Studying Assignment Size and Student Performance Using Propensity Score Matching

Teachers and instructors assign students homework of varying lengths. There is considerable evidence that factors such as cognitive load play a role in student performance and learning, but there has not been sufficient study of how these phenomena play out in the specific case of the length of homework. In this paper, we study the impact of assignment size on student performance. This paper represents the first attempt we are aware of to study how long assignments should be, in real-world data, in order to maximize student performance and learning. However, natural assignments of different lengths often vary in other ways. We control for this limitation using propensity score matching (PSM), an approach that helps to control for variables affecting outcome besides the intervention of interest. As such, we can conduct our analysis on large-scale data naturalistically collected through a digital educational platform. We use PSM to study the effect of assignment size on student performance while controlling for assignment difficulty, discrimination and reliability. We find that shorter assignments result in higher performance. These results can be used as a guideline for instructors and instructional designers when designing course assignments.