Employing Evidence-Based Practices for Children with Autism in Elementary Schools.

Data collectors were not blind to school condition, and teacher and research staff ratings could introduce bias. Usual-services schools also received materials, a program assessment report, and periodic check-ins, so the difference was mainly added training, coaching, and feedback rather than no support. Schools volunteered before randomization, so schools that did not agree to participate may differ. One coached school withdrew after pretest data were collected. The study was in one southeastern state, though the authors reported a generalizability index score of 0.92 suggesting the schools resembled a U.S. sample. The Vineland findings should be interpreted cautiously because the authors noted some items may be culturally biased. The intervention was multi-component, so the separate contribution of training, assessment, school planning, coaching, or feedback cannot be identified. Student outcomes relied on individually scaled goals and teacher or research staff ratings, not independent blinded child assessments. Race and ethnicity analyses were secondary and involved small groups, so those patterns are suggestive rather than definitive.

Do not read this as proof that the coaching model improves broad language, social, or adaptive test scores. Student goal attainment and teacher fidelity were higher in coached schools, but test scores did not change more in coached schools than in usual-services schools, and overall school quality did not change more than usual services. Also do not credit coaching alone. The coached group received training, assessment, school planning, coaching, and feedback together, while usual-services schools also received some materials and a program report.