Measurement Validity
An instrument records something specific. The claim built on it is usually broader. The gap between the two is where most arguments about evidence actually live, and it is almost always missing from the coverage.
What it is
Every measurement is a proxy for the thing somebody cares about. A questionnaire is a proxy for an experience. A sensor is a proxy for a behaviour. A click is a proxy for an interest.
The question is never whether the proxy is perfect. It is how much of the claim survives the difference, and whether anybody has tried to find out.