Correlation is definitely not causation: Following are six data sets, collected during each of the years 2000—2009. Each of the first three is highly correlated with one of the second three. Can you guess which is correlated with which?

A. German passenger cars sold in the United States (in thousands)

B. Per capita consumption of turkey (in pounds)

C. Power generated by nuclear power plants (billions of kilowatt-hours) 1. Total revenue generated by U.S. skiing facilities (millions of $) 2. Number of fatalities while boarding a bus 3.

Divorce rate in Connecticut per 1000 people

Students may guess which are correlated with which (one can only guess), then compute the correlations to see which are in fact highly correlated. Construct scatterplots for the highly correlated pairs of variables, and discuss for which pairs the correlation coefficient is an appropriate summary.
These correlations were discovered by considering thousands of pairs of variables and choosing a few whose correlations were large. In Section 1.1 we described statistics as a process in which the first step is to formulate questions and the second is to collect data. Did this method of finding correlations follow this process? Discuss how this might have led to the discovery of correlations that do not reflect a real relationship.

Year A B C 1 2 3 2000 863 13.7 753.9 1551 1 3.3 2001 837 13.8 768.8 1635 1 3.2 2002 930 14.0 780.1 1801 1 3.3 2003 830 13.7 763.7 1827 2 3.2 2004 810 13.5 788.5 1956 1 3.1 2005 923 13.2 782.0 1989 2 3.0 2006 1154 13.3 787.2 2178 8 3.1 2007 1183 13.8 806.4 2257 7 3.2 2008 1142 13.9 806.2 2476 7 3.4 2009 829 13.3 798.9 2438 1 3.0