An OSCE examiner training program has four parts. A curriculum built on seven learning outcomes. A delivery format that fits the examiner pool. A calibration exercise on the exam's own stations. A refresher cycle tied to changes in format or regulation. It is recorded in an examiner register. It reduces examiner variation. The evidence says it does not remove it.
What a calibration session does in the room is covered in how to train OSCE examiners. This page is about the program around that session: what it teaches, how it is delivered, how often it repeats, and how it is recorded.
Why a program and not a briefing
The AMEE guide on OSCE organization identifies untrained examiners as one of the four factors that lower an OSCE's reliability. It says training will reduce examiner variation. A briefing on the morning of the exam cannot do that work. It has twenty minutes and a room full of people who have not seen the stations. There is no time to score anything. A program is done before the day, and the briefing reminds people of it.
The program also answers a question the board and the regulator ask: who examined, and what were they trained to do? A register with names, dates and content is the answer. Without a program, there is nothing to register.
What must the curriculum cover?
The AMEE guide lists seven learning outcomes for examiner training. The table turns them into a curriculum with a delivery method for each.
Outcome | What the examiner can do afterward | How it is taught |
|---|---|---|
Scope and principles | Explain what the OSCE samples, why stations are short and many, and what the examiner's score is used for | Short pre-reading or e-learning module |
Professional conduct | Behave the same way to every candidate; say only the scripted prompts; stay neutral | Workshop discussion of scenarios |
Scoring-instrument use | Score a checklist item by item and give an independent global rating | The calibration exercise on real stations |
Written feedback | Write comments a candidate can act on, in the time available | Worked examples; a comment rewritten in the workshop |
Verbal feedback in formative exams | Give two-minute feedback that is specific and kind | Role-play, formative examiners only |
Confidentiality | Keep station content and candidate performance inside the exam | Signed agreement plus the briefing reminder |
Dangerous-behavior procedures | Recognize a critical error, record it as the policy requires, and know who to tell | Walk-through of the procedure with the form |
The middle row is the one that takes the time. The other six can be taught once and refreshed. Use of the scoring instrument has to be practiced on the stations the examiner will mark. That is why the calibration exercise is a designed component and not an optional extra.
Which delivery format?
Three formats are in use. A 2026 scoping review of 20 studies describes the common shape of the calibration part.
- Face-to-face workshop. Half a day. Examiners score recorded performances, compare marks, and discuss the differences with a facilitator. Best for a new examiner cohort and for a new exam format.
- Online module. Pre-reading, the principles, the instruments, and benchmark videos scored alone with feedback. Cheap to scale. Used on its own, it teaches the instruments rather than the standard.
- Blended. An online module and benchmark marking before a facilitated workshop. The workshop then focuses on the discrepancies rather than the basics. This is the design most large programs converge on.
The scoping review found two kinds of calibration. General calibration aligns examiners' internal standards. Case-orientation calibration is tailored to the stations they will mark on the day. Most interventions used frame-of-reference training: video benchmark performances at several levels, scored and then discussed. A program needs both kinds. General calibration once, case orientation before each sitting.
What is the calibration exercise?
The designed part. Its inputs are the exam's own stations and the marking scheme with its anchors. It also needs recorded or acted performances at three levels, including a borderline one. The exercise runs the same way each time.
- Examiners score a performance alone, checklist then global rating.
- The facilitator shows the spread of marks. Not the average: the spread.
- The group discusses the items where marks differed and agrees what the anchor means.
- Examiners score a second performance. The spread is shown again.
- The facilitator records the session: who attended, which stations, how far the spread closed.
The record is the point. A school that can show the spread closed on the borderline performance has evidence for the register. A school that can show it did not close has a reason to look at the anchors. That is a marking scheme problem, not an examiner one.
How often does training repeat?
The AMEE guide ties refresher training to change: a new format, a new instrument, a regulator's new requirement. To that add time. A program sets a period. An examiner who has not examined within it is retrained before they examine again. The program also sets case orientation before every sitting for every examiner, however experienced.
The register holds the cycle: examiner, initial training date and content, refresher dates, case orientations attended, sittings examined. The recruiting page uses the same register to plan the pool.
What can training not do?
The honest answer belongs in the program document because the board will otherwise expect more than training can deliver. The 2026 scoping review found the evidence on examiner training to be "limited and inconsistent". Some studies reported modest gains in reliability or accuracy. Others reported no change, or examiners who became more stringent after training. Where benefits appeared, they were most evident for borderline performances. That is the useful finding: training helps most where the decision is hardest.
Training does not turn a hawk into a dove. The examiner variance page carries the numbers on how much examiners move results. It says what an organizer does about the part training leaves. The short version is design: enough stations that no examiner dominates a candidate's result, decisions at exam level, and monitoring after the day.
The program checklist
Limits
The evidence base is small and mostly from medicine. The scoping review's 20 studies used varied designs and often reported the intervention too thinly to judge. A 2024 nursing study comparing simulation, online, and lecture training found simulation to be best, with 18 raters in all. Read these as guidance on shape, not as proof of effect size. The program's own before-and-after spread is the evidence that matters for the school running it.
Frequently asked questions
What should OSCE examiner training cover?
Seven learning outcomes from the AMEE guide. The scope and principles of the exam. Professional conduct. Use of the scoring instruments. Written feedback. Verbal feedback in formative exams. Confidentiality. The procedures for dangerous behavior. Instrument use is taught at the exam's own stations.
How long should examiner training take?
A blended program: an online module of about an hour, then a half-day workshop. Before each session, a shorter case orientation. Workshop time goes to scoring benchmark performances and discussing the spread of marks, not to lectures.
Does examiner training actually work?
It reduces examiner variation. It does not remove it. A 2026 scoping review of 20 studies found the evidence limited and mixed. Some studies showed modest gains and others none. Gains were clearest for borderline performances. Design and monitoring cover the rest.
How often should OSCE examiners be retrained?
After a change in the format, the instruments, or the regulator's rules, as the AMEE guide advises. Add a time rule for examiners who have not examined for a set period. Run case orientation before every sitting for everyone. The register records all of it.
What is frame-of-reference training for examiners?
A calibration method. Examiners score video benchmark performances at several levels. They compare their marks with the intended standard. Then they discuss the differences with a facilitator. It was the most common approach in a 2026 scoping review.