Watson Glaser Guide: How to Pass the Critical Thinking Test at Law Firms

If you are working through the law firm application process, our Application Process hub covers every stage from written applications and psychometric tests through to interviews and assessment centres.

The Watson Glaser Critical Thinking Appraisal is used by the majority of Magic Circle, Silver Circle, and US firms in London to screen applicants before written stages or interviews. Many candidates who are well prepared in every other respect fail it, not because they lack the analytical ability it tests but because they have never sat a test in this specific format and do not understand what each question type is actually asking them to do. Critical thinking in the Watson Glaser sense is a specific technical skill, and like any technical skill it develops through practice rather than through general intelligence or legal knowledge. This guide covers exactly how each question type works, where candidates go wrong on each one, and how to approach your preparation so that the test reflects your actual ability rather than your unfamiliarity with the format.

What the Watson Glaser test is and why firms use it

The Watson Glaser Critical Thinking Appraisal was developed in the 1920s and has been revised several times since. The version used by most law firms today is the Watson Glaser II, which consists of 40 questions to be answered in 30 minutes, covering five question types. Some firms administer a longer version with 80 questions.

Firms use it because commercial legal practice requires a specific kind of thinking: the ability to read a set of facts, distinguish between what those facts establish and what they merely suggest, identify the assumptions being made in an argument, and evaluate the strength of competing conclusions. These are not skills that GPA or degree classification reliably measures, and the Watson Glaser provides a consistent, comparable data point across a large candidate pool.

The pass mark varies between firms and between years. Most firms do not publish their cut-off scores, but the general understanding in the market is that the top firms expect scores in the upper quartile. Preparation significantly affects outcomes: candidates who practise regularly and understand the question types perform meaningfully better than those who sit the test cold, regardless of their general analytical ability.

The five question types, one by one

1. Assumptions

What the question asks: You are given a statement and asked whether a particular assumption is "made" or "not made" in that statement. An assumption is something that is taken for granted, accepted without proof, as a basis for the statement.

How it works: The statement presents a claim or recommendation. The assumption presented to you may or may not be genuinely implicit in making that claim. Your job is to decide whether the statement would make no sense, or would be pointless, unless the assumption were true. If the statement could be made without needing the assumption to be true, the assumption is not made.

Where candidates go wrong: The most common mistake is treating "possible" or "plausible" as the standard rather than "necessary." An assumption that might be true, or that would make the statement more reasonable, is not necessarily made in the statement. The test is stricter than it sounds: is this assumption genuinely required for the statement to make sense?

A second common mistake is importing background knowledge or personal views. The Watson Glaser tests your ability to reason from the information given, not your knowledge of whether assumptions are factually accurate in the real world. An assumption that is false in reality might still be made in a particular statement, and your job is to assess the logic of the statement rather than its factual accuracy.

Example:

Statement: "We should increase the number of electric vehicle charging points in city centres to reduce carbon emissions."

Proposed assumption: "Electric vehicles produce lower carbon emissions than petrol vehicles."

This assumption is made. The statement would be pointless unless this were true: if electric vehicles produced the same or higher emissions, increasing charging points would not reduce carbon emissions and the recommendation would not follow. The assumption is genuinely implicit in making the claim.

Proposed assumption: "Most people in city centres currently drive petrol vehicles."

This assumption is not made. The statement about increasing charging points could be made regardless of the current proportion of petrol vehicles. The claim stands even if most people already drive electric vehicles. The assumption is not required for the statement to be coherent.

2. Inferences

What the question asks: You are given a short passage of factual information and asked to evaluate a series of inferences drawn from it. Each inference is marked as True, Probably True, Insufficient Data, Probably False, or False.

How it works: True means the inference follows necessarily from the facts given. Probably True means it follows with a high degree of probability but is not certain. Insufficient Data means the passage does not give enough information to assess the inference. Probably False means the inference is unlikely given the facts. False means it contradicts the facts directly.

Where candidates go wrong: The Insufficient Data category trips up more candidates than any other. The temptation is to mark inferences as Probably True because they seem reasonable or consistent with general knowledge, when the correct answer is Insufficient Data because the passage simply does not provide the relevant information. The test is about what the passage tells you, not what you already know or what seems likely.

The True and False categories are also frequently confused. True means the inference necessarily follows: there is no way it could be wrong given the facts provided. If there is any scenario in which the inference might not hold even given those facts, the answer is not True but Probably True at most.

Example:

Passage: "Sales of hybrid vehicles in the UK increased by 23% between 2022 and 2023. During the same period, sales of petrol-only vehicles decreased by 11%."

Inference: "More people bought hybrid vehicles than petrol vehicles in 2023."

