TARA preparation guide

October or January: reading the official TARA sitting data

The TARA ran twice in the 2026 admissions cycle, October and January, and a candidate may sit it only once per cycle. That makes the choice of event one of the most asked and least reliably answered questions about this test. UAT-UK does publish the raw material: the technical report prints the score distribution for each scored module split by event, and the annual report prints bookings, UK and overseas, for each sitting. This page lays those figures side by side and then does the part that matters most, which is explaining why a difference in means between the two events is not a difference in how hard the papers were.
Practise 51 TARA papers (2 free)Full TARA guide
Papers
51
Questions
2153
Free to try
2
Quick answer: how the two sittings differ
Size of each event
976 candidates in October (37%), 1,650 in January (63%)
Critical Thinking mean
4.46 in October (SD 2.06), 4.06 in January (1.82)
Problem Solving mean
4.66 in October (SD 1.81), 4.76 in January (1.74)
One scale or two
One. Both events used the same scaling constants, anchored on October at 4.5 and 7.0

Every figure on this page comes from a printed table or sentence in UAT-UK's 2025/26 TARA technical report and its 2025-26 annual report. Nothing is read off a chart image, and no number here was produced by doing arithmetic on two tables.

In short
  • Both events share one set of scaling constants, anchored on the October cohort: median ability maps to 4.5 and the 90th percentile to 7.0. Within one cycle the two events' scaled scores are built to be comparable.
  • The two modules do not move the same way. Critical Thinking means are 4.46 (SD 2.06) in October and 4.06 (1.82) in January. Problem Solving runs the other way: 4.66 (1.81) in October against 4.76 (1.74) in January.
  • January was the bigger TARA event: 976 candidates against 1,650, a 37% to 63% split. That inverts the UAT-UK pattern overall, where October remained the larger window with 63.1% of all tests delivered.
  • The gap between events is not a difficulty gap. The report states that scaled scores are comparable within the cycle regardless of which form a candidate took, and both events used the same constants, so the difference sits with who turned up.
  • Only three tables in the technical report are split by event: the score distribution, and the raw-score and scaled-score reliability and standard error. Timing, unreached items, item analysis and every demographic breakdown are published for the whole cycle only.
01

The two sittings side by side

The TARA ran in only two events in the 2026 cycle: October 2025 and January 2026. The technical report prints the score distribution for each scored module split by event; the annual report separately prints bookings for each sitting with a UK and overseas breakdown. The table below pulls every per-sitting figure from those two documents into one place. Each cell is printed in the source, and none of it is arithmetic of ours.

976 / 1,650

Candidates who sat: October / January

37% / 63%

Share of the whole TARA cycle

4.46 / 4.06

Critical Thinking mean: October / January

4.66 / 4.76

Problem Solving mean: October / January

October 2025January 2026
Tests delivered9761,650
Share of the TARA cycle37%63%
Bookings (annual report)1,0451,761
TARA's share of that sitting's bookings4.8%13.7%
Bookings: UK / overseas508 / 537874 / 887
Critical Thinking mean (SD)4.46 (2.06)4.06 (1.82)
Critical Thinking 25th / 50th / 75th2.9 / 4.5 / 5.82.8 / 4.0 / 5.2
Critical Thinking 90th percentile7.06.6
Problem Solving mean (SD)4.66 (1.81)4.76 (1.74)
Problem Solving 25th / 50th / 75th3.5 / 4.5 / 5.83.5 / 4.8 / 5.7
Problem Solving 90th percentile7.07.0
Scaling constantsShared with JanuaryShared with October

The score rows come from Table 4 of the technical report; the headcount and geography rows come from Tables 2 and 3 of the annual report. Bookings and delivered tests are different counts, so read the two blocks separately.

02

Where the gap comes from: the scale is built on October

014.5 and 7.0

Anchor on the October cohort

UAT-UK requires the scale to be fixed to the October ability distribution: median ability maps to a scaled score of 4.5 and the 90th percentile to 7.0. The anchoring abilities are 1.2529 and 2.5610 for Critical Thinking, 0.5029 and 1.8528 for Problem Solving.

02Fixed once

Fit the line, read off the constants

A regression line through those two anchors yields the constants that turn ability into a reported score: 2.1055 and 1.9112 for Critical Thinking, 3.5686 and 1.8520 for Problem Solving. The result is capped at 1.0 and 9.0.

03One ruler

January reuses the same constants

The report states it plainly: the same scaling constants were used for both the October 2025 and January 2026 events, precisely so that scaled scores stay consistent and comparable across the two events inside the 2026 cycle.

