Project Risk Analysis Software and Project Risk Management Software Forum

We have been looking at alternatives to triangular distributions for modeling uncertainty in our tasks. One of the alternatives that has been mentioned is BetaPert. In what cases should it be used as opposed to triangular?

Both BetaPert and Triangular both require 3 pt estimates for Low, Base, and High. BetaPert is a variation of the original Pert distribution that was developed in the early 1950s to provide probability distributions for project management purposes. BetaPert is generally used as an alternative to Triangular Distribution in cases where the estimators believe that there is less possibility that an activity will have a duration that occurs at the boundaries. In practical terms, PetaPert generates a distribution with lower standard deviation and that is shown by a more probability density around the calculated mean. Lets take a look at a simple example and compare the Triangular and BetaPert distributions with a 3 pt estimate of 3,5, and 8 days.



What does this mean? BetaPert is less sensitive to the boundary estimates (in this case 3 and 8 days) and returns probabilities that closer to the original most likely estimate. In very simple terms, if you have that that suggests the low and high estimates have a very unlikely probability of occurring, BetaPert may be a better alternative than Triangular.

A point of interest. Some recent studies that we have seen seems to indicate that while BetaPert is appropriate in specific cases, the boundaries estimates are generally two narrow.