Does Practicing for Interviews Actually Work? What the Research Found

The Myth: "10,000 Hours" Oversold What Practice Actually Does

The idea that roughly 10,000 hours of practice turns anyone into an expert traces back to psychologist K. Anders Ericsson's 1993 research on violinists, later popularized — and considerably simplified — by Malcolm Gladwell's book Outliers. The simplified version became cultural shorthand: put in the hours, become great.

In 2014, psychologists Brooke Macnamara, David Hambrick, and Frederick Oswald published a meta-analysis in Psychological Science that put the actual claim to the test across 88 studies and more than 11,000 participants, measuring how much of the difference in performance between people was genuinely explained by how much they'd practiced.

The results varied enormously by domain, and not in the direction the popular version of the idea predicts:

• Games (like chess): practice explained about 26% of the performance difference between people
• Music: about 21%
• Sports: about 18%
• Education: about 4%
• Professions: less than 1%

That last figure is the one that matters most here. For professional skills — the category job-interview performance would fall into — the amount of time someone had practiced explained almost none of why some people were better than others. Something other than raw hours was doing nearly all of the work.

“In professional domains, how much someone had practiced explained less than 1% of the difference in how good they were at it.”

Why Practice Alone Often Doesn't Work: The Feedback Problem

If raw hours of practice aren't doing the work, what is? One strong candidate is feedback — except feedback has its own surprisingly rocky research history.

Psychologists Avraham Kluger and Angelo DeNisi reviewed decades of feedback research in a 1996 meta-analysis published in Psychological Bulletin, synthesizing 607 effect sizes across more than 23,000 observations. On average, feedback improved performance. But in just over a third of the cases they analyzed, feedback made performance worse — not neutral, worse.

The difference came down to what the feedback pointed people's attention toward. Feedback that kept someone focused on the specific task — what exactly to change, and how — tended to help. Feedback that pulled attention toward the self instead — comparisons to other people, a score with no explanation, anything that reads as a judgment of the person rather than the work — tended to hurt, by diverting mental effort into self-evaluation and defensiveness instead of the task itself.

This is the piece that's usually missing from advice telling you to "just do more mock interviews" — the number of interviews you complete matters far less than whether each one gives you something specific to fix.

“Over a third of the time, feedback made performance worse — the kind that judges the person instead of the task is the kind most likely to backfire.”

The Narrow Exception: What Happened When Researchers Tested Practice on Real Job Interviews

Here's where the picture changes. Most of the studies in the Macnamara meta-analysis measured broad expertise — chess mastery, musical skill, athletic performance built over years. Job interviewing is a narrower, more specific, more mechanical skill: a short, structured performance with a repeatable shape. And when researchers built a tool to practice that exact skill and tested it in randomized controlled trials, the results looked very different from the professions finding above.

A research team led by Matthew Smith (then at Dartmouth, later Johns Hopkins and the University of Michigan) built Virtual Reality Job Interview Training (VR-JIT) — simulated interview practice with an on-screen virtual interviewer and structured, task-specific feedback after each session. Pooling data across four separate randomized controlled trials (119 participants total, people with conditions including mood disorders, PTSD, schizophrenia, and autism spectrum disorder, randomized roughly 2:1 to training versus a comparison group), the researchers found that trainees completed an average of 14.8 practice interviews, improved measurably from their pre-training to post-training interview performance, and — critically — that improved post-training performance statistically mediated the link between completing more practice interviews and actually receiving a job offer.

A 2025 pragmatic randomized controlled trial extended this into a very different, high-stakes setting: men preparing for release from two Michigan prisons through a trades-focused employment program. 101 participants were randomized (roughly 2:1) to the standard program plus VR-JIT, or the standard program alone. The practice group showed a significant, moderate-to-large improvement in measured interview skill (effect size d = 0.60, a meaningfully large jump), and after adjusting for baseline differences, had more than triple the odds of being employed within six months and reached employment meaningfully faster.

• Interview skill (performance-based, not self-reported): improved significantly more in the practice group, d = 0.60
• Employment within 6 months: 82.3% (practice group) vs. 73.5% (comparison) — adjusted odds ratio 3.76
• Time to employment: adjusted hazard ratio 1.62 in favor of the practice group

“Across four randomized trials, improved interview performance after structured practice statistically explained the link to actually getting a job offer.”

