Late last week the Los Angeles Unified School District and its teachers union (UTLA) reached an agreement on including student test scores in teacher evaluations. The agreement does not include all of the details (UTLA has posted a summary and the full provisions), but it is clear that they will not be using a teacher’s individual value-added score. Instead, they will be using school-level value-added data and a teacher’s raw performance data. Using raw performance data in a teacher’s evaluation is an incredible mistake, and one that both reformers and union leaders should loudly denounce.
Randi Weingarten, president of the American Federation of Teachers, should lead this charge. As recently as August she wrote that, “AFT affiliates are working to transform teacher evaluations from momentary snapshots into meaningful personnel systems that promote continuous growth in student achievement and teacher skill.” Raw performance data, like Los Angeles will be using, are a “momentary snapshot.” Value-added measures (VAM), like the ones that UTLA is rejecting, are measures of “continuous improvement.”
In its statement, UTLA said it rejects the use of value-added data for personnel decisions and that, “individual VAM scores are not useful at the formative level, and research shows that they are an extremely unstable and unreliable method of measuring instructional outcomes or evaluating teacher effectiveness.” Do you know what is not useful at the formative level? What is an unstable and unreliable method of measuring teacher effectiveness? The answer to both questions is raw test scores!
As more states and districts adopt teacher evaluation policies that include student growth, there’s been a lot of concern about whether they adequately account for student differences. That’s a fair concern and worthy of more research, particularly as we learn more about what models can and cannot control for and whether there are differences across grades and subjects. But do you know what doesn’t even attempt to address these differences? Raw test scores! Just by looking quickly at L.A.’s latest report card we see that 76 percent of white students and 70 percent of Asian students were proficient in 5th grade English, compared to only 45 percent of Latinos and 44 percent of African-Americans. Value-added models attempt to take starting differences into account (although they don’t all use race to do so), but raw test scores make no such attempt. If we’re at all concerned about fairness or equity in teacher distribution, raw scores create terrible incentives for teachers to teach in already high-achieving classrooms.
In L.A., the fear over value-added has now gone too far. Teachers, unions, district leaders, and education advocates should all stand up for an evaluation system that is fair, one that is based on multiple measures including student growth (not mere snapshots of student performance), and provides teachers with the support they need to improve. The recent agreement fails that test.
Photo Credit: NorthJersey.com