04After the window

Post-equated, so no score on the day

Both scored modules are equated after the testing window closes, with items calibrated on a Rasch model before any scaled score exists. That is why no score appears on the day, and it applies to both events equally.

05Not the paper

So the gap measures people

Given one ruler, and given that form difficulty is already removed in the conversion, a difference in means between the events can only come from a difference in who sat them. That is the one sentence worth carrying away from this page.

The single easiest sentence to get wrong here is reading the distance between 4.46 and 4.06 as a statement about how hard the October paper was. The report is direct: a raw score depends on both candidate ability and form difficulty, while the scaled score strips form difficulty out and sits on one ruler within the cycle, and both events shared one set of constants. The report's own illustration is that a candidate scoring 6.5 in Critical Thinking, in either January or October, has higher ability than one scoring 4.2. One ruler, two different means, means two different groups of people.
One further limit is just as firm: because the scaling is fixed independently for each admissions cycle, this cycle's scaled scores cannot be compared directly with those from future events. So a TARA figure quoted from a different year, however confidently, is not a target you can set yourself against.
03

Who actually sat which event

October, delivered

976

37% of the TARA cycle; 1,045 bookings

January, delivered

1,650

63% of the TARA cycle; 1,761 bookings

October, TARA's share of the window

4.8%

508 UK bookings, 537 overseas

January, TARA's share of the window

13.7%

874 UK bookings, 887 overseas

The direction of that split is itself worth noticing. Across UAT-UK as a whole, October remained the larger window: the annual report puts 63.1% of the cycle's tests between 12 and 16 October 2025. The TARA runs the other way, with clearly more candidates in January. The annual report prints a year-on-year change for every ESAT and TMUA sitting but n/a for both TARA sittings, because the TARA was introduced as a new test for 2026 entry and has no prior cycle to be measured against. The report also states that the TARA was not used by the University of Cambridge this cycle, so candidates could choose either event freely, and that a candidate may sit only once per cycle. The two groups were never randomly assigned to their events.
Cost and format are identical across the two events and give you no reason to prefer either: 75 pounds for candidates in the UK and Republic of Ireland, 130 pounds everywhere else; six days of testing per window, with each test delivered across two days; three TARA modules of 40 minutes each, 22 multiple-choice items in each scored module, and a Writing Task of one prompt from three with a 750-word limit that is not placed on the scale.
04

The same score sits differently in each event

Comparable scores do not mean identical distributions. A 6.5 means the same level of ability in either event, but where a 6.5 sits inside each cohort is not the same thing. What follows lays out the percentiles, reliabilities and standard errors that the report prints event by event, and leaves the judgement to you.
Critical Thinking percentiles
October runs 2.9, 4.5, 5.8 and 7.0 at the 25th, 50th, 75th and 90th percentiles; January runs 2.8, 4.0, 5.2 and 6.6. The means are 4.46 against 4.06 and the SDs 2.06 against 1.82. October's distribution is wider and reaches higher. Its median of 4.5 and 90th percentile of 7.0 are the two values the scale was anchored at, which is no coincidence: the constants were fitted to this cohort.
Problem Solving percentiles
October runs 3.5, 4.5, 5.8 and 7.0; January runs 3.5, 4.8, 5.7 and 7.0. The means are 4.66 against 4.76 and the SDs 1.81 against 1.74. The two events almost overlap on this module: identical at the 25th and 90th percentiles, with January higher at the median. It moves in the opposite direction to Critical Thinking.
Reliability and standard error by event
Raw-score Cronbach's alpha: Critical Thinking 0.69 to 0.79 in October and 0.69 to 0.72 in January; Problem Solving 0.69 to 0.75 in October and 0.59 to 0.72 in January. On the scaled score: Critical Thinking 0.61 to 0.77 in October and 0.69 to 0.72 in January; Problem Solving 0.69 to 0.74 in October and 0.61 to 0.73 in January. Scaled-score SEM: Critical Thinking 0.94 to 1.24 in October and 0.87 to 0.98 in January; Problem Solving 0.95 to 0.98 in October and 0.94 to 0.97 in January.
How the report explains January's 0.59
That 0.59 is a raw-score figure, on one Problem Solving form in January. The report names it as an outlier and attributes it, at least in part, to the low number of candidates on that form (263) drawn from a restricted candidate population. The remaining forms sit between 0.68 and 0.72, which it calls very homogeneous. It also notes that each TARA module carries only 22 items, so reliabilities are expected to run lower than on other UAT-UK tests: the common rule of thumb asks for 0.80 or better, but at this module length the report treats anything above 0.70 as satisfactory. Of all the per-sitting figures used on this page, this is the only one the report gives a reason for.
What the report does not split by event
Test timing, unreached and low-time items, the item and DIF analyses, and every demographic breakdown the report carries (gender, residence, first language, school type, ethnicity, free school meals, parental education, health conditions) are printed for the whole cycle only, never split into October and January. If you want to compare time pressure or the international share between the two events, the official data does not answer that, and this page will not invent an answer. For the whole-cycle timing figures, see TARA timing and unreached items.
Read the percentiles with measurement error in view. Under classical measurement theory, the report notes, there is roughly a 68% chance that a candidate's true score lies within one SEM of the observed score and roughly 95% within two. The scaled-score SEMs here run between 0.87 and 1.24 depending on module and event, while scores are reported to a single decimal place. The error band is far wider than that last digit, so treating a difference in the last decimal place as real gets you nowhere, whether it sits between two of your own practice attempts or between the two events' means.
05

