The Watson Glaser Critical Thinking Appraisal is a timed test that measures how rigorously you can evaluate arguments and evidence. It is built from five sections (Inference, Recognition of Assumptions, Deduction, Interpretation and Evaluation of Arguments) each with its own decision rules. A typical version runs around 40 items in roughly 30 minutes. It is not a verbal reasoning test: it does not ask what a passage says, it asks what logically follows, what is being taken for granted, and whether an argument is strong. It is the standard screening test at law firms and is used widely in professional services and senior recruitment.
Treat each of the five sections as a separate test with its own rules, because that is exactly what it is. Suspend everything you know about the real world and reason only from the text. In Deduction, accept the premises as true even when they are absurd. In Evaluation of Arguments, an argument is strong only if it is both directly relevant and important. A true but trivial point is weak. Work at roughly 45 seconds per item, and answer everything.
What Is the Watson Glaser Test?
The test presents short statements, scenarios and arguments and asks you to make a series of tightly constrained judgements. The five sections appear in a fixed order and each has its own answer options: Inference asks you to rate a conclusion as True, Probably True, Insufficient Data, Probably False or False. Recognition of Assumptions asks whether an assumption is made or not made. Deduction asks whether a conclusion follows or does not follow. Interpretation asks whether a conclusion follows beyond reasonable doubt. Evaluation of Arguments asks whether an argument is strong or weak.
The critical point, and the one candidates find hardest, is that these five sets of rules are different. The word 'follows' means something stricter in Deduction than in Interpretation, and 'true' in Inference is a probabilistic judgement while 'follows' in Deduction is an absolute one. Candidates who treat the whole test as one long verbal reasoning exercise reliably underperform, because they apply the same standard of proof to five sections that demand five different standards.
What the Test Actually Measures
- Analytical rigour: whether you can hold a strict standard of proof and refuse to be moved by a plausible but unsupported claim.
- Separation of fact from assumption: recognising the unstated premises an argument depends on. This is the core legal skill the test proxies for.
- Formal deductive reasoning: whether a conclusion follows necessarily from given premises, regardless of whether the premises are actually true.
- Argument evaluation: judging relevance and importance, not agreement. Whether you personally agree with an argument is irrelevant to whether it is strong.
- Resistance to bias: several items are deliberately built around emotive topics, and the test measures whether your judgement survives contact with your own opinions.
Watson Glaser is the closest thing the assessment market has to a standard entry test for law. Many UK and international law firms use it as a training-contract or vacation-scheme sift, and it also appears in professional services, government and senior management recruitment. Pass marks are not publicly published and differ by firm and intake.
Section 1: Inference
You are given a short passage of facts and a proposed inference, and you rate it on a five-point scale: True, Probably True, Insufficient Data, Probably False, False. Unlike the rest of the test, this section is explicitly probabilistic. You are allowed, and expected, to use general world knowledge to judge how likely something is, provided you anchor it in the passage.
- True: the inference follows beyond reasonable doubt from the facts given.
- Probably true: more likely to be true than false, given the facts, but not certain.
- Insufficient data: the facts simply do not bear on the inference either way. This is the correct answer far more often than candidates expect.
- Probably false: more likely false than true, given the facts.
- False: the inference contradicts the facts or cannot possibly follow from them.
The passage: 'Northvale Logistics has announced a EUR 40 million investment in a new automated distribution centre near Rotterdam. In its statement the company said the site would allow it to serve customers in the Benelux region from a single hub for the first time.'
Rate the inference on the five-point scale.
The passage never mentions delivery times. But it does say the company will serve a whole region from a single hub for the first time, and it is calling that an advantage worth EUR 40 million. Reduced delivery time is a highly plausible motive for a regional hub, though other motives (cost, capacity, automation) are equally plausible. That plausibility, anchored in the passage's own framing of the hub as a benefit, makes the inference more likely than not, but nowhere near certain.
Probably True. Note that in the Deduction section, an identical gap in the text would produce 'Conclusion does not follow'. The standards genuinely differ by section, and this is the trap the whole test is built around.
Section 2: Recognition of Assumptions
You are given a statement and a proposed assumption, and you decide whether the assumption is made or not made. An assumption is something the speaker must be taking for granted for their statement to make sense. The test here is narrow: not whether the assumption is reasonable, but whether the statement actually depends on it.
You must decide whether the speaker is taking that assumption for granted.
Decide: assumption made, or assumption not made.
Test it by negation, which is the reliable technique for this entire section. Negate the assumption: 'Productivity cannot be influenced by working patterns.' If that were true, the speaker's claim that a four-day week would raise productivity collapses entirely. It becomes incoherent. So the statement genuinely depends on it.
Assumption made. Now test a near-miss: 'The team currently works five days.' Negate it ('the team does not currently work five days') and the proposal to move to four days still makes sense (they might work six). The statement does not depend on it, so that one would be Assumption not made.
