# Monte Carlo – better than Statistical PERT?

When I took the PMP exam in 2005, the book I used to help me prepare, Rita Mulcahy’s “PMP Exam Prep” which is still a popular PMP prep book, even though Ms. Mulcahy passed away a few years ago (her company keeps the book updated for the current iteration of the PMP exam).  In her book, Ms. Mulcahy said that “PMI suggests that Monte Carlo simulation will create a project duration that is closer to reality than CPM or PERT.” (page 105, 4th edition).

She’s right — but only if someone is skilled in creating a Monte Carlo simulation model, and only if that same person has access to good Monte Carlo simulation software.

I created SPERT because if you’re a project manager on a shoestring budget, and you don’t have access to Monte Carlo simulation, and you don’t want to figure out which probability distribution is the very best fit for every task on your project, you could still benefit from using a technique based in statistics and probabilities, and a technique that is “accurate enough” for most purposes.

SPERT is better than PERT because it provides an infinite number of estimates, each of which offer a corresponding probability that the estimate will be (or will not be) exceeded.  SPERT provides for the estimator’s subjective opinion, too, about the most likely outcome in a way that neither PERT nor Monte Carlo simulation can readily do.

But SPERT is just one technique, and as such, it has drawbacks just like other estimation approaches have drawbacks, too.  As a project manager, you want to have a lot of tools and techniques in your toolbox that you can pull out and use at any time.  SPERT is as easy-to-use, easy-to-understand technique that can help you gain insight into the estimation process, impacts, and it helps you to communicate in concrete terms what you very abstractly feel.

Monte Carlo simulation, when done correctly, requires thoughtful consideration of far more than a 3-point estimate and a subjective opinion like SPERT requires.  The pay-off for the added work of building a Monte Carlo simulation is that it will evaluate all the paths through the network diagram (not just the critical path) and the resulting normal curve created by the Central Limit Theorem can help you see what-if scenarios from every angle.   But not every estimation problem requires that level of rigor.  Sometimes, you need to put together an estimate very quickly, and be able to convey a sense of confidence in the estimate you provide a project sponsor or stakeholder.  Sometimes, using SPERT is an excellent choice; sometimes, a Monte Carlo simulation is.  Sometimes other estimation approaches are needed and are useful.

There are many more choices than PERT, SPERT and Monte Carlo simulation to estimate a project.   But if you have access to Microsoft Excel, you already have the necessary software to create SPERT estimates.  All you need now is the skill to do so.