Assessing the Characteristics of Preclinical Medical School Course Exams to Support Curricular Improvement: a Proof-of-Concept Analysis

Authors

  • Nicholas B. Sajjadi
  • Joseph Price III

Abstract

Background

Examinations consisting of multiple-choice questions are the mainstay of formative and summative assessments in preclinical medical education, however, exam items are subject to poor composition and require performance analysis to optimize. The primary objective of this study was to devise and implement a proof-of-concept analysis focusing on normality testing as useful for test-writing development, informing curricular decisions, and supporting student progress.

Methods

We examined 63 exam datasets from the 2019-2020 academic year at the Oklahoma State University College of Osteopathic Medicine that included 4 variables: exam scores, item difficulty, item discrimination index, and item point biserial. We aimed to assess normality for each variable graphically with Q-Q plots and numerically using the Shapiro-Wilk test and Benjamini-Hochberg procedure for adjustment.

Results

Q-Q plots analysis of the 63 exam data sets revealed evidence of non-normality for 57 exam score sets, 58 item difficulty sets, 37 item discrimination sets, and 9 point-biserial sets. Comparably, Shapiro-Wilk testing suggested non-normally distributed data for 58 for exam scores sets, 59 item difficulty sets, 40 item discrimination index sets, and 4 for point biserial sets.

Conclusions

Many exams at our institution had non-normally distributed variables, and the assumption of normality for future inferential statistical methods is not necessarily supported. We recommend that future investigators perform normality testing before proceeding with statistical analysis, the results of which will inform the choice of statistical methods. Further study is required to elucidate important causal factors that may lead to non-normal distributions, and consideration of when non-normality might be expected or even desirable.

Downloads

Published

2022-12-12

Issue

Section

Medical Education