Some indications to correctly use the estimations of an expert in the PERT methodology

The so-called Caballer distribution, that is, the beta distribution of parameters p = h±√(2) and q = h∓√(2) had its origin in the paper of Ballastero and Caballer and has been tabulated by Caballer. In this paper, this sub-family of beta distributions of the first type is studied, as well as its relation with the classical beta of the PERT, with the family of constant variance and with the mesokurtic family. These models have risen from the works of Sasieni, Gallagher and Littlefield and Randolph. The study of these sub-families finally allows us to state a criterion with which it is possible to select the most adequate sub-family for each case, starting from data a, b and m contributed by the expert. So a new answer to Sasieni's question is given.