TOOLKIT › TARA, LNAT, TMUA
STREAM LEARNING — ADMISSIONS TEST COURSES
TARA · LNAT · TMUA
2019
850
100%
THE ADMISSIONS TEST COURSES — TAUGHT SINCE 2019
Every concept taught. Every past paper solved.
We offer programmes for TARA, LNAT, and TMUA university admissions tests: complete courses include every concept in the specification taught on video, past paper solution walkthroughs, and original papers. The programme grew out of the TSA community that has run since 2019 — the largest of its kind — and it is taught by the people who write this site's original papers.
ON THIS PAGE
Section 2 essay — graded
"The story of ________, a ruler draped in gold, whose pilgrimage would weave chaos into Egypt's economy was what sparked my curiosity into the interplay between history, society, and economy."
It's unnatural to read a story about a rich ruler and immediately think "history, society, economics" are interconnected — not very organic.
Where is this analysis meant to have happened? Your content and conclusions are totally separate from each other.
How work is marked — the rubric ↓THE TEACHERS
10 years of teaching the papers
SJ read Economics & Management at Oxford, teaches the TARA programme and reasoning for LNAT.
Joe read Engineering at Imperial College London, leads TMUA and writes its original papers.
THE COMMUNITY
Open since 2019. Free.
The community began in 2019 and is now the largest TSA community. It now continues into TARA and TMUA. Membership is free.
1,200
MEMBERS EVERY YEAR
850
SOLUTION VIDEOS
100%
RELEASED PAPERS COVERED
TSA ultimate last-minute guide — Oxford
2022 · from the archive
TSA Problem Solving
2023 · from the archive
The Ultimate TSA Guide
2025 · from the archive
850 solution videos, 2019 to today — watch the full archive on YouTube →
THE COURSES
Concept galleries, solved archives, original papers
CONTENTS
- CONCEPT GALLERY
- Every concept in the published specification
- ORIGINAL PAPERS
- Written in the live three-module format
- SOLVED ARCHIVE
- Sixteen years of TSA papers
- WRITING TASK
- Marked against the published rubric
The TARA course — the first complete programme for the test
Every concept in UAT-UK's published specification has its own video; original papers are written in the live three-module format; and the writing task is taught against a published rubric — the same rubric used to mark submissions inside the course. Because UAT-UK built TARA deliberately close to the TSA, sixteen years of TSA past papers remain the closest practice available, and all of them are solved in the archive.
FREE — NO ACCOUNT NEEDED
The TARA writing rubric
A rubric is included for you to grade your writing against ten criteria.
Baseline: a 5 is a competent Year 12 writing under timed conditions — forty minutes, 750 words. It addresses all three parts, makes a clear argument, uses some specific evidence, and is organised. It does not have to be sophisticated to score 5.
CANDIDATE SELF-ASSESSMENT
Mark your own paper
Write the task under timed conditions first — forty minutes, no notes. Then complete the diagnostic below before you look at a single criterion.
STEP ONE — THE DIAGNOSTIC · ANSWER IN WRITING BEFORE YOU SCORE
01What were the three parts you were asked for?
02Did you answer that question, or an adjacent one?
03State your core argument in one sentence. If you cannot, that is the finding.
04Is that argument sound? Name the flaw if not — circular, unsupported leap, false premise.
05If you used a technical concept, is it applied correctly? Misuse is a logical flaw, not a knowledge gap.
06Which of the three parts are developed — not merely mentioned?
