A cohort larger than one circuit is split across sessions, parallel circuits, or days. Each split creates a group that sat a different exam in some way: at a different time, with a different set of examiners, or at a different site. The plan has to keep those groups comparable, and then check that they were. AMEE Guide 49 puts the check as a number. Variance between groups should stay below 30 percent of the total.
When does a cohort need more than one circuit?
One circuit holds one candidate per station. A 14-station circuit assesses 14 candidates per lap. A lap of 14 slots at 10 minutes runs about two and a half hours. A cohort of 150 needs eleven laps, and that does not fit in a day. The organizer has three ways to add capacity. Each has something to buy and something to cost.
Design | Adds | Costs | The question to check |
|---|---|---|---|
More sessions on one circuit | Capacity within a day | Time; later sessions start tired | Did the afternoon cohort know more than the morning one? |
Parallel circuits | Capacity within a session | A second set of examiners and patients; the same equipment twice | Did circuit A mark like circuit B? |
More days | Capacity without more rooms | Station exposure; a second run-sheet | Did day two candidates know the stations? |
The AMEE guide on OSCE organization plans for all three. Its tasks before the day include sending candidates to centers and providing transport instructions. It requires that the later cohort is on site before the earlier one leaves. The venue page covers what a parallel circuit needs in rooms and equipment.
How are candidates allocated?
Four rules cover most schools.
- Draw at random within the constraints that matter. Adjustments, declared conflicts with an examiner or a patient, and travel from another site are constraints. Everything else is random. A hand-drawn list invites the claim that one group was favored.
- Publish the list late. Candidates need their session time, not weeks in which to trade it. Two or three days is enough.
- Keep the later cohort naive. The later cohort registers and hands in devices before the earlier cohort leaves quarantine. The security page sets out the timing.
- Record the group with the results. Session, circuit, day and site are fields on every candidate's record. The check after the exam needs them, and so does an appeal.
How are examiners allocated across circuits?
Examiners are the largest source of difference between groups. A study of a single exam across sites found pass rates between 68.8 and 75.9 percent for candidates with the same ability. The difference was which examiner cohort they met. Homer's analysis of 442 sittings put examiner strictness at about 16 percent of the variance in domain scores. Both are on the examiner variance page.
The plan can spread that effect rather than concentrate it. Where two circuits run in one session, the same station has a different examiner on each. Where a school runs several sessions in a day, examiners can swap circuits between sessions. Then no group meets only the strict examiners. When sites differ, one examiner can score recorded performances from both sites. That gives the analysis a link between them. None of this removes the effect. It stops it landing on one group.
What is the check?
After the exam, before the pass mark is set, the analysis compares the groups. AMEE Guide 49 names the metric. The variance between groups at a station should be below 30 percent of the station's total variance. Above that, the groups did not sit the same station. The cause may be the examiners, the patients, or the timing. The quality section covers the metric in full.
The plan makes the check possible. It needs enough candidates per group for the numbers to mean anything. A circuit with six candidates on it is a problem in itself. It needs the group fields recorded. And it needs the rule agreed in advance: what the school does when a station fails the check. The usual answers are to decide at the exam level rather than at the station level, or to withdraw the station. The scoring section covers both.
How many days?
Every extra day exposes the stations for another night. The evidence on whether that changes scores is mixed, and the security page reviews it. The choice is between three positions. Run the same stations and accept the exposure, with quarantine and the check. Run different stations on each day and accept a harder equating problem. Or add a parallel circuit instead of a day. The third is usually cheapest in validity and dearest in staff.
A planning table
One row per group, before the day, is enough to see whether the plan can be checked afterward.
Group | Day | Session | Circuit | Candidates | Examiners | Patients | Check possible? |
|---|---|---|---|---|---|---|---|
A | 1 | Morning | 1 | 14 | 14 + 2 reserve | 9 + 2 reserve | Yes, against B |
B | 1 | Morning | 2 | 14 | 14 + 2 reserve | 9 + 2 reserve | Yes, against A |
C | 1 | Afternoon | 1 | 14 | Swapped from circuit 2 | Same, rested | Yes, against A and B |
D | 2 | Morning | 1 | 6 | Day 1 examiners | Same | Too small alone; pool with C |
The last row is the common failure. A small final group on a second day costs a run-sheet, a day of staff and a night of exposure. It gets six candidates whose scores cannot be checked on their own. Merging them into an earlier session, or into a parallel circuit, is usually the better plan.
Frequently asked questions
How many candidates can one OSCE circuit assess?
One per station per lap. A 14-station circuit assesses 14 candidates per lap. The laps per day depend on the slot length and the changeover time. Beyond that, add a parallel circuit, a session, or a day.
What is a parallel circuit in an OSCE?
A second copy of the same stations is run at the same time with its own examiners and patients. It doubles capacity within a session. Both copies need the same equipment. Their scores are compared before they are pooled.
How should candidates be allocated to OSCE sessions?
At random, within the constraints that matter: adjustments, declared conflicts, and travel. Publish the list two or three days before. Record the session, circuit, day, and site on each result.
What is the 30 percent rule in an OSCE?
AMEE Guide 49 advises that the variance between groups for a station should stay below 30 percent of the station's total variance. Above that, the groups did not sit the same station. The exam board needs a rule for what to do.
Is a two-day OSCE fair?
It can be, with quarantine, random allocation, and a check that day two scores did not rise. The evidence on leakage is mixed. A parallel circuit instead of a day avoids the question, at the cost of more staff.