How to Lie with Statistics

"The secret language of statistics, so appealing in a fact-minded culture, is employed to sensationalize, inflate, confuse, and oversimplify." Biased samples skew results.

"The secret language of statistics, so appealing in a fact-minded culture, is employed to sensationalize, inflate, confuse, and oversimplify." Biased samples skew results. Surveys and studies often rely on samples that are not representative of the entire population. For example, a study on the income of Yale graduates may only include those who responded to a survey, potentially excluding less successful alumni who are unwilling to report their earnings. This can lead to inflated average income figures that do not accurately represent the entire graduating class. Self-selection introduces bias. When participants choose whether to participate in a study, it can create a biased sample. For instance, people who are more satisfied with a product may be more likely to respond to a customer satisfaction survey, leading to overly positive results. To combat this, be wary of studies with low response rates or those that don't disclose their methodology. Size matters in sampling. Small sample sizes can lead to unreliable conclusions. A study based on just a handful of cases may produce results that are not statistically significant or representative of the larger population. Always look for information on sample size and statistical significance when evaluating claims based on data.

"There are three kinds of averages: the mean, median, and mode. The same set of figures can be used to demonstrate at least three different averages." Different averages tell different stories. The mean (arithmetic average), median (middle value), and mode (most frequent value) can each paint a different picture of the same data set. For example: Mean income in a community can be significantly skewed by a few high earners Median income often provides a more accurate representation of typical earnings Mode can highlight the most common income level Choose the right average for context. When presented with an average, always ask which type is being used and why. For income data, the median is often more informative than the mean, as it's less affected by extreme values. In contrast, the mean might be more appropriate for data sets with less variation. Beware of unspecified averages. When an "average" is presented without specifying the type, it's often chosen to support a particular narrative. Always be skeptical of unqualified averages and seek additional context to understand the full picture of the data.

"The Gee-Whiz Graph...is employed to give the impression of important change where there is relatively little." Truncated scales exaggerate changes. Graphs that don't start at zero can make small differences appear dramatic. For example, a graph showing stock prices over a short period might use a scale…