What Is a Good TARA Score? Official Data From Oxford's New Admissions Test

The TARA, the Test of Academic Reasoning for Admissions, is a computer-based test that Oxford and UCL use for some undergraduate courses from 2027 entry. Its owner, UAT-UK, sets no pass mark and publishes no grade boundaries. Each of the two scored modules, Critical Thinking and Problem Solving, is reported from 1.0 to 9.0 with no overall score, and the scale is built so that the median candidate scores about 4.5. The 2026 sitting runs from 12 to 16 October, and UAT-UK's first technical report, published in July 2026, gives the first real statistics. This guide sets out what they show, how accurate a score is, and why a TARA score is not an IQ.

What is the TARA and who uses it?

UAT-UK was set up in 2023 as a joint venture between Imperial College London and the University of Cambridge, and delivers its tests through Pearson test centres. The TARA was offered for the first time in October 2025. For 2027 entry, UAT-UK lists it for Oxford and UCL only; Cambridge does not use it. Oxford's own list covers Economics and Management, History and Economics, History and Politics, Human Sciences, Philosophy, Politics and Economics (and its Foundation Year), Experimental Psychology, and Psychology, Philosophy and Linguistics. UAT-UK's course list adds UCL computer science, mechanical engineering and robotics degrees. Several tutoring sites say the TARA replaces Oxford's Thinking Skills Assessment (TSA); we could not find that stated on Oxford's admissions-tests page, which labels the TARA as new for 2027 entry.

What is in the test, and when is it sat?

• Critical Thinking: 22 multiple-choice questions in 40 minutes, reasoning with everyday written language
• Problem Solving: 22 multiple-choice questions in 40 minutes, reasoning with numerical skills; no calculator, and basic maths is assumed
• Writing Task: one essay question chosen from three, up to 750 words, 40 minutes

Marks are for correct answers only, so a wrong answer costs nothing. The Writing Task is not scored by UAT-UK: a copy goes to each university, which uses it as it sees fit. The October 2026 window is 12 to 16 October, booking closed on 28 September, and results arrive on 16 November 2026. A second sitting runs from 4 to 8 January 2027 with results on 8 February, although Oxford says its undergraduate applicants must take their test in October 2026. The fee is £78 in the UK and Republic of Ireland and £133 elsewhere. Twenty-two questions in 40 minutes is about 109 seconds each, our arithmetic, and the test is fast: in the first cycle 8% of candidates left at least one Problem Solving item unreached, against 1% in Critical Thinking, and over 85% used between 35 and 40 minutes.

How is a TARA score calculated?

UAT-UK uses the Rasch item response model, which places every candidate on one ability scale whichever form they sat. The scores are post-equated at the end of each window, which is why no score arrives immediately. The reported scale is then anchored to the candidates themselves: the median candidate is fixed at 4.5 and the 90th percentile at 7.0, and scores are capped at 1.0 and 9.0. UAT-UK says typical candidates will score around 4.5 and that approximately 10% will score above 7.0. Because the scale is set from each admissions cycle, the report says scores from one cycle cannot be compared directly with another. A TARA score is therefore a rank within that year's cohort, not a fixed amount of reasoning. One inconsistency on UAT-UK's own results page is worth knowing: one passage fixes the 90th percentile at 7.0 and another says 9.0, while the technical report and the Explanation of Results both say 7.0.

What did the first 2,626 candidates score?

The technical report covers the 2025-26 cycle: 976 candidates in October 2025 and 1,650 in January 2026. Its scaled-score statistics for all 2,626 candidates are below, Critical Thinking first and Problem Solving second.

• Mean: 4.21 (standard deviation 1.92) and 4.72 (1.77)
• 25th percentile: 2.8 and 3.5
• Median: 4.0 and 4.7
• 75th percentile: 5.4 and 5.8
• 90th percentile: 6.7 and 7.0

The medians sit a little below and above the design target of 4.5 because the scale was set on the October sitting, where the Critical Thinking median was 4.5; January candidates scored lower in Critical Thinking (mean 4.06 against 4.46). UAT-UK publishes the full distributions only as unlabelled bar charts. Read by eye, about 8% of Critical Thinking candidates sit at the floor of 1.0, and the tallest bars are at 4.5 (about 12%) and 3.0 (about 11%), so the curve is lumpy, not smooth. With 22 questions there are only 23 possible raw scores, so results bunch on a limited set of values. The charts also show scores at 7.5 in both modules, so a tutoring site's claim that nobody scored 7.5 can describe only the first sitting.

