The impact of video-based intervention for improving attitudes towards autism in Chinese higher education: A pre-post study

This study assessed whether a brief video-based intervention could improve attitudes toward autism among Chinese university students and staff. A total of 1158 participants—including undergraduates, graduate students, and staff—completed the 17-item Autism and Neurodiversity Attitudes Scale (ANAS) before and after watching a 6-minute autism-awareness video. Paired-sample t-tests assessed pre-post differences. Exploratory factor analysis (EFA) examined whether the intervention altered the underlying attitudinal structure. To predict individual responsiveness, four machine learning models were trained using pre-intervention responses and demographics; TabPFN achieved the highest accuracy (73.4 %). The study revealed significant positive changes in attitudes towards autism following the educational video intervention (t = -13.30, p < 0.001, Cohen's d = 0.39). We identified three stable dimensions of autism perception: support for normalization, acceptance as natural variation, and empathy. SHapley Additive exPlanations analysis identified that pre-intervention responses to deficit-based items and pity-based attitudes are the strongest predictors of attitudinal change, while demographic variables had negligible influence. Higher baseline endorsement of normalization and pathologizing attitudes were associated with reduced responsiveness to the intervention, highlighting the role of entrenched beliefs in moderating intervention effectiveness. While a brief video intervention can enhance attitudes toward autism, its ability to transform deep-seated beliefs remains limited.