Video modelling and video self-modelling: what the evidence shows

Background: what video modelling is and why it matters

Video modelling means showing a child a short video of someone doing a skill or getting through a situation, so the child can watch, remember and then try it. The 2020 US national review of autism interventions defines it as a video-recorded demonstration of the target behaviour or skill, shown to the learner (Steinbrenner et al., 2020). It grew out of the wider psychology of learning by watching, covered in a separate review, and spread in autism education as cameras and editing software became cheap enough for schools and families.

Researchers test several forms separately:

• Other people as models. An adult, peer or sibling performs the skill on screen. An early review sorted studies into adult, peer, self, point-of-view and mixed models (McCoy & Hermansen, 2007).
• Video self-modelling. The child watches themselves succeed, in footage edited so that only competent behaviour remains. Positive self-review collects the child's own best existing moments. Feedforward edits together things the child can already do, so that the child appears to perform a new or harder skill. Peter Dowrick, a pioneer of the method, argued that this "learning from the future" lets existing component skills be recombined very quickly (Dowrick, 2012).
• Point-of-view modelling. The camera films at the child's eye level, showing the task as the child would see it, usually without showing a person at all (Hine & Wolery, 2006).
• Video prompting. A task is split into steps. The child watches one step, does it, then watches the next. It is commonly studied for multi-step daily living tasks (Aljehany & Bennett, 2019).

For hard moments, the appeal is that a video can show a situation before the child meets it, as often as needed, at a calm time. Early researchers also suggested that video holds some children's attention better than live demonstration (Charlop-Christy et al., 2000).

How the research has been done

Most studies use single-case experimental designs: one to five children are measured repeatedly before and after the video is introduced, often with staggered starts across children or skills. If behaviour changes only when the video arrives, each time, that suggests the video caused the change. A 2024 Campbell Collaboration evidence and gap map found 438 studies of video-based interventions for social behaviour in autistic children and young people aged 3 to 18: 394 single-subject designs, 25 randomised controlled trials and 23 other group or before-and-after studies. Video modelling was the most studied form (273 studies), followed by self-modelling (82), point-of-view modelling (61) and video prompting (57). Adults and peers were the usual models, and all-male samples outnumbered all-female ones by 238 to 17 (McConnell et al., 2024).

Outcomes are usually counts of one behaviour, such as social initiations or task steps completed correctly. Reviews summarise single-case data with overlap statistics. The percentage of non-overlapping data (PND) is the share of intervention points above the best baseline point. The improvement rate difference (IRD, from 0 to 1) compares the proportion of improved points between phases. Tau-U is a related index that can also adjust for a trend already present in the baseline. These measures do not always agree: in one 2025 review of 17 studies, one index suggested strong effects (0.7 to 1) and two others much weaker ones (0.3 to 0.5), which the authors judged to be underestimates (Mohanan et al., 2025). Our methods glossary explains these terms.

What the reviews and trials found

Meta-analyses of single-case studies

The first meta-analysis. Bellini and Akullian pooled 23 single-case studies with autistic children and adolescents. Video modelling and self-modelling were effective for social-communication skills, functional skills and behaviour; skills were maintained over time and transferred across people and settings; and both met the authors' criteria for an evidence-based practice (Bellini & Akullian, 2007).

Reviews by type of model. With other people as models, the average IRD was 0.83 for autistic participants, judged highly effective, and 0.68 for participants with other developmental disabilities, judged moderately effective (Mason et al., 2012). Point-of-view modelling had an overall IRD of 0.78; it had only been tested with autistic people or people with developmental disabilities, and its effects varied with age, disability and how the video was used (Mason et al., 2013). Self-modelling had strong effects for autistic children of preschool and primary-school age, especially on social-communication and behaviour; positive self-review produced significantly larger effects than feedforward; and the evidence for older students and other disabilities was thin (Mason et al., 2016).

Reviews by skill. Video modelling was moderately effective for functional living skills, with no significant differences by age, additional diagnoses or type of video (Hong et al., 2016). Video prompting for daily living skills (17 studies, 54 participants) had a moderate effect (Aljehany & Bennett, 2019). For social communication, a review of 24 studies concluded that video modelling meets What Works Clearinghouse standards for an evidence-based practice, while noting inconsistent results across children and across studies (Qi et al., 2018).

