Decision Making / 1 / 1.5 / 1.5.5

1.5.5 Confusion Over Correlation And Causality

Almost every day we hear news that drinking red wine reduces the incidence of heart attack, that Mediterranean diet, although relatively high in fats, is good for longevity. We could offer, and find evidence to support, a proposition that higher rates of aircraft crashes involve pilots with eyes of a particular colour. Whilst certain propositions might appear reasonable on the basis of available evidence, we must be careful not to confuse correlation and causality. Correlation is an indication of the extent to which two measures stack up statistically. What is observed may or may not be related to cause-and-effect. It is unfortunate fact that correlation is frequently and quite incorrectly interpreted as causality.

It is highly unlikely that brown-eyed pilots make human errors that result in aircraft crashes any more than green, blue or grey-eyed ones might. Sterman (2000: 14) makes the point that confusing correlation with causality can lead to terrible misjudgements and policy errors. This is certainly so when it comes to risk management.

So, we must train ourselves to recognise the difference between correlation and cause-and-effect. The arguments in this book rely on reliable identification of cause-and-effect relationships, that is, causality. We use correlation only as a guide to where we might look for cause-and-effect relationships.