Just had a conversation of sorts where it was stated, I look at the median EQF of a portfolio as one gauge of the quality of the overall underlying data. The problem is without the variances of any single point metric, that metric is pretty much worthless.
Here's a little exercise we use at conferences and training sessions.
- I'm going to send you on a mission to determine the Mode of a measurement. You'll get a clipboard, a hat, a folding chair, and a few other items. You'll go to two locations for 365 days and record the high temperature of the day.
- Let's start with Trinidad-Tobago. You've got your clipboard, a nice folding chair sitting on the beach, a good sun hat and sunscreen.
- Next, you'll go to Cody Wyoming, just north of my home and do the same thing
Now let's look at the Mode of those numbers. Why the Mode. It's the most recurring value in a time series. This is what we use when modeling - Estimating - variables in project planning. The Mode of a Task's work duration is the Most Likely value drawn from a sample of possible values in the Probability Distribution Function. It is the value that occurs most often in the Task's life. If we have a similar Task - through a Reference Class Forecasting process, we want to use the Mode of the duration. Not the Mean (average) or the Median (the middle most). The Mode is what the Task will take most of the time.
So now let's look at the data.
- For Trinidad-Tobago, the number that occurs most often across the 365 days of sitting in your chair writing down numbers is 84° F
- In Cody Wyoming, the number that occurs most often in the 365 days is - wait for it - 84°F
Now, these are the Mode numbers. The Most Likely To Occur. For temperatures, this is a bit of a stretch because temperatures are driven by a periodic cycle of seasons and weather.
But for task durations, this is a legitimate starting point for estimating. What is the most likley duration for this work? Without the variance on the Mode or the Mean, the Median is the middlemost number, there can be no confidence you wouldn't freeze to death in February if you go to Wyoming in your shorts and teeshirt