If you are applying for a training contract or vacation scheme at a leading UK law firm, there is a good chance the Watson Glaser test is standing between you and the interview stage. This guide explains exactly what the test involves, walks through a realistic example from each of its five sections, and shows you how to prepare so that your application is judged on your potential rather than your unfamiliarity with an unusual test format.
You can also put the theory into practice straight away. There is a free Watson Glaser practice test on this page, with no sign-up needed, so you can see where you stand before you read a word further or after you have finished the guide.
What is the Watson Glaser test?
The Watson Glaser Critical Thinking Appraisal (often shortened to W-GCTA, or simply "the Watson Glaser") is a psychometric test of critical thinking. It was first developed in the 1920s by the American psychologists Goodwin Watson and Edward Glaser and has been refined over the decades since. Today it is published by Pearson TalentLens and is widely regarded as the standard commercial measure of critical thinking used in graduate recruitment.
Unlike a verbal reasoning test, which checks whether you can read a passage accurately, the Watson Glaser tests how you think about information. It measures whether you can separate fact from assumption, follow an argument to its logical end, and judge the strength of a claim without letting your own opinions or general knowledge get in the way. Those are precisely the skills a trainee solicitor uses every day: reading a contract for what it actually says rather than what a client hopes it says, testing the logic of an opposing argument, and drawing only the conclusions the evidence supports.
Who uses it
In the UK, the Watson Glaser is most closely associated with law firm recruitment. Firms that have used it as part of their training contract and vacation scheme processes include Clifford Chance, Linklaters, Hogan Lovells and Freshfields, along with a number of other City and national firms. It also appears outside law, in government schemes and some finance and professional services roles, but law is where most candidates meet it.
One practical note: firms review and change their assessment providers from time to time, so always check the current application process for each firm you apply to. If a firm's process mentions a "critical thinking test", it is very likely to be the Watson Glaser or something closely modelled on it, and the preparation in this guide will serve you either way.
Why it matters
For most firms that use it, the Watson Glaser is an early filter. It typically sits just after the application form, and candidates who fall below the firm's benchmark are usually not progressed, regardless of the strength of their written application. That makes it one of the highest-stakes half hours in the entire application process. The encouraging news is that the test rewards familiarity. The question formats follow strict, learnable rules, and candidates who understand those rules before test day have a clear advantage over those meeting them for the first time under timed conditions.
The RED model
The publisher summarises the skills behind the test with the RED model:
Recognise assumptions. Notice when something is being taken for granted rather than stated, and separate what is known from what is merely supposed.
Evaluate arguments. Judge arguments on their logic and relevance, setting aside emotion and personal opinion.
Draw conclusions. Arrive at conclusions that the evidence actually supports, without overreaching beyond it.
The five sections of the test, covered one by one below, all map onto these three skills.
The five sections of the Watson Glaser test
The Watson Glaser typically contains around 40 questions split across five sections: Inference, Recognition of Assumptions, Deduction, Interpretation and Evaluation of Arguments. Each section has its own instructions and its own answer options, and the instructions differ in small but important ways. Most wrong answers on this test come from applying the rules of one section to another, so it is worth studying each format on its own terms.
A rule that applies across the whole test: base your answers only on the information given. The test deliberately includes statements you may personally disagree with, or that conflict with common knowledge. Your task is to reason from the passage, not from the world.
1. Inference
What it measures. An inference is a conclusion drawn from observed or supposed facts. This section measures how well you can judge the probability that a conclusion is true given a passage of information, and, just as importantly, whether you can recognise when the information is simply not there. You are asked to grade each inference on a five-point scale: True, Probably True, Insufficient Data, Probably False, or False.
This is widely considered the hardest section, because it deals in shades of likelihood rather than clean yes-or-no logic. The skill being tested is calibration: knowing the difference between what a passage proves, what it makes likely, and what it leaves open.
Worked example.
Passage: A survey of 400 trainee solicitors at large commercial firms found that 68 per cent had completed at least one practice critical thinking test before applying. Trainees who had practised reported feeling more confident on test day than those who had not.
Proposed inference: Practising critical thinking tests guarantees a higher score on the Watson Glaser.
Options: True / Probably True / Insufficient Data / Probably False / False
Correct answer: Probably False.
