Watson Glaser Tips: Five Sections, Three Different Standards of Proof

The Watson Glaser Critical Thinking Appraisal is the standard screen for training contracts and vacation schemes at major law firms, and it is the test where being generally able is least sufficient. Its five sections each ask a different question, and they apply three different standards of proof. Nearly every avoidable mark is lost by answering one section with another section's rules.

The three standards, in one place

• Strict necessity governs Deduction. The conclusion has to follow from the premises with no gap at all.
• Beyond reasonable doubt governs Interpretation. The conclusion does not have to be logically forced, but the evidence has to leave no reasonable alternative.
• Degree of support governs Inference. You are rating how far the passage pushes a conclusion, on a five-point scale, not deciding whether it is proven.

Recognition of Assumptions and Evaluation of Arguments do not test proof at all. They test dependency and relevance respectively, which is a different thing again.

Inference: rate the support, do not import knowledge

You read a factual passage and judge each proposed inference as True, Probably True, Insufficient Data, Probably False, or False.

• Decide direction before strength. First ask whether the passage supports the inference, contradicts it, or does neither. That collapses five options to a manageable choice within one direction.
• The extremes should be rare. True and False belong to inferences the passage settles outright. Most items live in the middle three, and candidates who use True and False freely are almost always over-reading.
• Outside knowledge is the trap. An inference can be entirely true in the world and still be Insufficient Data, because the passage is your only evidence. If you find yourself reasoning from what you know about the industry or the topic, you have already left the question.

Recognition of Assumptions: negate it and see what breaks

You are given a statement and a proposed assumption, and asked only whether the statement takes that assumption for granted. Not whether the assumption is true. Not whether it is reasonable.

The reliable test is negation. Negate the proposed assumption and reread the statement. If the statement no longer makes sense, or its point collapses, the assumption is being made. If the statement survives, it is not.

The trap in this section is plausibility. Assumptions that are perfectly sensible background beliefs about the world are still not made unless the statement actually depends on them. This section rewards a narrow reading, which is uncomfortable for people used to giving generous readings.

Deduction: the premises are true, even when they are silly

Given premises you must accept as true, decide whether a conclusion follows. Real-world truth is irrelevant in both directions: a conclusion that is obviously correct in reality does not follow unless the premises deliver it, and a conclusion that sounds ridiculous does follow if they do.

Three formal errors account for most losses:

• Reversal. All partners are lawyers does not give you all lawyers are partners.
• Over-negation. The denial of all clients agreed is not no clients agreed. It is that at least one did not.
• Reading some as not all. In logic some means at least one, and is perfectly compatible with all. A conclusion that some X are Y follows from a premise that all X are Y.

Interpretation: weigh the evidence, do not extend it

Here the standard drops from necessity to beyond reasonable doubt, which tempts people to relax too far. It is still a high bar, and the specific way candidates clear it wrongly is by generalising past what was measured.

If the passage reports a result for a group over one year, a conclusion about the group in general, or about a wider population, does not follow. If it reports a majority, a conclusion about almost everyone does not follow. Check the scope of the evidence against the scope of the conclusion before anything else; that comparison alone resolves a large share of the section.

Evaluation of Arguments: relevance and importance, not agreement

You judge each argument for or against a proposal as Strong or Weak. Strong means both directly relevant to the question and important. Weak means trivial, tangential, or concerned with something the question did not ask.

• Your own view is not evidence. An argument you agree with is frequently weak, and an argument for a position you dislike is frequently strong. This section is where personal opinion does the most damage.
• True but minor is weak. An argument can be factually correct and still fail on importance.
• Watch for scope drift. Arguments that answer a slightly different question, usually a broader or more emotive version of it, are the most common weak options.

Scoring and what firms look for

Each section is scored on correct answers, results are compared against a graduate or professional norm group, and there is no negative marking. Two things follow. Never leave a blank, since a guess is free. And a raw score means little on its own: law firms typically set a cut-off somewhere around the 60th to 80th percentile of the norm group, so what matters is your position relative to other graduates rather than the fraction you got right.

How to practise so it actually helps

Do the sections separately and diagnose which standard of proof you keep misapplying, because almost everyone has exactly one recurring confusion. The two most common are treating Interpretation like Deduction, which makes you far too strict, and treating Inference like a general knowledge question, which makes you far too generous. Once you can name your own error, a handful of focused sets fixes more than hours of mixed practice.

Each section is available as a separate timed drill so you can repeat the one you are weakest on rather than resitting a full appraisal every time.