National evidence classifications

The National Clearinghouse on Autism Evidence and Practice lists video modelling as one of 28 evidence-based practices, supported by 97 articles (35 published 1990 to 2011 and 62 published 2012 to 2017); live modelling is listed separately (Steinbrenner et al., 2020). The National Standards Project (Phase 2, 2015) classes "Modeling", which it defines as covering both live and video modelling, as one of 14 "Established" interventions for autistic people under 22 (National Autism Center, 2015). For childhood phobias, a review rated filmed and live modelling as "probably efficacious" and participant modelling (watching, then doing with guidance) as "well established", noting that much of the support came from laboratory or school studies with non-referred children (Ollendick & King, 1998).

Video versus live modelling

In a frequently cited comparison, five autistic children each learned one task by video and a similar task from a live model; video led to faster learning and also promoted generalisation (Charlop-Christy et al., 2000). Later comparisons are less tidy. In a preschool classroom, four autistic children responded differently to video and live modelling, which the author described as heterogeneous treatment response (Wilson, 2013). In a 2026 study of eight autistic children aged 4 to 10, live modelling produced quicker first responses, while video did better on generalisation and long-term maintenance, and the children's cognitive profiles affected which worked best (Badovini et al., 2026). In a small randomised trial with 31 autistic children aged 5 to 12, a video modelling app for self-care was compared with live modelling; the authors report significant self-care gains in the app group (Zohrabi et al., 2025).

The child as model, or someone else?

Head-to-head tests are small and mixed. With five autistic children aged 4 to 11 learning conversation answers, there was no overall difference between watching themselves and watching someone else; children who learned from video learned equally well from both (Sherer et al., 2001). With three autistic children learning letters, all three reached mastery from self-modelling, but only one from peer modelling (Marcus & Wilder, 2009). Among 67 children aged 5 to 9 learning front crawl, watching themselves (at a higher or their current level) or a peer all beat a control video, with no differences between them (Minkels et al., 2025).

Controlled trials in real situations

Randomised trials of video for a specific stressful moment are few and small. In a dental clinic, 40 autistic children aged 4 to 12 were randomised to watch or not watch a video beforehand; those who watched needed slightly fewer visits to complete the appointment (1.5 against 2 on average) (da Silva Moro et al., 2024). In a widely cited earlier trial, 60 children facing surgery who saw a film of a peer going through hospital showed less anxiety before and after the operation than children shown an unrelated film (Melamed & Siegel, 1975). But a double-blind trial with 24 toddlers with developmental delays found that a video of a hearing test did not increase the number of tests completed, although parents reported less stress (Roberts et al., 2020). And in a 2026 trial with 80 children aged 6 to 12 facing dental treatment, an AI-personalised self-modelling app and a standard modelling video both reduced fear; the personalised app was not better on the main fear measure, although it produced a larger drop in heart rate (Tasgaonkar et al., 2026).

Maintenance, generalisation and null results

Maintenance and generalisation are among the stronger claims for this approach (Bellini & Akullian, 2007); a review of 18 school-based self-modelling studies also reported evidence of generalisation and maintenance (Hitchcock et al., 2003). Individual studies show the limits. When video modelling and reinforcement taught perspective-taking to three autistic children, only two passed a task they had not been taught (LeBlanc et al., 2003). For dental compliance in participants with severe or profound intellectual disability, step-by-step desensitisation helped all five participants, but video modelling helped only one of three (Conyers et al., 2004). A 15-year-old autistic student with very limited imitation skills changed minimally with either a sibling or an adult as video model (Rayner, 2011).

For whom and under what conditions

Diagnosis. Most of the evidence concerns autistic children. Other-as-model effects were larger for them than for participants with other developmental disabilities (Mason et al., 2012), and self-modelling evidence is thinnest outside autism (Mason et al., 2016). For ADHD, a review of 11 original studies with 5- to 16-year-olds found preliminary evidence, mainly for video combined with other components (Wilkes-Gillan et al., 2021). Small studies have used self-modelling for selective mutism, within a package of gradual exposure and rewards (Bork & Bennett, 2020), reading fluency in learning disabilities (Decker & Buggey, 2014), compliance in a children's psychiatric unit (Axelrod et al., 2014) and swimming in children with spina bifida (Dowrick & Dove, 1980).

