The use of Pearson’s r, aka the correlation coefficient, to demonstrate a relationship between two sets of data has long been a standby of public relations measurement and social science research in general. Back in the day, when the calculation of a correlation coefficient required some actual effort at calculation…
…those who used this statistic had motivation for considering carefully what they were up to (“Think twice, calculate once.”). Not so nowadays, when, in the blink of a cursor, Excel will calculate the correlation coefficient between any two strings of numbers you point it at. Today it’s easy to dredge around in your data and throw as many correlations against the wall as possible, just to see what sticks.
So it’s refreshing to come across Spurious Correlations, a website full of hilarious examples that remind us that that correlations are common and often meaningless. There we find, for instance, that, based on annual data for the U.S., the number of honey producing bee colonies is inversely correlated with the number of juvenile arrests for possession of marijuana, yielding an r = -0.933389:
Also that, also based on annual data for the U.S., the total annual revenue generated by arcades is closely correlated with the number of computer science doctorates awarded, with an r of 0.985065:
Spurious Correlations is a humorous warning that correlation does not equal causality. Always take your statistics with a grain of salt.
Thanks to HFBoards for the cartoon.
Latest posts by Bill Paarlberg (see all)
- The Growing Demand for Online Privacy: Will It Hamper Communications Measurement? - July 21, 2017
- What Is the Meaning of Measurement, Part 2 - July 21, 2017
- Great Minds on Measurement: John Cage - July 18, 2017