn’t to provide the reader with the means to create statistics. Other introductory statistics texts give more detailed explanations of measuring samples and plotting data. Instead, the importance lies in teaching critical thinking itself. Huff positions a strong foundation of critical thinking as fundamental to navigating the realm of statistics and the contemporary world. Sensational claims fill advertisements, the news, financing, and other places. If someone does not use critical thinking when examining these claims in their everyday lives, they may fall victim to the deceptive practices used by all these sources. In Huff’s view, the adoption of critical thinking helps ensure the ultimate well-being of his reader.
Statistics Is an Art Form
One of Huff’s main themes in How to Lie with Statistics is that statistics are not only a science but also an art form. He writes, “The statistician must choose among methods, a subjective process, and find the one that he will use to represent the facts” (122). Statistics are based on demonstrable mathematics, but the party responsible for collecting, analyzing, and presenting the data still heavily influences the results.
Each of the body chapters covers how to create misleading statistics. That also means they touch on the numerous ways subjectivity can encroach on the process of making them. There are multiple types of averages the statistician must choose from. Each works best under certain circumstances to create the most accurate picture. However, using the “incorrect” average, while still technically correct, can dramatize and create a misleading result. Another of the methods used is the word choice of the presentation. A result can appear very different depending on its description. For example, the same figure might be described as “a one percent return on sales, a fifteen percent return on investment, a ten-million-dollar profit, an increase in profits of forty percent (compared with 1935-39 average), or a decrease of sixty percent from last year” (84). None of the options is incorrect, but each presents a different picture.
Huff also notes that human choice appears in statistics in the selection of what information to keep or omit in the finished product. For example, a study might cite only the positive results and ignore negative or inconclusive findings. Whole studies can be suppressed if the outcome isn’t favorable. The methods used to obtain the data can also be left out, or the amount of error may not appear in the results. The potential subjectivity of statistics also appears when discussing the visual representation of a result. Presenting a given statistic as a chart or graph instead of raw data is a deliberate choice by the person or entity presenting them. For example, on a line graph with a distorted scale, a slight difference in data can be manipulated to appear larger than it is. The shape and size of bars on a bar chart or of pictorial representations can distort proportions and prompt false conclusions.
Huff’s purpose in highlighting this subjective aspect of statistics is to dispel some of the air of authority that numbers have over those who are uneducated on the subject. Because modern society generally views the sciences as wholly objective, results may go unquestioned. Huff shows that because humans create statistics, they can never be totally objective. As Huff notes, people who are using statistics in advertising are “about as unlikely to select an unfavorable method as a copywriter is to call his sponsor’s product flimsy and cheap” (122), and “the man in academic work may have a bias (possibly unconscious) to favor, a point to prove, an axe to grind” (123). No statistic can entirely remove bias and subjectivity from itself. Huff doesn’t present this as wholly a bad thing, however. The true art of statistics lies in how they are used. While dishonest parties can use them to deceive, they can also be used to inform decisions and illuminate the truth.
The Importance of Proper Sampling
Another central theme throughout How to Lie with Statistics is the importance of proper sampling. The subject is the focus of the book’s first chapter but continues as a through line for the remainder of the text. Huff says that samples are “the heart of the greater part of the statistics you meet on all sorts of subjects” (15). The sample is the base with which a statistician constructs their work. Without a sample, there is no data. Huff also notes that “[t]he result of a sampling study is no better than the sample it is based on” (20). A sample is, at its core, a representation of a whole. Selecting the correct sample for a given statistic is critical to obtaining the best and most accurate measurements. The lack of a strong foundation causes the rest of the study to fall apart.
Throughout How to Lie with Statistics, Huff presents several ways samples prove faulty or inadequate. One way to ruin a sample is the presence of too much bias. Genuinely random samples are impossible to obtain, and as Huff notes about removing all possible sources of bias, “the battle is never won” (24). Despite the inability of total removal, the statistician should eliminate as many sources of bias as possible from their samples. If strong enough biases enter the sample, they color the statistical results before any other forms of manipulation appear. Biases can also be both intentional and unintentional. Huff notes that “unconscious bias […] is often more dangerous” (126). The danger of unintentional bias comes from being harder to spot than intentional ones.
In addition to bias, a sample that is too small for the population it seeks to represent distorts the data. With too few examples, potential variations within a subset of a population appear more significant than they are. This skewing is useful for creating a manipulative statistic but falls short of an accurate representation. The danger of the bad sample comes from its illusiveness compared to other statistical manipulation forms. If numbers are left off the side of a chart, or the person presenting the statistic makes a leap in logic from the tested data to the stated result, the viewer can immediately take note of the problem. However, Huff warns, “You cannot, as a casual reader […] come to exact conclusions as to the adequacy of a sample” (128). The responsibility of correct sample-taking lies with the entity running the study. The reader should, however, remain aware that this problem exists and be able to recognize it.
Unlock all 45 pages of this Study GuidePlus, gain access to 8,800+ more expert-written Study Guides.
Including features: