Confidence Interval
A statistical range that is expected to contain the true value with a stated probability, commonly 90% or 95%.
A confidence interval expresses a plausible range for a value together with a probability. A 95% lower bound, for example, is the value below which the true result is unlikely to fall — a natural anchor for conservative crediting.
Applying a lower-bound figure rather than a central estimate is a common way to build conservativeness directly into the result, since it uses the pessimistic end of the interval.
Because confidence intervals depend on the data and model, they must be computed honestly; narrow intervals from poor data would give false assurance.
Frequently asked questions
Why use a lower bound for crediting?
It ensures credits reflect a benefit that is very likely real, aligning with conservative accounting.
Does a wider interval always mean worse data?
Often, though genuine natural variability can also widen an interval; the cause should be understood.