The Research Foundation
Structured interviewing is not a recent innovation from Silicon Valley. It is grounded in decades of research in industrial-organizational (I/O) psychology, the branch of psychology dedicated to understanding human behavior in the workplace. The most influential study, the Schmidt and Hunter (1998) meta-analysis, synthesized 85 years of research and remains one of the most cited papers in personnel selection.
Their finding: structured interviews are twice as predictive of job performance as unstructured interviews. The validity coefficient for structured interviews is 0.44, compared to 0.20 for unstructured ones. In practical terms, this means a structured interview helps you identify a top-quartile performer roughly four times more reliably than an unstructured conversation.
Why Structure Improves Prediction
The scientific explanation for structure's effectiveness comes down to two factors: standardization and relevance.
Standardization Reduces Noise
In measurement theory, every assessment has signal (true information about the candidate) and noise (irrelevant variation). Unstructured interviews are dominated by noise: different questions per candidate, different evaluation criteria per interviewer, different moods across the day. Standardization, asking the same questions in the same order with the same rubric, strips away this noise and lets the signal emerge.
Google's internal research through the re:Work program confirmed this at scale. When they standardized their interview process, inter-rater reliability increased significantly, and the process saved interviewers an average of 40 minutes per interview in preparation time.
Relevance Increases Signal
Structured questions are designed to elicit information about job-relevant competencies. Each question maps to a specific skill or behavior that matters for the role. This targeted approach means every minute of the interview is generating useful data, unlike unstructured conversations where significant time may be spent on topics with no predictive value.
The Two Validated Question Types
Behavioral Description Interviews (BDI)
Developed by Tom Janz in 1982, BDIs ask candidates to describe specific past experiences. The theoretical basis is the "behavioral consistency" principle: how someone behaved in the past is the best available predictor of how they will behave in the future.
Research shows BDIs have a validity of 0.45 to 0.48, making them among the most predictive single interview techniques. They work because they force candidates to provide verifiable evidence rather than aspirational claims.
Situational Interviews (SI)
Developed by Latham, Saari, Pursell, and Campion (1980), situational interviews present hypothetical but realistic scenarios and ask candidates what they would do. The theoretical basis is goal-setting theory: behavioral intentions predict behavior.
Situational interviews have a validity of 0.35 to 0.40. They are particularly useful for entry-level roles where candidates may lack directly relevant past experience, and for assessing how candidates would handle scenarios specific to your organization.
What Research Says About Common Interview Practices
First Impressions
Research by Barrick, Swider, and Stewart (2010) showed that rapport-building during the first minutes of an interview has no predictive value for job performance but significantly influences the interviewer's final evaluation. In other words, first impressions add noise without signal. Structured interviews mitigate this by anchoring evaluation to specific responses rather than global impressions.
Interviewer Experience
Surprisingly, research shows that experienced interviewers are not significantly better at predicting job performance than novices when using unstructured interviews. Both are poor. However, experienced interviewers using structured formats perform significantly better. The structure, not the experience, is the active ingredient.
Panel vs. Individual Interviews
Meta-analyses show that panel interviews have higher validity than individual interviews (0.46 vs. 0.42) because they average out individual biases. However, the practical advantage of panels must be weighed against scheduling difficulty and candidate intimidation. An effective compromise is independent evaluation by multiple reviewers of the same async video response.
Follow-Up Probing
Research by Campion, Palmer, and Campion (1997) showed that structured probing questions (e.g., "Can you tell me more about the outcome?") improve validity because they help candidates provide more complete evidence. Unstructured probing (following interesting tangents) does not improve validity and may reduce it by introducing inconsistency.
Applying the Science: Practical Takeaways
- Use behavioral and situational questions exclusively. Eliminate brainteasers, self-assessment questions, and open-ended prompts like "tell me about yourself."
- Develop anchored rubrics for every question. The rubric should be created before any interviews are conducted, not adapted afterward to match preferred candidates.
- Minimize the influence of first impressions. Score each answer independently as it occurs. Do not form a global impression and retrospectively assign scores.
- Use multiple independent evaluations. Whether through panel interviews or async video reviews, more independent data points produce more accurate assessments.
- Train interviewers on the format, not on "reading people." Intuition-based interviewer training has no evidence base. Process-based training improves outcomes.
The Science-Technology Connection
The practical challenge of structured interviewing has always been enforcement. Interviewers know they should follow the rubric but drift toward conversational habits under time pressure. Technology solves this by making structure the default rather than the aspiration.
StormInterview embeds the science into the workflow: structured questions, anchored rubrics, independent scoring, and AI-assisted evaluation that applies your rubric consistently across every candidate. The research is clear. The technology makes it practical.
Try StormInterview and apply four decades of hiring science to your next interview round.