Why don't my staffing rate numbers add up to the same total shown on cumulative effort charts and reports?
First, it is important to understand the difference between the average staffing rate and cumulative effort charts and reports.
Average Staffing Rate charts and reports display the number of FTE staff used by the project during any given month, regardless of how much of the month is covered by that phase. When a phase starts or ends mid-month, the staffing rate for that month is not affected.
Cumulative Effort charts and reports display the total effort for each month. So here, when a phase starts or ends mid-month, the total effort reflects those partial months.
To see this more clearly, let's look at an example. If there 10 people work during December (and the phase begins on December 31st), then the Staffing Rate chart and report will show 10 people for December; but the Cumulative Effort for that month would only be .33 mm (10 people x .033 months). This logic holds true for other rate values like defects and cost.
A second factor that may cause discrepancies between cumulative effort or cost and the sum of the monthly effort/cost rates is really a calculus issue. Staffing numbers represent the average manpower rate for a given time period. These numbers are typically taken at mid-month.
For graphing purposes, it is more practical to break up the smooth theoretical curve used to derive cumulative effort/cost into slices (like bars in a histogram) where the top of each bar corresponds to the average staffing rate for each month. Because each bar has a flat top, each month a small amount of error is introduced as there will always be a small area under the curve that is not covered by the bars.
As we add more and more months (bars) to the staffing chart, the bars become narrower and narrower and the amount of error decreases. If the intervals are small enough, the sum of the rates should be very close to the cumulative effort. However, when the intervals are large (for instance, monthly for a 3-month project), the approximation will not be as close.