In this article argue that too much previous research has tended to assess the effects of student, classroom, and school variables in isolation from other variables and has often used statistical techniques that ignored the nested nature of the 3 classes of factors.

Then argue that a more educationally oriented framework should be used to assess the effects of various student, classroom/teacher, and school variables on student learning, particularly student learning gains, and we identify several variables within each class of factors that research so far has identified. In the article, we suggest some standard hierarchical linear modeling models that could be used to conduct analyses that do account for the nested nature of all variables.