Class size reduction is one of the American public’s most preferred education policies (1,2). The vast majority of research on class size reduction policy has focused on academic outcomes. However, to the best of our knowledge, only two papers have looked at the relationship between class size and health experimentally or quasi-experimentally, and both used the Tennessee STAR data.  One found no association between class size and disability status in adulthood (3), and another study found, counterintuitively, higher mortality among those who had been in small classes (4).

It is plausible that class size has health benefits beyond the education and labor outcomes already documented through several possible pathways. More generally, observational data suggest the quality of the educational experience is associated with health outcomes across the lifespan (5).  In particular, students in smaller classes report stronger engagement in their school (6).  Increased school engagement is in turn associated with positive health outcomes (7,8). Students in smaller classes display on average more prosocial behavior (9), which in turn is associated with improved health (10). Additionally, in smaller classes, teachers may have stronger relationships with each student (11), and having a positive relationship with a non-parental adult is associated with positive youth development (12), which is in turn improved adolescent health (13).

In this study, I seek to examine the association between class size and health behaviors in a large-scale policy implementation setting. In particular, I use observational data from the most recent decade of students attending North Carolina public schools to quasi-experimentally assess the impact of K-3 class size on physical activity and substance use behaviors in adolescence.

In particular, I apply Angrist and Lavy’s (1999) Maimonides rule approach to the North Carolina context. In North Carolina, all kindergarten-3^rd grade classrooms must be 24 students or less, and the average class size in the local educational authority must be no higher than 21.  (Notably, this rule goes into effect after the first 40 days of school, and the state provides funding to the district to support an average class size of 18.)

I use the number of students per grade to calculate the expected number of classes given the number of students in the grade and the expected class size accordingly.  Some school districts have rules requiring smaller class sizes than what the state requires; when this is the case, the district policy will be applied. 

Using this instrument for class size, I look at sedentary behavior and physical activity. To assess sedentary behavior, I use student-reported TV watching (reported annually at end of grade tests), which has been associated with adverse health outcomes in adolescence (14) and into adulthood (15). After school sports participation, which is also reported annually at the end of grade tests, will serve as a proxy for physical activity. 

Complete references list available upon request.