Validity and Fairness Evidence | The Rigors

Current validity register

Inspect Standard v1
Review governance

Define the question before seeing the result.

Before a statistic is produced, the operating owner records the question, population, denominator, included and excluded records, standard version, measure, collection method, missing-data rule, subgroup protection, minimum disclosure threshold, reviewer, analysis date, limitation, and publication decision. A target pass rate or marketing need cannot choose the method after results are known.

• scoring agreement and the kinds of criteria producing material reviewer divergence;

• stage completion time, abandonment, defects, and accommodation experience;

• correction, appeal, recusal, integrity-review, and reassessment patterns;

• protected fairness review appropriate to lawful, consented, sufficiently sized data;

• the relationship between assessment evidence and later client-specific review; and

• only with an adequate, permissioned record, whether any assessment signal relates to later delivery evidence.

Small groups are suppressed or combined under an approved privacy rule. Country, nationality, accent, Academy status, and unnecessary accommodation or diagnosis information remain excluded from merit scoring.

Publish limits beside results.

Any future result must show the period, denominator, version, measure, limitations, owner, review date, and correction path. A percentage without its denominator or a comparison without a defined population does not publish. No “top X%,” selectivity, performance, equality, or predictive claim is available today.

The Rigors is modeled on the structure of ISO/IEC 17024, the international standard for bodies that certify persons. IT Modality is not accredited to it and does not claim to be.