Medical Student Academic Performance Is Not Normally Distributed

Exam score distributions are classically used to introduce students and novice statisticians to the normal distribution. Basing academic assessment on normality in medical education influences tracking student progress and determining curricular efficacy. This study sought to determine whether exam scores of medical students at the Oklahoma State University College of Osteopathic Medicine were normally distributed and to quantify deviations from normality. Exam distribution data for 58 exams from the Class of 2022 and the Class of 2023 were assessed for normality graphically using Q-Q plots and numerically using the Shapiro-Wilk Test for Normality. The resulting p-values from the Shapiro-Wilk test were controlled for the false discovery rate using the Benjamini-Hochberg procedure. Skewness and kurtosis were then determined using the SAS method to minimize bias. 91.37% of exams had statistically significant deviations from normality evidenced graphically and numerically. The average skewness was -0.9484 and the average excess kurtosis was -0.6543. 74.14% of the exams were moderately-to-highly skewed and 89.66% were platykurtotic. All statistical analyses were performed using R (version 4.0.2). The results indicate a curve slumped heavily to the right containing a long, thin left-tail. These findings suggest that there are many high scores on average or that there is a subset of exam scores consistently far below the class average. Characterizing exam performance will enable more robust evaluations of student progress and course efficacy, permitting future investigations exploring causes and implications of non-normal exam score distributions regarding educational outcomes for medical students.