The old-fashioned practice of rating instructors by watching them teach is tricky, labor-intensive, potentially costly and subjective — but perhaps the best way to help them improve, according to a study released Friday by the Bill & Melinda Gates Foundation.
The findings highlight the importance of teacher observations, but also pinpoint why they frequently don't work. The old way — observing a teacher once a year, or once every five years in some cases — is insufficient. And the observers, typically the school principal, frequently don't know what to look for anyway.
But that doesn't mean teacher observations should be tossed aside. The best way to evaluate teachers, while also helping them improve, is to use several measures — including data-based methods that rely on students' standardized test scores, along with an updated teacher observation system, the report found.
Separately, research by Harvard and Columbia economists found that students of teachers who improve their test scores derive quantifiable long-term benefits, including higher incomes as adults. The analysis tracked 2.5 million students over 20 years.
The latest Gates study is part of a teacher-effectiveness initiative undertaken by the Seattle-based foundation. Earlier research examined the value of having students fill out surveys about their classrooms, with such questions as whether time is wasted in class or if there is feedback on homework. Even more hotly debated has been the Gates work on value-added measures, which use student progress on standardized tests to gauge the quality of their teachers.
Using these methods to evaluate teachers is "more predictive and powerful in combination than anything we have used as a proxy in the past," said Vicki Phillips, who directs the Gates project.
A multifaceted strategy isn't perfect either, Phillips added, but far surpasses traditional practices, which frequently offer little evaluation of any kind. Under those, about 98% of teachers rate as effective, while instructors receive little feedback that would help them improve and little opportunity to share proven techniques with colleagues.
The traditional approach in the Memphis, Tenn., school system had long included teacher observations — once every five years, said Tequilla Banks, head of Teacher Talent and Effectiveness for Memphis City Schools. Principals would receive some initial training, but no one certified their evaluation skills or monitored their success.
Under a new system, supported by $90 million from Gates, teachers are observed four to six times annually by more than one person.
Evaluators must pass a certification program that involves analyzing sample lessons to see if their skills match those of a 17-member committee that included teachers, principals and other administrators.
Teachers get detailed feedback on observations within seven days.
"This process is neither quick nor easy," said Banks. "Getting classroom observations right is tough work but a critical component."
In Memphis' still-evolving system, 40% of a teacher's evaluation is based on observations, 35% on a value-added formula, 15% on other measures of achievement, 5% on student surveys and 5% on the teacher's demonstrated knowledge in the subject being taught.
A teacher's value-added rating should be part of the equation because it best predicts how future students of that teacher will perform on standardized tests, said Tom Kane, the Gates foundation's deputy director of education. But year to year, a value-added score can fluctuate and it's less helpful in guiding improvement, Kane said.
Banks credited the teachers union for playing a key role in developing the system.
In the Los Angeles Unified School District, teachers union leaders have accepted that test data should be used to improve instruction, but side with experts who question the use of student test scores in teacher evaluations. District officials side with experts who endorse value-added measures as a portion of a teacher's evaluation.
L.A. Unified is moving ahead with plans for such a system. The union insists that the components of an evaluation must be agreed upon through labor negotiations.
The study by the Harvard and Columbia economists found that students enjoyed tangible long-term benefits from teachers who consistently recorded high value-added ratings.
Replacing a poorly rated teacher with an average one, for example, would raise a single classroom's lifetime earnings by about $266,000, the economists estimated. The students also are more likely to go to college and less likely to become teenage mothers.
Researchers have not released the name of the school system in the study — at the request of that district, said co-author John Friedman, a Harvard assistant professor of public policy. The 20-year period of the analysis, which awaits formal peer review, began in 1989.
"Our study shows that great teachers create great value," the authors wrote.