“…Second, especially in the case of retention's effects on adolescent deviancy, the stable unit treatment value assumption (Rubin, 1986) cannot be maintained. Previous research has hinted that structural school characteristics, such as the tracking system (Van Houtte, Demanet, & Stevens, 2012), and compositional characteristics, such as socioeconomic status (SES) and ethnic composition (Demanet & Van Houtte, 2011), may shape students' frame of reference for making social comparisons. Similarly, we may expect the percentage of retained students at a school to shape the nature of the social comparison process, which may affect the relation between grade retention and school misconduct.…”