About CorrelationIsNotCausation.co.uk
If you have come here, then you probably already know that statstics can (and often do) lie. This is not the first web site to examine this topic. I took a more than a little inspiration from the excellent Spurious Correlations website, but I wanted to use data which was a bit more relevent to the British public, and anyway, I just fancied seeing if I could do it.
A few notes on the data itself: The data is organised by category, and data is only correlated against data from other categories to reduce the number of boring correlations. So far, I have got all of the data from freely availible sources on the web. In a lot of cases, I found out that less is more. I started with a few hundred timeseries of Visitor Attaction data and lots of Weather data broken down by region, but due to the magic of statistics, the correlations between these data sets overwhelemed the more interesting ones, so I took them out. I would like to include more "Quacks and Quackery" data, so if you know of any relaibale sources, please do let me know (contact details are at the bottom of the page). Also a note of thanks to Alan Henness at Zeno's Blog, whose work I shamelessly reused in order to compile the homeopathy datasets. Final point about the graphs: It's probably best not to link directly to them for now. Every now and then I reload all of the data and the Urls may change.
Some new notes on the data as a result of feedback given to me by various friends and colleagues: Thanks to Derek for pointing out the difference between negatively correlated (-1) and uncorrelated (0). Duly noted and corrected! Also thanks those who pointed out that the weather data doesn't show the absolute maximums or minimums, but the average maximums or minimums. You can read a lot more about all of that in the Met Office Website Archives if you are interested.
The site is a typical 'lamp' setup, coded in PHP and MySQL and runs on a Linux server. The graphs were produced using JQPlot. At the moment, it there is no advertising, but if it gets a lot of traffic, I might add some banners, just to pay for the hosting.
You can contact me via info@correlationisnotcausation.co.uk, sorry but I can't promise to answer every email, but I'll try!