Section 3: Deduction
You are given premises and a conclusion, and you decide whether the conclusion follows necessarily. This is formal logic. The premises are to be accepted as true even when they are factually absurd, and world knowledge is strictly forbidden. If the conclusion could be false while all the premises are true, it does not follow, full stop.
Accept both premises as true and decide whether the conclusion follows.
Decide: conclusion follows, or conclusion does not follow.
Partners are a subset of qualified solicitors. Some qualified solicitors do tax. But the tax specialists could all sit entirely outside the partner subset. Nothing in the premises forces any overlap. Because it is possible for both premises to be true while the conclusion is false, the conclusion is not necessary. The fact that in a real law firm some partners almost certainly do tax is exactly the reasoning the section forbids.
Conclusion does not follow. This is the classic 'some' fallacy, and it is the single most productive trap in the Deduction section. A conclusion that merely could be true does not follow.
Section 4: Interpretation
You are given a passage, usually containing data or a described situation, and a proposed conclusion. You decide whether the conclusion follows beyond reasonable doubt. The standard is stricter than Inference and looser in feel than Deduction, but in practice it is close to Deduction: if the conclusion generalises, adds a cause, or extends beyond the group the passage describes, it does not follow.
Decide whether the conclusion follows beyond reasonable doubt.
Decide: conclusion follows, or conclusion does not follow.
The passage reports what 68% of 500 surveyed employees said. The conclusion generalises to 'employees at Northvale' as a whole. Unless the 500 are the entire workforce (which the passage never states) the leap from a sample to the population is exactly the kind of extension Interpretation rejects. A second, subtler problem: 'said they would prefer' is a stated preference; 'prefer' is a claim about actual preference.
Conclusion does not follow. Watch for the same trap in reverse: a conclusion that only restates what the passage says, with no extension at all, usually does follow, and candidates reject it because it feels too easy.
Section 5: Evaluation of Arguments
You are given a question and an argument for or against, and you decide whether the argument is strong or weak. A strong argument must be both directly relevant to the question and important. An argument can be perfectly true and still weak, because it is trivial or because it addresses a different question. Your personal agreement with the position is irrelevant, and the test contains items on deliberately divisive topics to check exactly that.
| Formula | Expression |
|---|---|
| Strong argument | Directly relevant to the question AND addresses something important |
| Weak: irrelevant | True, but answers a different question than the one asked |
| Weak: trivial | Relevant, but the consequence is minor compared with what is at stake |
| Weak: circular | Restates the question as its own justification |
| Weak: unsupported analogy | Rests on a comparison the question does not license |
| Not a criterion | Whether you agree with it. Agreement and strength are unrelated in this section |
The Decision Rules, Section by Section
Print this table and internalise it. Almost every point lost on the Watson Glaser comes from applying one section's standard of proof inside another section.
| Publisher | Questions | Time | Difficulty | Notes |
|---|---|---|---|---|
| Inference | Passage + proposed inference | 5 options: True to False | Medium | Probabilistic. World knowledge is permitted, anchored in the passage. 'Insufficient data' is common |
| Recognition of Assumptions | Statement + proposed assumption | 2 options: made / not made | Hard | Use the negation test. Ask only whether the statement depends on it, not whether it is reasonable |
| Deduction | Premises + conclusion | 2 options: follows / does not follow | Hard | Formal logic. Accept absurd premises as true. World knowledge is forbidden |
| Interpretation | Passage + conclusion | 2 options: follows / does not follow | Medium-Hard | Beyond reasonable doubt. Reject anything that generalises, adds a cause, or extends the group |
| Evaluation of Arguments | Question + argument | 2 options: strong / weak | Hard | Relevant AND important. True but trivial is weak. Your own opinion is irrelevant |
Our logical reasoning category covers deductive, inductive and critical-thinking item types with AI explanations on every answer. The free preview gives you 4 questions per test plus 5 AI coach messages.
Start the free testWho Uses It and Why
The Watson Glaser is used most heavily by law firms, where the skill it proxies for (separating what a document establishes from what a client wishes it established) is the daily work. Beyond law it appears in professional services, consulting, government and senior management selection, usually alongside a numerical test and a situational judgement test.
- It is a critical thinking test, not a verbal test. The difference is explained in our verbal reasoning guide, and if you are sitting both, prepare them separately.
- Firms that use it typically pair it with a situational judgement test and sometimes a numerical reasoning test in the same online stage.
- Deepen the deductive component with a free deductive reasoning test, which drills the exact syllogistic reflexes Section 3 rewards.
9 Strategies That Move Your Score
- 1Know which section you are in. Before each item, remind yourself of the standard of proof for that section. This single habit is worth more than any other technique on this test.
- 2In Deduction, obey the premises absolutely. 'All lawyers are cats. Some cats are grey.' You accept it and reason from it. The moment world knowledge enters, you lose points.
- 3Use the negation test in Assumptions. Negate the proposed assumption. If the statement collapses, the assumption is made. If the statement survives, it is not made. This converts an intuitive section into a mechanical one.
