next up previous contents
Next: 10.5.2 Climatology Up: 10.5 Some dangers in Previous: 10.5 Some dangers in

10.5.1 Unrecognized causes

A correlation between two sets of observations does not necessarily indicate a causative relationship between the two measured characteristics. It is often the case that X and Y are correlated because both are related to Z, and neither X nor Y influence the other. For example, rainfall on a given day is highly correlated with rainfall on the preceding day, but this does not reflect a causative relationship so much as a tendency for weather patterns to persist for several days. When designing experiments that search for causative relationships, one must consider alternate relationships that will produce correlations like this one.




NCAR Advanced Study Program
http://www.asp.ucar.edu