This paper proposes a behavioral model to study how schools influence students’ educational behavior and academic achievements. The school quality is then defined into two dimensions: the amount of market-valued skills schools impart and how well schools cultivate an educational identity. Using data from Add Health in the US, I test the major hypotheses from the theoretical model. On the one hand, school resources (average class size and teacher supply) and student-level curriculum have some effects on the math GPA scores. On the other hand, educational identity indicators (school-level happiness and participation at school teams, clubs or organizations) and the previous math GPA scores are significant determinants in students’ observable effort level such as absenteeism behavior, and through this channel both determinants indirectly influence math GPA achievement. These empirical results inform us that an identity-based behavioral model adds to a rational expectation educational choice model in understanding the widening academic achievement gap between adolescents from different socioeconomic backgrounds. The paper presents the limitation of using school resources to study the school quality and advocates a richer set of school quality measures.