Some of the factors that bear on important decisions we make are linked. For example, we might not know how much demand there will be for an online service, but we know that if it is at the high end of our expectations then the load on the servers and networks used to deliver it are likely to be high too. Modelling such relationships in quantitative risk analysis must be done with great care. Ignoring them can result in outcomes that are wrong, sometimes significantly so, and that have the potential to mislead the users of an analysis.
This tutorial note examines the topic of correlation, the statistical term for the degree to which the variation in one factor will be related to variation in one or more other factors. It suggests some practical ways to incorporate correlations, or the relationships that give rise to them, in quantitative risk analysis models.
Read the full tutorial here.