Authors C. Kirabo Jackson and Elias Bruegmann based their findings on an analysis of 11 years of data on North Carolina schoolchildren. The study is due to be published in October in the American Economic Journal: Applied Economics, a peer-reviewed journal.
The authors and some independent experts said the study results are important, because they carry implications for school staffing practices and debates going on now at the national level over how to structure merit-pay plans for teachers.
“If it’s true that teachers are learning from their peers, and the effects are not small, then we want to make sure that any incentive system we put in place is going to be fostering that and not preventing it,” said Mr. Jackson, an assistant professor of labor economics at Cornell University in Ithaca, N.Y. “If you give the reward at the individual level, all of a sudden my peers are no longer my colleagues—they’re my competitors. If you give it at the school level, then you’re going to foster feelings of team membership, and that increases the incentive to work together and help each other out.”
Studies outside of education have long shown that effective workers can have a spillover effect on their colleagues. Supermarket checkers, for instance, work faster when they are in the line of sight of a productive colleague and berry-pickers tend to calibrate their working speed to that of friends laboring nearby. But studies up until now have not noted the same pattern in teaching, a profession in which it’s long been thought that peers work mostly in isolation.
‘Big Enough’ Effects
For their study, Mr. Jackson and Mr. Bruegmann focused on mathematics and reading test-score data for students in 3rd through 5th grades, most of whom would have had the same teacher for all of their core academic subjects. They measured teacher quality in two ways: by tracking “observable” characteristics, such as whether teachers were experienced or certified, and by calculating how effective teachers were at raising the test scores of their students. The latter, a “value-added” calculation, was figured using data from teachers’ previous students.
Either way, the researchers found, student achievement rises across a grade when a high-quality teacher comes on board. The effects were twice as strong, though, for the value-added calculations. They show that, for the average educator teaching in a grade with three other teachers, replacing one peer with a more effective one has a spillover effect of .86 percent of a standard deviation on students’ test scores.
For math, that equates to roughly one-tenth to one-fifth the size of the impact that is estimated to come from replacing the students’ own teacher with a better one, the paper says.
“He [Mr. Jackson] has some pretty good evidence, as good as you can get in an observational study, that when a good teacher shows up in your grade it seems to have a positive impact, and that impact stays around,” said Douglas O. Staiger, an economics professor at Dartmouth College in Hanover, N.H., who was not part of the study on peer effects.
Another outside expert, Jonah Rockoff, an assistant professor of labor economics at Columbia University’s business school in New York City, concurred. “The effects are big enough that they would matter,” he said. “If we think about rewarding teachers based on student outcomes, teachers are going to care about who’s teaching alongside of them.”
They said the question now is: Do the test scores rise because the new teacher’s arrival is motivating peers to do better, because that teacher is helping out other teachers by doing some of the teaching, or because teachers are learning from their new colleague?
In their paper, Mr. Jackson and Mr. Bruegmann argue that peer learning is the likely explanation, mostly because they find that the effects persist over time. In both math and reading, the quality of a teacher’s peers a year or two before affects his or her students’ achievement, according to their report.
“If it’s motivation, when you’re no longer surrounded by the peer who’s making you work harder, then you shouldn’t still be working harder,” Mr. Jackson said.
The study also finds that good teachers seem to have the most impact on beginning teachers, as well as those who are certified or have regular teaching licenses.
Mr. Rockoff said the idea that teachers, especially beginners, are learning from more-effective colleagues on an informal basis could explain why recent studies, including one released this week by Mathematica Policy Research, are finding that formal teacher-induction programs don’t seem to be having any effect.
In the new study, which looks at comprehensive teacher-induction programs in 17 districts across the country over two years, researchers at the Princeton, N.J.-based research group found that such programs are not any more effective than business as usual at reducing teacher turnover or boosting student achievement.
In an as-yet-unpublished working paper of New York City teachers, Mr. Rockoff said he also finds “weak” evidence of effects from a formal mentoring program on teacher absences, teacher retention, and student achievement. He did, however, find a link between more hours of mentoring and high student achievement in reading and math.
“It’s not that new teachers are showing up before and no one is helping them do anything,” he said. But another problem, he added, may also be that highly structured mentoring programs—ones that might, for example, require mentors to spend a specified amount of time with all new teachers—might be taking away valuable time that more-skilled beginners might be able to use doing more productive activities, such as planning lessons.
Mr. Jackson and Mr. Rockoff said the new peer-effects findings raise questions about the way schools are staffed, particularly urban schools. Studies have shown, for instance, that teachers in urban schools tend to be less experienced and hold fewer credentials than their suburban counterparts.
“A lot of beginning teachers end up in inner-city schools and move to suburban districts,” Mr. Jackson said. “Sending the teachers who need the most guidance to be surrounded by teachers who are the least well-equipped might be a problem. We need to make sure we have some high-quality teachers in inner-city school districts.”
Experts said more research is needed to figure out exactly how peer effects work among teachers.
“Can we take stellar teachers and move them around?” Mr. Staiger asks. “I’m not sure this paper actually says that. This is kind of the most compelling evidence we have to date that there are these spillover effects, and now we have to try to understand them.”
A version of this article appeared in the September 16, 2009 edition of Education Week