ABSTRACT
This chapter introduces the third level of structure, classroom, which is nested within schools to the earlier discussed two levels of data structure, students within schools, and along with associated predictors of reading achievement at each level. It shows that, using Mplus, it is possible to estimate models with three levels of a nested structure and that the syntax for defining and fitting these models is very similar to that in the two-level case. When using Mplus for multilevel models, it is important that all ID numbers be unique. The chapter discusses the application of multilevel models to longitudinal data. Of key importance is that the core ideas, including fitting of the null, random intercept, and random coefficients models, as well as inclusion of predictors at different levels of the data, do not change when there is longitudinal data. By recasting longitudinal data in this manner, we make available to ourselves the flexibility and power of multilevel models.
