LYING WITH STATISTICS
The American University
The misuse in communication of data in print or presentation, either intentionally or unintentionally, the result of which misleads those to whom the communication is directed.
A classic example of associating statistics with lying is attributed to British Prime Minister Disraeli, who declared, "There are three kinds of lies: lies, damned lies, and statistics." More contemporary and book length efforts to explore lying and statistics include Darrel Huff's (1954) How to Lie with Statistics and Robert Hooke's (1983) How to Tell the Liars from the Statisticians. Neither of these books were intended to be primers for lying but, rather, light-hearted guides for nonstatisticians in how to distinguish between complete statistical disclosure and good statistical reasoning on the one hand and misleading or malicious reporting of data on the other.
Increasingly, we rely on statistics to determine trends, to judge public opinion, and even to learn which toothpaste reduces cavity production. Hooke (1983) distinguishes between statistics in the plural and in the singular. Most people think of statistics as plural—as sets of numbers and figures and data. Statisticians think of it as singular—a subject matter that allows one to understand chance, cause and effect, correlation, and the scientific method. People who gather data (statistics plural) are not necessarily statisticians. If these "data pushers," as Hooke refers to them, use the data in an incomplete or uninformed manner, then their manipulation, intended or not intended, is considered lying.