Common or Random Causes - Causes of Variability

Consider throwing a pair of dice. The possible outcomes range from 2 to 12, but with different probabilities of occurrence (see figure below). Over a long series of throws, '2' would result in about 3 per cent (1/36) of the throws, and '7' would result in about 17 per cent (6/36) of them. However, the result of a particular throw is unpredictable. It will be determined by a number of variables which interact in a manner of which we have no detailed knowledge at the time of the throw e.g. the trajectories of the dice as they leave the hand, the microscopic contours of the surfaces of the dice and landing surface, the mechanical properties of the surfaces at the points of contact, etc. Such variables are called common causes.

When, in the long run, the outputs of a process can be expressed in probabilistic terms, the process is described as a common cause system. Manufacturing processes are typically common cause systems. That is, even when great care is taken in the set up and control of the process, variability of the output occurs. This variability cannot be reduced unless the process is changed. Statistical methods provide a means of registering common-cause variations and distinguishing them from the second type, assignable causes.