The correct answer is Insufficient Data. The passage tells us about the rate of change in sales, not the absolute numbers. Hybrid vehicle sales increased by 23% from a starting point we do not know. Petrol vehicle sales decreased by 11% from a starting point we do not know. Without knowing the starting points, we cannot determine which category sold more in absolute terms in 2023.

Inference: "Sales of hybrid vehicles increased between 2022 and 2023."

The correct answer is True. This follows necessarily and directly from the passage, which states exactly this.

3. Deductions

What the question asks: You are given a short passage and asked whether a conclusion "follows" or "does not follow" from the information in it. A conclusion follows if it must be true given the premises, using only the information provided and basic logic.

How it works: This question type tests formal logical reasoning. The conclusion must follow necessarily from the premises: if the premises are true, the conclusion must also be true. If the conclusion could be false even if the premises were true, it does not follow.

Where candidates go wrong: The most common mistake is accepting conclusions that are probably true or reasonable rather than conclusions that necessarily follow. If there is any scenario in which the premises could be true but the conclusion false, the conclusion does not follow in the Watson Glaser sense.

Candidates also frequently accept conclusions that follow from their background knowledge rather than from the passage. The rule is strict: only the information in the passage can be used to evaluate the conclusion.

Example:

Passage: "All solicitors in this firm have completed the SQE. Some of the people in this firm who have completed the SQE are partners."

Conclusion: "Some solicitors in this firm are partners."

This conclusion follows. The premises tell us all solicitors have completed the SQE, and that some people who have completed the SQE are partners. It is therefore possible that some solicitors are among those partners, and indeed this necessarily follows from the combination of premises.

Conclusion: "All partners in this firm have completed the SQE."

This conclusion does not follow. The passage tells us some people who completed the SQE are partners. It does not tell us anything about whether all partners completed the SQE. There could be partners who became partners through other routes.

4. Interpretations

What the question asks: Similar to deductions, but the standard is slightly different. You are given a short passage and asked whether a conclusion "follows beyond a reasonable doubt" or "does not follow beyond a reasonable doubt." The phrase "beyond a reasonable doubt" means the conclusion follows with sufficient confidence to justify acting on it, not that it is logically certain.

How it works: The interpretations section uses a slightly lower standard than deductions. A conclusion that probably follows from the evidence, even if not with absolute logical necessity, may still follow beyond a reasonable doubt. The key question is whether a reasonable person reading the passage would be confident that the conclusion is warranted.

Where candidates go wrong: Many candidates treat this section identically to deductions, applying the same strict logical standard. The "beyond a reasonable doubt" framing is meaningful: the question is asking for practical confident reasoning rather than formal logical proof, and conclusions that are highly probable given the evidence may follow even if they are not logically certain.

Example:

Passage: "A survey of 500 law students found that 78% said they had experienced significant stress during their first year of study. 65% said they had considered leaving the course at some point during that year."

Conclusion: "Many law students find the first year of their degree stressful."

This conclusion follows beyond a reasonable doubt. The data strongly supports it: 78% of a reasonably large sample reporting significant stress is compelling evidence that many students find the experience stressful, and a reasonable person would be confident in this interpretation.

Conclusion: "Most law students who experience stress in their first year go on to leave their course."

This conclusion does not follow beyond a reasonable doubt. The passage tells us that 65% considered leaving, not that they actually left. The passage does not give us data on actual attrition rates, and "considered leaving" and "left" are very different things.

5. Evaluation of arguments

What the question asks: You are given a question and a series of arguments for or against a position. You are asked to identify whether each argument is "strong" or "weak." A strong argument is both important in relation to the question and directly addresses the main point. A weak argument is either not important, does not address the question directly, or makes a claim that is vague, irrelevant, or based on personal opinion rather than substantive reasoning.

How it works: The key distinction is between arguments that genuinely bear on the question and arguments that are technically relevant but minor, circular, or based on emotional appeal rather than substantive reasoning. An argument can be factually accurate and still be weak if it does not really address the question being asked.

Where candidates go wrong: The most common mistake is treating personal views or emotional responses as a factor. An argument that feels persuasive to you personally is not necessarily strong in the Watson Glaser sense. The question is whether the argument directly and substantively addresses the issue, not whether you agree with it or find it compelling.

Candidates also frequently mark circular arguments as strong. An argument that essentially restates the question as its own answer ("We should do X because X is a good thing to do") is always weak regardless of how reasonable the position is.

Example:

Question: "Should law firms be required to publish their gender pay gap data annually?"

Argument for: "Yes, because transparency about pay disparities is the first step toward addressing them, and without publicly available data there is no external accountability for firms to make progress."

This is a strong argument. It directly addresses the question, identifies a specific mechanism (external accountability), and makes a substantive claim about why the policy would achieve its intended purpose.

Argument against: "No, because it would create additional administrative burden for firms."

