For hiring managers

What is a situational judgment test (SJT)? Formats, scoring keys and how employers use them

On this page
  1. How does a situational judgment test work?
  2. Does it matter whether the test asks "what would you do" or "what should you do"?
  3. How is a situational judgment test scored?
  4. How is an SJT different from a situational interview, a work sample or a personality test?
  5. How well do SJTs predict performance, and what about adverse impact?
  6. What legal rules apply to situational judgment tests?
  7. How do you build a situational judgment test for your own roles?
  8. What should staffing agencies know?
  9. A copy-ready SJT item template
  10. Questions people ask

A situational judgment test (SJT) is a hiring test that describes a realistic work problem, in text or video, and asks the candidate to choose, rank or rate possible ways of handling it; the answers are scored against a key built from experienced job holders' judgments of the best and worst responses. The U.S. Office of Personnel Management classes SJTs as "low-fidelity simulations," because candidates judge a situation rather than act it out, and says they are "particularly effective measures of managerial and leadership competencies." As of November 2026, OPM offers a Supervisory Situational Judgment Test through its USA Hire program, and the UK Civil Service runs the Civil Service Judgement Test, an online SJT; private employers buy SJTs from test publishers or build their own.

Not legal advice. The legal points on this page summarize the federal statutes and regulations linked below as checked in November 2026. The wider law on employment tests is on are personality tests legal in hiring.

How does a situational judgment test work?

Every SJT item has four parts: a scenario, a set of response options, an instruction that says what the candidate is to do with the options, and a scoring key. OPM's situational judgment tests page says the scenarios and options "are typically based on critical incidents generated by subject matter (i.e., job) experts," and lists five common ways candidates may be asked to respond:

  1. what they would do in the situation;
  2. what they would be most and least likely to do;
  3. which response is the best among several options;
  4. which responses are the best and second-best;
  5. what would most likely happen next, in the situation or as a result of a decision.

Tests can be linear, where everyone sees the same items in the same order, or interactive, where later scenarios branch depending on earlier answers. They can be delivered on paper, by computer or by video. OPM publishes a sample SJT item.

The UK Civil Service's guidance on its test shows a current format in detail (GOV.UK): candidates see three scenarios for each behavior being assessed, rate each of four possible actions as Counterproductive, Ineffective, Fairly Effective or Effective, and are told to judge each action on its own, since more than one can get the same rating. Some scenarios are short videos, which come with subtitles, British Sign Language and written transcripts. The test is untimed, though the guidance says most people take two to four minutes per scenario, and the score is reported as a percentile against a representative group of applicants.

OPM's Supervisory Situational Judgment Test is a multiple-choice SJT for first-line supervisors in the federal government that targets up to eight competencies: accountability, conflict management, customer service, developing others, flexibility, interpersonal skills, problem solving and strategic thinking. Candidates take it unproctored at the time of application, and OPM describes it as a hurdle whose results can be complemented by structured interviews or other valid assessments later in the process.

Does it matter whether the test asks "what would you do" or "what should you do"?

Yes. A meta-analysis by McDaniel, Hartman, Whetzel and Grubb (Personnel Psychology, 2007) compared two kinds of instruction: behavioral tendency instructions, which ask how the candidate would likely behave, and knowledge instructions, which ask how effective each response is. The publisher's abstract reports that the instruction changed what the test measured: tests with knowledge instructions correlated more with cognitive ability, and tests with behavioral tendency instructions correlated more with personality. The instruction had little effect on how well the tests predicted job performance, and SJTs added prediction beyond cognitive ability and the Big Five personality traits.

In practice, pick one instruction per test and keep it. "Which response is most effective?" asks whether the candidate knows what good handling looks like; "what would you do?" asks about their habits. Both are legitimate, but a test that mixes them is hard to interpret and harder to defend.

How is a situational judgment test scored?

OPM says SJT scores "are based on subject matter experts' judgments of the best and worst alternatives." A panel of experienced job holders or their managers rates every option independently, and the options they agree on become the key. Candidates then earn points for agreeing with the experts. Three common ways to do that:

  • Pick the best: one point when the candidate chooses the option the experts rated most effective.
  • Most and least: a point for matching the experts' most effective option and another for matching their least effective one.
  • Rate every option: points for how close each of the candidate's ratings is to the experts' rating. This suits the rate-each-action response format the Civil Service test uses.

