Class attendance and learning outcome

Abstract

This paper presents new evidence on the dynamic treatment effects of class attendance on academic performance. The analysis is based on a dataset from a large introductory statistics course and a dynamic modeling framework of Ding and Lehrer (Rev Econ Stat 92(1):31-42, 2010). The course had seven progressive assessments spread across a thirteen-week semester. Assessment test scores were matched to individual student attendance records. We use a panel dataset to study the dynamic interactions over time and between learning activities including lectures and tutorials, while accounting for reverse causality and self-selection without resorting to instruments for attendance or discontinuity design. Class attendance is found to have a test score return rate of 1.3 percentage points per lecture and 1 percentage point per tutorial. For both lecture and tutorial attendances, the contemporaneous effect dominates the lagged effect, with the effects accumulating over time. We also find a substitution rather than complementarity effect between lecture attendance and tutorial attendance, but the former has a stronger effect on test scores than the latter. Our results also show these effects are stronger for under-performing students.