The Stability of Value-Added for Teachers of Special Education Students

• In Event: Unpacking Value-Added Measures of Teacher Performance: Implications for Disadvantaged Students and the Parents and Teachers Who Support Them

Abstract

Special education students are a heterogeneous group, and they have specialized academic and instructional needs. Although students with specific learning disabilities make up the largest share of special education students in the United States, students with disabilities such as autism or traumatic brain injuries are also classified as special education. Thus, there is no typical student that receives this classification. In addition to variation in disability type, special education students are placed into heterogeneous learning environments. While many are mainstreamed into regular classes for a significant portion of the school day, some receive exclusive instruction from a designated teacher in a separate classroom for the entire day. Ultimately, differences in student disability type, the proportion of special education students in a class, and the type of instruction—mainstreamed or separate—may have implications for the way value-added scores of special education teachers should or should not be used in a teacher evaluation system. This paper provides insight into the properties of value-added scores of special education teachers by examining the stability of teacher value-added scores across student groups—special education and non-special education—and across student disability types. It also provides a descriptive comparison among the value-added stabilities of mainstream teachers and teachers who exclusively teach special education students over time.

The data for this study are from a large, urban school district. In the administrative dataset, students are matched to teachers, and we have the end of year test scores for students in elementary and middle school. Because of the No Child Left Behind Act, almost all special education students are tested. To our advantage, the data set defines whether a student has a disability; it also categorizes the student’s disability. We have 7 years of data from which we compute value-added scores for teachers.

The paper addresses the following research questions:

1. How stable are value-added scores among teachers who teach special education students and non-special education students?
2. How stable are value-added scores among teachers who teach students with a variety of disabilities?
3. Are teacher value-added estimates more stable for mainstream teachers or teachers who exclusively teacher special education students?

As this paper is still in progress, results are not yet available. However, we can discuss the implications of different findings. On the one hand, if value-added measures are highly stable (i.e., highly correlated) among teachers who teach special education and non-special education students, this provides some evidence that these teachers perform well with both groups of students, on average. On the other hand, a lack of stability (i.e., low correlation) might suggest that teacher performance varies by student group or that the value-added scores contain significant measurement error. We examine the extent of measurement error by understanding some of its causes. For example, we analyze the underlying trends in student scores to examine whether the test effectively captures student growth at the lower end of the distribution. Overall, the results can better inform district leaders about the properties of value-added for an important subset of their teaching force.

Author

• Christopher Andrew Candelaria, Stanford University