The parameters of the classical PERT – an assessment of its success

In classical PERT, concrete values are assigned to the parameters determining the expected value and the variance (of a task, cash flow, etc.) based on the least possible, greatest possible and modal values provided by the expert. The use of these parameters has led to successful conclusions in numerous practical applications; nevertheless, their theoretical foundations have been debated from the start. This article attempts to find an explanation for the practical success of the classical PERT model, with an in-depth study of its theoretical premises. A logical path towards the classical PERT parameters is presented; furthermore, this is shown to be the only constant variance model that has the same kurtosis as the normal one. A method is proposed for a better fit to the data provided by the expert, improving on that of the classical model.