The studies that match how AcceptMe works

Practice, watch yourself back, read the feedback. Ratings rose. Practice alone did not do it.

Ninety MIT undergraduates each did a mock interview with a professional career counsellor, then were randomly split three ways. One group watched interview-tips videos. One practiced with an automated interviewer and watched their own recording. One practiced with the automated interviewer, watched the recording, and got automated feedback on it: smiles, speaking rate, pauses, filler words, head movements. A week later, everyone did a second interview with the same counsellor, who did not know which group they were in.¹

The feedback group’s ratings for overall performance rose significantly more than the tips group’s (p = .008) and the practice-and-watch group’s (p = .010). Watching yourself without being told what to change did no better than watching tips. One thing to know: when the authors split the sample by gender, the gain was carried by the women in the study; the men’s ratings did not differ across groups.¹

In AcceptMe: this is the loop. A timed run on camera, then your recording beside a transcript, with the coach’s notes on what to change.

Automated feedback beat a classroom session on the same material

Seventy students at two German universities were split between a simulated interview with a virtual interviewer that analysed their behaviour and fed it back in real time, and a conventional training session covering the same points. Interviewers who did not know the condition then rated each student in a mock interview. The simulation group scored higher on interview performance (d = 0.60) and on whether the interviewer would hire them (d = 0.55), and reported less interview anxiety.²

In AcceptMe: every answer comes back with what worked, one main thing to fix, and why. Nobody has to book a session to get it.

In recorded video interviews, an unguided practice run changed nothing. A short training did.

This is the format many programs now use: you see a question, get a moment to think, and record your answer on a clock with nobody on the other end. In the first study, 202 people were randomly given a short training video, a practice run, both, or neither, before a five-question recorded interview. The practice run on its own had no measurable effect on anything. The training raised rated performance, and it did so because trained people gave longer, more structured answers. A second study with 156 active job seekers, senior students among them, found the same thing.³

In AcceptMe: a run is never just a recording. The coach tells you what to change before the next one, and the structure it pushes you toward is the structure that raised ratings here.

Students told no complete stories until they were trained and practiced weekly

Twenty-one students, all under 25 and none with full-time work experience, wrote one interview answer a week for four weeks in response to “tell me about a time” questions, with a storytelling training after week two. Before it, not one answer was a complete story with a situation, an action and a result. After it, 62% were, and the rated quality of the narratives rose by about two points on a ten-point scale.⁴ Twenty-one people is a small trial, so take the direction from it, not the exact size.

In AcceptMe: the Story Bank is that weekly practice, coached. Each conversation ends in a written story with the situation, your action and the result in it.

AI notes on your own draft made the next draft better

A hundred and forty-four university students each wrote a short piece, then revised it. Half were given individualised feedback on their own draft, generated by GPT-4 with a fixed prompting procedure; half revised without it. Controlling for how good the first draft was, the feedback group produced higher-quality revisions, and they spent longer revising and changed more of the text to get there.⁵

In AcceptMe: the essay coach pins its notes to your own sentences, tagged as something that works, needs tightening, or has to be fixed. It never writes a line for you.