The total is usually turned into a percentile or band against a comparison group; what norms and percentiles are, and why they matter, is on what is a psychometric test.

A worked example item (made up, for illustration)

ROLE: Customer support team lead
SCENARIO: A customer emails that an order promised for Tuesday has not
arrived. One of your agents promised the date without checking the
warehouse system. The customer wants the shipping fee refunded and says
they will cancel a larger order. Rate how effective each action is.

A. Apologize, check the warehouse system for the real delivery date,
   give the customer that date and offer the refund your policy allows;
   coach the agent privately afterwards.
B. Forward the email to the warehouse team and ask them to reply.
C. Tell the customer the agent made a mistake and has been spoken to.
D. Refund the shipping fee straight away, without checking the order.

EXPERT KEY (4-point scale):  A = Effective   B = Ineffective
                             C = Counterproductive   D = Fairly effective
SCORING RULE (example): 3 points for an exact match, 2 for one step
away, 1 for two steps, 0 for three. Maximum for this item: 12.

A candidate who rates A Effective, B Fairly effective, C Counterproductive and D Effective scores 3 + 2 + 3 + 2 = 10 of 12. The numbers are an example only; a real key comes from your own experts.

How is an SJT different from a situational interview, a work sample or a personality test?

MethodWhat the candidate doesHow it is scoredTypical stage
Situational judgment testChooses or rates given responses to a written or video scenarioAutomatically, against an expert keyEarly, at volume
Situational interview questionDescribes in their own words what they would doAn interviewer, against written anchors (scoring rubric)Phone screen to final round
Work samplePerforms a task from the jobA rubric or right/wrong (work sample tests)Late, for finalists
Assessment centerActs out several exercises, such as role plays and in-basket tasksTrained assessors (assessment centers)Late, senior roles
Personality questionnaireRates statements about themselvesTrait scales against norms (personality tests)Early

The SJT sits between the personality questionnaire and the work sample: more job-specific than the first, much cheaper to run than the second. If you want the candidate's reasoning in their own words, use the interview version; the question bank for that is on interview questions for judgment.

How well do SJTs predict performance, and what about adverse impact?

OPM rates SJTs as having a high degree of content validity, because the scenarios mirror the job, and a "moderately high degree of criterion-related validity." It reports that applicants "often perceive SJTs as being very fair," and that subgroup differences are "typically moderate," with racial differences that "may be smaller than those typically observed for cognitive ability tests." OPM also notes the return on an SJT is high when new hires need strong social and interpersonal skills on day one, and "significantly lower" when those skills can be learned on the job. Average validity and group-difference figures from the 2022 re-analysis of selection research are on types of pre-employment assessments.

Two design choices change those numbers in your own process. First, reading load: a long, dense scenario can turn a judgment test into a reading test, so keep the language at the level the job needs. Second, culture-bound scenarios: a "correct" answer that depends on workplace customs unrelated to the job measures familiarity, not judgment. Check pass rates by group after launch either way.

  • Disparate impact. If an SJT screens out a race, color, religion, sex or national-origin group at a higher rate, the employer must show it is "job related for the position in question and consistent with business necessity" (42 U.S.C. 2000e-2(k)), and it may not use different cutoffs or adjusted scores by group (2000e-2(l)). Run the four-fifths check at the SJT stage, not only at the offer.
  • Validation of "judgment". The Uniform Guidelines list "judgment" and "leadership" among the constructs for which a content-validity strategy "is not appropriate" (29 CFR 1607.14(C)(1)), so an SJT sold as a measure of judgment is expected to have criterion-related or construct evidence behind it. The Guidelines are still printed in the eCFR as of November 2026, but the Justice Department's Office of Legal Counsel found them unlawful in part on June 9, 2026, and OPM has removed references to them from its own rules; where that leaves employers is on cognitive ability tests in hiring.
  • Disability. An SJT built to measure job judgment is not designed to identify a mental disorder, which is the line the EEOC draws for psychological tests that count as medical examinations. The accommodation duty still applies: tests must measure the skill rather than "the impaired sensory, manual, or speaking skills" of a candidate with a disability (42 U.S.C. 12112(b)(7)). Captions and transcripts for video scenarios, screen-reader support and extra time are the usual adjustments; see ADA accommodations in interviews.
  • Automated scoring of open answers. Some newer SJTs ask candidates to type or record an answer and score it with AI. That brings in the rules on automated tools; see how AI interview scoring works and what is an automated employment decision tool.

