Correctness

A measure of how closely data matches reality. This is the most difficult measure of Data Quality to validate.

There is an argument that data can never be said to be correct. Like a scientific theory it is much easier to identify when a piece of data is wrong than to prove that it is correct. Tests can be defined to identify data that is incomplete, late or inconsistent, but any element that passes all the current tests could well fail the next one.

One way to overcome this is to explicitly explain the limits of our confidence. We can make statements like "this location is accurate to within 3 metres", or "the closest HTML 4 colour is yellow", or "this is the album called 'Led Zeppelin IV' in Wikipedia".

Some aspects of reference integrity can be enforced by the Physical Data Model.

Links to this page

The following pages link to here: Consistency, Data Quality, Data Steward