Change pay, change teaching?

Taylor Betz will make a lot more as a high school math teacher this year than her normal salary might suggest.

There's the $2,300 bonus she gets for working at a "hard-to-serve school," the $2,300 for filling a "hard-to-staff position," the $2,300 that all teachers at her school are likely to get for raising student scores on state tests, the $2,300 "beating the odds" bonus she gets for significantly raising the math scores of her own students, and a few smaller bonuses.

Given the extra money, it's easy to see why a teacher like Ms. Betz would be an enthusiastic supporter of the "pay for performance" system that Denver has adopted. But even though such systems are proliferating, they're still both highly controversial and little understood.

Performance pay is one of several areas getting attention right now as education reformers zero in on high-quality teaching as the key to helping students learn. The thinking goes like this: It takes good teachers to improve student achievement, and it will take better pay to lure and keep good teachers.

Not only that, advocates of these plans say, but pay should be more directly linked to how well teachers do. And one way to make that link is by looking at students' scores on standardized tests.

Critics, including many unions, point to several issues. It's difficult to determine which teachers are most effective, and it's particularly unfair to tie pay partly to student test scores, the critics say. Also, there's a lack of solid evidence so far that changing the pay structure really improves teaching.

Proponents, meanwhile, insist that one thing holding the teaching profession back is its vastly outdated pay system.

"The single most important [in-school] determinant of a student's success in the classroom is the teacher," says Matthew Springer, director of the National Center on Performance Incentives at Vanderbilt University in Nashville, Tenn. "Yet the ways in which we compensate teachers – years of experience and degrees held – are not strongly correlated with student achievement gains.... That's driving some advocates [of compensation changes] to say there must be a better way."

Where this is happening

Florida, Minnesota, and Texas, along with several other states to a lesser degree, now have policies promoting what's sometimes called merit pay. Houston and New York City have recently developed districtwide programs. And even in the midst of the recession, a portion of schools in dozens of districts are trying experiments through grants from foundations and the US Department of Education's Teacher Incentive Fund.

Denver's system, called ProComp, is the longest-running performance-pay system for a big district in the United States. Its pilot project began in 1999. Now, some 70 percent of Denver teachers – and all new teachers – participate. The cost is funded entirely by a property-tax hike of $25 million per year that voters approved.

Already, backers there say they've seen a change in the culture, with many teachers welcoming the new pay system and approaching goals differently. And preliminary results from an internal study indicate that ProComp teachers are driving higher student performance – but the reason for the improvement isn't definitive yet. An external evaluation is due later this year.

"There needs to be a connection between what [Denver is] doing and increased student achievement," says Phil Gonring, a program officer at the Rose Community Foundation in Denver who was heavily involved in ProComp's inception.

Advocates of plans like ProComp see a couple of ways that such systems could lead to improved student achievement. One way is if the extra pay becomes an incentive for existing teachers to improve. Such improvement could happen because these programs usually provide more professional-development opportunities, as well as more tools to analyze performance.

Another possibility: The change in pay structure attracts and retains better-qualified people who might not otherwise enter teaching.

"Teachers are going to get paid a lot more under these performance-pay plans," Mr. Gonring says. "What we'll see is a transformation in the labor market. It's going to become more economically viable for young people to come into the profession and stay for a good period of time. The No. 1 education issue is human capital management. And money plays a huge role in getting people to stay."

Still, many teachers and unions hang on to a gut reaction against anything that ties their pay to standardized test scores. (See story on linking test scores.)

In addition, none of these pay programs answers the question of what to do with underperforming teachers, who are usually protected by tenure. "I guess your salary stays low, and maybe that sends the message that you should look at another career," says Paul Teske, dean of the School of Public Affairs at the University of Colorado in Denver. "But ProComp doesn't directly address that," says Mr. Teske, who is conducting the external review of Denver's program. [Editor's note: The original version contained an incorrect title for Teske.]

Among the questions that administrators and researchers are trying to sort through:

•What amount of money makes a difference to teachers?

•Should more money go toward group bonuses (to teachers at an entire school, for instance) or be available for individual teachers?

•Which works better: salary increases or yearly bonuses?

•And most important, if the goal is to reward the best teachers, is there a fair way to determine who the best teachers are?

Denver's approach to these questions resulted in a complicated system with more than 10 categories that can affect salary.

"ProComp remains something you can fault for being overly complex, but as you're trying to win over the hearts and minds of teachers, you err on the side of complexity," says Brad Jupp, a senior academic policy adviser in Denver's district. He was previously a union representative who helped design the program.

One of the most striking aspects of Denver's system, in fact, is that it was done with the full support of the local teachers union – a sharp contrast to districts like Washington, D.C., where efforts to push a new pay program, among other reforms, have antagonized the union.

Denver teachers initially opposed

Back in 2004, when the Denver teachers union voted on ProComp, many teachers had a deeply ingrained opposition to "merit pay." One poll about a month before the vote showed that just 19 percent of them were in favor. The district undertook a public-relations blitz and massive information campaign, and it ended up winning the support of 59 percent of teachers.

That backing, say Mr. Jupp and others – along with the fact that the union was involved in designing the program – has been crucial to its success.

For any district, collaboration with teachers can be key, research indicates.

"You can't be a dictator and expect things to be done successfully," says Kim Ursetta, president of the Denver Classroom Teachers Association. "There has to be that commitment to doing it collaboratively."

Part of the resistance to any changes in teacher pay stems from the history of the current system. In the early 20th century, when it was common for women and minorities to be paid less, unions fought hard for the so-called single-salary schedule, in which all teachers are paid based on years of experience and levels of education.