How do you build a situational judgment test for your own roles?

  1. Start from the job analysis. Pick the three to five competencies where judgment in people situations matters most, for example handling an upset customer or a missed deadline.
  2. Collect critical incidents. Ask experienced job holders and their managers for real situations that were handled well or badly. OPM describes this as the usual source of scenarios.
  3. Write each scenario around one decision. Keep it short, specific to the job, and free of cues about anyone's age, sex, ethnicity, religion or health.
  4. Write four or five plausible options. Every option should be something a real employee might do; an obviously silly option adds nothing.
  5. Key it with experts. Several experts rate each option independently. Keep the options they agree on, rewrite or drop the rest, and record who keyed what.
  6. Pilot it. Try the items on current employees, check how long they take and which items nobody gets right, and fix the reading level.
  7. Decide how the score is used (a pass band, a ranking or a weighted input) before candidates take it, and check pass rates by group after launch.
  8. Refresh items on a schedule: items used online for a long time get shared.

What should staffing agencies know?

An SJT is only as good as its key, and the key belongs to the job. When an agency screens for a client, the scenarios should come from the client's real situations and the client's experienced staff should help key them; a generic "customer service judgment" test may not match how that client wants problems handled. The agency administers the test, so it carries the same duties as an employer for consistency, accommodations and pass-rate monitoring. Send the client the band or score with the test name and version, so both sides can show later which instrument was used.

A copy-ready SJT item template

SJT ITEM (example; one per scenario)
Competency: _____________________  From job analysis item: ______
Source incident (who supplied it, date): ________________________
Instruction: [ ] rate each action  [ ] pick best  [ ] most/least likely
Scenario (one decision, under 120 words, no protected-trait cues):
_______________________________________________________________
Options:
 A. __________________________  Expert rating: ____
 B. __________________________  Expert rating: ____
 C. __________________________  Expert rating: ____
 D. __________________________  Expert rating: ____
Experts who keyed it (min. 3): ____________  Agreement: [ ] yes
Pilot: median time ____ s   Reading level checked [ ]
Accessibility: [ ] text version  [ ] captions/transcript for video
Scoring rule: ________________________________________________
Retire or review by: __/__/____

The template is an example to adapt. Keep the completed items, the key and the pilot results together; they are the record that shows the test came from the job. The same scenarios can be turned into an integrity or honesty measure only with care, because some states restrict honesty tests; see what is an integrity test.

Questions people ask

What is a situational judgment test in a job application?

It is an assessment, usually online, that describes realistic problems from the job in text or video and asks you to choose, rank or rate possible responses. Your answers are compared with a scoring key built from experienced job holders' judgments of the most and least effective responses, and your score is often reported against a group of other applicants.

How is a situational judgment test scored?

Against a key. Subject matter experts, usually experienced job holders or their managers, rate how effective each response option is, and a candidate earns points for agreeing with them: for picking the option they rated best, for matching their most and least effective choices, or for giving each option a rating close to the experts' rating. The total is often converted to a percentile against a comparison group.

What is the difference between a situational judgment test and a situational interview question?

Both present a work scenario. A situational judgment test gives the candidate fixed response options and scores them automatically against a key, so it suits early, high-volume stages. A situational interview question asks the candidate to describe in their own words what they would do, and an interviewer scores the answer against written anchors, which allows follow-up questions but takes interviewer time.

How can a candidate prepare for a situational judgment test?

Take the employer's own practice test if it offers one, read the instructions carefully, and answer the question asked: what you would do and what the best response is are different questions. The UK Civil Service, for example, tells candidates to review each action on its own and notes that more than one action can deserve the same rating.

Are situational judgment tests legal in hiring?

Yes. They are ordinary employment tests, not medical examinations, but the usual rules apply: if a test screens out a race, sex or national-origin group at a higher rate it must be job related and consistent with business necessity, scores may not be adjusted by group, and candidates with disabilities must get accommodations, such as captions or transcripts for video scenarios.