Internal verification, from brief approval to the signed log
Internal verification is the centre's own quality gate: a second, trained pair of eyes on briefs before learners see them and on assessment decisions before they're finalised. Done well, it catches problems while they're cheap to fix. Done badly — or late — it turns a class set into a June crisis. Here is the whole job.
Stage 1 — Verifying assignment briefs (before delivery)
Before a brief goes live, the IV checks it against the unit's specification:
- Fit: does the brief cover all the learning aims and criteria it claims to, with no criteria smuggled in and none missing?
- Sufficiency: can a learner actually generate the required evidence in the scenario given — is the data there to analyse, the case there to evaluate?
- Validity window and version: correct PSA version (V1/V2), current cycle — see the PSA guide.
- Accessibility: is the language readable for the cohort, including second-language learners in international centres?
A brief that fails here fails cheaply — you fix a paragraph instead of re-assessing 28 scripts in March.
Stage 2 — Verifying assessment decisions (after marking)
Once the assessor has marked, the IV samples. For each sampled learner:
- Follow the criterion trail. For each criterion the assessor says is met: does the cited evidence exist, and does it satisfy the verb at that level?
- Check the record, not the memory. The assessment record must show per-criterion decisions with evidence locations. A record that summarises ("overall a good project") verifies nothing.
- Confirm band arithmetic. The unit grade follows the all-or-nothing band rule — the IV re-applies it to the verdicts.
- Spot-check authentication where it's a concern — drafts, version history, in-class evidence.
Sampling: how much is enough?
There is no universal number; sampling must be risk-based and recorded. The pattern most centres converge on:
| Situation | Sensible sample |
|---|---|
| First round with a new assessor | Large — up to 100% until trust is established |
| Experienced assessor, established unit | 10–25%, always including borderline and all-Distinction scripts |
| Any resubmission | Every one, every time — high-risk by definition |
| Before standards verification | Refresh across assessors so records look alike |
Two rules regardless of size: sample every criterion across the sample (not just the first learning aim), and always include at least one script the assessor found hard to decide.
The IV record
Every verification leaves a log: what was sampled, what was checked against what, what was agreed, what was returned to the assessor and why, and sign-off dates. This log — not anyone's confidence — is what a Standards Verifier reads as evidence your centre self-regulates. See what a good log entry contains.
The Lead IV's job
The Lead Internal Verifier owns the assessment standard across the programme: standardisation meetings, resubmission authorisations (theirs alone), verifier allocation, and the centre's assessment plan. The role deserves its own page — see the Lead IV guide.
Where checking tools fit
IV time is scarce, and the most mechanical part of stage 2 — confirming every criterion against cited evidence for every sampled script — is exactly what a criterion-level checker produces. Teams run checkb.tech on the whole batch, use its per-criterion verdicts and quotes as the initial screen, and spend human IV time on the disagreements and borderline calls. The triage queue is built for that pass, and the export lands as verification-ready paperwork.
FAQ
Can the assessor also be the IV for their own unit?
No — the verifier must be different from the assessor whose work they verify. Small departments cross-verify across subjects.
Does IV happen before or after learners get feedback?
Decisions are verified before they're finalised — normally before formal feedback and records are locked. Informal feedback during the process is fine and encouraged.