If estimates of mean and variance are needed and only experts' opinions are available, the literature agrees that it is wise behaviour to ask only for their (subjective) estimates of quantiles: from these, estimates of the desired parameters are calculated.Quite a number of methods have been suggested up to now, using different sets of quantiles.Four of these methods are compared here by means of an experiment in which 56 students acted as experts on four distributions.The accuracy of the estimation of both the quantiles and the parameters are discussed.
|Place of Publication||Tilburg|
|Number of pages||31|
|Publication status||Published - 2002|
|Name||CentER Discussion Paper|