A comparison of peer video modeling and self video modeling to teach textual responses in children with autism.

Marcus et al. (2009) tested two kinds of video modeling with three children with autism. One group watched videos of a peer reading letters. The other group watched videos of themselves reading letters. The team switched the conditions every few days to see which format taught faster.

The goal was simple: teach the kids to name printed letters. The researchers filmed each child after they mastered a letter, then showed that clip back during self-modeling sessions.

Every child reached mastery when they saw themselves on screen. Only one child learned with the peer video, and even that child learned faster with self-modeling. The data lined up clearly: self beats peer for teaching textual responses.

Skills also stuck around. The children kept naming letters correctly after the videos stopped.

Richman et al. (2001) looked at the same self-versus-peer question eight years earlier. They found no difference when teaching conversation skills. Alonna’s team shows the choice does matter for reading letters. The skill you teach seems to decide which model works best.

Mulder et al. (2020) repeated the contest with social tasks and got the same winner: self-video modeling came out on top. Together the studies build a pattern: self-modeling holds an edge across both academic and social goals.

Meta-analyses agree video modeling works in general. Hong et al. (2016) and Storch et al. (2012) both report moderate to large gains for learners with autism. The new detail is that filming the learner himself can speed things up even more.

If you run video modeling for reading, typing, or labeling, start with self-modeling. Film the child doing the target skill correctly, edit out errors, and play the clip before practice. Check timing data: if mastery drags, switch to self clips even if peer videos are easier to make. Track daily correct responses to see which format wins for each learner.