Does video interviewing actually work? What the research says

Ask a video interview vendor for proof and you’ll get the same citation almost every time: Schmidt and Hunter’s meta-analysis found structured interviews predict job performance at up to .51, one of the strongest tools personnel psychology has ever measured. It’s a real number, and it’s the most cited paper in the field. It just isn’t a number about video.

Schmidt and Hunter never tested a webcam. Their research, like the decades of work before and after it, measured structure: the same questions, asked the same way, scored against the same criteria, for every candidate. Video is a way of delivering that structure. It didn’t earn the number by itself.

So does video interviewing actually work? The honest answer needs two different bodies of research: the deep, well-replicated literature on structured interviews in general, and a newer, thinner literature that specifically tests whether recording answers instead of taking them live changes anything. Both say something real. Neither says quite what the sales page implies.

The number every vendor quotes was never about video

Structured interview validity is one of the best-documented findings in industrial psychology. Schmidt and Hunter’s 1998 meta-analysis (Psychological Bulletin, 124(2), 262-274) put structured interviews at about .51 against job performance, versus .38 for the improvised, unstructured kind. We’ve written about what structure actually means in more depth, but the short version is four properties: the same questions, in the same order, scored against the same rubric, by evaluators who score before they debrief.

In 2022, the field corrected its own homework. Sackett, Zhang, Berry, and Lievens (Journal of Applied Psychology, 107(11), 2040-2068) found that older meta-analyses had over-corrected for range restriction, a statistical adjustment that assumed you were hiring from a much wider pool than you actually were. Once removed, most validity estimates dropped. Structured interviews landed at .42 and kept the top spot. Cognitive ability, the old front-runner at .51, fell to .31. We broke down the full corrected table elsewhere, but the headline holds here too: structure still wins, and the gap between it and everything else actually widened.

Neither paper mentions a camera. The candidates in these datasets sat across a table or a phone line from a trained interviewer. The number describes what happens when you standardize the questions and the scoring. No study anywhere found that a webcam makes an interview more valid. Every study that found a validity number found it in the question set and the scorecard.

What the video-specific research actually asked

Because the classic number is silent on delivery format, a separate, much younger body of research has tried to answer the real question: does an interview conducted asynchronously, recorded on a candidate’s own time and reviewed later, preserve what made a live structured interview valid in the first place.

The mechanism gives a reason to expect yes. Levashina, Hartwell, Morgeson, and Campion’s 2014 review in Personnel Psychology traced the source of structured interview validity to four design choices: the same interview questions, the same order, a fixed rubric, and evaluators scoring independently before they compare notes. None of those four require a person in the room. If anything, removing the interviewer removes the one part of a live interview most likely to drift off script, warm up to one candidate, or ask a harder follow-up of somebody they’ve decided they don’t like.

The direct evidence is newer and thinner, but it’s starting to show up. A 2025 study in Human Performance (Renier, Schmid Mast, He, and Garner) put the same past-behavior questions in front of candidates face-to-face and through one-way video, then checked both against actual work-sample performance. Overall interview performance predicted the work sample in both formats. So did performance on the specific question aimed at that work sample. It’s one study, not a fifty-year meta-analysis, so read it as early and encouraging rather than settled. The field has grown enough to earn its own dedicated research now too. The International Journal of Selection and Assessment ran a full special issue on asynchronous video interviews in 2025, which is usually a sign a topic has moved from niche curiosity to a real research agenda.

Where the evidence actually runs thin

The recording part of the transfer looks solid so far. What’s still short on evidence is the layer riding on top of it: the AI model scoring what got recorded, the exact thing every vendor pitches as their edge.

A 2024 design-science paper on asynchronous video interview systems put this plainly: AI-driven assessment products have moved faster than the evidence base for how to build them responsibly. The authors’ fix wasn’t to abandon automated scoring. It was a set of design principles, score behavioral cues that are hard to fake, train on validated labels instead of whatever the model happened to learn, and keep the scoring tied to criteria a human defined rather than criteria the model invented on its own.

That distinction matters more than it sounds. “The research says video interviewing works” often quietly means “the research says structured interviews work, and we added a model on top, trust us.” Those are two different claims with two very different amounts of evidence behind them. It’s the same distinction we walk through when people ask whether AI can screen candidates fairly: the honest version keeps the AI in the surfacing role. It transcribes, scores against the criteria you set, and shows its reasoning. It doesn’t get to invent the bar a candidate has to clear. That’s not a hedge. It’s the part the newest research says is still being figured out, so it’s the part that should still have a person checking the work.

The “candidates won’t do it” objection, checked against the data

Even if the science holds up, the practical worry an owner-operator has isn’t academic. It’s losing a good candidate because they wouldn’t sit for a recorded video, the same worry behind most of the adoption numbers you’ll see quoted elsewhere.

