In this paper, the methodology of formulation, analysis, and interpretation on data analysis models for successive cohorts using three-level multilevel models are suggested. Student outcome was modeled by student background in the first-level model. In the second-level model, parameters of the first-level model, which represented the adjusted school mean and background effect, were modeled by school-grade. As a result, parameters of the first-level were separated into two groups of components: entire-grade average components and growth components. In the third-level model, long-run average components and growth components were modeled as a function of school-level variables which represented school context. At each level, intraclass correlations were derived for justification of the implementation of multilevel model and η² form indexes for examining goodness-of-fit. In addition, mean growth curve for each school, overall mean growth curve were derived and finally the evaluation strategies of school effect on these parameters were suggested.