Typically developing children. The evidence comes mainly from sport and medical preparation. Among 33 children learning to swim, an edited video of themselves performing well beat both unedited footage and practice alone (Clark & Ste-Marie, 2007), and a feedforward video improved children's trampoline performance (Ste-Marie et al., 2011). The surgery trial and the 2026 dental trial above also involved children without a diagnosis (Melamed & Siegel, 1975); (Tasgaonkar et al., 2026).

Age. Self-modelling effects were strongest at preschool and primary age (Mason et al., 2016), a functional-skills review found no significant age differences (Hong et al., 2016), and age moderated point-of-view effects (Mason et al., 2013).

Prerequisites. Video modelling assumes the child will watch the screen and can copy what they see. Children who succeeded with video seemed to have stronger visual learning skills, although this was an anecdotal observation (Sherer et al., 2001), and a teenager with very limited imitation barely changed, leading the author to call imitation an important prerequisite (Rayner, 2011).

How the video is used. Self-modelling alone outperformed self-modelling bundled with programmed rewards or other components (Mason et al., 2016), and extra prompting did not significantly change video prompting effects (Aljehany & Bennett, 2019). The best number and frequency of viewings are still open questions (Mohanan et al., 2025).

What this means in practice

• Choose one moment and one observable step. Almost all of this research targets specific behaviours, such as answering a question or sitting through a dental check, rather than broad change.
• Check attention and imitation first. If a child does not watch short videos or copy simple actions, start with live demonstration and guided practice (Rayner, 2011).
• Any credible model can work. Adults, peers and the child themselves have all been effective, and no model type wins for every child (Sherer et al., 2001); (Minkels et al., 2025).
• Show success, in brief clips, close to the moment. Videos of the child's own best moments did especially well (Mason et al., 2016). In one hospital study, children watched brief clips of themselves just before the sessions where the behaviour was needed (Axelrod et al., 2014).
• Pair watching with doing. For fears, guided practice in small steps has the strongest evidence; a video prepares for that practice rather than replacing it (Ollendick & King, 1998); (Conyers et al., 2004).
• Measure and adjust. Count the target behaviour before and after. Responses vary a lot (Qi et al., 2018), so if nothing changes after a fair trial, change the approach.
• Know when to get help. If fear or avoidance is severe, long-lasting or disrupting daily life, talk to a GP, paediatrician, psychologist or the school's special educational needs lead.

What this means for short personalised cartoons

Some features plausibly carry over to a short animated story about one child and one moment: a visual demonstration of one situation before it happens, a character who gets through it, repeated calm viewing, and a model the child identifies with. A cartoon version of a child sits somewhere between a self-model and a peer model, and head-to-head studies show no consistent advantage for either (Sherer et al., 2001); (Minkels et al., 2025).

Other features do not carry over. Self-modelling uses real footage of the child's own behaviour, and its theory rests on seeing oneself (Dowrick, 2012); an animated likeness is not that. Animation itself is neither clearly better nor worse: of eight autistic children in one study, two learned more from human video, three from animated video and three showed little difference (Bloh et al., 2025) (see cartoons versus real video). The closest trial to a personalised video found an AI-personalised self-modelling app no better than a standard modelling video on the main fear outcome (Tasgaonkar et al., 2026). Personalisation is a hypothesis to test, not an established benefit (see personalisation).

Cala Luna's personalised cartoons are designed around these findings, but nobody has trialled them. They are preparation stories, not therapy: research-based in design, untested in effect.

Limits of the evidence

• Small, mostly single-case studies. Of 438 studies mapped, 394 were single-subject designs and only 25 were randomised trials (McConnell et al., 2024). Such designs show that change happened for these children, not how often it would happen in general.
• Mostly autistic boys. All-male samples outnumbered all-female ones by 238 to 17 (McConnell et al., 2024). Evidence for anxious children without a diagnosis is thin.
• Effect sizes depend on the statistic chosen (Mohanan et al., 2025).
• Packages blur the picture. Many studies combine video with prompting, rewards or exposure, so the video's own contribution is hard to isolate (Bork & Bennett, 2020).
• Narrow settings. All 17 studies in one 2025 review were conducted in Western countries (Mohanan et al., 2025).
• Few studies of hard moments. Most research teaches skills; the few trials on distress in stressful situations are small and mixed (Roberts et al., 2020).
• Publication bias is likely. Successful single-case studies are more likely to be published than failures.

