Situational Judgement Tests: How They Work, How They’re Scored and How to Practise

A situational judgement test puts you in front of a realistic work problem and asks what to do about it. It isn't testing what you know; it's testing how you weigh up a situation. That makes it different from an aptitude test, and it means preparing for it is different too.

What an SJT looks like

Each question describes a scenario, often a few sentences, sometimes a short video:

a colleague has missed a deadline that affects your work;
a customer is angry about something that wasn't your fault;
your manager has asked for two things that can't both be done by Friday.

You then see several possible responses, typically four or five, and judge them.

The answer formats

Different tests ask you to judge the options in different ways:

Rate each option, for example from Counterproductive to Effective.
Choose the best and the worst, or the most and least likely.
Pick the best, or the best and second best.
Rank all the options.

Research summarised by HumRRO found that rating every option gives the most information and the most reliable results, and that candidates tend to like it better too.

Cohesyve

Practise this before it counts

Run the same kind of AI job simulation employers use and get a scored report on where you stand. 5 free assessments a month, no card required.

Start practising free →

"Should" or "would"?

Read the instruction carefully, because two common wordings ask different questions:

"What should you do?" asks what's best: your knowledge of good practice.
"What would you do?" asks what you'd actually do: your habits.

Experts recommend "should" wording for high-stakes hiring, and one study found people scored lower when asked what they "would" do. If the instruction says "should", answer with the best response, even if it isn't your usual one.

How SJTs are scored

There's no universal answer key. Most SJTs are scored against a key built from the judgements of experienced people in the role, who agree which responses are best and worst; the research summary recommends 10 to 12 such raters. Some keys are built from data instead, by looking at which answers good employees give, and these work about as well.

How the answers are scored matters. In one study, researchers scored the same SJT 28 different ways and found its reliability ranged from about 0.33 to 0.73 depending on the method.

Who uses them

UK Civil Service: the Civil Service Judgement Test (CSJT).
US federal hiring: listed as a selection method in the Office of Personnel Management's guidance.
Military selection and promotion.
Medical school admissions: the UK UCAT includes an SJT section.
Employers: Amazon's Work Simulation, part of its online assessment, uses the SJT format.

Two examples in detail

The UK Civil Service Judgement Test. Three scenarios per behaviour being assessed, some on video. For each scenario you rate four actions as Counterproductive, Ineffective, Fairly effective or Effective. It's untimed, at about two to four minutes per scenario, and a self-rating section makes up 15% of the score. Results are a percentile, each job's pass mark is set after applications close, and you can take it again on a future application.

The UCAT's SJT section. 69 questions in 26 minutes, reported in Bands 1 to 4, where Band 1 is the best.

How to prepare

Learn what the organisation values. SJT keys reflect what the employer considers good judgement. For the Civil Service that's its Success Profiles behaviours; for Amazon, its Leadership Principles; for a hospital, its stated values. Read them before you start and think about what they mean in practice.

Practise the format. SHL Direct, the UK Civil Service and the UCAT all publish free practice SJTs. The first few are disorienting; after a dozen, you'll recognise the structure.

Judge each option on its own. In rating formats, options aren't competing with each other. Two responses can both be effective, and none may be perfect.

Look for the responses experienced people trust. Across most SJTs, the strongest options tend to deal with a problem directly and early, involve the right person, follow policy, keep the customer or patient in mind, and stay within your own authority. The weakest tend to ignore the problem, pass it to someone else without reason, or overreach.

Don't overthink the scenario. Answer from the information given. Inventing details ("maybe the manager is on holiday") changes the question you're answering.

Be consistent. Some tests include a self-rating section or similar scenarios. Answers that come from a clear view of good practice stay consistent without effort.

If you're building one for hiring

Base scenarios on real situations from the job, the ones that separate good performers from the rest. Write four or five plausible responses for each; obviously wrong options add nothing. Have experienced people in the role rate the responses to build the key, use "should do" instructions for high-stakes decisions, and pilot the test before relying on it.

A static SJT can also be shared once it's in use. One alternative is a simulation in which the scenario unfolds as a conversation and adapts to what the candidate says, which is how Cohesyve, which publishes this guide, builds its chat and voice role-plays.

Sources

Checked 2 October 2026.