Moderating factors of video-modeling with other as model: a meta-analysis of single-case studies.

The team pooled every single-case study that used video modeling with another person as the model. They looked at kids with autism and kids with other delays.

They ran a meta-analysis to see how big the effect really is and what makes it bigger or smaller.

For children with autism, watching a peer or adult on video gave a large boost in learning new skills.

Kids with other developmental disabilities still gained, but the effect was medium, not large.

Certain set-ups, like clearer videos and closer practice trials, made the gains even stronger.

Hong et al. (2015) later stamped video modeling as "evidence-based" for daily living skills after checking study quality. That review covers many of the same papers Storch et al. (2012) pooled, so the two reports back each other up.

Hong et al. (2016) zoomed in on just daily-living goals and still found solid, moderate effects. Their 2016 numbers sit inside the range A et al. reported, showing the result holds when you narrow the lens.

Marcus et al. (2009) seems to clash: they saw self-modeling beat peer modeling for teaching letters. But their sample was tiny and the skill was very small. Storch et al. (2012) averaged across many skills and kids, so both can be true—self may win for tiny tasks, yet other-model still works well overall.

You can film a typical peer or staff member once and reuse the clip for many learners. That saves prep time while still giving autism learners a strong push. If a child isn’t progressing, check the video quality and how often they practice—those levers matter.