Education reform: Can poor test scores get a teacher fired?

In Houston, a controversial education reform measure allows teachers to be fired based on their students' test scores. Some parents back the policy, but teachers unions have reservations.

Imagine if a computer could identify the weakest-link teachers – the ones who should be told it's time to get out of the classroom.

It's not quite so simple, but a new policy in Houston allows teachers to be fired based on data that some experts say isolates a teacher's effect on his or her students' test-score gains.

Reform advocates say school districts should improve teacher quality in part by using such "value added" data. Dozens of districts, including Houston's, have already incorporated the concept into "pay for performance" systems. Education leaders in New York City and the District of Columbia are moving toward linking it to tenure or dismissals. But none has gone ahead as boldly as the Texas district.

"The worst teachers in a school really drag down achievement," says Eric Hanu­shek, a senior fellow at Stanford's Hoover Institution in California. "The biggest tension is: How much do you rely upon objective statistical information from test scores, and how much do you rely on other measures of teacher performance?"

A number of parents backed the Houston decision at a packed February board meeting. But the local teachers union is planning a legal challenge, claiming, among other concerns, that the formula is not public and leaves teachers in the dark about how they're judged.

The district defends it as a tool to help principals ensure that each classroom has an effective teacher. No one has been let go yet under the new policy, but at the end of the school year, the data could be cited as one criterion for not renewing a teacher's contract.

The controversy highlights a broader debate over how to improve teacher evaluations – which are "largely broken," US Secretary of Education Arne Duncan said recently before a House committee.

Secretary Duncan presented the blueprint of the No Child Left Behind overhaul to the House education committee Wednesday. The plan includes a number of provisions for improving teacher quality.

For all the potential flaws, linking teacher evaluations to student achievement data is a move in the right direction, says Kate Walsh, president of the National Council on Teacher Quality in Washington. Under the status quo, she says, teachers tend to be fired only if they are abusive or break the law. Studies in a sample of districts across the United States have found that less than 1 percent of teachers earn an unsatisfactory rating on their evaluations. "We do not fire teachers because they aren't good at teaching math [or other subjects]," she says.

Even Randi Weingarten, president of the American Federation of Teachers (AFT), is open to some use of student test data in teacher evaluations, but, she says, "there's the right way and the wrong way to do it." In Houston and New York, she says, "it's high-stakes 'gotcha,' as opposed to using data to inform [classroom] instruction."

The Houston AFT-affiliated union has multiple objections to the new policy. One is the time lag: The teachers don't find out how the formula scored them, based on their students' performance on two standardized tests, until the middle of the following school year. That timing is not very useful, they say.

"If you want to come up with a magic number ... and send it to me six months after I give the test and tell me, 'Oh, that was your goal, and – oops – you've missed it,' you haven't done anyone any good at that point.... You haven't helped the kids," says Gayle Fallon, president of the Houston Federation of Teachers.

The move also decreases the trust of the union, Ms. Fallon says, which suspected this kind of "slippery slope" when the data were first used for performance bonuses in 2007.

The Houston union's skepticism is legitimate, says Sean Corcoran, an assistant professor of educational economics at New York University who has been researching value-added methods. Factors such as many students moving in or out of the district during the school year, or even a distraction outside the window on test day, could make the numbers less reliable, he says. "Their high-stakes use is almost impossible, and I think anyone who thinks otherwise either doesn't understand these models or is willing to put aside a lot of uncertainty," he says.

Greg Meyers, president of the Houston Independent School District's Board of Education, defends the district's new policy. Using the data in making decisions about teacher-contract renewals is just a natural progression after Houston's forays into performance pay and other reforms, he says.

"We're probably one of the most data-driven districts in the country.... Teachers and students know which teachers are less effective. This will help pinpoint and quantify it," Mr. Meyers says.

The broader point is to use the data to target professional development, he says.

Groups such as the Education Equality Project – a national coalition that advocates closing achievement gaps among racial and economic groups – argue that if value-added data meet certain criteria and are gathered for several years, they can fairly tie student gains to the individual effect of teachers.

While some object to particular data systems, 55 percent of teachers nationally say that, in general, student growth over the course of an academic year is a "very accurate" measure of teacher performance, according to a survey released this month by Scholastic Inc. and the Bill & Melinda Gates Foundation.