comparing expected vs. actual

Home > Statistics Every Writer Should Know > The Stats Board > Discusssion

comparing expected vs. actual
Message posted by mike whitehead on December 13, 1999 at 12:00 AM (ET)

I want to generate a statistic that will help me compare two numbers-- an predicted result vs. an actual result. I seem to remember using a % difference calculation in college:
% diff = (|(predicted - actual)| / (predicted)) x 100%.

I have 2 questions:
1) Is this the right formula?, and
2) Is this an appropriate function in the following situation?--

A manager estimates a budget for a project. At the end of the project, the final cost of the project is determined. I want to compare "how good" the prediction was and eventually,
I would like to compare THIS result to other similar projects for this manager, as well as OTHER managers.

I realize that a simple percent would be one alternative [(actual/predicted) x 100%],
but I would like to see what other functions may be appropriate.

READERS RESPOND:
(In chronological order. Most recent at the bottom.)

Re: comparing expected vs. actual
Message posted by JG on December 13, 1999 at 12:00 AM (ET)

I think you should divide by the actual. But this is only a stating point. The 'goodness' of a prediction and meaningful comparisons are complex problems for which there is no clear cut answer.

Statistics

Your $5 contribution helps cover part the $500 annual cost of keeping this site online.