The evidence
Why rehearsal with feedback works.
Rehearsal Room rests on a simple, well-tested premise: rehearsal with specific feedback improves performance. Below is the research behind it — independent, peer-reviewed studies, each stated plainly with its statistic, source, and a link to the primary record.
Every finding here maps to something the product actually does. The deliberate-practice and mock-interview work is why the core of the app is a live rep against an AI counterpart. The feedback research is why every rep ends in a forensic debrief that names the exact moment and dimension that slipped. The exposure-and-anxiety studies are why you run the real scenario before the real thing. The retrieval-practice work is why the app is built for repeated reps and a skill identity across sessions, not a single perfect take. We don't present studies as standalone — they are the reason each mechanism exists.
An honest note on scope: some of these — particularly the mock-interview randomized controlled trials — were conducted in specific populations (veterans, returning citizens, autistic youth, people with schizophrenia or substance use disorders), often using one research group's proprietary training tool. They are cited here as supporting evidence for the underlying principle that rehearsal with feedback improves interview performance, not as proof of this product's outcomes. Effect sizes vary, samples are sometimes small, and we have flagged the strongest caveats rather than hide them.
Group one
Deliberate practice & the power of feedback
The foundational deliberate-practice study argues that expert performance is built through prolonged, effortful, structured practice — and that differences even among elite performers track accumulated deliberate practice, not just innate talent.
Two empirical studies of 30 violinists and 12 pianists of differing ability levels (Berlin).
A large cross-domain meta-analysis confirms deliberate practice meaningfully predicts performance — strongest in skill domains like games, music, and sports — while leaving substantial room for other factors.
Deliberate practice explained 26% of variance for games, 21% for music, 18% for sports, 4% for education, and less than 1% for professions.
Even a critical replication of the original violin study still finds a substantial, positive relationship between accumulated deliberate practice and skilled performance — real, though smaller than the landmark estimate.
Deliberate practice accounted for 26% of performance variance in the replication (η² = 0.26) versus 48% in the original 1993 study (η² = 0.48).
Targeted feedback is one of the most powerful levers on learning and performance — but its effectiveness depends heavily on how it is delivered.
Overall effect of feedback on student learning d = 0.48 (medium), from 435 studies, 994 effect sizes, N > 61,000.
Feedback improves performance on average, but the effect is not universally positive — well-designed, task-focused feedback helps, while poorly targeted feedback can backfire.
Average feedback-intervention effect d = 0.41, but over one third of feedback interventions decreased performance.
Group two
Immediate, specific feedback in skill acquisition
Immediate, informative feedback combined with repeated, revised attempts is foundational to the deliberate-practice theory of expert skill.
Deliberate practice is defined by tasks with immediate informative feedback and knowledge of results, combined with repeatedly performing similar tasks with reflection between attempts.
Feedback delivered immediately, and feedback that adapts to the response rather than being generic, produces measurably better learning outcomes.
Immediate feedback beat delayed feedback, F(1,151) = 4.046, p < .05 (η²ₚ = .024); adaptive, response-specific feedback outperformed elaborated and knowledge-of-correct-response feedback, F(2,151) = 4.069, p < .05 (η²ₚ = .049). Effects are significant but small.
Feedback is among the highest-impact interventions on learning, but its power depends on the specificity of the information conveyed rather than feedback merely being present.
Overall effect d = 0.48 across 435 studies; impact substantially moderated by information content, with stronger effects on cognitive and motor-skill outcomes.
Group three
Mock interviews measurably improve interview performance
These randomized trials were run in specific populations using one group's training tool. They evidence the principle — rehearsal with feedback improves interview performance — not this product's results.
In a randomized controlled trial, adults with schizophrenia who completed a virtual mock-interview simulator had roughly nine times the odds of a job offer within six months, and their role-play interview scores improved while controls' did not.
Odds ratio 8.73 for a job offer by 6 months (p = 0.04); role-play improvement p = 0.001; more training linked to fewer weeks to an offer (r = −0.63, p < 0.001); n = 32.
Among job-seekers with substance use disorders, those given mock-interview training were far more likely to land competitive employment within six months and found jobs weeks faster.
78.6% of trained versus 44.4% of controls reached competitive employment (OR = 5.67, p = 0.043, 95% CI 1.07–30.04 — wide, on a small sample); fewer weeks searching (9.3 vs 16.7).
In a feasibility RCT inside prison-based employment services, returning citizens who added mock-interview training were over seven times as likely to be employed six months after release, with large gains in interview skills and reduced interview anxiety.
Employment by 6 months OR = 7.4 (p = 0.045); interview skills, motivation, and anxiety all improved with large effects (η²ₚ > 0.15); N = 44.
A larger, pragmatic RCT in a prison re-entry program — the most robust in this set — found trainees had nearly four times the odds of post-release employment and were hired faster.
Employment within 6 months OR = 3.76 (p = 0.032); faster employment HR = 1.62 (p = 0.037); interview skills p < 0.001; N = 101.