Open questions for research

1. Does video modelling reduce avoidance and distress in common hard moments, such as starting school or a haircut, for anxious children without an autism diagnosis, in randomised trials with blinded ratings?
2. With content held constant, do children learn as well from an animated likeness of themselves as from real self-modelling footage, a peer or an adult?
3. How many viewings are needed, and how close to the real situation should the last one be?
4. Can a brief check of attention to video and imitation predict who will benefit?
5. For fear-related moments, does showing the model's initial worry (a coping model) help more than calm competence throughout?
6. Does video add anything to graded real-world practice, or is its main value in getting a child to try the first step?
7. Do effects last beyond three months and generalise across settings and people?
8. What do the 25 randomised trials and other group studies show when pooled? The planned meta-analysis of control-group designs has yet to appear (McConnell et al., 2024).

Frequently asked questions

References

1. Bellini, S., & Akullian, J. (2007). A meta-analysis of video modeling and video self-modeling interventions for children and adolescents with autism spectrum disorders. Exceptional Children, 73(3), 264–287. doi.org/10.1177/001440290707300301
2. McCoy, K., & Hermansen, E. (2007). Video modeling for individuals with autism: A review of model types and effects. Education and Treatment of Children, 30(4), 183–213. doi.org/10.1353/etc.2007.0029
3. Dowrick, P. W. (2012). Self model theory: Learning from the future. WIREs Cognitive Science, 3(2), 215–230. doi.org/10.1002/wcs.1156
4. Hine, J. F., & Wolery, M. (2006). Using point-of-view video modeling to teach play to preschoolers with autism. Topics in Early Childhood Special Education, 26(2), 83–93. doi.org/10.1177/02711214060260020301
5. Aljehany, M. S., & Bennett, K. D. (2019). Meta-analysis of video prompting to teach daily living skills to individuals with autism spectrum disorder. Journal of Special Education Technology, 34(1), 17–26. doi.org/10.1177/0162643418780495
6. Charlop-Christy, M. H., Le, L., & Freeman, K. A. (2000). A comparison of video modeling with in vivo modeling for teaching children with autism. Journal of Autism and Developmental Disorders, 30(6), 537–552. doi.org/10.1023/A:1005635326276 · PubMed
7. McConnell, K., Keenan, C., Storey, C., & Thurston, A. (2024). Video-based interventions promoting social behavioural skills for autistic children and young people: An evidence and gap map. Campbell Systematic Reviews, 20(2), e1405. doi.org/10.1002/cl2.1405 · PubMed
8. Mohanan, R., Veena, K. D., Parsekar, S. S., Bellon-Harn, M. L., & Aithal, V. U. (2025). Effectiveness of video-based instruction with additional components to improve conversational skills in autistic children: A systematic review. American Journal of Speech-Language Pathology, 34(3), 1567–1584. doi.org/10.1044/2025_AJSLP-24-00301 · PubMed
9. Mason, R. A., Ganz, J. B., Parker, R. I., Burke, M. D., & Camargo, S. P. (2012). Moderating factors of video-modeling with other as model: A meta-analysis of single-case studies. Research in Developmental Disabilities, 33(4), 1076–1086. doi.org/10.1016/j.ridd.2012.01.016 · PubMed
10. Mason, R. A., Davis, H. S., Boles, M. B., & Goodwyn, F. (2013). Efficacy of point-of-view video modeling: A meta-analysis. Remedial and Special Education, 34(6), 333–345. doi.org/10.1177/0741932513486298
11. Mason, R. A., Davis, H. S., Ayres, K. M., Davis, J. L., & Mason, B. A. (2016). Video self-modeling for individuals with disabilities: A best-evidence, single case meta-analysis. Journal of Developmental and Physical Disabilities, 28(4), 623–642. doi.org/10.1007/s10882-016-9484-2
