# Structured interviews: what the research says and how to run one

> Structured interviews are the best-supported way to interview candidates, and a small team can build one in an afternoon. What structure means, why it works, and a worked example with questions and rating anchors.

Jennifer at SortTheStack, October 9, 2026. Original: https://sortthestack.com/blog/structured-interviews

A structured interview asks every candidate the same job-related questions, in
the same order, and rates each answer against examples of strong, acceptable
and weak answers written before the first interview. **It is the best-supported
way to interview, and a small team can build one in an afternoon.** You do not
need an HR department, software or a consultant. You need your list of
requirements and a few hours of preparation.

This guide covers what "structured" actually means, what the research says
about it, and how to build and run one, using a worked example. There's a
printable [planning and rating sheet (PDF)](https://sortthestack.com/downloads/structured-interview-guide.pdf)
to build your own, free, no sign-up.

## Why structure is worth the effort

In a 2022 re-analysis of decades of selection research, structured interviews
came out as the single best predictor of job performance among the methods
studied, with an estimated validity of .42. Unstructured interviews came in at
.19. We cover that research, and what those numbers do and do not mean, in
[what actually predicts job performance](https://sortthestack.com/blog/what-predicts-job-performance).

This is not one surprising result. A 2014 review of the structured interview
literature by
[Levashina, Hartwell, Morgeson and Campion](https://doi.org/10.1111/peps.12052)
called it "one of the most consistent findings in the history of research on
the employment interview": structured interviews are "much more reliable and
valid than unstructured interviews". The authors counted twelve meta-analyses
reaching the same conclusion.

The same review found that structure greatly reduced differences in interview
ratings based on race, gender and disability, while noting that much of that
evidence comes from laboratory studies and more field research is needed.

## What "structured" actually means

Most people asked to describe a structured interview say "the same questions
for everyone". That is one part of it. The most widely used framework, from
Campion, Palmer and Campion (1997) and summarised in the 2014 review, lists 15
components in two groups: what you ask, and how you evaluate the answers.

| What you ask (content)                              | How you evaluate (evaluation)                 |
| --------------------------------------------------- | --------------------------------------------- |
| Base the questions on an analysis of the job        | Rate each answer, or rate on several scales   |
| Ask every candidate the same questions              | Use rating scales with examples (anchors)     |
| Limit prompting and follow-up                       | Take notes                                    |
| Use better types of question                        | Use more than one interviewer                 |
| Use a longer interview, or more questions           | Use the same interviewers for every candidate |
| Control ancillary information (resume, test scores) | Don't discuss candidates between interviews   |
| Take the candidate's questions after the interview  | Train the interviewers                        |
|                                                     | Combine ratings by formula, not by discussion |

You do not have to use all fifteen. Across the research the review analysed,
studies used about six on average. The six used most often were: basing
questions on the job, asking the same questions, using better types of
question, anchored rating scales, rating each question, and training. Those six
are a sensible core for a small team, and they are what the method below is
built on.

## Build one in an afternoon: a worked example

We'll use the same fictional hire as our
[guide to screening a pile of resumes](https://sortthestack.com/blog/how-to-screen-a-pile-of-resumes):
Larkspur Partners, an accounting firm hiring a Senior Accountant. Screening
has produced five finalists.

### Step 1: Start from the requirements you screened on

The requirements you wrote for screening are your job analysis in miniature.
Larkspur's were:

- **Must have:** CPA designation; runs a month-end close independently.
- **Nice to have:** prepares working papers for an external audit; experience
  with a cloud accounting system.

A resume can show the CPA. It cannot show _how_ someone runs a close, or what
they do when it goes wrong. That is what the interview is for. Don't spend
interview time re-confirming what the application already proved.

### Step 2: Write five or six questions, two kinds

There are two well-studied question types:

- **Past-behaviour questions** ask what the candidate actually did: "Tell me
  about a time when…"
- **Situational questions** describe a realistic problem from the job and ask
  what they would do.

The 2014 review concluded that both have acceptable validity, that situational
questions mainly measure job knowledge and past-behaviour questions mainly
measure experience, and that they work best used together.

