For decades, hiring advice repeated one ranking: general cognitive ability tests predict job performance best, so start there. A 2022 review changed that ranking, and the change matters for anyone choosing how to assess candidates.
This article explains the findings in plain terms and what employers commonly do with them. Hiring assessments are subject to employment law. Before you change a selection process, review it with your employment counsel.
What the research found
Sackett, Zhang, Berry and Lievens, writing in the Journal of Applied Psychology in 2022, revisited the research on hiring predictors. They found that structured interviews had the highest mean operational validity (r = .42), followed by job knowledge tests (.40), empirically keyed biodata (.38) and work sample tests (.33), with cognitive ability at .31.
Here, r is a correlation: a number that shows how closely scores on a hiring tool track later job performance. Zero means no relationship. Higher numbers mean scores and performance move together more closely. Operational validity is the researchers’ estimate of that relationship as the tool is used in real hiring.
None of these numbers is close to perfect. Every tool on the list misses something about performance. That is a reason to combine tools with care, not to give up on measurement.
What not to conclude
It is tempting to read the list as a shopping order: buy structured interviews, drop everything else. The research does not say that.
The numbers are averages across many studies, jobs and organizations. Your role, your candidates and your version of each tool will differ. A cognitive ability test still predicts performance to a degree, and a poorly built structured interview can do worse than the average suggests.
The gaps between the top tools are also small. Structured interviews, job knowledge tests and biodata sit close together. The practical message is about quality and fit: choose tools that measure what your job requires, build them carefully, and check how they work for your own hires over time.
Structure is the common thread
Look at the top of the list. Structured interviews and job knowledge tests share one trait: they are built from the job, and every candidate gets the same questions judged the same way.
An unstructured interview is a conversation. Each candidate gets different questions, and each interviewer weighs answers by their own sense of a good fit. A structured interview fixes the questions, the order and the scoring rules in advance.
The lesson is less about which tool to buy and more about how to build whichever tool you use. A work sample with a vague rubric loses much of what makes it useful. A structured interview with anchored scores keeps it.
Build a structured interview
A structured interview can be built in a few days for one role. Most follow the same pattern.
- Start from the job. List the five to eight tasks or situations that matter most, from a job analysis or interviews with people who do the work.
- Write the questions. For each task, write one or two questions about past behavior (“Tell me about a time…”) or a realistic situation (“What would you do if…”).
- Anchor the scores. For each question, describe what a weak, adequate and strong answer looks like, in concrete terms.
- Train the interviewers. Walk through the anchors together and score two or three sample answers until scores agree.
- Score independently. Each interviewer rates every answer alone, before any discussion.
- Combine by rule. Add or average scores by a set formula. Discussion can resolve wide gaps, not override the scores.
Keep notes on each answer. They help explain a decision later and show interviewers where their scores drift.
Where work samples fit
A work sample asks the candidate to do a slice of the real job: diagnose a fault, draft a client letter, reconcile an account, respond to a mock call. It scored .33 in the review, below structured interviews and knowledge tests but above cognitive ability.
Work samples have other strengths worth weighing. Candidates tend to see them as relevant. They give the hiring team a direct look at the work. And they double as a preview, so candidates learn what the job involves before accepting it.
They also take more time to build and to score. A short, well-chosen task scored with a clear rubric usually beats a long one scored by impression. Our article on designing a work sample test covers the build in detail.
Combine tools with a plan
Most hiring processes use several steps. The research gives a reason to put structured, job-based measures at the center and to think about what each step adds.
- Screen on requirements. Licenses, certifications and legal eligibility, checked consistently.
- Test knowledge the job needs. A short job knowledge test where the role depends on specific knowledge.
- Observe the work. A work sample for the tasks that matter most.
- Interview with structure. Anchored questions on situations the test cannot cover.
- Decide by rule. Weights for each step, set before anyone is scored.
Write the weights down before the first candidate. A rule set in advance is easier to explain than a judgment reached after.
Example: a field service hire
Example scenario. Say a regional utility hires 20 field technicians a year. Its current process is a resume review and a 30-minute conversation with a supervisor.
It rebuilds the process in four steps. A licensing check. A 40-item job knowledge test on electrical safety and meter types. A 45-minute work sample: diagnose a fault on a training panel and write the service note. A structured interview with six anchored questions on customer situations and safety calls. Each step has a set weight, agreed with HR and counsel before the first candidate.
The supervisors who once ran free-form interviews now score anchored answers. After the first round, they compare scores on a recorded sample answer and find two of them scored a full point apart. A short calibration session follows. This is a scenario, not a client, and the results of any process depend on how it is built and used.
Watch fairness and the law
Any hiring tool can affect groups of candidates differently. Programs commonly monitor pass rates by group at each step and review steps where differences appear, with help from counsel and a qualified industrial and organizational psychologist.
Job relevance sits at the center of how hiring tools are commonly reviewed, and it is also the center of validity. The Standards for Educational and Psychological Testing (AERA, APA and NCME, 2014) define validity as “the degree to which evidence and theory support the interpretations of test scores for proposed uses of tests.” A tool that is valid in one company’s research is not automatically valid for your role. Keep the evidence that ties each step to your job.
Accommodations apply in hiring too. Tell candidates how to request one, and configure approved accommodations before the session.
What to do this week
- Write down every step in one role’s hiring process and what each step is meant to show.
- Pull the questions your interviewers ask. If they vary by interviewer, that is the first fix.
- Pick the three tasks that matter most in the role and ask whether any step observes them.
- Draft anchors for one interview question: weak, adequate and strong answers.
- Book time with counsel to review any change before it reaches candidates.
- Read up on inter-rater reliability in our assessment glossary, since it applies to interviews as much as to essays.
Our skill evaluation systems build job-specific work samples, simulations and structured interview kits for one role at a time (indicative, from $4,500 per role). The Role Readiness Assessment program runs that build with your team, and our page for employers and HR teams shows how it fits a hiring process.
