While controversial, VAM has played a powerful and increasingly important role in K-12 assessment. It’s become the centerpiece of governmental accountability initiatives at the city, state, and federal levels. The reason is that the focus of teacher assessment has shifted from its historic concern with how well teachers taught to how much students learned. It’s a profound change, reflective of the nation’s transition from an industrial economy focused on the process of education to an information economy concerned with the outcomes of that education.
Arthur Levine
For this reason, I have long been an advocate of VAM. At the Woodrow Wilson Foundation, we have used state data to determine VAM scores for all of the teachers prepared through our Woodrow Wilson National Teaching Fellowship program in Indiana and Michigan. However, there are severe practical limitations in using VAM measures to assess teacher effectiveness. For example, states do not test in many subjects, making it impossible to gauge faculty performance in those areas. Even in those subject areas in which states do test, the number of students assessed in a class must be large enough to assure significant results, which is challenging in small urban and rural schools. And assessing teachers in classes that have had more than one teacher is problematic.
Outcomes-based assessment is a relatively new phenomenon in teacher evaluation. So it’s not surprising that current VAM models are flawed. I suspect when we look back at current practice in the decades ahead, we will view it as primitive. It will only improve by successive approximations. The question is, what do we do now?
First, measure what matters. Continue to focus on student learning as the key measure of teacher effectiveness.
Second, recognize the limitations in VAM methodology. Don’t abandon VAM but instead supplement it. The Bill & Melinda Gates Foundation found, through its Measures of Effective Teaching project, that quality can best be evaluated using a combination of student achievement gains, multiple teacher observations, and student evaluations. There are a host of other effective measures that can be added to the mix, including tests of subject-matter knowledge, the Classroom Assessment Scoring System (CLASS) for teacher practice, the Danielson competency framework tools, National Board for Professional Teaching Standards assessments, the edTPA teacher-licensing exam, and much more.
Third, improve state data systems for assessing teacher effectiveness. In most states, the data is not available to permit comprehensive or timely teacher evaluation. Too often, policy is ahead of feasibility. That is, government mandates outcome-based assessment of teachers but the data to accomplish this are not being collected or have severe limitations in applicability. States need to better link policy and data systems. They need to create the data systems required for outcome-based assessment. And they need to test them in practice before using them statewide.
Fourth, we need to create more advanced, agile, and sensitive 2.0 versions of VAM assessment. Today, we know far more about what is required and how to collect and analyze student performance data than we did when VAM was first launched. We need to apply the needs and the knowledge to address the shortcomings of VAM 1.0. That means addressing insufficient data in areas like high school physics, chemistry, earth science, or advanced math. It means addressing small class sizes. It means determining the contributions of educators who may replace a teacher of record midyear.
Outcomes-based assessment cannot be stopped, but it can be substantially better. If we accomplish these things, we can better serve our teachers, the programs that educate them, the schools and most importantly, our children.
Arthur Levine is the president of the Woodrow Wilson National Fellowship Foundation in Princeton, N.J. He served as the president of Teachers College from 1994 to 2006.