New Study: “Dog Mentors” Show Promise Supporting Autistic Students in UK Schools

Introduction

Dimolareva et al. (2026) examine a UK-based program called The Dog Mentor, which trains dog handlers and assesses school dogs for integration into classrooms supporting neurodivergent students, particularly those on the autism spectrum. Children with autism face well-documented barriers to learning, compounded in the UK by long diagnostic waitlists and delays in receiving Educational Health Care Plans. While Animal-Assisted Services and Interventions (AAS/AAI) have shown promise for autism-related outcomes in prior research, session structure and dosage vary widely across studies, making it hard to draw general conclusions about what works. This study offers the first evaluation of The Dog Mentor program specifically, drawing on data from 58 of the organization’s 600+ registered schools.

Methods

The Dog Mentor team distributed a questionnaire to registered schools, asking dog handlers—typically teaching assistants, teachers, or SENCOs with handler duties added to their primary role—about the activities and outcomes they observed working with autistic and otherwise neurodivergent students. Fifty-eight schools responded (primary, secondary, and specialist provision settings). Two researchers used content analysis, combining deductive tallying with inductive coding, to identify themes across responses, achieving 90% inter-rater agreement.

Results

Ninety-three percent of schools reported the program successfully supported their goals for autistic students. Three themes emerged: facilitated learning, mental health, and healthy relationships. The most frequently used approaches were therapeutic/regulation support (40 schools, 85% success rate) and one-on-one individual support (25 schools, 84% success rate). Eighty-eight percent of schools reported improved emotional self-regulation, and 98% reported reduced stress and anxiety, particularly during transitions like entering school or changing classrooms. Variability in how the program is delivered—rather than a fixed protocol—appears to be a strength, since handlers tailored dog-assisted sessions to each child’s needs, and reported success across a wide range of approaches.

Discussion

The authors frame the program’s flexibility as consistent with the heterogeneity of autism itself, arguing that a single standardized session structure would likely underserve the range of needs across the sample. They note, however, that all findings rest on handler perception rather than validated diagnostic or physiological measures. The authors connect the program’s training philosophy—positive reward, empathy, and requiring self-regulation before interacting with the dog—to its outcomes. Because the dog’s wellbeing depends on the child’s behavior, calming becomes concrete practice rather than abstract instruction, which may reinforce the perspective-taking and empathy themes reported in 84% and 81% of schools. Mental health findings centered on transitions, with dogs deployed at predictable stress points—arrival, classroom changes, exams. The authors link this to earlier physiological work from their group (Meints et al., 2022) showing dog-assisted interventions can lower salivary cortisol, though this study collected no physiological data itself.

Implications for Mentoring Programs

For school-based support programs broadly, the clearest lesson is that flexibility in delivery should be a design feature, not a compliance problem. Programs may do better defining core principles—positive reinforcement, self-regulation first—while giving staff real latitude in how a resource is deployed for a given child. The transition-point finding has value beyond animal-assisted programs: any support resource may have outsized impact when deliberately positioned at moments of predictable stress rather than spread evenly across the day. Programs should also build in honest assessment of fit, since this approach worked less well for at least one student with significant developmental delay, and should consider tracking trauma-related outcomes even when current sample sizes are too small for firm conclusions.

Read the full article here