Type A and Type B uncertainty components

For a plain-language introduction and worked example, start with Measurement uncertainty.

Type A repeated components derive uncertainty from ordered observations. Metra retains every decimal reading and calculates the mean, sample standard deviation, standard uncertainty of the mean, and n - 1 degrees of freedom. Too few observations create a blocking finding.

Type A entered components use an event-specific or controlled statistical magnitude produced outside the repeated-reading workflow. Record its provenance; do not use it to disguise raw observations that should have been retained.

Type B components use other available information, including reference-standard snapshots, resolution, stability, environmental effects, specifications, reproducibility, and controlled entered values.

Classification describes how a component was evaluated, not whether it is more or less important.

Each result freezes its input, distribution, resolved divisor, sensitivity coefficient, degrees of freedom, standard uncertainty, contribution, unit, and source-record identity and hash where applicable.