3 min read

What a mock interview can and cannot measure

A scored rehearsal evaluates the answer in front of you. That's useful and narrower than it sounds.

  • mock interview
  • interview practice
  • interview feedback
  • rubric

A mock interview is a measurement, and the first question about any measurement is what is being measured. In a rehearsal with a rubric, it is measuring the answer: whether the incident is specific, whether the trade-off is named, whether the result is stated, and whether the delivery holds together. It is not measuring the candidate.

What the rubric can see

Structured interviews are built from a job analysis, and the rating scale is anchored to observable behaviour rather than to a general impression. The US Office of Personnel Management's guide is explicit that interviewers should rate what the candidate describes, not how the candidate behaves during the conversation. That distinction is the reason a rehearsal can be useful at all: if the score depended on chemistry, practising the answer would be pointless.

A rubric scores the answer on dimensions the posting already implies. A question about a production incident is graded based on what broke, what you did, what you rejected, and what changed. These are facts about the incident, and a rehearsal can tell you if you said them.

What it cannot see

The score cannot see the internal candidate, the panel member's former colleague, or the role that was re-scoped after the posting went live. Those are the reasons a good answer can still lose, and a mock interview that implied otherwise would be dishonest. The honest interpretation of a low score is that the answer has a specific flaw, not that the interview was lost. A rejected application is not a failed interview is the longer version of the limit.

A rehearsal also cannot assess the interviewer's private weighting. One panel values a crisp summary; another wants the mechanism. The rubric makes the criteria visible, which is the best available approximation, not a copy of the room.

Feedback is the active ingredient

The evidence on practice is about feedback, not repetition. A meta-analysis of 435 studies and 994 effects found a small to moderate average effect of d = .48, with heterogeneity high enough that the kind of feedback matters more than whether feedback exists. Practising the same answer without a reader is repetition. Practising it against a scored rubric is feedback.

Interview training also reduces anxiety and raises self-efficacy, and that is a real benefit. It is not the same benefit as a higher score. A study of 15 participants found a mock interview lowered salivary cortisol and marginally raised confidence, and a separate line of work found interventions can reduce anxiety without improving evaluator-rated performance. Confidence is worth having. It is not evidence that the answer has improved.

How to use a rehearsal

Rehearse one question at a time, with the posting open. Name the requirement the question is testing, answer the question, then read the score against the requirement rather than against your memory of how it felt. Reading the interview questions off the job posting is the mapping step; STAR is the wrong half of STAR is the reminder that the structure is a thinking tool, not a script.

Then change one thing. A rehearsal that yields five findings and no next steps is entertainment. The useful output is a single sentence: name the trade-off you did not name, add the number you could not find, or cut the preamble before the incident.

The honest limit

No published source measures how much a scored rehearsal improves a real interview outcome. The feedback evidence measures task performance in other contexts, the anxiety evidence measures how people feel, and the mock-interview literature is limited. What is documented is more specific: feedback with a specific target alters behaviour, and a rubric makes the target specific. That's enough to justify the practice. It is not enough to promise the offer.

Where the evidence stops, say so. The rehearsal is a tool for identifying the next thing to fix, not a prediction.

Related