For Larkspur:

| #   | Requirement            | Type           | Question                                                                                                                                                          |
| --- | ---------------------- | -------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| 1   | Runs month-end close   | Past behaviour | Walk me through the last month-end close you were responsible for, from the first day to sign-off. What did you own, and what did others do?                      |
| 2   | Runs month-end close   | Situational    | It's day three of close. A bank reconciliation is out by $14,000 and the partner wants the numbers tomorrow morning. What do you do, in order?                    |
| 3   | Audit working papers   | Past behaviour | Tell me about the last time you prepared files for an external audit. What did the auditors ask for that you hadn't prepared, and what did you change afterwards? |
| 4   | Cloud accounting       | Past behaviour | Describe a time you had to learn a new accounting system or change a process because of one. How did you approach it?                                             |
| 5   | Reviewing others' work | Situational    | A junior accountant's journal entries keep needing correction. You're not their manager. What do you do?                                                          |

Question 5 is about a duty of the job, not a requirement you screened for. That
is fine in an interview: a duty is something you want to see someone handle,
even though you shouldn't reject a resume for not mentioning it.

### Step 3: Decide your follow-up questions in advance

Follow-up questions are where interviews quietly become unstructured again:
one candidate gets gentle prompting, another gets challenged. The research on
probing is still unsettled. The 2014 review found it unclear whether probing
makes ratings more accurate or introduces bias.

The practical compromise, which follows the
[U.S. Office of Personnel Management's guidance on structured interviews](https://www.opm.gov/policy-data-oversight/assessment-and-selection/structured-interviews/structured-interviews.pdf),
is to write the allowed follow-ups down and use the same ones for everyone. OPM
groups them as:

- **Situation:** "Who was involved?" "What led up to it?"
- **Action:** "What was your role?" "What did you do first?"
- **Result:** "What was the outcome?" "What would you do differently?"

Two rules from the same guidance are worth keeping. Don't use leading follow-ups
that hint at the answer you want. And when a candidate says "we did…", ask what
_they_ did.

### Step 4: Write the rating anchors

This is the step most interviews skip, and the one that makes the rest work.
For each question, write what a strong, an acceptable and a weak answer would
sound like, before you meet anyone. The 2014 review concluded that anchored
rating scales improve both reliability and validity. OPM's guidance suggests a
scale of three to seven levels with at least three labelled.

Here is question 1 on a 1–5 scale, with three anchors:

| Rating        | What the answer sounds like                                                                                                                                                                                 |
| ------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| 5: Strong     | Describes owning the close end to end: a checklist or calendar they maintained, reconciliations they performed and reviewed, how they handled a late or wrong item, and sign-off. Clear about who did what. |
| 3: Acceptable | Describes a close they took a substantial part in, with specific tasks they owned, but someone else coordinated it or signed off.                                                                           |
| 1: Weak       | Describes the close in general terms, or only tasks done for someone else, and cannot say what they personally owned.                                                                                       |

Ratings of 2 and 4 are for answers between two anchors. Write anchors for every
question. If two of you are interviewing, write them separately first, then
agree on them together. That is a small version of what OPM recommends with a
panel of experts.

## Run it the same way for everyone

- **Tell candidates the format in advance.** The 2014 review notes evidence
  that applicants react better to structured interviews when they know the
  questions or topics beforehand. Telling everyone the same thing keeps it fair.
- **Same questions, same order, same allowed follow-ups, same length of time.**
- **Take notes on what was said**, not your impression of it. Notes are what you
  rate from, and what lets you explain a decision later.
- **Rate each answer straight after it is given**, using the anchors, before
  moving to the next question. Waiting until the end lets one strong or weak
  answer colour the rest.
- **Take the candidate's questions at the end.** That conversation matters, but
  keep it out of the ratings.
- **Don't ask about anything unrelated to the job.** OPM's guidance lists age,
  sex, race, national origin, religion, marital or family status, and
  disability or health status, and tells interviewers never to ask or comment
  about any of them. Employment law differs by country, state and province, so
  check what applies where you hire.

