Chapter 11: Direct Modeling of Student Growth With Multilevel and Mixture Extensions

Student academic growth is the center of the new era of educational reform. Kolen and Brennan (2004, p. 376) summarize that growth can be defined in two ways: One is a domain definition of growth and the other is a grade-to-grade definition of growth. The domain definition of growth relates changes in performance over all content in the domain. It is usually operationalized as one test covering all content in the targeted domain. Most often, repeated measures can be utilized to collect relevant data. However, this is not a very practical approach as some items are too easy or too difficult for some students and for some grades, which leads to a waste of resources. The grade-to-grade definition of growth regards student performance over the on-grade/on-level content. Under this framework, content standards, curricula, and test blueprints need to be designed to have hierarchical content strands with substantial between-grade overlap (Yen, 2007). This approach targets student performance in a more limited way. In educational settings, this second approach is most often adopted.