How to Lie with Statistics

Chapter 9 covers the general practice of using statistics to mislead an audience to false conclusions. “Statisticulation”—a combination of the words statistics and manipulation—is the term Huff coins to describe this act. He clarifies that not all inaccurate statistics result from intentional deception by the statistician; they may also result from salespeople or journalists who present the data after its creation. He notes that most of these manipulative statistics are made to exaggerate the result, which makes it harder to argue that this deceptive framing is unintentional.

Huff continues by discussing other methods of manipulation he didn’t cover in the previous chapters. Using maps, particularly of the US, without accounting for the facts that some states have a large area and a small population can distort whatever statistic they claim to represent.

In the next section, Huff warns the reader to look for decimals. Many statistics cannot have results that are accurate enough to justify their use, so their inclusion is an exaggeration to make the result look more precise. This is especially true for statistics created from small sample sizes. Huff continues with a warning about percentages. He notes that confusion stems from the number used as a base and the method used to get the percentage. This often results in numbers that appear more dramatic than they really are. For example, he cites a government study that identified 4.9% of workers as receiving a particular wage, although the study failed to mention that this percentage represented only two people.

The following section covers the problem with data that seems to add up but doesn’t. For example, Huff lists reports of money lost during strikes that include all expenses, including those that are irrelevant to the strike. Another example of this involves adding up the percentage of price increases on items in a way that artificially inflates the cost of living. Huff returns to percentages and common confusion with percentage points to make results look better or worse than they are; he notes that percentages should be treated with caution. He also warns about the use of percentiles, as the difference between certain ratings is more significant than others due to the tendency of data to form a bell curve. Huff continues by turning to the statisticians themselves or what can happen when a rivalry appears between them. Here, the problem of deceptive statistics often becomes evident due to the dueling conclusions reached by statisticians on opposing sides of an issue.

The final problem Huff talks about is the manipulation of index numbers. He highlights their importance due to their connection to wages and the cost of items. In his examples of how they can be manipulated, he returns to the average and adds another type: the geometric average. Using it to examine the price differences in the example he gives, the reader finds no real change, despite efforts to skew the data to demonstrate an increase or decrease in elements such as the cost of living, depending on the agenda of the person interpreting the data.

Huff concludes that Statistics Is an Art Form as much as a science. The process by which data is gathered, analyzed, and presented comes from human decisions and biases. He says readers should remember this when examining any statistic, but they should also not wholly reject statistics as completely bad. He compares this to “refusing to read because writers sometimes use words to hide facts and relationships rather than to reveal them” (123). The problem is not the tool but the way it is used.