Panel Paper: Measuring Teacher Quality: Does the Common Core Make a Difference?

Saturday, November 4, 2017
Water Tower (Hyatt Regency Chicago)

*Names in bold indicate Presenter

Jessalynn James1, Thomas Dee2, Aaron Robert Phipps1 and James Wyckoff1, (1)University of Virginia, (2)Stanford University


Much has rightfully been made of the potential for the Common Core State Standards (CCSS) and exams aligned with those standards to improve the depth of students’ understanding (Conley, 2014). These standards and tests require a different level of understanding and practice by teachers. As a result, many school districts are scrambling to adjust their professional development to better equip teachers with the skills and practices that improve student performance on Common Core exams. Given these changes, we would expect that student performance gains on different assessments might be attributable to different teaching practices. Indeed, prior research demonstrates that teachers’ value-added ranking can vary according to the test used to assess student outcomes (Backes, et al., 2016; Lockwood, et al., 2007; Papay, 2011). The teaching practices that drive learning on the CCSS—as measured, for example, by the Partnership for Assessment of Readiness for College and Careers (PARCC) exam—may differ from those that drive learning on less rigorous standards; similarly, teachers who are effective at teaching to the less-rigorous standards may be less effective at teaching deeper, higher-order content. As a result, a teacher whose students showed relatively strong gains on a traditional state achievement might not on a Common Core exam. Thus, teacher value-added may sort teachers differently depending on the depth of the learning standards on which the exam is based.

The District of Columbia Public School system (DCPS) provides a unique setting for researching these issues. DCPS, like many districts across the U.S., recently transitioned from local learning standards to the CCSS. DCPS adopted the new learning standards in the 2012-13 academic year and by the 2014-15 academic year had transitioned from the District of Columbia Comprehensive Assessment System (DC CAS) to PARCC for assessing student achievement. Throughout this period, DCPS maintained a common teacher evaluation rubric, the Teaching and Learning Framework (TLF), which addresses nine core teaching practices. These data allow us to investigate: first, whether the two assessments rank teachers differently according to their value added; and second, how performance on the TLF relates to student achievement gains across the two assessments. For example, are some components of the TLF more highly correlated with teacher value-added on the DC CAS than PARCC and vice versa?

We estimate value-added using a method similar to the high-stakes teacher evaluation model used operationally within DCPS (see Isenberg & Walsh, 2014). Teacher effects are estimated quasi-experimentally using regressions that include teacher-by-grade-by-year fixed-effects. Distributions of these value-added estimates are analyzed descriptively; we explore how teachers’ relative value-added rankings compare before and after adoption of the CCSS and the PARCC exam, as well as how teachers’ value-added compares to each of the teaching standards measured by TLF when teachers are differentially more effective on one test than the other. Preliminary results suggest that, on average, teachers rank similarly across the two exams, but certain teaching practices are better associated with high performance on one test than the other, potentially allowing us to identify high-leverage practices for generating student achievement gains.