Los Angeles teacher appraisals won't be based on 'value added,' increasingly being used across the country. Now a key question is how test scores will figure in.
By Teresa Watanabe and Howard Blume, Los Angeles Times | http://lat.ms/UQ1JWH
Randi Weingarten, president of the American Federation of Teachers, praised L.A. Supt. John Deasy and union President Warren Fletcher, above, for agreeing to use a rich mix of data to evaluate teacher effectiveness. (Irfan Khan, Los Angeles Times / December 9, 2012)
December 9, 2012, 8:52 p.m. :: The recent groundbreaking agreement over evaluations for educators in the Los Angeles school district is a major victory for the teachers union because it limits the use of a controversial — but increasingly widespread — measurement of teacher effectiveness.
The tentative pact puts the nation's second-largest school system at odds with a national trend to gauge the effect of teachers on student achievement by using a value-added analysis. That method, known in Los Angeles Unified as Academic Growth Over Time, is opposed by many teacher unions as unreliable; but it is being used in Illinois, New York, Texas, Florida, Washington, D.C., and elsewhere.
The restrictions on using that method will force L.A. Unified to alter its new teacher evaluation system, which every principal and one volunteer instructor in all of the district's 1,300 schools were to be trained on this year. The new system was to include an individual growth rating as a key measure of teachers, along with a rigorous new observation process, parent and student feedback and an instructor's contribution to the school community.
Instead of the growth rating for individual teachers, the district and United Teachers Los Angeles agreed to use a mix of individual and schoolwide data, such as raw state test scores, district assessments and high school exit exams, along with rates of attendance, suspension, graduation, course completion and other indicators.
Los Angeles Mayor Antonio Villaraigosa, who wrote legislation in 1999 requiring the use of state test scores in teacher evaluations while a California assemblyman, said the district's agreement to drop the use of individual growth ratings was "a major concession, no doubt about it."
"But I ultimately supported it because overall it was a step forward from what we had before," he said. "This was an opportunity to get a better tool to measure success in the classroom."
The tentative accord, which will be voted on in January by teachers, was hailed as a creative breakthrough by the chief of a leading national teachers union. Randi Weingarten, president of the 1.5 million-member American Federation of Teachers, praised L.A. Supt. John Deasy and union President Warren Fletcher for agreeing to use a rich mix of data to evaluate teacher effectiveness rather than what she called the "junk science" of value-added methods.
The Academic Growth Over Time measure uses a complex mathematical formula to estimate the progress students make, based on state standardized test scores and controlling for income, race, English language ability and other outside factors that can affect learning.
"They broke through the ideology to really figure out a multiple-measure process to evaluate teachers," Weingarten said.
A teacher's growth rating will still be used to help set goals at the beginning of the school year but cannot be used in the teacher's final evaluation. Schoolwide growth ratings, however, will be used.
These limitations will require L.A. Unified to wrestle with still-unanswered questions on how to use a mass of unfiltered test scores and other data to determine a teacher's effect on student learning, as required by a court order issued by Los Angeles County Superior Court Judge James C. Chalfant. The judge ruled this year that the district was violating state law by not using test scores in teacher performance reviews.
"What is not clear is exactly how this will be implemented in practice," said Judith Perez, president of Associated Administrators of Los Angeles.
District officials say they are working on guidelines for principals but could not yet offer concrete information.
The agreement has raised a host of other questions. It does not say, for instance, how much raw test scores will count in a teacher's final rating. Elsewhere, value-added ratings are used. They make up 40% of the reviews in Washington, D.C., 35% in Tennessee and 15% in Chicago elementary schools.
Deasy has said that for Los Angeles, 25% would be a reasonable figure for test scores and other data. The agreement says only that these elements will not be "the sole, primary or controlling factors" in the final review.
Fletcher used that ambiguity in selling the agreement to the union's governing House of Representatives last week, saying the L.A. deal minimized the effect of test scores on teacher evaluations.
L.A. would not be alone in failing to specify how much test scores and other data count in a teacher review, said Stanford education professor Linda Darling-Hammond. Massachusetts does not specify fixed percentages and still won a Race to the Top federal grant, which requires the use of objective achievement measures, she said.
Fletcher said that because state test scores aren't released until late summer, they can't be used for final evaluations the previous spring. Data from past tests, such as student performance in specific skill areas like reading comprehension, can be used to analyze a student's weaknesses and a teacher's shortcomings — to set goals for instructors, he said.
In contrast, Deasy said the actual scores themselves also matter. They aren't merely a guide for improving instruction. A teacher whose students consistently perform poorly on state tests must improve or face dismissal, he said.
Some teachers expressed misgivings about the use of raw standardized test scores rather than value-added ratings. Kyle Hunsberger, a math teacher at Johnnie L. Cochran Jr. Middle School, said some instructors might be more reluctant to take on low-income minority students if they are evaluated on test scores, which do not account for poverty and other factors teachers cannot control but that are known to affect learning.
Some experts expressed similar concerns.
"I definitely think value-added is better for getting a clean estimate of a teacher's contribution to student learning, but it may not be the best system if teachers distrust it," said University of Washington professor Dan Goldhaber.
Despite the national trend toward value-added, U.S. Secretary of Education Arne Duncan declined to state a preference for that method over raw test scores. What mattered to him, he said, was an evaluation process that motivated all teachers to help all students progress academically.
"I want to know how much students are improving each year," he said. "And are there incentives so teachers and schools and districts are focused not just on a smaller number of kids but on 100% of students?"
He added that the L.A. agreement should not leave unanswered the question of how much weight scores and other data will have.
"Those are tough, complex, difficult issues, but many other districts have come to agreement," Duncan said. "It can't be undefined."
Former state Sen. Gloria Romero, a proponent of using standardized test scores and value-added calculations in teacher evaluations, said she was pleased to see the union accept the use of data but was concerned because the agreement makes all individual value-added ratings confidential. "It does raise questions with the public's right to know about performance," Romero said.
Despite the questions, Deasy praised the union for reaching a deal and said it would ultimately help all students learn from an effective teacher.
"We've moved from none, zero, over my dead body to a remarkably sophisticated and nuanced position," he said. "The debate needs to shift from what's in the agreement to how do we do it well in this gigantic school system. And we need all the voices that pushed us to get to where we are to help us do it well."