C.16.RM:2 - Problem

A disagreement can arise from the measured subject, its description, the instrument, the instrument model or the calculation. Improving the wrong contribution consumes effort while leaving the wanted answer unavailable. Repeated observations can reduce some random effects and preserve a loading error, background contribution or computational mistake.

A change can also alter the quantity being measured. Cooling a specimen to stabilize an instrument may change the specimen property whose value was wanted at the original temperature. Better agreement in the changed situation therefore needs a relation back to the intended use.

The problem is to locate a useful repair, carry it through the affected relations and establish what the revised result supports. Explaining the historical cause of a fault and obtaining an adequate measurement can require different amounts of work.