Understanding Variability in Teacher Performance Measurements

With federal support from the United States Department of Education’s Institute of Education Sciences, researchers at the National Center on Performance Incentives are studying various aspects of value-added measurement. This particular study provides greater understanding about the variability in teacher performance measurements over multiple school years. NCPI affiliates Daniel McCaffrey and J.R. Lockwood, along with Tim Sass, will examine the relative contributions of potential sources of variability and how alternative methods may be more or less sensitive to them. With a more complete base of knowledge about the sources of year-to-year instability in the different kinds of performance measures, research will be able to balance the competing goals of bias reduction and variance control in the next generation of value-added measures.

This supplemental study has four distinct objectives:

  • Describe the sources of variance in teacher effects across school years.
  • Characterize the implications of stratification for estimating teacher effects for widely used longitudinal estimation methods through a detailed state-level case study.
  • Develop tools that allow analyses to determine which teachers can be compared without confounding or bias, and which comparisons are sufficiently precise to support inferences about relative teacher performance.
  • Offer practical advice to help practitioners make informed decisions about how different sources of variance can be managed when creating performance measures for high-stakes decisions.

Please revisit this site in coming months for additional information and forthcoming publications related to this research on value-added measurement.