Educators and researchers on both sides of the value-added debate met in Washington last week to discuss the role of student-performance data in evaluating teachers.
Speaking at a forum organized by the nonprofit Center for American Progress, Dan Goldhaber, an education researcher and professor at the University of Washington-Bothell, presented findings from his new paper, “When the Stakes Are High, Can We Rely on Value-Added?” Goldhaber began by declaring himself “an advocate of using value-added measurements carefully to inform some high-stakes decisions.” He said he understands there are downsides to value-added estimation, which measures a teacher’s impact by tracking student growth on test scores from year to year, but that “we shouldn’t let the perfect be the enemy of the good.”
Opponents of value-added initiatives cite measurement error as a concern—but that’s an inevitable part of any teacher evaluation process, said Goldhaber. In fact, the current systems for evaluating and rewarding teachers have measurement-error potential that he calls “more opaque.” For example, having a master’s degree is not a good predictor of how well an educator will teach reading or math, Goldhaber said. And classroom observations can miss the mark as well: “You can catch teachers on a good day or on a day where all the kids are sick and there’s chaos.”
Overall, he said, his research indicates that value-added estimation “does a better job of predicting achievement than a whole host of teacher characteristics,” including experience and credentials.
But panel participant Angela Minnici, an associate director at the American Federation of Teachers, said there’s a lot of confusion around value-added systems and that it’s “really a concern for teachers.”
In conversations with teachers around the country, Minnici said she often hears about “perceptions of fairness"—for instance, that tests don’t assess the breadth and depth of student learning. And, as Goldhaber himself pointed out, the potential for cheating and misclassification can cause distrust of value-added measures, too.
“Teachers don’t have a problem with being held accountable,” Minnici said, “but they want to make sure it’s fair and that it will help them improve and be successful with their students.”
Some educators’ resistance to the focus on value-added measures stems from the fear that policy changes will lead them down the test-obsessed No Child Left Behind road again, according to Minnici. Teachers are also “reform-fatigued,” she said. They’ve seen numerous reform efforts come and go, and many may be just “biding their time” until the value-added fad has passed.
Quality Control
Jennifer Steele, a policy researcher at the RAND Corporation who spoke at the event, has been studying the challenges schools face when factoring student-achievement data into teacher evaluations. She offered findings from a recent report she co-authored, “Incorporating Student Performance Measures into Teacher Evaluation Systems.” Steele’s team looked at evaluation systems used statewide in Tennessee and Delaware, and those used at the district level in Denver; Washington, D.C.; and Hillsborough County, Fla.
The paper concludes that student-performance measures need to be reliable (internally consistent and without a big margin of error), valid (aligned to the curriculum, but not so closely to require teaching to the test), and “vertically scaled” (reflecting students’ absolute growth rather than a change relative to peers) before they are used for high-stakes decision-making.
But these kinds of quality control measures aren’t cheap, Steele noted. “The challenge is finding resources [at the state and district level],” she said. “We’re advocating doing validation studies, and that demands resources.” She is hoping implementation of the common core standards, which 44 states so far have adopted, will ease the burden on state and district central offices.
The RAND report also looked at the different approaches systems take in overcoming another major challenge—measuring student growth in non-tested subjects and grades. Some systems, for example, hold all teachers accountable for schoolwide performance, said Steele. Denver and the District of Columbia allow teachers in non-tested areas to choose their own student growth targets. Hillsborough district leaders took a more laborious approach, developing several hundred end-of-course exams to use for value-added tracking. Steele’s group does not support any one method, but says states and school systems learn from one another.
The report also recommends that evaluation systems use multiple measures of teacher effectiveness, including value-added data, and consider multiple years of student achievement data in calculating value-added estimations.
Goldhaber agreed that value-added measurements should not be used in isolation for evaluating teachers—but they should “be used to identify groups of teachers we should be concerned about,” he said.
“I’m not wedded to the idea of using value-added,” Goldhaber asserted. “I am wedded to the idea that we need to do more to differentiate between teachers.”
