As structural engineers, we often make assumptions based on descriptive statistics which is very shallow in my opinion. In recent statistical test, it revealed that not all of the strongly correlated relationships in the British Standard are accurate.

The concrete cover thickness is often associated with its capacity to deter the ingress of corrosion agents such as carbon and chloride but very little to our understanding, this relationship may not be true in case-specific event.

When running statistical analyses, the covariance returned with exceptionally low power relationship and you have to probe harder and diligently for the missing link(s). Worst, when it is proven that the initial mediation and moderation does not work.