Recent Study Shows Two Random Statistics are Strongly Correlated
May 2014
I am sure most people have seen the new site, Spurious Correlation . The site has a database of random statistics and continually compares the data to generate correlations between two sets of statistics. While there are relationships found by the site that are likely legitimately linked, most just seem ridiculous.
A random correlation I noticed was the “Number of People who Starve to Death in the US” and the “Per Capita Consumption of Margarine.” A correlation coefficient of 0.97 is nothing to sneeze at.
I can just picture a butter commercial encouraging people to dump margarine and save a life. More than anything, the site serves to show that with enough comparisons, you can find a statistically significant result somewhere.
This website could be a gold mine for the disingenuous reporter. The Daily Mail could run a daily article grabbing a result from the site, and probably have enough material for years. Just take two random things that are strongly correlated and speculate wildly about the reasons. This differs little from how many news stories—and even some scientific studies—are written now.