How do you perform a simple uncertainty budget for a verification?

• A. Measure random errors only
• B. Identify major sources, estimate their contributions, and combine them (usually quadratically).
• C. Use fixed values for all sources
• D. Ignore systematic biases

Answer: (B)

The main idea is to quantify uncertainty by breaking it down into the pieces that can affect the measurement and then combining those pieces into a single overall estimate. In a verification, you start by listing the major sources of uncertainty—things like instrument resolution, calibration uncertainty, environmental conditions, and any method-related variability. For each source, you estimate how much it could contribute, typically as a standard deviation or an equivalent measure of spread.

Once you have those contributions, you combine them to get the total uncertainty. For independent sources, the standard approach is to use the root-sum-square method, which adds the squares of the individual contributions and takes the square root. This gives the standard uncertainty of the result and helps highlight which sources are the main drivers.

This approach is practical because it reflects real variability rather than relying on a single fixed value. It also accommodates including systematic biases if they are known and can be quantified; simply ignoring biases would underestimate the total uncertainty. Conversely, focusing only on random errors misses important, sometimes dominant, systematic effects. Using fixed values for all sources or ignoring biases would both fail to capture the true uncertainty in the verification. If some sources are correlated, you’d account for those correlations rather than just quadrature, but for a simple budget independent contributions and quadrature is a reasonable starting point.