12. Hong, E. R., Ganz, J. B., Mason, R., Morin, K., Davis, J. L., Ninci, J., Neely, L. C., Boles, M. B., & Gilliland, W. D. (2016). The effects of video modeling in teaching functional living skills to persons with ASD: A meta-analysis of single-case studies. Research in Developmental Disabilities, 57, 158–169. doi.org/10.1016/j.ridd.2016.07.001 · PubMed
13. Qi, C. H., Barton, E. E., Collier, M., & Lin, Y.-L. (2018). A systematic review of single-case research studies on using video modeling interventions to improve social communication skills for individuals with autism spectrum disorder. Focus on Autism and Other Developmental Disabilities, 33(4), 249–257. doi.org/10.1177/1088357617741282
14. Steinbrenner, J. R., Hume, K., Odom, S. L., et al. (2020). Evidence-based practices for children, youth, and young adults with autism. National Clearinghouse on Autism Evidence and Practice, University of North Carolina. ncaep.fpg.unc.edu/
15. National Autism Center. (2015). Findings and conclusions: National Standards Project, Phase 2. Randolph, MA: National Autism Center. nationalautismcenter.org/national-standards/phase-2-2015/
16. Ollendick, T. H., & King, N. J. (1998). Empirically supported treatments for children with phobic and anxiety disorders: Current status. Journal of Clinical Child Psychology, 27(2), 156–167. doi.org/10.1207/s15374424jccp2702_3 · PubMed
17. Wilson, K. P. (2013). Teaching social-communication skills to preschoolers with autism: Efficacy of video versus in vivo modeling in the classroom. Journal of Autism and Developmental Disorders, 43(8), 1819–1831. doi.org/10.1007/s10803-012-1731-5 · PubMed
18. Badovini, C., Giannatiempo, S., & Vascelli, L. (2026). Improving imitation skills in children on the autism spectrum through video modeling. Tizard Learning Disability Review, 31(1), 82–95. doi.org/10.1108/tldr-10-2025-0055
19. Zohrabi, S., Karamali Esmaili, S., Sheikhtaheri, A., & Zayeri, F. (2025). Effectiveness of a therapist-managed video-modeling smartphone application for self-care training for children with autism and their parents: A randomized controlled trial. Journal of Autism and Developmental Disorders. Advance online publication. doi.org/10.1007/s10803-025-07161-8 · PubMed
20. Sherer, M., Pierce, K. L., Paredes, S., Kisacky, K. L., Ingersoll, B., & Schreibman, L. (2001). Enhancing conversation skills in children with autism via video technology: Which is better, "self" or "other" as a model? Behavior Modification, 25(1), 140–158. doi.org/10.1177/0145445501251008 · PubMed
21. Marcus, A., & Wilder, D. A. (2009). A comparison of peer video modeling and self video modeling to teach textual responses in children with autism. Journal of Applied Behavior Analysis, 42(2), 335–341. doi.org/10.1901/jaba.2009.42-335 · PubMed
22. Minkels, C., van der Kamp, J., & Beek, P. J. (2025). Learning the front crawl by observation: Comparing self- and other-models. Human Movement Science, 103, 103398. doi.org/10.1016/j.humov.2025.103398 · PubMed
23. da Silva Moro, J., Rodrigues, T. D., Kammer, P. V., de Camargo, A. R., & Bolan, M. (2024). Efficacy of the video modeling technique as a facilitator of non-invasive dental care in autistic children: Randomized clinical trial. Journal of Autism and Developmental Disorders, 54(2), 501–508. doi.org/10.1007/s10803-022-05820-8 · PubMed
24. Melamed, B. G., & Siegel, L. J. (1975). Reduction of anxiety in children facing hospitalization and surgery by use of filmed modeling. Journal of Consulting and Clinical Psychology, 43(4), 511–521. doi.org/10.1037/h0076896 · PubMed
25. Roberts, M. Y., Rosenwasser, J., Phelan, J., & Hampton, L. H. (2020). Improving pediatric hearing testing for children with developmental delays: The effects of video modeling on child compliance and caregiver stress. Journal of the American Academy of Audiology, 31(5), 310–316. doi.org/10.3766/jaaa.18070 · PubMed
