Re-using questions in classroom-based assessment: An exploratory study at the undergraduate medical education level

To alleviate some of the burden associated with the development of novel quality questions on a regular basis, medical education programs may favour the use of item banks. This practice answers the real pragmatic need of having to create exams de novo at each administration while benefiting from using psychometrically sound questions to assess students. Unfortunately, programs cannot prevent trainees from engaging in cheating behaviours such as content sharing, and little is known about the impact of re-using items.

We conducted an exploratory descriptive study to assess the effect of repeated use of banked items within an in-house assessment context. The difficulty and discrimination coefficients for the 16-unit exams of the past 5 years (1,629 questions) were analyzed using repeated measure ANOVAs.

Difficulty coefficients increased significantly (M = 79.8% for the first use of an item, to a mean difficulty coefficient of 85.2% for the fourth use) and discrimination coefficients decreased significantly with repeated uses (M = 0.17, 0.16, 0.14, 0.14 for the first, second, third and fourth uses respectively).

The results from our study suggest that using an item three times or more within a short time span may cause a significant risk to its psychometric properties and consequently to the quality of the examination. Pooling items from different institutions or the recourse to automatic generated items could offer a greater pool of questions to administrators and faculty members while limiting the re-use of questions within a short time span.