## Score and decide

With one interviewer, add up the ratings. With two or more, each person rates
alone first, then you combine the ratings by a rule you chose beforehand, such
as the average. Then discuss. Our
[candidate scorecard template](https://sortthestack.com/blog/candidate-scorecard-template) is built for
exactly this step.

The order matters. A meta-analysis by
[Kuncel and colleagues (2013)](https://doi.org/10.1037/a0034156) found that
combining information mechanically, by formula or fixed rule, predicted job
performance substantially better than holistic judgement, even when the people
judging were experts. Discussion is still useful for catching what someone
misheard or missed. If you disagree by more than a point on a question, go back
to your notes and the anchor for that question, not to your overall impression
of the candidate.

## What a structured interview won't do

- **It won't make every interview good.** Its estimated validity varies a lot
  from study to study (see
  [what predicts job performance](https://sortthestack.com/blog/what-predicts-job-performance)), which
  suggests the quality of the questions and anchors matters. A
  loosely structured interview will not get you the average result.
- **It won't stop candidates presenting themselves well.** The 2014 review found
  applicants use impression management extensively in structured interviews,
  though structure reduces its effect on ratings.
- **Some people won't like it.** The review found applicant and interviewer
  reactions to structure are not consistent. Explaining the format up front
  helps.
- **It is one input, not the decision.** Pair it with a short work sample if
  the job allows, and with what screening showed. No single method predicts
  everything.

One more finding is worth knowing. A study of U.S. federal court cases by
[Williamson and colleagues (1997)](https://doi.org/10.1037/0021-9010.82.6.900)
found that more structured interviews, especially those that were objective
and job-related, were associated with verdicts favouring the employer in
discrimination claims. That is an association from U.S. cases decided decades
ago, not legal advice, but it points the same way as the rest of the evidence.

## Where SortTheStack fits

SortTheStack doesn't run interviews. It does the step before: it reads every
application against your requirements and quotes the evidence it found. Where
a requirement is marked unclear, that is often the best interview question you
have. "Your resume mentions the year-end close at your current firm. Did you
run it, or support it?" is question 1 above. You can
[see how the screening works](https://sortthestack.com/product).

## Sources

- Levashina, J., Hartwell, C. J., Morgeson, F. P., & Campion, M. A. (2014). The
  structured employment interview: Narrative and quantitative review of the
  research literature. _Personnel Psychology, 67_(1), 241–293.
  [doi.org/10.1111/peps.12052](https://doi.org/10.1111/peps.12052)
- Campion, M. A., Palmer, D. K., & Campion, J. E. (1997). A review of structure
  in the selection interview. _Personnel Psychology, 50_(3), 655–702.
  [doi.org/10.1111/j.1744-6570.1997.tb00709.x](https://doi.org/10.1111/j.1744-6570.1997.tb00709.x).
  The 15 components above are as summarised in Levashina et al. (2014).
- Sackett, P. R., Zhang, C., Berry, C. M., & Lievens, F. (2022). Revisiting
  meta-analytic estimates of validity in personnel selection. _Journal of
  Applied Psychology, 107_(11), 2040–2068.
  [doi.org/10.1037/apl0000994](https://doi.org/10.1037/apl0000994)
- Kuncel, N. R., Klieger, D. M., Connelly, B. S., & Ones, D. S. (2013).
  Mechanical versus clinical data combination in selection and admissions
  decisions: A meta-analysis. _Journal of Applied Psychology, 98_(6),
  1060–1072. [doi.org/10.1037/a0034156](https://doi.org/10.1037/a0034156)
- Williamson, L. G., Campion, J. E., Malos, S. B., Roehling, M. V., & Campion,
  M. A. (1997). Employment interview on trial: Linking interview structure with
  litigation outcomes. _Journal of Applied Psychology, 82_(6), 900–912.
  [doi.org/10.1037/0021-9010.82.6.900](https://doi.org/10.1037/0021-9010.82.6.900)
- U.S. Office of Personnel Management,
  [Structured Interviews](https://www.opm.gov/policy-data-oversight/assessment-and-selection/structured-interviews/structured-interviews.pdf)
  (training guide), accessed 9 October 2026.