<- file stat 97refdes.html ->

REF for firemen correlation

=======================Neil Henry, 28 Apr 1997==========ssc,sse From: neil w henry <nhenry@saturn.vcu.edu> Subject: Re: firemen correlation example Message-ID: <3364BB47.D85@saturn.vcu.edu> Carol A. E. Nickerson wrote: > > There is a classic example of a positive correlation between A and B > being caused by C that involves the number of firemen at a fire, > the damage done by a fire (I think), and the size of the fire. Does > anyone know exactly how this example goes and have a cite for it? > > The example was used by Paul F. Lazarsfeld in a 1946 address to illustrate what is meant by a spurious relationship. "It has been found that the more fire engines that come to a fire, the larger is the damage. Because fire engines are used to reduce damage, the result is startling and requires elaboration. As a test factor, the size of the fire is introduced. The partials then become zero and the original result appears as the product of two marginal relationships; the larger the fire, the more engines -- and also the more damage." "Interpretation of Statistical Relations as a Research Operation", in Lazarsfeld and Rosenberg, THE LANGUAGE OF SOCIAL RESEARCH, Free Press, 1955. This book was very influential on writers on research methods in sociology, and the example appears in many social statistics texts, usually without attribution since it seems so obvious. Indeed I would not be surprised to learn from an enterprising historian of statistics that Yule had used it earlier. You might compare Lazarsfeld's discussion of spurious correlation, which is very much in the Yule tradition, with Herbert Simon's in "Spurious Correlation: A Causal Interpretation" (1954) which appeared at about the same time. A very nice piece by John Aldrich appeared in Statistical Science in 1995 ("Correlations genuine and spurious in Pearson and Yule"). * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Document by Rich Ulrich. E-mail to wpilib+@pitt.edu

FAQ top.

Ulrich home page.

Ulrich FAQ. http://www.pitt.edu/~wpilib/stats99.html