"If you torture the data long enough, it will confess” has come to mean that if you manipulate the data long enough, you’ll finish the the result you want. Some people has asserted that the word “data” is plural, so it should be “If you torture the data long enough, they will confess.”
Economist Ronald Coase is credited with the saying, but a 1977 source has his quotation as “If you torture the data long enough. nature will confess.” This is not the same meaning; the “nature” addition means that if you torture the data long enough, nature will confess its truth.
Wikipedia: Ronald Coase
Ronald Harry Coase (born 29 December 1910) is a British economist and the Clifton R. Musser Professor Emeritus of Economics at the University of Chicago Law School. After studying with the University of London External Programme in 1927–29, Coase entered the London School of Economics where he took courses with Arnold Plant. He received the Nobel Prize in Economics in 1991.
Coase is best known for two articles in particular: “The Nature of the Firm” (1937), which introduces the concept of transaction costs to explain the nature and limits of firms, and “The Problem of Social Cost” (1960), which suggests that well-defined property rights could overcome the problems of externalities (see Coase Theorem). Coase is also often referred to as the “father” of reform in the policy for allocation of the electromagnetic spectrum, based on his article “The Federal Communications Commission” (1959) where he criticizes spectrum licensing, suggesting property rights as a more efficient method of allocating spectrum to users. Additionally, Coase’s transaction costs approach is currently influential in modern organizational economics, where it was reintroduced by Oliver E. Williamson.
Coase coined the well known, but often misquoted adage “If you torture the data long enough, it will confess."*
*Gordon Tullock, “A Comment on Daniel Klein’s ‘A Plea to Economists Who Favor Liberty’”, Eastern Economic Journal, Spring 2001, note 2 (Text: “As Ronald Coase says, ‘if you torture the data long enough it will confess’.” Note: “I have heard him say this several times. So far as I know he has never published it.")
Issues in Public Utility Regulation:
Proceedings of the Institute of Public Utilities Tenth Annual Conference
Michigan State University. Institute of Public Utilities
Edited by Harry Martin Trebing
In a remark that deserves broad circulation, it has been said that if you torture the data long enough, it will confess, even to crimes it did not commit.
Proceedings of the Business and Economic Statistics Section
American Statistical Association
As the economist Ronald Coase said “If you torture the data long enough it will confess”. Perhaps he should have said that they will confess.
By Stanley Lieberson
Berkeley, CA: University of California Press
As R. H. Coase put it, “If you torture the data long enough nature will confess” (cited in Wallace 1977: 431).
By Christopher Code
London: Whurr Publishers
The late John C. Oglivie, Professor of Psychology and Statistics at the University of Toronto, had a sign in his office which read: ‘If you torture the data long enough, they will confess anything’.
Companion to Contemporary Economic Thought
Edited by David Greenaway, Michael Bleaney and Ian M. T. Stewart
New York, NY: Routledge
Working from a common data set, we can only perform different statistical tests on the same data, and, as Coase is reported to have remarked, if you torture the data long enough they will confess.
15 October 1991, Chicago (IL) Sun-Times, “U. of C. professor wins Nobel economics prize” by Bill Rumbler, pg. 1:
Colleages said Coase was known for his painstakingly slow research and writing, once commenting, “If you torture the data long enough, it will confess.”
Modern American Usage: A Guide
By Wilson Follett
Revised by Erik Wensberg
New York, NY: Hill and Wang
Pg. 99 (data):
True, we continue to read statements here and there like Additional information...will be reported as the data is gtransmitted; and the following from an elegant magazine, mocking statisticians: ‘The results [of certain calculations] could then confirm the Statistician’s Law: “If you torture the data long enough, it will confess.”
New York (NY) Times
Economic Scene; Which party in the White House means good times for stock investors?
By Hal R. Varian
Published: November 20, 2003
Of course, there is always the possibility of a spurious correlation. As econometricians say, ‘’If you torture the data hard enough, it will confess to anything.’’ People have been looking for economic predictors of stock market behavior for decades, so it’s not surprising that every now and then we find some correlations.
A Guide to Econometrics
By Peter Kennedy
Cambridge, MA: MIT Press
This methodology is described eloquently by Coase: “if you torture the data long enough, Nature will confess.”
What Wall Street Doesn’t Want You to Know:
How You Can Build Real Wealth Investing in Index Funds
By Larry E. Swedroe
New York, NY: Truman Talley Books/St. Martin’s Press
The first problem is a result of what is known as “data mining.” If you look hard enough for a correlation, you will probably find one. In other words, if you torture the data enough, eventually it will confess.
New York City • Banking/Finance/Insurance • (0) Comments • Monday, January 31, 2011 • Permalink