How TMUA Scoring Works Since 2024: Raw Marks, Equating, 1–9 Scale

Since October 2024 your TMUA score is not a fixed conversion of your raw mark out of 40, and none of the UAT-UK pages and documents we read prints a table that turns one into the other. It counts your correct answers, uses the Rasch model to place every version of the test on one scale, then sets the scale so the median October candidate scores 4.5 and the 90th percentile scores 7.0. A raw mark only becomes a score once a whole sitting has been analysed. This guide walks through each step from UAT-UK’s own documents, shows the published scaling constants, and explains what the old 2017–2020 conversion tables did and did not show. ProSyllabus is an independent study site, not affiliated with UAT-UK or any university. UAT-UK’s pages and documents were read on 23 October 2026; the university pages and UAT-UK’s past-paper listing on 18 October 2026.

UAT-UK TMUA page, Test Results page and TMUA Technical Report 2025-26, read 23 October 2026.

UAT-UK’s technical reports print the two ability values each scale was fixed on and the line that follows. Ability (theta) runs from about −3.0 to +3.0, centred on zero.

The line is steeper in 2024/25 (1.7998 per unit of ability) than in 2025/26 (1.6369). An ability of 0 scored 4.5 in the first cycle and 4.6 in the second; an ability of 2.0 scored 8.1 and 7.9. Those are our sums on the published numbers. The point they make: even with the ability known, the score depends on the cohort that fixed the scale.

The ability values in the table are not raw marks. The technical reports say that, following item analysis, the item difficulties are used to generate a raw score to theta (or ability) table, but the reports we read do not print it, so we found nothing that allows an official conversion of a raw mark into a score.

From 2017 to 2020 Cambridge Assessment Admissions Testing printed a conversion with each year’s answer keys (headed “Grade conversion” in 2017–2019 and “Score conversions” in 2020): Paper 1, Paper 2 and overall (it also printed one for the 2016 practice papers). Here is the overall column at seven raw marks, exactly as printed. It is history: the 2019 Explanation of Results says its scale was designed so that approximately one third of candidates would score above 6.5.

Two things stand out. Up to a raw mark of about 24 the four years are almost identical. Above it they split: 32 out of 40 was the maximum 9.0 in 2017 and 2018 but 7.5 in 2019 and 2020. The same raw mark meant something different in different years, which shows why a conversion should not be assumed to hold for another sitting. The 2019 scale was designed to put about a third of candidates above 6.5; on today’s scale about one candidate in ten scores above 7.0.

UAT-UK’s specimen and sample tests on the Pearson site mark each answer right or wrong but give no score. UAT-UK: “This is intentional because the score scale in any particular year is based on the ability distribution of the cohort, with the median fixed at 4.5 and the 90th percentile fixed at 7.0.” It adds that “performance on a practice test will not necessarily be indicative of performance under exam conditions.”

That is how our TMUA practice tests report a mock: correct answers by paper and by topic, with the working for every question, and no invented 1–9 score. Five questions are free to try. One of them:

The reports explain the standard error of measurement (SEM) this way: there is about a 68% chance that a candidate’s “true” score lies within one SEM of the score reported, and about 95% within two. Our rough reading, not the report’s: with an SEM near 0.6, two scores a few tenths apart say little about who is stronger, while two scores a point and a half apart say much more.

To read your score once you have it, use our TMUA percentile checker; for what each university publishes, see TMUA scores by university.