English Test Score Conversion - Why the Tables Mislead

Apply cluster. Mechanism only: no conversion grid, no score requirements and no fees appear on this page.

The most-searched object in English-test admissions is a score conversion table, and it is also the least reliable. This page explains why converting between English test scores is not arithmetic, how to recognise the one kind of conversion worth using, and how to choose between the four tests on mechanism rather than on a number someone put in a grid.

Why conversion tables exist, and why they mislead

They exist because institutions historically set requirements against one test and then had to deal with applicants holding another. Somebody needed a rule of thumb, so a rule of thumb was made. Then it was copied, stripped of its caveats, republished by preparation companies and agents, and eventually presented as if it were a unit conversion.

It is not a unit conversion. Converting metres to feet is exact because both measure the same physical quantity on a linear scale. Converting between English tests is none of those things, for four separate reasons that compound.

1. The tests do not measure quite the same thing

All four assess academic English across reading, listening, writing and speaking, but they weight and elicit those skills differently. One tradition uses integrated tasks, where you read and listen before writing, so a listening weakness leaks into the writing score. Another keeps them separate. One assesses speaking through a live interview, another through recorded answers, another through automated scoring of recorded answers. A candidate who is strong in conversation and weak on long audio does not have one true level that the tests estimate with error. They have genuinely different performances on genuinely different tasks.

2. The scales have different granularity

Some tests report on a short, compressed band scale where one step covers a wide range of underlying performance. Others report on a long point scale assembled from section scores. Mapping a wide step onto a narrow one loses information in one direction and invents it in the other. A single band on a compressed scale legitimately corresponds to a range on a long scale, and any table that gives you one number in that cell has thrown away the range.

3. A real concordance is statistical, not definitional

Where a legitimate concordance exists, it is produced by having a sample of candidates sit both tests and then modelling the relationship. That yields a central tendency and a spread. The spread is wide, because of reason one: individuals move around inside it depending on their skill profile. Published concordances say so. The copies circulating on the internet drop that sentence and keep the grid.

4. Most tables in circulation are nobody’s concordance

Ask of any table you find: who produced this, from what data, and when? A large share of what circulates is a third party’s reconstruction, or one publisher’s comparison of its own test against a competitor’s, which is a marketing document with a research shape. A test owner comparing its product favourably to a rival is not a neutral source about either.

How to recognise a conversion worth using

There is exactly one authoritative answer to the question of what score you need, and it is not a conversion at all: it is the requirement published by the institution you are applying to, expressed in the test you intend to sit. If the institution lists a requirement for your test, use it and ignore every table. If it does not list your test, the institution probably does not accept your test, and a conversion will not persuade an admissions office that has not published one.

Conversion is legitimate in one narrow place: your own planning, before you have chosen a test, treated as a range rather than a value. Knowing you are probably in one band, which is probably around a certain region on the other scale, plus or minus a lot, is useful for deciding whether you are close to a threshold. It is not useful for deciding to skip a retake.

The four tests, described by mechanism

Rather than a grid of numbers, here is what actually differs. Any of these can be the right choice; which one is right depends on your profile and, far more, on what is accepted.

• The live-interview academic test. Human examiner for speaking, mixed question machinery in reading, separate writing tasks. Suits people who perform better with an interlocutor and who read carefully. Widely accepted.
• The integrated academic test. Recorded speaking, long lectures and passages, writing tasks that require you to combine what you read and heard. Suits strong listeners and note-takers. Widely accepted.
• The computer-delivered, automatically scored test. Fully machine-scored across all four skills, generally with a faster turnaround. Suits people who are comfortable with recorded delivery and who want a shorter results queue. Acceptance is broad but less universal, and it varies more by country than the first two.
• The short adaptive online test. Taken remotely, adaptive item selection, much shorter, results quickly. Acceptance grew rapidly and is now common at institutions but is markedly less common with permission authorities. This is the one where acceptance, not ability, is most likely to be the binding constraint.

There are two acceptances, and people check only one

This is the most consequential thing on the page. An English test result may need to satisfy two different bodies with two different rules, and satisfying the first tells you nothing about the second.

  YOUR RESULT
      |
      +===  INSTITUTION
      |       accepts: which tests?
      |       requires: overall AND per-skill minimums?
      |       accepts: which product version?
      |       validity: measured at which date?
      |
      +===  PERMISSION AUTHORITY   (a separate rulebook)
              accepts: often a narrower list
              requires: a specific approved product
              may require: a test taken at an approved centre
              validity: its own window, its own start date

  BOTH must pass. Passing one is not evidence about the other.

The failure mode is ordinary and expensive: a candidate sits a test the university accepts, receives an offer, and then discovers the permission route requires a different product from an approved list. The remedy costs a second test and, often, an intake. Check both rulebooks before booking, not after. Our student visa requirements page explains where the second rulebook sits in the process, and how to get a student visa covers the sequence.

Choosing, in order

1. List what the institution accepts, from its own programme page, including product version and whether per-skill minimums apply.
2. List what the permission authority accepts, from the authority own page, including whether an approved centre is required.
3. Intersect the two lists. If the intersection has one member, you are finished choosing.
4. If it has more than one, choose on format fit, using the IELTS versus TOEFL comparison for the two academic formats.
5. Then check availability, because a test you cannot book in time is not in the intersection either. Put the booking on your application timeline.

Test-choice worksheet

Complete this before you pay for anything. Every line is a fact you have confirmed at a publisher, not an assumption.

Why we publish no conversion table

We could produce one in an afternoon by copying what everyone else has copied. It would attract traffic and it would occasionally cost a reader an intake, because the number in the cell is a central tendency stripped of its spread and republished as a fact. Publishing it would also contradict what this site is for: we earn nothing from where you apply or what you sit, so we have no reason to give you a confident number in place of an accurate range.

If you need a concordance, take it from a test owner own research publication, read the paragraph about the confidence interval, and treat the cell as a region. Then go and read the requirement the institution actually published, which is the only number that decides anything.

Where to go next

• IELTS versus TOEFL, by format rather than difficulty
• How to prepare for the test you choose
• Student visa requirements, the second rulebook
• How the student visa process works
• Building your application timeline
• Studying in English outside the English-speaking world
• The application document checklist
• Our editorial policy, for why this page carries no scores