A Baseline for Multiple-Choice Testing in the University Classroom

There is a broad literature in multiple-choice test development, both terms of item-writing guidelines, and psychometric functionality as measurement tool. However, most the published concerns testing context expert-designed high-stakes standardized assessments, with little attention being paid to use technique within non-expert instructor-created classroom examinations. In this work, we present quantitative analysis large corpus tests deployed classrooms primarily undergraduate university Canada. Our report aims establish three related things. First, reporting on functional operation 182 variety courses at all levels education establishes much-needed baseline for actual as-deployed tests. Second, motivate modified statistical measures—such item-excluded correlation measures discrimination length-normalized reliability—that should serve useful parameters future comparisons psychometrics. Finally, empirical data from our survey update widely used item-quality guidelines.