An examination technical file is the controlled evidence set from which the examination body can explain, operate, evaluate and change a named examination. It is an indexed system of current records, not a single report prepared for an audit.
Identify the product and intended decisions
State the examination owner, product name, current versions, languages, skills, levels, target populations, delivery modes, territories and intended score interpretations and uses. Distinguish the technical subject from related products, preparation courses and services that are outside the claim.
Connect construct, framework claim and specifications
Define the target-language-use domain and the knowledge, skills and processes scores are intended to represent. Specifications should translate that construct into content coverage, task types, response conditions, timing, weighting, scoring rules and form-level constraints. Any framework-related level claim needs its own maintained rationale and evidence.
Control development, review and form assembly
Retain item and task briefs, writer and reviewer competence, review criteria, sensitivity and accessibility checks, pretesting or trial evidence, item decisions, bank status and approvals. For every released form, show how content and statistical constraints were met and how exposure, reuse and retirement were controlled.
Document administration and security as an end-to-end chain
Map responsibilities from registration and identity checks through secure distribution, venue or platform controls, invigilation, access arrangements, incident handling, return of materials, data transfer and retention. Include partner and centre controls, authorisation, training, monitoring and escalation where delivery is distributed.
Make scoring and results reproducible
Record scoring keys and rubrics, marker or rater selection, training and standardisation, allocation and double-marking rules, quality indicators, adjudication, score calculation, grading or cut-score decisions, result checks and release authorisation. Keep a traceable route from a candidate response to the issued result and any later correction or appeal outcome.
Maintain the validity and fairness evidence chain
Bring together content-review evidence, response-process studies, item and task performance, rater behaviour, reliability or generalisability, measurement error, classification consistency, external relationships, subgroup outcomes, access data, consequences, complaints and appeals. State samples, uncertainty and limits, then record the decisions taken in response.
Control versions, changes and the master index
The master index should name each record, owner, approved version, location, access restriction, retention rule and relationship to the examination version. A change log records rationale, impact analysis, approval, implementation date and revalidation needed. Superseded evidence remains retrievable where it supported historical results or decisions.