"Having a single-salary schedule was a human rights victory," says Gonring of the Rose Community Foundation. "But now that we've moved beyond that, the civil rights issue has shifted to students and what we're going to do so that students have teachers who produce results."

A number of merit-pay programs were tried in the 1980s, but they generally failed or didn't last long because of budget cuts. They left a bad taste in many teachers' mouths.

But in Denver, some teachers – especially young teachers – have welcomed the change. It makes teaching more attractive, they say, knowing it might be treated more like other professions, in which their performance affects their pay, they get tangible feedback for outstanding results, and they have the opportunity to earn a better salary.

"Telling me I'll have some great pension plan 20 years down the road isn't going to keep me here now," says Ben Jackson, an Advanced Placement English teacher at Bruce Randolph School, the same school where Betz works. Mr. Jackson is young and energetic, in his second year of teaching, and he's eager to work with low-income kids and embrace new reforms.

But he also acknowledges that the traditional teacher career path – with few opportunities to augment his salary and no reward for outstanding performance beyond personal satisfaction – has little appeal for him. Many teachers from his generation, he notes, leave the classroom for policy work after just a few years, despite their love of teaching.

ProComp doesn't solve all Jackson's concerns about stagnating as a teacher, but it helps address some of them. This is where the profession is headed, he says.

"To be able to say, 'I know I'm doing a great job, and I want to be paid for doing a great job,' is paramount," he says.

One thing many systems are trying to do is reward not just performance, but also the decision to work in a school like Bruce Randolph, where 95 percent of students receive free or reduced-price lunches and it can be tough to attract high-quality teachers. When it opened in 2002, the school had so many problems that its entire administration was replaced in the first month. After the first three years, it was the lowest-performing middle school in Colorado.

When Kristin Waters came in as principal in 2005, she asked to have the school included in the ProComp plan. She hoped it would help her recruit new, talented teachers.

Betz, Jackson, and other teachers say the $2,300 bonus they receive for working there isn't the only – or even the primary – reason they chose the school. But the extra money doesn't hurt, and teachers should be rewarded for working in such schools, they say.

Most contentious issue

Even if more teachers are coming on board with the idea of rewarding the best among them, the most contentious piece of performance pay remains how to determine who those teachers are. Most are resistant to emphasizing student test scores. For starters, standardized tests typically cover only certain subjects in certain years, meaning that some teachers could be left out. But on the other hand, scores can rarely be tied to one teacher, since they're usually viewed as the result of a longer, broader process.

Voters and policymakers, meanwhile, want to make sure that performance pay is tied to student achievement. And test scores are often the most neutral – albeit imperfect – measure.

Although Denver has tried to create a mix of incentives to keep people happy, everyone still agrees the system is far from perfect.

Doug Moehle, who teaches sixth- and 10th-grade social studies at Bruce Randolph, is one of the last teachers in the school not to enroll in ProComp. (New teachers in Denver automatically participate, but veteran teachers still have a choice.) The criteria aren't there to make evaluations fair, he argues. Denver, like most districts, tries to use "value-added" models when looking at test scores – looking at improvement in scores rather than overall performance. But Mr. Moehle still sees lots of ways that such systems can fall short.

"Test scores are the worst possible way to judge teachers," he says.

Still, Moehle is considering joining ProComp this year if he decides it's financially in his interest.

Denver recently went through its first major overhaul of ProComp, and the resulting changes led to a major battle with the union (which was still included in the process, but not as much as before). Among other things, the district decided to give more of the money to teachers just starting out, in an effort to retain them and avoid turnover. It also shifted more money to group incentives and increased the size of many bonuses.

"The larger the incentive, the greater the attractive power it has," says Jupp, the academic policy adviser, explaining some of the reasoning. "And we learned that when [raising] student performance is so difficult to do, lowering the stakes for individual incentives and raising the stakes for whole-faculty incentives makes sense."

Other districts have also adjusted after discovering unintended consequences. In Hillsborough County, Fla., the district instituted a performance-pay system in the fall of 2007, under a state directive with many limits on how the program could be structured.

In the first round, only 3 percent of the $2,100 bonuses went to teachers in high-poverty schools, despite sophisticated formulas to try to avoid such inequities, according to an analysis by the St. Petersburg Times. This year, separate award categories were created for those schools.

It's "a big improvement ... [but] I view myself as the teller of the cautionary tale," says Jean Clements, president of the Hillsborough Classroom Teachers Association.

It's hard to know, meanwhile, how the economic crisis will affect the momentum of these pay systems. Denver is still committed, but in Utah, the Legislature might trim some of the $20 million designated for teacher bonuses this year. Florida lawmakers tried to cut back the state's merit-pay program for teachers, but the governor vetoed the reductions. And in Minnesota, the governor has said he wants to expand the touted Q Comp program, but lawmakers are balking at an increase without more data showing that it's improving student achievement.

The problem is that until recently, these pay changes have rarely been paired with strong research evaluations, though several are now in the pipeline. A few studies have shown a positive effect between teacher incentives and student achievement, but it can be difficult to ensure that other variables weren't in play.

"We really don't know what the impact of a teacher pay-for-performance program is going to be on student achievement, teacher behavior, teacher attitudes," says Mr. Springer of the National Center on Performance Incentives. Research does show some factors that seem to work better, he adds. "You have to have a broad set of stakeholders involved in the decisionmaking ... [and] you have to make sure there's a funding stream for a long period of time."

• Part 2 will appear next week.