Technology-enhanced learning for statistical graph interpretation: An item response theory analysis of learning outcomes

This study aimed to examine the effectiveness of technology-enhanced learning (TEL) in improving students’ statistical graph interpretation skills through a rigorous Item Response Theory (IRT) analysis. Employing a quasi-experimental pretest–posttest control group design, the research involved 120 undergraduate students from four classes, equally divided into experimental and control groups. The experimental groups received TEL-based instruction featuring interactive graph visualizations and automated feedback, while the control groups followed conventional lectures and exercises over seven sessions. Data were collected using a 60-item multiple-choice test covering bar charts, histograms, boxplots, and scatterplots, which was content-validated by experts and trialed for clarity, yielding high reliability (Cronbach’s α = 0.833). Construct validity was ensured through unidimensionality and invariance testing, confirmed by eigenvalue and DETECT analysis. Data analysis applied IRT to calibrate item parameters discrimination (a), difficulty (b), and guessing (c) and to estimate students’ latent abilities (θ). Model comparison identified the 3PL model as the best fit, capturing both difficulty variation and guessing behavior. Calibration results showed that most items exhibited satisfactory psychometric quality, supporting the robustness of the instrument. Findings revealed that TEL groups achieved a nearly one-logit gain in ability from pretest to posttest, significantly higher than the minimal improvement observed in the control groups, as confirmed by independent t-tests and normalized gain analysis. These results indicate that TEL substantially strengthens students’ ability to interpret statistical graphs while demonstrating the diagnostic value of IRT in evaluating both item quality and learning effectiveness.