Why. The passage tells us that most surveyed trainees practised and that practising was associated with confidence. It says nothing about scores at all, and it certainly establishes no guarantee. Words like "guarantees" should always put you on alert, because almost nothing in an inference passage supports an absolute claim. The inference is not outright False, since the passage does not contradict it directly, but a guarantee goes so far beyond the evidence that Probably False is the best grading. If the proposed inference had said "practising may help some candidates feel more confident", the answer would have been True, because the passage states the association directly.
The five-point scale takes some getting used to, and the fastest way to internalise it is repetition. You can try a free Watson Glaser practice test on this page, no sign-up needed, and see how your instincts on Insufficient Data compare with the model answers.
2. Recognition of Assumptions
What it measures. An assumption is something taken for granted, an unstated belief the statement relies on to make sense. This section presents a short statement followed by a proposed assumption, and asks a single question: is that assumption made in the statement, or not? The options are Assumption Made or Assumption Not Made.
The useful mental test is negation: imagine the proposed assumption is false, and ask whether the original statement still makes sense. If the statement collapses without it, the assumption is made. If the statement survives perfectly well, it is not.
Worked example.
Statement: "We should apply to firms in Leeds as well as London, because competition for London training contracts is so intense."
Proposed assumption: Competition for training contracts in Leeds is less intense than in London.
Options: Assumption Made / Assumption Not Made
Correct answer: Assumption Made.
Why. Apply the negation test. Suppose competition in Leeds were just as intense as in London. The statement's reasoning would then fall apart, because "apply to Leeds as well" is being offered as a response to London's intensity, which only makes sense as advice if Leeds offers relatively better odds. The speaker never says Leeds is less competitive, and that is precisely the point: the section is about what a statement silently relies on, not what it declares. Note what would make this an Assumption Not Made instead: a proposed assumption such as "Leeds firms pay higher salaries than London firms" plays no role in the statement's logic, and the statement stands or falls without it.
3. Deduction
What it measures. This section presents premises followed by a proposed conclusion, and asks whether the conclusion follows necessarily from the premises. The options are Conclusion Follows or Conclusion Does Not Follow. Unlike Inference, there is no probability scale here. Deduction is strict: the conclusion must be unavoidable given the premises, with no exceptions possible. "Probably" is not good enough.
This is also the section where the instruction to ignore real-world knowledge matters most. The premises may be untrue in reality. Your job is purely to test the logic.
Worked example.
Premises: All partners at the firm are qualified solicitors. Some qualified solicitors at the firm work in the Manchester office.
Proposed conclusion: Some partners at the firm work in the Manchester office.
Options: Conclusion Follows / Conclusion Does Not Follow
Correct answer: Conclusion Does Not Follow.
Why. The premises tell us the partners sit inside the larger group of qualified solicitors, and that the Manchester office contains some members of that larger group. But nothing forces any overlap between the partners and the Manchester solicitors. It is entirely possible, consistent with both premises, that every solicitor in Manchester is an associate and every partner sits in London. Since a scenario exists in which the premises are true and the conclusion is false, the conclusion does not follow. This is the classic deduction trap: the conclusion is plausible, and in the real world might well be true, but plausibility is irrelevant. If you can construct even one counterexample, the answer is Does Not Follow.
4. Interpretation
What it measures. Interpretation looks similar to Deduction at first glance: a short passage, a proposed conclusion, and the options Conclusion Follows or Conclusion Does Not Follow. The difference is the standard of proof. In Deduction, the conclusion must follow with absolute logical necessity. In Interpretation, you are asked whether the conclusion follows beyond reasonable doubt, a slightly softer standard that lawyers will recognise. A conclusion can follow in Interpretation even if a far-fetched exception is imaginable, provided the passage makes it unreasonable to doubt.
Worked example.
Passage: A firm received 3,000 applications for its vacation scheme this year and made 60 offers. Every applicant who received an offer had passed the firm's online critical thinking test.
Proposed conclusion: Most applicants to the scheme this year did not receive an offer.
Options: Conclusion Follows / Conclusion Does Not Follow
Correct answer: Conclusion Follows.
