Risk management - Correlated events part 2
Correlated events part 2
We have spoken about correlation as though it is all or nothing.
Clearly, we can have no correlation for activities where there are no common underlying events influencing their outcome.
We have mentioned fully correlated activities [see The Risk management process - evaluate - positive correlation] where a HIGH value in one member of the group drives HIGH values in the other members and similarly for LOW or MEDIUM.
Where correlation exists it is not usually 100% but can be very high or only partial if the correlation is lower.
For example, if you examined the correlation between the HEIGHT and WEIGHT of individuals it would be PARTIAL. In the main tall people will be heavier than small people.
However, it is obvious that some small people will be heavy and some tall people thin and light.
For other areas the correlation may be a lot higher. If you looked at the WEIGHT of wood against its LENGTH, (for a uniform cross section), you would expect the correlation to be very high.
PARTIAL correlation has a few problems inhibiting its use. Its very easy for people to understand the relationship if the correlation is ZERO or 100% but more tricky if it is somewhere in between.
When the correlation is PARTIAL it is hard to say whether it is say, 30% or 65%. In the main, it is easier to just use ZERO or 100% with very little detrimental affect.