Planning and analyzing the quality of a descriptive statistics assessment in health science courses

Resumen

Traditional statistics education in the health sciences often relies on transmissive, decontextualized methods that prioritize rote memorization over statistical thinking, the investigative process essential for problem-solving and decision-making. To bridge these gaps, we developed and evaluated an assessment tool comprising seven multiple-choice questions and one short-essay question, grounded in the GAISE guidelines and centered on a real-world public health problem. We enrolled twenty students from a health-related course at a public university in São Paulo. To establish validity evidence (across cognitive, instructional, and inferential dimensions), we assessed students’ perceptions of difficulty and calculated difficulty and discrimination indices for each item. We found strong alignment with classroom content and the ability to identify conceptual misconceptions and discriminate between varying levels of students’ knowledge. Analysis of the essay responses and student reports revealed that synthesis capacity and the visual complexity of graphs act as cognitive barriers, independent of a student’s underlying conceptual understanding. While the small sample size limits the findings to this course context, this study contributes to the continuous improvement of the teaching-learning process by offering a practical framework for aligning statistics assessment with evidence-based pedagogical practices.