A randomized trial with veterans who have PTSD showed mock-interview practice produced significantly greater improvement in actual role-play interview performance, with a large effect.
Greater role-play improvement versus controls (p = 0.04); large within-subject effect d = 0.76. (Self-confidence d = 0.58 did not reach significance, p = 0.09.) n = 33.
In a pilot RCT with autistic transition-age youth, those who practiced with a virtual mock-interview tool showed significantly better measured interview performance than controls, with a large effect.
Interviewee performance improvement F(1,12) = 6.07, p = 0.030, partial η² = 0.336 (large effect); N = 14 (a small pilot sample).
Group four
Rehearsal reduces performance anxiety
Clinically significant social anxiety is widespread in the U.S., establishing that fear of speaking or performing in public is a mainstream problem, not a fringe one.
Lifetime prevalence of social anxiety disorder is 12.1%; 12-month prevalence is 7.1%.
A U.S. government health authority independently reports the same prevalence (from the same NCS-R survey), giving the figure citable weight.
An estimated 7.1% of U.S. adults had social anxiety disorder in the past year, and an estimated 12.1% experience it at some point in life.
Repeated, structured exposure to speaking situations produces large, statistically significant reductions in public speaking anxiety — direct support for the premise that rehearsal works.
In vivo (real-world) exposure reduced public speaking anxiety with a large effect size of −1.41 (Z = 7.51, p < .001); virtual-reality exposure reduced it with an effect of −1.39 (Z = 3.96, p < .001).
Simulated, rehearsal-style exposure delivered through technology — no live audience required — meaningfully lowers public speaking anxiety, validating a practice-tool approach.
Virtual-reality exposure produced a significant pooled reduction in public speaking anxiety; effective studies averaged about six sessions of ~37 minutes, and VR was statistically as effective as treatments such as CBT.
Group five
Simulation transfers to real performance
In a randomized, double-blinded trial, surgical residents trained on a virtual-reality simulator performed dramatically better in the actual operating room — demonstrating that simulator practice transfers to real performance.
VR-trained residents dissected 29% faster and made 1.19 errors per case versus 7.38 for controls (P < .008); untrained residents were five times more likely to injure tissue (P < .04).
A systematic review of randomized trials found that surgical skills built in simulation transfer to the real operating room across multiple procedures.
Across 18 RCTs, simulation-trained surgeons showed better operative time, accuracy, fewer errors, and fewer complications — and no study found them performing worse than controls.
A meta-analysis pooling randomized trials confirms that pre-operative virtual-reality simulation improves real operative performance.
Across 8 RCTs, VR training reduced operating time by a mean of 8.35 minutes (95% CI 3.60–13.10, P < 0.001); technical-skill scores favored VR training (MD 6.22, P < 0.00001).
In aviation, a meta-analysis of flight-simulator research found simulator practice combined with aircraft training reliably improves pilot performance versus aircraft-only training.
From 247 articles, 26 experiments had sufficient data; simulator-plus-aircraft training consistently produced improvements for jets over aircraft training alone.
In sports, a meta-analysis shows that mental rehearsal (motor imagery) produces a meaningful improvement in real motor performance — especially when combined with physical practice.
Overall effect of imagery interventions on performance was medium, d = 0.431 (95% CI 0.298–0.563); imagery plus physical practice beat physical practice alone.
Group six
Structured-interview preparation changes outcomes
Structured interviews substantially outpredict unstructured ones for job performance, so preparing for structured and behavioral formats targets the highest-validity hiring instrument employers use.
Corrected validity of ρ = .44 for structured interviews versus ρ = .33 for unstructured (245 coefficients, N = 86,311).
An updated, range-restriction-corrected meta-analysis ranks structured interviews as the single strongest predictor of job performance — above general cognitive ability.
Operational validity of .42 for structured interviews versus .31 for general cognitive ability. (The .42 has a wide credibility interval; it is not a fixed value for any single hire.)
Coaching and preparation that focuses candidates on relevant interview content measurably raises their structured-interview scores — direct evidence that targeted preparation changes performance.
Participants in more comprehensive coaching (N = 144) received significantly higher interview ratings, with interviewee knowledge a partial mediator.
Content-focused coaching boosts scores without degrading — and may even improve — the interview's validity, countering the worry that prepping candidates just teaches them to game the process.
Predictive validity and reliability were relatively greater among coached interviewees than uncoached ones, in a sample of public-safety incumbents. (Descriptive, from one field sample.)
Group seven
Retrieval practice: doing beats reviewing
Repeated retrieval practice (active recall) produces far better long-term retention than repeated study — practicing the skill itself beats passive review.
At a one-week delay, the tested group recalled 56% of material versus 42% for the repeated-study group (d = 0.83). On the immediate 5-minute test the pattern reversed (75% vs 81%).
Once a skill or fact can be performed correctly once, continued repeated practice is what drives durable retention; dropping practice after a single success collapses long-term performance.
Repeated retrieval yielded about 80% recall on a one-week final test versus 36% and 33% when items were dropped after one correct response (d = 4.03, non-overlapping distributions).