Association for Public Policy Analysis & Management

While missing school has clearly shown to negatively predict achievement, little research has been dedicated to understanding the compositional effects that peer attendance has on student achievement. Classmates clearly contribute to opportunities to learn (OTL) in school through pedagogical activities such as peer tutoring, class participation, and group work; thus, one might consider that peer absenteeism might negatively associate with a student’s achievement, though little research serves to reinforce this claim. In fact, a major limitation of the existing studies on the spillover effects of classmate absenteeism is that rates of missing school are aggregated to the classroom-level (i.e. classroom rates of unexcused absences), which predict individual-level achievement. As such, no research to date has illustrated how peer absenteeism at the student-level (i.e. peer rates of absences) predicts student-level achievement. Moreover, research has yet to address the potential for unobserved omitted variable bias regarding general peer effects as it relates to absenteeism. In other words, while a relationship between peer absenteeism and student test performance may be established, there is still the potential that unobserved peer-related factors influence both peer absenteeism as well as student-level test performance that bias estimates (i.e. peers’ motivation to learn). This study explores the role of peer absenteeism on individual test performance while controlling for individual absenteeism, lagged achievement of students, and unobserved heterogeneity and omitted variation in student test performance through an instrumental variable.

This study uses administrative data from a small, urban district in California to address the role of peer absenteeism on student achievement in elementary schools. For each year of data, the dataset contains records of daily absences for every student, student demographics, exam scores and the dates of examination, and teacher and classroom characteristics. Overall, the analytic sample includes 33 third grade classrooms, 34 fourth grade classrooms, and 33 fifth grade classrooms within 13 public elementary schools over three contiguous years consisting of N = 4,037 student observations in grades three through five in 2011-13. The dependent variables were state-administered standardized end-of-year exams for English language arts and mathematics. The key measures in this study were derived from daily logs of absences for students and peers. Other independent variables include student and classroom-level demographic characteristics. To supplement the baseline model, school fixed effects and a two-stage instrumental variable strategy are employed, where peer sick days from the previous school year serve as the instrumental variable. This IV was chosen given that peer absenteeism due to illness from the previous school year has no relationship to individual test performance the following school year. Additionally, peer rates of missing school due to unforeseen illness are likely to remain stable from year to year.

The findings from this study suggest that peer absenteeism negatively predicts individual student test performance throughout all models. When more rigorous models were used to control for unobserved heterogeneity and omitted variable bias, estimates grew in magnitude yet remained statistically predictive of test performance. This study highlights the importance of attendance on achievement through an OTL framework.