On a certain approach to ``risk-in-risk'' estimator (Q1390666)

The authors study the problem of optimal use of the opinions of a number of experts about some unknown distribution function. These opinions are in the form of (subjective) estimates of a number of quantiles of the distribution function. The proposed approach is Bayesian. An explicit solution is obtained under two assumptions: 1) that the unknown distribution is from the log-normal family and 2) that the estimates given by the experts are jointly independent. The authors have implemented their method in a program system which they use to illustrate their approach. They give an example concerning the evaluation of the seismic stability of a construction.