26. Tasgaonkar, A., Rathi, N., Tasgaonkar, M., Agrawal, P., Tapkir, A., & Mehta, V. (2026). A randomised controlled trial comparing effectiveness of audio-visual aid and AI-personalised video self-modelling interventions to reduce dental fear and anxiety in paediatric patients. European Archives of Paediatric Dentistry, 27(5), 1561–1571. doi.org/10.1007/s40368-026-01241-8 · PubMed
27. Hitchcock, C. H., Dowrick, P. W., & Prater, M. A. (2003). Video self-modeling intervention in school-based settings: A review. Remedial and Special Education, 24(1), 36–45. doi.org/10.1177/074193250302400104
28. LeBlanc, L. A., Coates, A. M., Daneshvar, S., Charlop-Christy, M. H., Morris, C., & Lancaster, B. M. (2003). Using video modeling and reinforcement to teach perspective-taking skills to children with autism. Journal of Applied Behavior Analysis, 36(2), 253–257. doi.org/10.1901/jaba.2003.36-253 · PubMed
29. Conyers, C., Miltenberger, R. G., Peterson, B., Gubin, A., Jurgens, M., Selders, A., Dickinson, J., & Barenz, R. (2004). An evaluation of in vivo desensitization and video modeling to increase compliance with dental procedures in persons with mental retardation. Journal of Applied Behavior Analysis, 37(2), 233–238. doi.org/10.1901/jaba.2004.37-233 · PubMed
30. Rayner, C. (2011). Sibling and adult video modelling to teach a student with autism: Imitation skills and intervention suitability. Developmental Neurorehabilitation, 14(6), 331–338. doi.org/10.3109/17518423.2011.603369 · PubMed
31. Wilkes-Gillan, S., Cordier, R., Chen, Y.-W., Swanton, R., Mahoney, N., Trimboli, C., Yule, E., & Tam, E. (2021). A systematic review of video-modelling interventions for children and adolescents with attention-deficit hyperactivity disorder. Australian Occupational Therapy Journal, 68(5), 454–471. doi.org/10.1111/1440-1630.12747 · PubMed
32. Bork, P., & Bennett, S. (2020). Video self-modeling, fading, and reinforcement: An effective treatment package for children with selective mutism. Clinical Child Psychology and Psychiatry, 25(2), 446–455. doi.org/10.1177/1359104519855110 · PubMed
33. Decker, M. M., & Buggey, T. (2014). Using video self- and peer modeling to facilitate reading fluency in children with learning disabilities. Journal of Learning Disabilities, 47(2), 167–177. doi.org/10.1177/0022219412450618 · PubMed
34. Axelrod, M. I., Bellini, S., & Markoff, K. (2014). Video self-modeling: A promising strategy for noncompliant children. Behavior Modification, 38(4), 567–586. doi.org/10.1177/0145445514521232 · PubMed
35. Dowrick, P. W., & Dove, C. (1980). The use of self-modeling to improve the swimming performance of spina bifida children. Journal of Applied Behavior Analysis, 13(1), 51–56. doi.org/10.1901/jaba.1980.13-51 · PubMed
36. Clark, S. E., & Ste-Marie, D. M. (2007). The impact of self-as-a-model interventions on children's self-regulation of learning and swimming performance. Journal of Sports Sciences, 25(5), 577–586. doi.org/10.1080/02640410600947090 · PubMed
37. Ste-Marie, D. M., Vertes, K., Rymal, A. M., & Martini, R. (2011). Feedforward self-modeling enhances skill acquisition in children learning trampoline skills. Frontiers in Psychology, 2, 155. doi.org/10.3389/fpsyg.2011.00155 · PubMed
38. Bloh, C., Bacon, L., Begel, B., Madara, K., & Koller, B. (2025). Comparing human video modeling to animated video modeling for learners with autism. The Analysis of Verbal Behavior, 41(2), 262–279. doi.org/10.1007/s40616-025-00224-y · PubMed

Spotted an error, or know a study we should add? Email hello@calalunakids.com. Every study we draw on is in the research library; research terms are explained in our methods glossary.