R. Achterkamp, H.J. Hermens, M.M.R. Vollenbroek-Hutten
8 min
Mobile technology-supported coaching programs for physical activity are becoming increasingly common, but they often lack integration with established behavioral science principles. A core psychological construct used to drive behavior change is self-efficacy—the individual's belief in their ability to succeed in a specific situation. One proven way to boost self-efficacy is through vicarious experience, which involves observing a similar peer successfully perform a task.
This study investigated whether incorporating short, instructional video demonstrations into a mobile-based feedback loop could effectively increase self-efficacy and improve physical task performance.
Researchers recruited 36 participants to perform a novel, standardized task: walking a distance of eight meters while wearing scuba fins and a blindfold, aiming to complete the task in specifically timed intervals (14, 16, or 18 seconds). Participants were randomized into two groups. The experimental group viewed a mobile video of a peer successfully performing the walk before every trial, while the control group performed the task without the video.
Statistical analysis revealed that the vicarious experience provided through the smartphone video did not result in a significant difference in self-efficacy or task performance compared to the control group. Although many participants reported that the videos were helpful for understanding the task and identifying with the model, these perceptions did not translate into measurable improvements in confidence or actual outcome metrics. A secondary observation noted that most participants in the video group adopted the specific physical 'walking strategy' shown by the model, suggesting that while the intervention failed to impact general self-efficacy, it did influence specific motor imitation.
While this study did not support the hypothesis that simple video-based vicarious experience consistently boosts self-efficacy for novel tasks, the researchers suggest that the experimental design—specifically the artificial nature of the task and the lack of external reinforcement—may have obscured potential benefits. Future research should explore whether such interventions are more effective when implemented in real-world, ecologically valid settings where users engage in genuine physical activity routines.
Sam: That's exactly the problem they solved, and the solution is a bit absurd. Participants had to walk across the lab, about eight meters, wearing scuba fins and a blindfold. The goal was to arrive in exactly fourteen, sixteen, or eighteen seconds. A pilot study had matched those times to fast, normal and slow walking. Nobody has practised that, so everyone starts with the same experience.
Alex: That does level the field. How did each trial actually run?
Sam: Everyone got six trials, two at each target time, in random order. Before each one, the video group watched a phone clip of a same-sex model doing the task successfully. Then an automated voice asked everyone how confident they were, on a scale from zero to a hundred. Then came the target time, a countdown, and the walk. Laser gates at each end recorded the time.
Alex: If people heard how close they got after each walk, that's feedback, and it's mastery experience sneaking back in. Did they control for that?
Sam: They did. No one got any feedback, not how close they were, not whether too fast or slow. That was deliberate, to isolate the video's effect. The experimenter only stepped in to correct someone's course if they veered too far.
Alex: So, the results. Did the video move confidence?
Sam: Not significantly. The video group averaged about sixty-six out of a hundred. The control group averaged about sixty-one. That gap was well within the noise, with a p-value around point four. Confidence also didn't change over the six trials in either group.
Alex: And performance? Confidence aside, maybe the video simply taught people how to do it.
Sam: Also not significant. There's a trend the authors point to. On average, the video group was almost dead on the target time, while the control group drifted, ending up nearly a second off on average. But individual spread was several seconds in both groups. So that difference could easily be chance.
Alex: One obvious explanation is that people just didn't relate to the model. Can they rule that out?
Sam: Reasonably, yes. Video viewers rated how much they identified with the model at around seventy out of a hundred. They mostly mentioned same gender, similar age and similar build. And twelve of the eighteen said the video helped. They said it helped them count steps, judge step length, and catch the rhythm of the fins slapping the floor.
Alex: So they liked it and felt it helped, but the confidence numbers didn't budge.
Sam: That's the tension. And there's one more finding, which came about by accident. The male model shuffled forward with his fins on the floor. The female model lifted her knees high to step. Fifteen of the eighteen video viewers copied their model's style on every trial.
Alex: Hold on. Men shuffling and women high-stepping could just be how men and women walk in fins. Is that copying, or is it sex?
Sam: You can't tell from this design, and the authors say so. Model style and participant sex line up perfectly, and there's no control group for this measure. They suggest a follow-up that randomly assigns people to watch one style or the other. So call it suggestive that people picked something up, not evidence.
Alex: So why did the main effect fail? What's their own explanation?
Sam: They offer several. The task is artificial, low in ecological validity, meaning it doesn't resemble real life much. The model was never shown being rewarded, which Bandura says makes modeling stronger. That was left out on purpose, to keep mastery out of it. Then the small groups and high variability. And confidence started fairly high, so there may have been a ceiling effect.
Alex: They also checked whether confidence predicted performance at all, didn't they? That seems relevant.
Sam: They did, as a post-hoc check. Splitting people at the median confidence score, high and low confidence groups performed no differently. That's odd given the literature they cite, and to me it hints the task may not engage self-efficacy the way real behavior does.
Alex: So how much should I trust the null?
Sam: Lightly, in my view. Eighteen people per group with that much spread can't detect modest effects. A null here means no evidence of an effect, not evidence of no effect. The authors themselves frame it as partial success and call for tests in more realistic settings.
Alex: So who should actually read the full paper, and where should they start?
Sam: Anyone designing mobile behavior-change interventions, or anyone studying modeling and self-efficacy. Start with the discussion section's list of reasons the effect may have failed. It's a handy checklist for designing a better study. Then look at the qualitative answers on what viewers took from the video. The authors also float a future comparison of instructional videos against text-message coaching in daily life.
Alex: And for everyone else, the one line to remember?
Sam: People copied what they saw on the phone, but their confidence didn't move. Imitation isn't the same as belief.
Alex: Something to keep in mind before adding a demo video to your app.