Engineering judgement in reliability and safety and its limits: what can we learn from research in psychology

Engineering judgement has an important role in safety or reliability assessment. This paper focuses on the use of engineering judgement for integrating diverse evidence into an assessment of the safety or reliability of a product. In many cases of stringent safety requirements, this form of engineering (or "expert") judgement, i.e., "informal inference from complex evidence", is the crucial resource for the decision maker, for lack of more solid, objective evidence. This dependence on judgement is especially evident in the assessment of the unreliability due to possible design faults in complex products, and computer software in particular. Although engineering judgement plays an essential role in the assessment, there are good reasons to doubt the ability of experts in some of the judgement tasks in which they are usually employed. Experimental research both about the way humans think and integrate evidence, and about the performance of experts in tasks similar to engineering judgement, support the idea that the ability of experts may be overrated. This paper summarises some literature about common fallacies and ways to guard against them, and argues for a more disciplined use of expert judgement.