STEP TWO — TEN CRITERIA · CLICK THE BAND YOU HONESTLY SIT IN
| CRITERION | 8–10 | 5–7 | 3–4 | 1–2 | /10 |
|---|---|---|---|---|---|
| 01 Task completionAll three parts, at proportional depth? | All three substantive; the word count is distributed proportionally; no part is an afterthought. | All three present, but one is noticeably underdeveloped. | One part barely addressed, or two lack real substance. | A part is missing, or the three-part structure is not attempted. | – |
| 02 Engagement with the statementThe specific wording, or the general topic? | Key terms unpacked and interrogated; every argument stays tethered to the precise claim. | Some terms addressed, others taken at face value; parts drift to the topic. | Mostly the topic area, with occasional reference to the statement. | Could have been written without reading the statement. | – |
| 03 Counter-argument qualityHow many real lines of attack, how developed? | Three or more distinct lines hitting different aspects; the strongest version is granted first. | Two clear lines, or three that overlap; may attack a weaker version. | One argument, restated; assertion instead of reasoning; possibly a straw man. | No coherent argument against — disagreement without reasoning. | – |
| 04 Use of examplesConcrete, and doing work for the argument? | Named events, laws, cases or data, each tied explicitly to the point it supports. | Present but tending to the generic; the link is implied rather than stated. | Vague, or used as the argument itself; "a study" with no specifics. | Absent, wholly generic, or irrelevant to the arguments made. | – |
| 05 Nuance and balanceIs the complexity genuinely acknowledged? | The force of the opposing case is granted; concessions are real; the view is qualified. | Some awareness of complexity, but concessions feel token — raised, then dismissed. | Largely one-sided; claims made in near-absolute terms. | Entirely one-sided; nothing qualified anywhere. | – |
| 06 Clarity, precision, economyDoes every sentence earn its place? | Direct and specific; abstract claims are followed by concrete clarification; no filler. | Generally clear, with some vague phrasing, filler or redundancy. | Frequent vague or circular statements; noticeable padding and repetition. | Pervasive vagueness; much of the word count is wasted. | – |
| 07 Structure and organisationCan the reader follow the architecture? | The three parts are demarcated, points build, related ideas are grouped. | Parts identifiable but boundaries blur; some points sit in the wrong section. | Interleaved; the reader has to work out which part they are in. | No discernible structure; paragraphs could be reordered without loss. | – |
| 08 Personal viewStated, justified, and more than a summary? | Explicit, justified from the arguments made, and adds a synthesis or distinction. | Stated and connected, but mainly a summary, or thin, or under-justified. | Stated but disconnected, one sentence, or contradicting the body. | No view, or one so vague as to be meaningless. | – |
| 09 Logical rigourDo the conclusions follow from the premises? | Claims follow; assumptions are explicit; generalisations qualified; causal claims supported. | Sound direction with some unsupported leaps; occasional conflation of concepts. | Conclusions that don't follow; one case treated as proof; terms shifting mid-argument. | Circular reasoning, false premises, internal contradictions. | – |
| 10 Argument coherenceOne connected argument, or three separate answers? | The explanation informs the counter-argument; the view engages the counter-arguments made. | Loosely connected; the view may not reference the earlier reasoning. | Parts disconnected; the view ignores or contradicts the body. | Three separate essays; the conclusion does not follow from the body. | – |
Not yet scored
0 OF 10 CRITERIA MARKED
Two rules make self-marking worth doing. Grade what is on the page, not what you meant. And use the whole scale — a paper that clusters itself at 5 and 6 on every line has not been marked, it has been flattered.
A MARKED PAPER — IN FULL
What a 5 looks like
Sample essay graded by SJ.
INTRODUCTION
In the case that artificial intelligence were to govern aspects of society this would mean that it would control administrative aspects such as traffic controls and resource allocation as well as having political governance. Artificial intelligence would make better decision than humans in the sense of efficiency as the algorithms can quickly conduct cost-benefit analyses and can detect things quicker, through surveillance by machines. We should overall only allow artificial intelligence to govern aspects of society to ensure that the outcomes that we have decided reflect our fundamental values occur. This is because governance requires human moral and democratic legitimacy and we have a responsibility to ensure that this occurs as we are the species in control of what occurs and by ensuring that we have a genuine awareness of maintaining our values, we thereby maintain our humanity rather than giving away our decision making to algorithmic machines.
