Evidence completeness scoring model

Dimension	Weight	Calculation	What it cannot establish
Current test coverage	60	Each of four tests: current ≤30 days = 15; ageing >30–90 days = 10; expired or missing = 0.	Execution quality, profitability or future test outcomes.
Linked documents	20	Five points per distinct test with an evidence document linked to an active sample. Maximum four tests.	Independent authentication or completeness of the underlying document.
Account context	10	Two points each for entity, tester country, account type, platform and deposit context.	Availability of the same account to another reader.
Sample provenance	10	Proportion of active samples naming tester, method and method version × 10, rounded to the nearest integer.	Independent verification of a tester’s identity or the claimed observation.

Dimension

Weight

Calculation

What it cannot establish

Current test coverage

Each of four tests: current ≤30 days = 15; ageing >30–90 days = 10; expired or missing = 0.

Execution quality, profitability or future test outcomes.

Linked documents

Five points per distinct test with an evidence document linked to an active sample. Maximum four tests.

Independent authentication or completeness of the underlying document.

Account context

Two points each for entity, tester country, account type, platform and deposit context.

Availability of the same account to another reader.

Sample provenance

Proportion of active samples naming tester, method and method version × 10, rounded to the nearest integer.

Independent verification of a tester’s identity or the claimed observation.

Read the four dimensions separately

A file can have fresh measurements and no published documents. A numerical total compresses this distinction, so each report publishes its components. The score falls as tests age and rises when relevant documentation is supplied.

No commercial weighting

Commercial placements cannot change the score, calculation or ordering. Ties share a position in the completeness table. The default audit register remains alphabetical.

Editorial ratings stay separate

If a desk-supplied editorial rating arrives, its value, scale and update date appear as a separate comparison row. It is never inferred from timing tests and never relabelled as a regulatory or safety score.

Limits and corrections

Completeness is not truth verification. A source can be documented and still be wrong. Corrections remain visible in the measurement log, and documents linked only to retracted samples do not count toward the score. This model uses active records at the time the page is generated.

A score for the evidence. A visible calculation.

Four dimensions. Published weights.

Read the four dimensions separately

No commercial weighting

Editorial ratings stay separate

Limits and corrections

Follow the record from every angle.