So how do you actually choose

October or January

What the official data settles, and what it cannot

The data answers thisPrinted directly in the report and the annual report
  • Both events share one set of constants, so scores are comparable within the cycle
  • Headcounts, the UK and overseas split, and each module's percentiles
  • The reliability and standard error bands for each event
The data does not answer thisNo comparable per-sitting figure is printed
  • Which paper was harder: form difficulty is already removed by the scaling
  • How time pressure differed: timing is whole-cycle only
  • Any score threshold: UAT-UK publishes no pass mark
Only you can answer thisNothing to do with the data, and decisive anyway
  • What your target courses require, and by when
  • Whether your preparation is actually ready by October
  • One sitting per cycle, with no second attempt

Tap a branch to unfold

The hard constraint is that you get one attempt. The report states that candidates may sit only once within an admissions cycle, and because the TARA was not used by the University of Cambridge this cycle, both October and January were open. What should drive the decision is what your target courses require and by when, plus an honest read on whether your preparation will be ready by October. The gap between the events' means should not enter that judgement, because it describes the people who sat, not the paper you will face.
To place a score against the whole cohort first, see TARA score statistics. For how hard the clock bites in each module, see TARA timing and unreached items. Every number above comes from a printed table or sentence in those two official PDFs, and not one of them was read off a chart image. FrontierVue is an independent preparation platform and is not affiliated with UAT-UK or any university.
FAQ

Frequently asked questions

Which sitting is the easier one to score on?
The question points the wrong way. Both events used the same scaling constants and scaled scores are comparable within a cycle, so there is no easier sitting to find: the difference in means is not a difference in paper difficulty but a difference in who sat. What decides your score is how ready you are on the day.
January had more candidates. Does that make it more competitive?
January was larger: 976 candidates in October against 1,650 in January, a 37% to 63% split. But the report carries candidate counts and no university outcomes anywhere in it, so competitiveness cannot be read out of those two numbers. Treating headcount as a competition index invents a statistic the universities have never published.
Can the two events' scores be compared? What about across years?
Within a cycle yes, across cycles no. The report states that both 2026 events used the same scaling constants precisely so they would be comparable, and equally that the scaling is fixed independently for each admissions cycle, so this cycle's scaled scores cannot be compared directly with those from future events.
If October goes badly, can I sit again in January?
No. The report states that candidates are allowed to sit only once within an admissions cycle. Because the TARA was not used by the University of Cambridge this cycle, you could choose either October or January freely, but once chosen that is your only attempt, which is what makes the choice worth taking seriously.
Does the report say what score is good enough?
No. UAT-UK publishes no pass mark and the report contains no threshold of any kind, for either event. What it gives is a distribution: the Critical Thinking 90th percentile is 7.0 in October and 6.6 in January, while Problem Solving is 7.0 in both. Whether those positions become your target is a matter between you and the courses you are applying to.
SOURCES

Official sources

MORE GUIDES

Other TARA preparation guides

FrontierVUE is an independent practice platform. It is not affiliated with or endorsed by UAT-UK, Pearson, OCR, the University of Cambridge, Imperial College London, or any official admissions-test owner.

FrontierVUE 是独立的备考练习平台，与 UAT-UK、Pearson、OCR、剑桥大学、 帝国理工学院或任何官方入学考试主办方均无隶属或背书关系。

Put it into practice

51 TARA papers on FrontierVUE, including 18 Frontier Original mocks built to the current format, every question with a worked solution. The first 2 papers are free.

Practise TARA papersSee pricing