Let us conduct a psychological experiment. Which formula do you think will yield a more precise and accurate result:
I guarantee that the majority of the readers will select the second formula as the one that would provide the best result. Unfortunately nothing can be further from the truth because any formula is as good as the data going into it. Some of you may have heard the following expression used by statisticians:
Garbage in, garbage out
It basically means that, no matter how complicated the formula is, if you “infuse” it with inaccurate data, the model will produce an inaccurate answer. Having said that take a closer look at the formulas above: the first model only has two ingredients, while the second one has multiple variables. Hence with a larger number of variables we end up with more room for estimation errors.
Why did I start this article with such a strange example? I have recently taught a project portfolio management course in London, UK and the following conversation took place between a Portfolio Manager of a large utility company and your humble servant:
PM: Oh, our executives will never sign up for PPM. We tried it before and it didn’t work
Me: Why didn’t it work?
PM: Well take a look here … (proceeds to show me a very complicated spreadsheet that incorporates not one, not two, but three scoring models for different types of projects at the same company. In order to assess the project proposal one has to answer several dozen questions by selecting appropriate choices from the dropdown menus. Once the choices are made, the spreadsheet “tells” you which scoring model to use. The models are also far from simple; close to a dozen scoring categories with five different potential scores for each variable …)
Me: So, did you use that model?