Fifty-one sittings of an entrance exam, assembled from eight FOI answers
The regulator published one pass rate a year, for three years. Its replies to individual requests give every sitting from 2013 to 2016 — 1,663 candidates at Level 1, 426 passes, and a rate that swings from 0% to 68% between one month and the next.
By Rose Traviss · 29 September 2026
Nobody joins the regulated immigration advice scheme without passing the Level 1 competence assessment. The regulator’s annual reports gave a pass rate for it in three years — 25%, 36%, 38% — and then stopped.
Individual people kept asking. Reading every Freedom of Information request made to that regulator turns up eight separate answers about assessments, and between them they give 51 individual sittings across 2013 to 2016.
| Level | Sittings | Sat | Passed | Rate | Covering |
|---|---|---|---|---|---|
| Level 1 | 35 | 1,663 | 426 | 26% | Aug 2013 – Oct 2016 |
| Level 2 | 13 | 235 | 99 | 42% | Aug 2013 – Aug 2016 |
| Level 3 | 3 | 19 | 4 | 21% | Aug 2013 – Dec 2013 |
These are not every sitting in the period — they are the ones somebody thought to ask about. Where the coverage overlaps a year the regulator did publish, the two agree closely: 26% here against a published 25% for Level 1 in 2015/16.
The rate is not stable
The striking thing is not the average. It is that the same assessment, administered by the same regulator to a self-selected pool of applicants, produced wildly different results from one month to the next.
| Sitting | Sat | Passed | Rate |
|---|---|---|---|
| August 2013 | 17 | 0 | 0% |
| June 2014 | 48 | 5 | 10% |
| November 2014 | 98 | 10 | 10% |
| March 2014 | 45 | 19 | 42% |
| October 2014 | 52 | 22 | 42% |
| June 2016 | 23 | 11 | 48% |
| September 2016 | 38 | 26 | 68% |
Nine of the 35 Level 1 sittings had a pass rate under 15%. Four were above 45%. In August 2013, seventeen people sat the assessment and none passed.
A four-fold swing between sittings of the same exam is the sort of thing an awarding body is normally expected to explain, and none of these documents explains it. The regulator’s own commentary in its annual reports treats the low rate as a problem for applicants — “disappointing”, with steps taken “to encourage increased success” — rather than as a question about the assessment.
What this took
Four different shapes, from eight requests:
- A table inside a Word document. “OISC Level 1 Pass rates March 14 April 15.doc” — thirteen rows of date, candidates, passes. Converted to plain text it comes out as 66 characters: its own title. The table is simply gone. Converting to HTML instead keeps it.
- Prose inside another Word document. “Competence Assessments 2013.doc” gives each month in a sentence: “October 2013 — Level 1 Assessments only / Level 1: 41 candidates — 8 passes.”
- Figures typed into a reply, in two house styles. “January 2017: 37 sat
- 11 passed” from one officer; “January Level 1: 45 took the test, 3 passed” from another, with no year at all — that one takes its year from the date of the request.
Two mistakes worth recording
Attributing a figure to the right level got it wrong twice, in opposite directions, and both times the numbers were real and only the heading was false.
The first attempt searched backwards from each figure for the word “Level” and defaulted to Level 1 when it found none. That put five sittings from a file called OISC Level 2 Assessment results under Level 1.
Fixing that by reading the file name introduced the reverse. The name
arrives URL-encoded: OISC%20Level%201%20Pass%20rates.doc. A pattern loose
enough to step over the % reads the 2 of %20 as the level, and
thirteen sittings from a Level 1 table came out as Level 2.
Both are now pinned by tests, and nothing is assumed: a figure whose level cannot be established from an explicit marker is recorded as unstated rather than guessed. None currently is.
What it says about the published record
The regulator holds this data at sitting level. It answered eight separate people with it, in eight different formats, over four years. It has never published it.
That is worth separating from a refusal. Nothing here was withheld — it was given, promptly, to everyone who asked. It simply was not published, so it exists only as correspondence attached to eight request pages on a third-party website, in Word documents that lose their own tables when converted, findable only by someone who knows to look.
The one thing that was refused is more telling. Asked for monthly application and approval figures from January 2013, the regulator said the work would exceed eighteen staff hours because “a new program would need to be set up and run to extract this information from our database” — and that “this database cannot tell us the number of applicant advisers for 2013 and 2014”. That is not a decision to withhold. It is a statement that the number does not exist in a form anybody can produce.
The figures
Read from eight requests to the Office of the Immigration Services Commissioner on WhatDoTheyKnow, collected 29 September 2026, each linked with the figures it produced on sources and licences. Pass rates are computed from the candidate and pass counts in each answer, never quoted from a percentage the answer states.