The Honest Caveats — Read Before You Over-Apply This

A few limits matter here, and a no-fluff answer has to include them. The pooled VR-JIT trials studied people with diagnosed mental health conditions or autism spectrum disorder in supported-employment programs — not a general population of job seekers, so the size of the effect may not transfer directly to every candidate. The 2025 prison study's authors describe their own unadjusted results as "not definitive": the raw difference in employment rates wasn't statistically significant on its own (p = 0.15), only after statistical adjustment for baseline differences between groups, and the trial enrolled fewer participants (101) than it was originally designed for (150) because of COVID-19 disruptions.

None of that erases the finding — effect sizes on the actual interview-skill measurement were consistent and significant across studies — but it means you should read "practice improves interview outcomes" as well-supported and real, not as a guaranteed, universal formula with a single fixed multiplier.

So Which Is True: Does Practice Work, or Doesn't It?

Both findings are true, and they're not actually in conflict once you notice what each one measured. The professions figure (less than 1%) came from studies of broad, long-horizon expertise — the kind built over a career, where "practice" often meant loosely defined, self-directed repetition with no structured feedback loop. The job-interview trials measured something much narrower and far better specified: a short, repeatable performance, practiced in a structured format, with task-specific feedback delivered immediately after each attempt.

Put differently: the research doesn't say practice is a myth. It says vague, unstructured "practice" is a much weaker lever than the 10,000-hour story implies — and that the lever that actually moves the needle is structure plus specific, task-focused feedback aimed at one well-defined skill, not hours logged in the abstract.

This is the actual mechanism behind the broader claim this blog has made before — that repetition, not information, is what changes how you perform under pressure — and it's worth being specific about which kind of repetition the evidence actually backs.

“Practice isn't a myth. Vague, unstructured practice is a much weaker lever than the 10,000-hour story implies.”

What This Means for How You Should Actually Practice

A few concrete adjustments follow directly from lining these studies up against each other:

• Don't count repetitions — count specific fixes. A mock interview you walk away from without one concrete thing to change next time is close to the "practice with no feedback loop" condition the professions data found barely worked at all.
• Prefer feedback that names the task, not just a verdict. "Your answer to the conflict question didn't name a specific action you took" is the kind of task-focused feedback the Kluger & DeNisi research found helps. A bare score with no explanation is closer to the kind that can backfire.
• Practice the specific, mechanical shape of an interview — not the general topic. VR-JIT's effect came from simulating the actual structured back-and-forth of an interview, not from reading or watching interview advice.
• Repeat across varied questions and scenarios. The VR-JIT trials had participants complete an average of nearly 15 separate practice interviews — a single run-through is a much weaker test of the mechanism than repeated, varied attempts.

This also lines up with why it's worth treating interview prep as training rather than studying — the shift from consuming interview advice to actually rehearsing it is the difference the research above is describing in more rigorous terms.

Key Takeaways

If you remember only a few things, remember these:

• A 2014 meta-analysis (Macnamara, Hambrick & Oswald, Psychological Science; 88 studies, 11,135 people) found deliberate practice explained less than 1% of performance differences in professional skills — far less than the popularized "10,000 hour rule" implies.
• A separate 1996 meta-analysis (Kluger & DeNisi, Psychological Bulletin; 607 effect sizes, 23,663+ observations) found feedback helps performance on average, but made it worse in over a third of cases — task-focused feedback tends to help, feedback that judges the person tends to hurt.
• Randomized controlled trials of structured interview-specific practice (Virtual Reality Job Interview Training, pooled across 4 RCTs plus a 2025 prison-reentry RCT) found real, measurable gains: a d = 0.60 improvement in interview skill, and in the prison study, triple the adjusted odds of employment within 6 months.
• The reconciliation: broad, unstructured "practice" is a weak predictor of expertise, but narrow, structured practice with task-specific feedback on one well-defined skill — like interviewing — has real randomized-trial evidence behind it.
• The VR-JIT evidence comes mainly from clinical and reentry populations in supported-employment settings, and the 2025 prison trial's authors call their own unadjusted results "not definitive" — real effects, not a universal guarantee.