ENPAIR 2025

Given the increasing demand for evidence-based diagnostics in special education, strengthening teachers’ visual analysis skills is crucial for accurately assessing learning progress and informing instruction. Visual analysis of learning-progress graphs is a widely used method for evaluating single-case data, yet it is prone to systematic judgment errors. Building on prior findings (Bosch et al., under review), this study investigates the effectiveness of a video-based training designed to improve the accuracy of teachers’ visual analysis. We focused on graph types that previously led to high error rates—specifically, graphs showing a data trend before the intervention. In Bosch et al., such graphs were misclassified as showing an effect in 84% of cases, compared to a 98% correct classification rate for graphs with both trend and intervention effect. To examine whether this pattern results from a heuristic based on comparing only the first and last data points, the current study aligned start and end values across both trend conditions. Additionally, A and B phases were matched in length to ensure comparable opportunities for trend detection. A total of N = 129 pre-service teachers participated in a randomized controlled study. Participants analyzed 40 graphs before and after receiving either the training or a control video. Results show that trend information impaired judgment accuracy, but training led to a 16% reduction in misclassifications in this critical condition. An exploratory comparison revealed lower ratings for graphs without trends but with an intervention effect, suggesting that specific design features influenced interpretation.