To use constructed-response questions, or not to use constructed-response questions? that is the question
Advocates of Constructed Response (CR) questions argue that CR questions provide a different assessment of student knowledge than is available from Multiple Choice (MC) questions. If that is the case, and if the benefit in terms of improved assessment is substantial, then it follows that grade outcomes using CR questions should be different from those using MC questions. We investigate this using a large dataset composed of individual assessment results from thousands of students in introductory economics classes at a large public university. Empirical analysis of our large sample of students indicates that a switch to an all-MC format would result in grade changes that are in the “small” to moderate range when compared to grade changes that occur between assessments. This evidence suggests that CR questions could be abandoned at relatively little cost in grading accuracy. However, there are other arguments in favour of keeping CR questions. In particular, it has been suggested that students perceive a mix of CR and MC as “fairer” than an assessment composed exclusively of one or the other question type. Further, some instructors believe that CR questions encourage students to study harder. We provide survey evidence that supports both arguments.