69传媒

Math teachers trained through the Boston Teacher Residency program are, on average, initially less effective at raising student scores in that subject than other novice teachers. But within five years, their instruction in that subject improves rapidly enough to surpass the effectiveness of their colleagues, a new study concludes.

For English/language arts, the residency-trained teachers were no more effective at improving student achievement than other new teachers.

The Boston program did, however, succeed in drawing a more ethnically diverse group of teachers to the profession than is typical; its candidates were more likely to teach the hard-to-fill subjects of math and science, and they were also much more likely than other new teachers to stay in the classroom for at least five years.

Over the long run, the study suggests, the program should have a modest positive impact on student achievement in Boston when longer retention rates are balanced against teachers鈥� initial weak performance.

Officials at the Boston Plan for Excellence, the nonprofit organization that oversees the residency and that commissioned the study, vowed to use the results to improve their programming.

鈥淚 was disappointed,鈥� Jesse Solomon, its executive director, said of the mixed findings. 鈥淚n my mind, there鈥檚 no way we should be doing worse in those first years. We鈥檒l do what we have got do to make sure that first-year result goes away.鈥�

The study鈥檚 sample for the effectiveness calculations, performed using a value-added method, was relatively small, but the findings were consistent when examined through several lenses.

鈥淲e think this provides reliable evidence on the effectiveness of BTR graduates to date,鈥� said Martin R. West, an assistant professor at the Harvard Graduate School of Education, and one of four scholars who conducted the study. 鈥淭he question is whether we can generalize based on those results to BTR graduates who will later have four to five years of experience, much less to graduates of other residency programs and other settings.鈥�

Residency Context

The study, published this week as a working paper by the Cambridge, Mass.-based National Bureau of Economic Research, is the first independent, empirical study of the teacher-residency approach to training.

In a residency, teacher-candidates have a yearlong apprenticeship under a mentor teacher, from whom they gradually take on increased teaching responsibilities. They also earn a master鈥檚 degree, typically while taking a slimmed-down set of coursework.

The Boston Teacher Residency, begun in 2003, was one of the first examples. It has attracted philanthropic support, spawned similar programs in other universities and school districts, and influenced federal teacher-quality policymaking. Residency programs have also been highlighted as promising models by the National Education Association and the National Council for Accreditation of Teacher Preparation, among others.

For the study, the researchers examined administrative data provided by the Boston school district on the residents who began serving as teachers in their own classrooms in the 2004-05 school year and beyond.

To examine retention, they compared those teachers hired through 2010-11 with other Boston teachers beginning during the same time period.

The data show that 80 percent of residency teachers stayed through year three, compared with 63 percent of their colleagues, and that 75 percent stayed to year five, compared with 51 percent of new, nonresidents.

At the middle and high school levels, the study found that the residency teachers were also more likely to have taught math and/or science than non-residency teachers.

And nearly half of the residents hired through 2010-11 were minority candidates, compared with a third of traditional hires.

Effectiveness Calculation

Because of the limitations of tested grades and subjects, the sample size was restricted for the teacher-effectiveness calculation to only about 50 residents.

Value-added methods attempt to control for demographic factors to isolate the impact the teacher has had on improving his or her students鈥� scores. The method has gained in favor among researchers, though it has proved controversial in policymaking, and questions linger about how sensitive the calculations are to potential bias.

Across several different model specifications, the residency teachers performed less well, on average, in their first year on the job than other new teachers, to a degree that the paper characterizes as equivalent to about two months of learning.

鈥淭he difference there is not trivial in magnitude. It鈥檚 larger than most of the findings in the literature comparing teachers entering through different preparation programs,鈥� Mr. West said.

But by year five, the residents were outperforming other teachers with the same level of experience by nearly the same degree. What鈥檚 more, they had improved rapidly enough to best veteran teachers with more than six years of experience.

Research has shown teachers tend to get better for at least their first five years on the job, but the pace of the residents appears to be accelerated.

The study found no differences between the groups of teachers in English/language arts. In general, reading scores seem to be less responsive to differences in instructional quality, as measured by value-added, than math scores.

The researchers ran a few additional tests to examine whether the presence or absence of mentor teachers influenced the results, but found no evidence either way. They also discussed other hypotheses for the findings鈥攆or instance, Boston may have a more selective labor market in general than many of the places residencies now operate鈥攂ut did not come to any firm conclusions.

To estimate the overall effect of the program, the researchers also ran several hypothetical scenarios, staffing one school with 50 traditional new teachers and another with 50 residency teachers, taking into account the different retention rates of the groups. They concluded that, after four years, the latter group would boost student achievement more.

For Mr. Solomon, the findings show that many of the program鈥檚 goals have been met, but the program has more work to do on the most important one鈥攊mproved learning outcomes. In its early years, the program had focused on recruitment and retention efforts, but over time has made producing effective teachers a priority, he said.

鈥淲e just felt like, look, we can study and measure retention and principal satisfaction 鈥榯ill the cows come home, but you want to know how kids are doing,鈥� Mr. Solomon said.

Beginning with this year鈥檚 group of residents, the Boston Plan for Excellence has made some alterations to the program. It has updated its recruiting practices, and teacher-candidates must now pass a series of performance assessments over the course of the year. They are also now placed in schools in teams.

Study Limitations

It鈥檚 difficult to know how far to extrapolate the findings to residency programs in general.

While the programs share the longer student-teaching experience, they differ in such details as the selection of mentors, sequencing of coursework, and program administration.

鈥淚t鈥檚 really clear that what the Boston data is saying is that we need more research, to look more deeply, and to see what the highest impact components of the model are that we think are linked to student achievement,鈥� said Anissa LIstak, the executive director of Urban Teacher Residency United, a Chicago-based technical-assistance provider to a network of teacher-residency programs.

On the other hand, the study raises more questions about the scope and sequence of student-teaching.

鈥淛ust having one year of clinical experience under a relatively effective mentor, in and of itself, does not ensure that graduates of the program will enter at a level above other novices,鈥� Mr. West said.

Other studies have found a similar pattern of apparently delayed effectiveness. For instance, a 2008 study on New York City鈥檚 Teaching Fellows, an alternative-route program without a residency year, found that teachers had weaker first-year performance than others, followed by stronger gains.

Similarly, an impact study by the Princeton, N.J.-based Mathematica Policy Research on teacher induction found that results on student achievement showed up some time after the induction services had ended.

Mr. Solomon pointed out, meanwhile, that the value-added data don鈥檛 represent the full breadth of teachers trained through the Boston residency, such as high school teachers or those in subjects other than reading and math.

Still, the study adds to the research literature on teacher preparation鈥攁n area in which hard information about which practices work under which conditions is almost embarrassingly limited.

More research on residency programs is forthcoming. The U.S. Department of Education鈥檚 Institute of Education Sciences has embarked on a five-year effort to study teacher-residency programs. Preliminary results from that study, which will look at both federally financed and independent programs, are due out in fall 2013.