Class Attendance, Self-Study and Academic Performance. A Panel Data Study

We analyze matched administrative-survey data collected on Economics students enrolled in two Econometrics courses offered in adjacent terms of different academic years at the Facultad de Ciencias Sociales y Jurídicas of Universidad Carlos III de Madrid to assess the impact of class attendance and self-study on academic performance. Using "proxy variables" in cross-sectional regressions to capture the effect of unobservable factors possibly correlated with attendance and selfstudy hours, we still find a positive and significant effect of attendance and self-study hours. However, when we use panel data fixed effect estimators to eliminate time-invariant individual specific unobservables and add proxy variables to control for potentially time-varying individual specific unobservables, the attendance effect disappears while the study-hours effect becomes more significant economically, although only marginally significant statistically. These results suggest, on the one hand, that the positive effect of attendance commonly found in the literature may still be upward biased, being unable to capture the impact of time-varying unobservables on academic performance, on the other hand, that self-study hours may be considered as an important causal determinant of academic performance.