How accurate is a TARA score?

Less accurate than most candidates assume, and UAT-UK says so. The technical report gives raw-score reliabilities (Cronbach's alpha) of 0.69 to 0.79 for Critical Thinking and 0.59 to 0.75 for Problem Solving, and notes that a reliability above 0.70 is still satisfactory for modules of 22 items, though the usual rule of thumb is 0.80. The standard error of measurement of a scaled score is 0.87 to 1.24 points in Critical Thinking and 0.94 to 0.98 in Problem Solving. Under classical theory, the report explains, the true score lies within one standard error about 68% of the time and within two about 95% of the time.

On a 9-point scale with a standard deviation of about 1.8 or 1.9, a standard error near 1 is roughly half a standard deviation. Take a reported Critical Thinking score of 5.4, the 75th percentile of the first cohort. Plus or minus one point spans 4.4 to 6.4, which by interpolating the official percentiles is about the 57th to the 87th percentile. Two candidates on 5.0 and 6.0 are therefore not reliably different. For comparison, our guide to the IQ score margin of error puts the standard error of a full-scale adult IQ at about 2 to 3 points on a scale with a standard deviation of 15, which is 0.13 to 0.2 of a standard deviation. The TARA is far shorter than an IQ battery, and it is not designed to rank candidates finely.

How does the TARA scale compare with the old TSA scale?

Oxford's earlier Thinking Skills Assessment reported Section 1 on a scale of 0 to 99, which Cambridge Assessment said was designed so that typical applicants to the most selective UK courses scored around 60, with 70 a comparatively high score and only a few very exceptional applicants above 80. The two scales are not interchangeable. A TSA 60 described typical applicants to selective courses; a TARA 4.5 is the median of that year's candidates, and 7.0 marks the top tenth.

Is the TARA an IQ test?

UAT-UK never calls it one. In the documents we read, the TARA is described as testing generic thinking skills and academic reasoning, and its stated purpose is to judge a candidate's potential on a demanding degree. The Oxford PPE department describes it as a test of verbal, spatial and numerical reasoning plus an essay. That overlaps with the content of IQ-style reasoning tests, but a TARA score cannot be turned into an IQ. The candidates are already a selected group, the score is a rank among them, and the scale resets each year. Our guide to achievement tests and IQ tests explains the wider distinction, and our guide to the average IQ of MIT, Oxford and Yale students covers what published scores imply about selective admissions.

Nor is there yet evidence that the score predicts degree results. The 2025-26 technical report covers reliability, timing and item statistics, and bias checks, but no validity study. An older Cambridge Assessment report on the predecessor TSA described its correlations with university examination results as weak, from very small samples. The technical report does record group differences: men outscored women in both modules, by 0.27 points in Critical Thinking and 0.94 in Problem Solving, and candidates whose first language is English scored 0.95 points higher in Critical Thinking but only 0.11 higher in Problem Solving, which UAT-UK attributes to the module's greater language load.

What should candidates do with a score?

Treat it as one input. Oxford says tests are used alongside a range of other information, and UAT-UK says most universities will use results alongside the rest of the UCAS application. Neither publishes a cut-off, and we found no statement of how Oxford uses the Writing Task. Compare the two module scores with the percentile table above, remember the one-point margin of error, and do not read a half-point gap as a difference in ability. To try similar reasoning formats, our critical thinking test and logical reasoning test are the closest on this site. Our guide to Watson-Glaser scores explains a comparable critical-thinking assessment, and our guide to UCAT scores explains another admissions-test scale.

To see how an IQ-scale result compares with the population, take the IQ Metrics test and read it against the percentile calculator.

Common questions