PARAGRAPH ONE — THE CASE FOR
One compelling reason to allow artificial intelligence to govern aspects of society would be to maximise welfare, in particular by removing human bias. This could be very helpful in cases of healthcare triage because it would maximise the efficiency of doctors' time as they would not need to spend time making often difficult and uncomfortable decisions about who to treat and thus, the doctors can maximise the time that they spend treating the patients who need their treatment the most. This would also remove any potential bias that the doctors may unwittingly make in these scenarios, which will thus ensure that justice is ensured. However, despite allowing artificial intelligence to make these decisions, the biases and injustice could still exude. This is due to the likely potential for the AI to be fed with decision data that is coated with unfair decisions being made by humans, and the AI thus reflecting that same bias in its future outputs. Therefore, we may not see a promising outcome from allowing AI to govern decision making when it comes to critical situations, when human bias can have massive impacts.
PARAGRAPH TWO — THE CASE AGAINST
A key reason why we should not allow AI to govern aspects of society is because AI has the absence of both moral and political legitimacy. AI lacks the capacity for to truly feel the emotions that humans feel when making decisions that rest upon morality. Whilst AI can behave intelligently, it cannot truly understand which renders its decisions as not truly meaningful. This strips people who vote away from the connection that they have with those who they vote for because if AI made decisions they would not have a representative to admire or to criticise and have dialogue with because the AI would supposedly be making the ideal decisions. This may lead to the voices of those who the outcomes are directly impacting to be silenced. Furthermore, as AI will not have political legitimacy since this depends on the consent of the governed and any political authority comes from the general will of the people which AI cannot represent, AI's governance may lead to a tearing of the social contract and may read to political unrest and anarchy which diminishes the utility and stability of society.
CONCLUSION
To conclude, AI should have a limited role in governing aspects of society, by assisting us in decision making and we should control its influence on society rather than allowing it to control us. It should be a tool as part of a piece in the steps to inform our political decision making, but moral decision making can only come from humans because it is our subjective experience which makes the decisions meaningful. Most importantly, it is necessary that we have people to hold accountable and responsible for any decisions that are made and this would not be the case if we had AI governing our society. When using AI as a tool, dialogue and consultation should take place so that decision makers and leaders cannot attribute any blame to the AI only.
ON THE INTRODUCTION
You've missed the nuance of the question here. It's saying "IF AI can make decisions better than humans, should we allow it to govern ASPECTS". You're sort of answering … should we allow AI to govern any aspects of society and would the decisions made by them be better.
In your first sentence, why would AI necessarily have to control traffic, resource allocation, and political governance? That's not the question.
Also, your concluding point is that we should only allow AI to govern aspects of society which we choose so technically they are not governing?
ON PARAGRAPH ONE
Starts off well with the healthcare example but again you're veering away from the question. If you are choosing AI to govern healthcare you are assuming they make better decisions than humans as noted in question. You argue that AI might still make bad decisions because of inherited biases but that defeats the point of it being a better decision maker. You are arguing the wrong thing by the end.
ON PARAGRAPH TWO
First part starts well — good logic. Bit extreme at the end there — the jump is too much without justification.
ON THE CONCLUSION
This should have been the introduction tbh you have understood the question much better here. It does contradict your intro slightly.
THE OVERALL VERDICT
This is much stronger than the first essay that you gave me to look at! Length is good, style much better, language much better (occasionally awkward e.g. "exude" in context doesn't work). Some of the logic is good. But overall there are major problems with answering the questions, consistency throughout, logical leaps, lack of justification.
Also, I would expect to see more real-world examples … you're much more likely to have good content as well as good arguments if you use what has actually happened!
YOUR ARGUMENT, DRAWN OUT — USE THIS TO FIND THE WEAKNESSES
Intro
AI governance = controlling admin and political aspects · should only allow AI to govern when it reflects our values · need human moral and democratic legitimacy.
Paragraph one
AI removes human bias, increases efficiency (healthcare triage) → AI inherits bias from training data → still produces biased outcomes.