Why. With 3,000 applicants and 60 offers, at least 2,940 applicants did not receive one, which is far more than half. The conclusion follows beyond reasonable doubt directly from the figures given. Contrast it with a different proposed conclusion: "Passing the critical thinking test guarantees an offer." That would be Does Not Follow. The passage says every offer-holder passed the test, which means passing was (in this cohort) necessary; it does not say that everyone who passed received an offer. Confusing "all offer-holders passed" with "all passers got offers" is a reversal error, and Interpretation passages are full of opportunities to make it. Read the direction of every claim carefully.
5. Evaluation of Arguments
What it measures. The final section presents a question of policy or judgement, followed by a proposed argument for or against it. Your task is to classify the argument as Strong or Weak. A strong argument is both important and directly relevant to the question. A weak argument is one that is of minor importance, relates only to a trivial aspect of the question, or does not address the question at all.
Two instructions define this section. First, assume every argument is true; you are judging quality of reasoning, not accuracy. Second, set your own opinion on the underlying question entirely aside. An argument can be strong for a position you disagree with, and weak for a position you support.
Worked example.
Question: Should large law firms require all trainee solicitors to spend a seat in a pro bono or community legal clinic?
Proposed argument: No; some trainees would find clinic work less glamorous than working on large corporate deals.
Options: Strong argument / Weak argument
Correct answer: Weak argument.
Why. Remember, we accept the argument's content as true: no doubt some trainees would find clinic work less glamorous. But glamour is a trivial consideration when weighed against the purpose of a training seat, which is professional development and client service. The argument addresses a minor, superficial aspect of the question, so it is weak. By contrast, an argument such as "No; mandatory clinic seats would reduce the supervised time trainees spend in the practice areas in which they will qualify, and firms have a duty to prepare trainees for their intended specialism" engages directly with an important dimension of the question, and would be classified as strong. Notice that both example arguments point the same way; strength and direction are independent, and the test often pairs a weak argument with a conclusion you happen to agree with, precisely to see whether your opinions leak into your judgement.
If you have followed the reasoning in all five sections, the best next step is to test it under time pressure. The free practice test on this page mirrors these formats and takes only a few minutes, with no sign-up needed.
Scoring and pass marks
The Watson Glaser is not scored like a school exam, and this is the single most misunderstood aspect of the test.
Your raw score is converted to a percentile. When you finish, the number of questions you answered correctly (your raw score) is compared against a norm group, a large sample of previous test-takers chosen to resemble the candidate pool, for example graduates or professionals in a given field. Your result is then expressed as a percentile. A result at the 70th percentile means you scored better than 70 per cent of the comparison group. It does not mean you got 70 per cent of the questions right.
Firms set their own benchmarks, and they do not publish them. Each firm decides what percentile a candidate must reach to progress, and these cut-offs are treated as confidential. They can also shift from year to year with the strength of the applicant pool and the number of places available. You will see figures quoted around the internet, commonly a raw score of around 75 to 80 per cent, or a percentile somewhere in the 60s or 70s, but you should treat all such numbers as unofficial estimates rather than facts. No published source confirms any specific firm's threshold, and this guide will not pretend otherwise.
What this means for your preparation. Because the scoring is relative, "good enough" is defined by other candidates, most of whom are also capable graduates and many of whom have practised. Aiming vaguely to "pass" is the wrong mindset; the practical goal is to be accurate and quick enough to sit comfortably above the pack. The realistic way to know whether you are there is to practise under timed conditions and track your accuracy over multiple attempts, which is exactly what a percentile system rewards.
One more scoring point worth knowing: there is no publicly documented negative marking on the Watson Glaser. With that in mind, you should answer every question, because a blank scores nothing while an educated elimination often finds the right answer.
How to prepare: practical tips
Preparation for the Watson Glaser is unusually effective compared with other psychometric tests, because the difficulty lies mostly in the format rather than in any knowledge you lack. Here is what actually moves the needle.
1. Learn the five rule-sets before you practise at volume. Most lost marks come from blurring the sections: applying Deduction's strict necessity to an Interpretation question, or letting real-world knowledge creep into Inference. Before you sit any full-length test, be able to state from memory what each section asks and what standard of proof it applies. The five worked examples above are a good template.
