14 May, 2014

(Slightly Off Topic) Spurious Correlations

During the course of the job, we find, discover and analyze "data" and come up with "information".  Sometimes we find correlations and "discover" causes.  We say "Event 'A'  caused Result 'X'".   However, it can  so happen that some "discovered" correlations are not "causal correlations" --- i.e. "Event 'A' has NOT really caused Result 'X'".  The mathematical correlation ("coefficient of correlation") may be high but there really is no logical or physical association between the two.

Here are some examples of Spurious Correlations.

The next time you say "I find a high correlation between the two", stop and think.  For a study of common biases and fallacies, I recommend "the art of thinking clearly" by rolf dobelli.


No comments:

Preview Popular Posts