Paragraph two
AI lacks moral legitimacy (can't feel emotions, can't truly understand) + lacks political legitimacy (no connection to voters, violates the social contract) → could lead to anarchy.
Conclusion
AI should be a tool to assist, not govern · need human accountability.
Written out like this, the break is visible in one glance: paragraph one argues that AI decides worse, when the question has already granted that it decides better. Do this to your own paper before you score it.
HOW THE 5 IS MADE UP — THE TEN CRITERIA
CONTENTS
- READING
- Taught by PLEW
- ESSAY
- Trained against the section’s marking expectations
- PRACTICE SETS
- Full solutions: vocabulary, logical flow, simplification, strategy
The LNAT course
Teaches the reading by the same method that generates the firm's LNAT-format questions — PLEW, the patented approach described on the Exam Question Engine page — and pairs it with essay training against the section's marking expectations. Every practice set carries full solutions: vocabulary, logical flow, simplification, strategy.
THE CURRICULUM
Every concept the course covers
Concept list to be added
THE PRACTICE PROBLEM
There is almost nothing to practise on
LNAT does not release past papers from real sittings. The official material is two practice tests, and the downloadable papers are the 2010 set — LNAT's own page notes that they still carry five options per question where the live test now has four. So the published official practice is sixteen years old and no longer in format. For comparison, the TARA course leans on sixteen years of released TSA papers.
That is why this course generates its own. The Exam Question Engine writes passages and questions in the current LNAT format and ships each one with a full solution — vocabulary, logical flow, simplification, strategy. Difficulty is set from measured indicators rather than guessed, so once you have worked through the two official tests there is still new material at the real standard.
Sources: the LNAT website's own practice-test page and its 2010 practice papers, checked at publish. Figures on this site carry a year and a denominator.
How the engine works →IN PREPARATION
The concept gallery, solved archive and marking materials for this test are published here as they are prepared. In the meantime, write to team@examrizz.com.
CONTENTS
- CONCEPT GALLERY
- Complete
- SOLVED ARCHIVE
- Every released TMUA paper
- METHOD
- Worked on camera, question by question
The TMUA course
Pairs the full concept gallery with the complete solved archive: every released TMUA paper, worked on camera, question by question.
ONE CONCEPT, IN FULL
An example concept video
Concept video to be added
THE CURRICULUM
Every concept the course covers
Concept list to be added
IN PREPARATION
The concept gallery, solved archive and marking materials for this test are published here as they are prepared. In the meantime, write to team@examrizz.com.
WHAT THE TESTS ARE
Three reasoning tests
Each decides progression at some of the most selective courses in the UK.
TARA
UAT-UK · REPLACES THE TSA AT OXFORD, OCT 2026
FORMAT
Three modules of forty minutes each: 22 multiple-choice questions in critical thinking, 22 in problem solving, and a writing task of up to 750 words, chosen from three prompts.
LNAT
NINE LAW SCHOOLS · OXFORD, CAMBRIDGE, UCL, LSE
FORMAT
Section A — 95 minutes: 42 multiple-choice questions on 12 argumentative passages.
Section B — 40 minutes: one essay from a choice of three.
TMUA
UAT-UK · MATHEMATICS
FORMAT
Paper 1 — Applications of Mathematical Knowledge: 20 multiple-choice questions, 75 minutes. Applying known mathematics in new situations.
Paper 2 — Mathematical Reasoning: 20 multiple-choice questions, 75 minutes. Mathematical reasoning and simple ideas from elementary logic.
DATES AND JOINING
TARA 2026
REGISTRATION CLOSES 29 SEPTEMBER · FIRST SITTING 12–16 OCTOBER · SECOND WINDOW 4–8 JANUARY 2027 · RESULTS 16 NOVEMBER
Enrolment for the October and January sittings is open; a family unsure which test applies to their course is welcome to write and ask.
Enquiries: team@examrizz.com