The evidence here is mixed, which is the honest answer rather than a marketing one. A 2016 study in Frontiers in Psychology found candidate attitudes toward asynchronous video interviews were driven mostly by how useful and how easy to use candidates found the format, not by personality traits, and rated it comparably to accepted methods like personality inventories, though on a small sample of students in a hypothetical scenario. A broader 2024 study in Human Resource Management (Tilston and colleagues) found reactions to asynchronous video interviews skew more negative than traditional interviews on average, but traced most of that gap to specific design decisions rather than the format itself.

That tracks with what we hear from teams running one-way interviews at volume, and it’s the same trade-off we walk through in our guide to one-way video interviews for small business hiring. The format earns its place on roles that pull a big, loosely-qualified crowd, where a short recorded answer replaces a phone screen instead of adding a new round on top of it. On a role with a thin candidate pool, where you want to talk to almost everyone anyway, skipping the recording and going straight to a conversation is usually the better trade. The research doesn’t say video always beats a phone call. It says structured beats unstructured, and video is one way to run structured at a volume one or two people can handle.

What actually makes a video screening step trustworthy

If the validity belongs to the structure and not the camera, the practical test for any video interview tool is simple: does every candidate get the same questions, and is the scoring applied the same way to all of them.

That’s the design question behind how we built Truffle’s one-way video interviews into a wider candidate screening workflow. Resume review cuts the obvious mismatches before anyone records anything. Candidates who clear that bar answer a one-way video interview built from a shared question set the employer approved before the first candidate applied, not a live conversation that drifts wherever the interviewer’s curiosity goes. Each question carries a scoring criterion the employer set at intake, an interview rubric in everything but name, so the score isn’t a gut read on how someone came across. AI transcribes each response, summarizes it, and scores it against those criteria, the same way for every candidate, then surfaces the match percentage and a Candidate Shorts highlight clip so a reviewer isn’t watching every recording start to finish. For roles where communication style and capability both matter, a talent assessment, a validated Big Five personality read or a situational judgment scenario, sits alongside the video and the resume, one more signal built on watching someone work rather than a self-report you have to take on faith.

None of that is the AI deciding who’s good. It’s the AI holding the criteria steady across every candidate and showing its work, which is the part the structured-interview literature has always said mattered. You still watch the shortlist. You still make the call.

The due-diligence question worth asking instead

The next time a vendor cites Schmidt and Hunter, or the corrected 2022 numbers, or any research at all to sell you on video interviewing, the useful response isn’t to argue with the study. It’s to ask two questions that decide whether the study’s number has anything to do with their tool: does every candidate answer the same questions, and does the scoring apply the same criteria to everyone, every time. If the answer to either is no, the citation is decoration. The number belongs to the process the researchers actually ran, not to whichever product happens to mention it in a sales deck. It’s worth asking before you read any one-way interview software review, because the star rating won’t tell you either.

Every time the field corrects its own numbers, the correction moves the same direction: further reward for the tools that built structure in from the start, and further exposure for the tools that only ever sold the camera. That’s less a fact about video and more a fact about where the evidence is headed next. The vendors who can answer the two questions above will look better with each correction, not worse. The ones who can’t were never really making a research argument. They were making a video argument and hoping nobody checked the difference.

Frequently asked questions about video interviewing research

Does video interviewing actually work for hiring?

It works to the extent it’s structured. The research most vendors cite (Schmidt and Hunter’s structured interview validity of roughly .51, corrected to .42 by Sackett et al. in 2022) tested the same questions and scoring applied to every candidate, not video specifically. Newer research that directly tests one-way video interviews finds the transfer mostly holds, but the format alone doesn’t earn the number. The structure behind it does.

Is one-way video interviewing backed by the same research as regular interviews?

Indirectly, yes, and directly, a smaller but growing body of research says mostly yes too. A one-way interview inherits the structured-interview literature because it forces the same questions in the same order for every candidate. A newer body of research on asynchronous video specifically, rather than structured interviews in general, has started testing whether that inheritance actually holds, and so far the signal points the same direction.

Do one-way video interviews predict job performance as well as live interviews?

A 2025 study in Human Performance compared candidates answering the same past-behavior questions face-to-face and through one-way video, and found overall interview performance in both formats validly predicted work-sample results. It’s one study, not a decades-deep meta-analysis, so treat it as early, positive evidence rather than a settled verdict.

Will candidates actually complete a one-way video interview?

It depends more on the role and the design than on video itself. Research on candidate reactions is mixed: some studies find attitudes comparable to accepted methods like personality tests, others find reactions run more negative than traditional interviews on average, with the gap tied to specific design choices rather than the format. One-way interviews earn their place on high-volume, loosely-qualified roles. For a thin candidate pool, skipping video and going straight to a conversation is often the better trade.