State to use a "value-added" growth model without calling it that

State test scores won’t count more toward the evaluations of elementary and middle school teachers next year, according to an amended proposal that a Board of Regents committee passed unanimously on Monday.

The proposed model, which was formally approved on Tuesday, included a methodology to calculate student growth that was nearly identical to the “value-added” model that State Education Commissioner John King brought to the board in April. Both models add new data points to the formula used to approximate how much each teacher has contributed to students’ growth.

But under state law, any model termed “value-added” would have required, controversially, that its weight increase from 20 to 25 percent on some teacher evaluations. King’s alternative this month was for the state to adopt an “enhanced growth model” that adds virtually all of the same data points but doesn’t have the value-added moniker. Spurning the name allows the state to avoid increasing the weight of test scores until all districts have at least one year of implementation under their belts, something the state teachers union has asked for.

“I would have thought that adding all these factors would qualify as ‘value-added,’ but this distinction was always opaque,” said Jonah Rockoff, a Columbia University economist who advised the state on its methodology “If the commissioner wants to keep the weight at 20 percent for another year then staying within the ‘student growth’ framework seems like the simplest way to do it.”

Both models will factor in more complex student and classroom characteristics to calculate learning on those state tests. But the April proposal was met with resistance from members who questioned the methodology’s reliability and opposed increasing the weight from 20 percent to 25 percent. The increase was one that lawmakers envisioned would take place for the current school year when they enacted the teacher evaluation legislation in 2010.

King’s proposal would stay in effect through end of the 2013-2014 school year, until New York City, which lags behind the rest of the state, has a chance to implement its teacher evaluation system for the first time.

With barely a peep of opposition, the board voted to adopt the “value-added” model at 25 percent beginning in the 2014-2015 school year. The changes affect only elementary and middle school teachers who teach English and math, who together make up less than 20 percent of the state’s teachers.

Chancellor Merryl Tisch first signaled that King’s proposal in April was in jeopardy last week, saying that forgoing the added weight of state test scores was a concession the Regents were willing to make.

“This is not the stuff that I feel we go to war over,” Tisch said.

The state teachers union was quick to claim credit for the amended proposal, issuing a lengthy statement that praised Tisch and the Board of Regents. In a sign that the chilly relations between the union and the State Education Department have not thawed, the statement does not mention King.

“NYSUT, as the voice of teachers across the state, remains committed to working as partners with the Regents and other policymakers toward our shared goal of raising standards and improving outcomes for all students,” said Dick Iannuzzi, president of the New York State United Teachers.

New York City Department of Education Chief Academic Officer Shael Polakow-Suransky said the city did not stake out a strong position on the issue.

“Honestly, it makes no difference,” he said. “I know this was a big issue for the state teachers union, but it really makes no difference between 25 and 20 [percent]. It isn’t what’s going to impact the teacher’s final score. It doesn’t change the data in a substantive way, so we don’t have strong opposition or support for the change.”

The state teacher’s union release refers to the “untested, unreliable Value-Added Model,” but the growth model that the Board of Regents formally approved today isn’t all that different. The basic growth model in place currently takes into account just four individual student variables: previous years’ test scores, whether a student is learning English, whether he has a learning disability, and whether his family is in poverty.

The “enhanced” growth model — as with the value-added model — takes into account many new and more complex variables. Now, the test scores of a student who was held back a year or who is new to a school will be weighted differently. Previous test scores in other subjects will also be considered.

Growth ratings will also be weighted based on how many special education or English language learning students a teacher has in her classroom.

The state will also count the test scores of students who were in attendance at least 60 percent of the school year, a change from the 2011-2012 school year that could account for nearly 150,000 students. Their scores will be weighted in part based on their rate of truancy.

Two factors that were part of the “value-added” proposal that got nixed from the newer version were a teacher’s class size and the number of overage and under-credited students. Those factors will be considered in 2014-2015 when state tests incrase to 25 perent. King said that there might be other data points to review with its contractor, American Institute of Research, to further refine the growth model.