Comparación de indicadores psicométricos en evaluaciones de opción múltiple entre estudiantes de Medicina Humana

Abstract

Introduction

Multiple-choice tests are essential in educational assessment, especially in medicine. Peruvian studies show that 3 and 5 alternatives have similar effectiveness, but the optimal number of alternatives remains debated globally.

Methods

We conducted a cross-sectional study with students from the assignment “Metodología de la Investigación Científica II”, who took assessments with four and five alternatives. Difficulty and discrimination index were calculated using Excel and Jamovi. Kruskal-Wallis test and concordance correlation coefficient where used fot he analysis, and chi-square test was used to compare distributions.

Results

A total of 55 students were included. The difficulty index showed similar medians for both 4 and 5 alternatives (p = 0.824), as did the discrimination index, which also did not identify differences between the medians (p = 0.654). The concordance values for the difficulty index indicated good consistency in both the midterm and final exams, with values of 0.925 (95% CI: 0.866 to 0.959) and 0.889 (95% CI: 0.822 to 0.932), respectively. The concordance for the discrimination index was low, with values of 0.318 (95% CI: −0.009 to 0.585) and −0.006 (95% CI: −0.247 to 0.259) for the midterm and final exams, respectively.

Conclusion

We did not find significant differences in the difficulty and discrimination of questions with 4 and 5 options. We suggest the use of 4 alternative questions in objective exams in research courses.