Daryl Huff, in his cleverly written and humorously illustrated classic ‘How to Lie with Statistics (1954)’, probes at the different ways in which statistics can mislead. He looks at biased sample studies (leading questions, ‘random’ selection, too-small samples); the different ways in which to represent averages and their drastically varying outcomes; and how, contrary to what certain statistics try to show, correlation does not necessarily imply causation. Huff shows how graphs can be used to distort and dramatize reality, by for example truncating the bottom of the vertical axis in a line or bar chart.
In the final part of the book, “How to talk back to a statistic”, Huff advises the reader on what to look out for before drawing conclusions from statistics. Huff argues that statistics can be skewed to show virtually any desired result, either through data collection methods or through the way in which the information is presented. “Many a statistic is false on its face. It gets by only because the magic of numbers brings about a suspension of common sense”, says Huff.
References
HUFF, D. (1954) How to lie with statistics. London: Penguin books
Information Distortion 