An alternative for robust estimation in Project Management

Recently, Hahn (2008) has proposed the mixture between the uniform and the beta distributions as an alternative to the beta distribution in PERT methodology which allows for varying amounts of dispersion and a greater likelihood of more extreme tail-area events. However, this mixture lacks a closed cumulative distribution function expression and its parameters remain a difficult interpretation. In addition, the kurtosis limit of the beta distribution is 3. Due to their higher kurtosis and easier elicitation we consider the Two-Sided Power and the Generalized Biparabolic distributions good candidates for applying Hahn’s idea (2008) and for building the Uniform-Two Sided Power (U-TSP) and the Uniform-Generalized Biparabolic (U-GBP) distributions. Using the same example from Hahn (2008) we are going to demonstrate that we can obtain more accuracy and a greater likelihood of more extreme tail area events. These distributions could be applied in other heavy-tailed phenomena which are usual in business contexts.