Chapter 10: Educational Value-Added Analysis of Covariance Models with Error in the Covariates

One of the more popular modeling approaches for educational value-added assessment is a type of regression known as analysis of covariance (ANCOVA). In this approach, a student’s current-year achievement level (as measured by a test score) is predicted from one or more previous test scores (indicating achievement level prior to the current year), perhaps along with some demographic/socioeconomic predictors, plus the value-added “effect” of the educational entity (e.g., teacher, school, district) responsible for the student in the current year. For expository purposes, in this chapter it is assumed that educational entity of interest is the teacher. Mathematically, the ANCOVA model is