2. Master the boundary between Insufficient Data and the "probably" options. In the Inference section, the most common error is choosing Insufficient Data when the passage makes something likely, or choosing Probably True when the passage says nothing at all. A useful habit: ask first "does the passage mention this subject in any way?" If not, lean towards Insufficient Data. If it does, ask which direction the evidence points.
3. Use the negation test for assumptions. For every Recognition of Assumptions question, flip the proposed assumption to its opposite and re-read the statement. If the statement now sounds absurd or unsupported, the assumption was being made. This single technique converts the vaguest-feeling section into a mechanical check.
4. Hunt for counterexamples in Deduction. Do not ask "is this conclusion sensible?" Ask "can I imagine any scenario where the premises hold and the conclusion fails?" One counterexample settles the question. Sketching quick mental diagrams of groups (partners inside solicitors, and so on) helps.
5. Suspend your opinions deliberately in Evaluation of Arguments. Before classifying an argument, notice whether you agree with the position it supports, then consciously set that aside. Candidates reliably misgrade arguments that flatter their own views. The test is designed to catch exactly this.
6. Practise under timed conditions. With roughly 40 questions in around 30 minutes, you have well under a minute per question on average. Untimed practice builds understanding; timed practice builds the pacing that the real test demands. Do both, in that order. A sensible rhythm is to keep moving, flag nothing as precious, and remember that every question carries the same weight whether it took you ten seconds or ninety.
7. Review your wrong answers properly. The value of a practice test is mostly in the debrief. For each error, identify which rule you broke: wrong section standard, outside knowledge, reversed a claim, opinion leakage, or misread. Most people discover that their errors cluster into one or two repeat offences, which makes them fixable.
8. Read argument-dense material in your daily life. Quality newspaper editorials, court judgments and well-argued essays give you free reps at spotting assumptions and weighing arguments. This is a marginal gain compared with structured practice, but over weeks it sharpens exactly the muscles the test measures.
9. Prepare your test environment. For an online sitting, that means a quiet room, a reliable connection, a charged laptop, and rough paper if the firm's instructions permit it. Check the firm's invitation email for the rules that apply to your sitting, since conditions vary.
10. Start with a diagnostic, not a mountain of theory. Ten minutes of realistic questions will show you which sections need work far faster than any amount of reading. If you have not yet taken one, the free practice test on this page requires no sign-up and gives you exactly that baseline.
Versions of the test: Watson Glaser III and online sittings
The test has been through several editions, and the one you are most likely to meet today is the Watson Glaser III. Key things to know:
It is usually taken online and unsupervised. Most firms send a link and a deadline, and you complete the test remotely. The Watson Glaser III draws questions from a large item bank, so two candidates sitting the same firm's test will typically see different questions of equivalent difficulty. This is one reason memorising specific questions is pointless, while learning the formats pays off.
Timing is typically around 30 minutes for around 40 questions. Exact conditions vary by firm and sitting, and some older or longer versions exist, so treat the invitation email from the firm as the final word on your timing.
Verification and follow-up testing exist. Because the first sitting is usually unsupervised, publishers provide shorter supervised retests that firms can use later in the process to confirm a candidate's result, for example at an assessment centre. The practical implication is simple: do the test yourself, honestly, first time. A score you cannot reproduce under supervision does not help you, and firms treat integrity seriously.
Some sittings are proctored. A number of employers use online proctoring (webcam monitoring or browser lockdown) even for the first sitting. If your invitation mentions proctoring, allow extra time for identity checks and system tests before the clock starts, and make sure your room meets the stated requirements.
If you meet a version with slightly different question counts or timing, do not be thrown. The five sections and their rules are stable across versions, and preparation transfers fully.
Next steps
The Watson Glaser rewards candidates who arrive knowing the rules, and you now know them: five sections, five distinct standards of proof, percentile scoring, and roughly 45 seconds per question. What remains is converting that knowledge into timed accuracy, and that only comes from realistic practice.
Start with the free practice test on this page, no sign-up needed, and use it as your diagnostic. Then, when you are ready to prepare properly for a firm deadline, a full account gives you the complete Watson Glaser question bank, timed mock tests in the real format, worked solutions for every question, and score tracking so you can watch your percentile position improve attempt by attempt. Candidates applying to competitive firms are preparing this way; the question bank makes sure you walk into your test at least as ready as they are.
