3.9 Quality measure element

What this heading covers

The QME is the smallest unit of a quality measure and the place where most programmes go wrong. The definition has three parts: a property of the target entity, a measurement method, and an optional transformation. Leave the method out and the element is not reproducible. Leave the property out and the element is a number looking for a meaning. Annex B of the draft lists thirty-one QMEs by name, all of them count or duration primitives: functions, requirements, failures, faults, hazards, test cases, restarts, jobs, tasks, user actions, inputs, data items, records, responses, results, user operations, system operations, languages, modules, components, interfaces, information structures, user interfaces, interface items, operational usage scenarios, events, input conditions, hardware environments, software environments and operational environments, plus duration. Those names are public in the contents even though the Annex B text is not.

Record position: measurement concept 2 of 3, from the fifteen terms in clause 3.

What the heading asks of a team

What to put in place before publication

1. Keep a QME catalogue with the property, the method, the unit and the data source for each entry.
2. Use the thirty-one names in Annex B as a starting vocabulary, and mark any element you add as yours.
3. Record the collection window for every count, because a count without a window cannot be compared.
4. Name the tool or register that supplies each element, and test that it supplies the same population twice.
5. Audit one element a quarter against its source, and record the difference.

Where teams usually fall short

• A count taken from a tool whose filter changed silently, so the element moved and nobody knows.
• Elements mixed across products in one total, which destroys the population the measure depends on.

What happens next

Annex B lists all thirty-one QMEs the draft names. The measurement function page shows how elements are combined into a quality measure.