ABSTRACT
Chapters 2 and 3 introduced the multilevel modeling framework and demonstrated the use of the nlme and lme4 R packages in fitting two-level models. In Chapter 4, we will expand upon this basic two-level framework by fitting models with additional levels of data structure. As described in Chapter 2, it is conceivable for a Level 1 unit such as student to be nested in higher level units such as classroom. Thus, in keeping with our examples, we may assume that at least a portion of a student’s performance on a reading test is due to the classroom in which he or she learns. Each classroom may have a unique learning context that may contribute to student performance, for example, the quality of the teacher, the presence of disruptive students, and time of day when students are in the class, among others. Furthermore, as we saw in the earlier chapters, the impacts of fixed effects on a dependent variable can vary among Level 2 units, resulting in a random slope model.
