Implications of the research on expert overconfidence and dependence

In cases where interval estimates are obtained from expert opinion, the literature on expert overconfidence and dependence suggests that caution is in order. In particular, it may be inadvisable in such cases to interpret any areas of overlap among the intervals provided by the various experts as strong evidence that the quantity of interest falls within the region of overlap. The available literature on methods for assessing and compensating for overconfidence and for combining intervals or probability distributions from multiple experts is discussed. An ad hoc method of combining multiple expert opinions, suitable for use in relatively small projects, is also presented.