- 4Attack the quantifiers. All, some, none, only. Most Deduction errors are 'some' errors: 'some A are B' and 'some B are C' never entitles you to 'some A are C'.
- 5In Evaluation, separate strength from agreement. Ask two questions in order: does this address the actual question, and does it matter? Never ask whether you believe it.
- 6Do not reject an easy answer for being easy. Interpretation items that simply restate the passage do follow. Test writers rely on candidates over-thinking them.
- 7Watch for emotive topics. Items on immigration, taxation, healthcare and education are placed there to test whether your reasoning survives your opinions. When you feel something, slow down.
- 8Budget roughly 45 seconds per item. With about 40 items in around 30 minutes there is no room to dwell. Flag, guess, move on.
- 9Answer every item. Negative marking is not standard here. A blank is a certain zero.
Mistakes That Cost the Most Marks
- Carrying one section's standard into the next. The most expensive error on the test, and the reason strong verbal reasoners often underperform on it.
- Importing world knowledge into Deduction. Especially damaging for experienced professionals, who cannot help knowing how the real world works.
- Confusing a reasonable assumption with a necessary one. In Recognition of Assumptions, the only question is whether the statement depends on it.
- Rating an argument strong because you agree with it. Agreement is not a criterion. Relevance and importance are the only two.
- Treating 'Insufficient Data' as a cop-out in Inference. It is a legitimate and frequently correct answer, not an admission of defeat.
- Preparing with generic verbal reasoning practice. It builds the wrong reflexes. Practise the five sections separately, and pair the work with logical reasoning practice.
Your 14-Day Preparation Plan
- 1Days 1-2: Baseline and diagnosis. Sit a full timed practice test. Score each of the five sections separately: almost every candidate has one section that is dragging the total down.
- 2Days 3-4: Deduction. Drill syllogisms untimed until the quantifier rules are automatic. This is the most mechanical section and therefore the fastest to improve.
- 3Days 5-6: Recognition of Assumptions. Practise the negation test on every item until you stop reasoning intuitively and start reasoning procedurally.
- 4Days 7-8: Inference and Interpretation. Work the two probabilistic sections back to back, specifically to train yourself to switch standards of proof between them.
- 5Days 9-10: Evaluation of Arguments. Deliberately practise on emotive topics. Write down, for each item, whether it is relevant and whether it is important, separately, in that order.
- 6Days 11-12: Full timed sets. All five sections in sequence, at 45 seconds per item, so switching between the rule sets becomes automatic under pressure.
- 7Day 13: Error review only. Redo only what you got wrong and name the rule you broke on each item.
- 8Day 14: One rehearsal, then stop. A full timed run in the morning, then rest. This test punishes fatigue heavily.
MakingMoves.ai covers 50+ test categories with 113,000+ practice questions, and the AI coach explains which decision rule an item turns on rather than just marking it wrong, which matters more on Watson Glaser than on any other test, because the rules differ by section. The free preview gives you 4 questions per test and 5 AI coach messages. Paid plans are GROW at EUR 19,95 per week, PRO at EUR 49,95 per month, and MAX at EUR 79,95 per month.
Frequently Asked Questions
Scores are reported as a percentile against a norm group (commonly a graduate or professional norm) rather than as a raw mark. Law firms and professional services firms set their own thresholds and do not publish them, and the bar differs by firm and by intake. The workable goal is to be comfortably above the median of the relevant norm group.
A typical version runs around 40 items in roughly 30 minutes, which works out at about 45 seconds per item. Some employers use shorter or untimed variants. Your invitation email will state the format, and it is worth checking, because the strategy for a generously timed version is different from the strategy for a tight one.
No, and treating them as the same is the most common preparation mistake. A verbal reasoning test asks what a passage says. The Watson Glaser asks what follows, what is assumed, and whether an argument is strong, with five different standards of proof across five sections. Verbal practice alone will not prepare you.
Only in the Inference section, and only anchored in the passage. In Deduction it is explicitly forbidden: you must accept the premises as true even when they are factually false. Interpretation and Evaluation of Arguments are judged on the text and the argument, not on what you know to be the case.
It is the standard critical thinking screen at many law firms for training contracts and vacation schemes, and it also appears in professional services, consulting, government and senior management recruitment. If your invitation names a critical thinking appraisal, assume this format and prepare the five sections separately.
Yes, and typically more than on other cognitive tests, because most of the difficulty is procedural rather than intellectual. Candidates lose marks by applying the wrong standard of proof, not by failing to understand the material. Learning the five rule sets removes that error class entirely.
Our logical reasoning category covers deduction, assumptions and argument evaluation with AI explanations that name the rule behind every answer, plus timed mocks that match real durations.
Explore logical reasoning practiceIngmar van Maurik is the founder of MakingMoves.ai and Assessment-Training.com. With 10+ years in psychometric assessment design, he has helped over 1 million professionals prepare for job assessments.