This is a weak argument. While technically relevant, administrative burden is a minor consideration compared to the substantive policy question being asked. It does not engage with whether the policy is otherwise justified.

Argument for: "Yes, because equality is important and firms should be fair to all their employees."

This is a weak argument. It is vague, it does not address the specific policy being discussed, and it essentially restates a general value without making a substantive case for why this particular measure would advance that value.

The preparation approach that actually works

Understand the question types before you practise

Reading this guide before you start practising is the right order. Candidates who start practising without understanding what each question type is testing tend to develop bad habits early, particularly on assumptions and inferences where the most natural instinct is consistently wrong. Understanding the standard for each question type first means your practice reinforces the right approach rather than the wrong one.

Practise under timed conditions from early on

The 30-minute time limit for 40 questions gives you 45 seconds per question, which is genuinely tight. Many candidates who understand the question types and apply the right reasoning still underperform because they have never built up the speed to work through the format efficiently. Practising individual question types without timing yourself is useful early in preparation, but it should give way to full timed tests as soon as you are confident in the approach for each type.

The Watson Glaser Practice Hub gives you unlimited access to full-length timed mock tests that mirror the real format, with worked explanations for every question showing not just the correct answer but why it is correct and why the other options are not. There is no cap on attempts, which means you can build up the volume of timed practice that actually makes the difference.

Review every question you get wrong, and some you get right

The temptation after a practice test is to focus only on the questions you answered incorrectly. This is better than nothing but misses a significant part of the learning. Some questions you answer correctly through the right reasoning and some through guessing or lucky application of the wrong logic. Going through your working on questions you got right as well as ones you got wrong identifies which approach you are actually using and where instinct is leading you astray even when it happens to produce the correct answer.

The worked explanations in the Watson Glaser Practice Hub make this process straightforward, because they explain the reasoning behind every answer rather than just confirming whether you were right.

Focus extra time on your weakest question type

Most candidates have one question type that they consistently find harder than the others. For many it is assumptions, where the standard of "necessary" rather than "plausible" is genuinely counterintuitive. For others it is inferences, where the Insufficient Data category is repeatedly underused. Identifying which type is weakest after your first few practice tests and directing extra preparation toward it is more efficient than spreading your time evenly across all five.

Do not prepare in isolation

Explaining your reasoning on a question to someone else, or having someone explain theirs to you, is one of the most effective ways to identify where your logic is going wrong. If you cannot explain clearly why an assumption is made or not made in a particular statement, that is a sign that you are applying the right answer without fully understanding the reasoning behind it, which will not hold up reliably under pressure.

Common mistakes across the whole test

Bringing in outside knowledge. The Watson Glaser tests reasoning from the information provided. An assumption that is factually accurate, an inference that you know to be true from general knowledge, or a conclusion you believe to be correct based on real-world experience may still be wrong in the context of the specific passage. Treat each passage as a closed system.

Applying the wrong standard to interpretations and deductions. Deductions require logical necessity. Interpretations require confidence beyond a reasonable doubt. These are different standards and applying the stricter deductions standard to interpretations will cause you to mark too many conclusions as not following.

Over-using Insufficient Data in inferences. Insufficient Data is the correct answer when the passage genuinely does not give enough information to assess the inference. It is not a default safe answer for inferences that seem uncertain. If the passage clearly makes a particular inference probable or improbable, marking it as Insufficient Data is wrong.

Letting personal views affect argument evaluation. An argument you find personally compelling may still be weak in the Watson Glaser sense if it is vague, circular, or does not directly address the question. An argument you find personally unconvincing may still be strong if it makes a substantive, direct, and important point.

Spending too long on individual questions. With 45 seconds per question, dwelling on a single difficult question is costly. If you are genuinely unsure after a reasonable attempt, move on and return if time allows. The questions do not increase in difficulty as the test progresses, so questions later in the test are not inherently harder.

A note on the firms that use it and when

The Watson Glaser is used at different stages by different firms. Some send it immediately after the initial application as a screening tool before reading written answers or inviting candidates to interviews. Others use it as part of an online assessment that also includes situational judgment tests, numerical reasoning, or verbal reasoning. A small number use it at assessment centre stage.

If you are applying broadly across the Magic Circle and Silver Circle, you will almost certainly sit this test more than once, which makes thorough preparation even more valuable: the investment pays off across multiple applications rather than just one.

Ready to practise?

Reading this guide gives you the framework. Applying it under timed conditions is what builds the skill. The Watson Glaser Practice Hub gives you unlimited free access to full-length timed mock tests with worked explanations for every question, no cap on attempts, and questions refreshed regularly so you are not just memorising answers. For candidates preparing for applications to top commercial law firms, it is the most direct preparation available.

For broader interview preparation covering motivational, competency, commercial awareness, situational judgment, and ethics questions, the Future Trainee Academy covers the full application process and is also completely free.