Situational Judgement Test (SJT)

Scenarios come from the job, not from imagination. Developers run critical-incident interviews with people doing the work, collect situations that actually happened and mattered, then write response options drawn from what real employees did. A panel of experienced performers rates those options, and the agreed ratings become the scoring key. Validation follows: pilot the items on a sample, drop any that fail to discriminate, check that the test relates to job performance, and examine results across demographic groups for adverse impact. That last step is not optional, and it repeats whenever the test or the role changes materially. Off-the-shelf instruments carry technical manuals documenting this work. Ask for the manual before buying, and be sceptical of any publisher who cannot produce one for the role family you are hiring into.

A situational interview question is live and open-ended. An interviewer describes a hypothetical, the candidate answers in their own words, and the interviewer scores that answer against a rubric in the moment. An SJT is fixed and pre-scored: everyone sees identical scenarios, picks from identical options, and marking involves no human judgement at the point of assessment. The trade-off is straightforward. Interview questions capture nuance, follow-up and reasoning a candidate volunteers. SJTs give consistency, throughput and comparability no human panel can match at volume. They complement rather than substitute for each other. A common pattern uses an SJT early to rank a large applicant pool, then situational questions later to probe the reasoning behind the pattern of answers somebody gave.

High-volume hiring where the job is well understood and behaviour matters more than credentials. Customer service, retail, contact centres, healthcare support, graduate schemes and early-career programmes are the usual homes. Applicants in those settings often have thin work histories, so resume review has little to read, while an SJT gives every candidate the same chance to show judgement. They fit poorly where roles are highly individual, where the right answer depends on context a short scenario cannot carry, or where senior candidates will resent a multiple-choice exercise. Volume also has to justify the build cost, since a properly developed test is a project rather than a purchase. Where the volume exists, an SJT paired with automated resume screening does ranking work no team can do by hand.

Tell them what the test measures, roughly how long it takes, whether it is timed, how the result feeds the decision, and that accommodations are available. Publish sample items so nobody is guessing at the format. Openness costs nothing here, because these are not memorisation tests: a candidate who has seen the format simply performs closer to their real level instead of being penalised for unfamiliarity. Say plainly whether the score is a hard cut-off or one input among several, since candidates rejected by an unexplained number talk about it publicly. Give results within a stated window, and automate the invitation and follow-up through your applicant tracking system rather than sending them by hand at volume.