Investigating the Sensitivity of Value-Added Model Teacher Classifications

• In Event: 63.018 - Promising Scholarship in Education: Dissertation Fellows and Their Research

Abstract

Traditional methods of teacher evaluation that are based primarily on principal observations suggest little variation in teacher quality; most teachers are classified as effective or above average (Cantrell & Kane, 2013; Goldhaber, Walch, &Gabele, 2014). In contrast, when teacher effectiveness is characterized on the basis of evidence of student learning using standardized test performance, there appears to be considerable variation in teacher quality (Gordon, Kane, &Staiger, 2006). Concern regarding the validity of both traditional observations and standardized test outcomes have led researchers to focus on the use of multiple measures in the teacher evaluation (Braun, Chudowsky, & Koenig, 2010; Cantrell & Kane, 2013). One such study is the Measures of Effective Teaching (MET) project which investigated multiple classroom observation protocols, Value-Added Model (VAM) estimates based on state assessments as well as alternative assessments, and student perception surveys as methods for teacher evaluation (Bill & Melinda Gates Foundation, 2013). Although findings from the MET study stressed the importance of using multiple measures to evaluate teachers, there are still many questions left unanswered.
Many states are collecting VAM estimates and scores from observation protocols for teacher evaluations. States generally produce two sets of teacher classifications: one based on their VAM estimates and a second based on observation scores. The intersection of these classifications on some form of matrix typically determines the final evaluation. However, there are questions regarding how to set the cut points for these classifications, how to specify the underlying model, as well as what is likely to change in these systems now that the tests used across the country are starting to change.

Author

• Jessica Alzen, University of Colorado - Boulder