A new study comparing test scores among elementary school students who had Teach for America instructors and who had other teachers finds it's a wash.

The leaders of the controversial organization, which recruits graduates of selective universities and places them in classrooms after only several weeks of training, are pleased with the results. Chief executive officers Elisa Villanueva Beard and Matthew Kramer write online that the other teachers had an average of 13.6 years of experience, compared to only 1.7 years among the Teach for America group.

Teach for America was established on the premise that highly qualified young people could do as well or better than experienced educators in the nation's most disadvantaged schools after only minimal training. While the organization has recently lengthened its training schedule and begun recruiting older people to teach as well, the results are arguably another vindication of Teach for America's model. The group has become a focal point in the larger debate over whether the existing system of education colleges and certifications are doing enough to meaningfully prepare teachers for the classroom

On the other hand, the fact that Teach for America's teachers did no better than their more experienced colleagues in this latest study points to how difficult it is to improve students' test scores by improving the quality of instruction. Other research has found that while a year spent in a gifted educator's classroom can have a profound effect on a child's career prospects, identifying those talented people is hard.

So is getting them to enter teaching and to remain in the poorly compensated profession. Teach for America is a very selective program that accepted only 15 percent of applicants last year, but choosing those prospective teachers whose students will perform consistently better on tests than other teachers' students has proven a challenge for the group, which is now having some trouble attracting new teachers as well.

Previous research has found that Teach for America staff give their students an advantage over their peers who had regular teachers on math tests. The authors of the new study weren't able to replicate those results. They could not confidently identify a difference in the test results of the two groups of students.

That might have less to do with Teach for America and more to do with the other teachers in the schools, said Melissa Clark, one of the authors of the new report from Mathematica Policy Research.

In schools in impoverished neighborhoods where principals have the most trouble hiring and retaining talented teachers, Teach for America staff would be more likely to have an advantage over their colleagues. As the organization expands, Clark said, Teach for America might be moving beyond the schools where it can do the most good. The organization might be more likely to place its recruits in somewhat less disadvantaged schools, where they are no more effective in raising their students test scores than their colleagues.

Teach for America begun a major expansion several years ago, funded by a grant from the federal government and likely facilitated by a weak labor market for college graduates, who may have increasingly turned to teaching after failing to find other work.

Clark noted that the teachers outside of Teach for America in her group's study seem to be better qualified and more experienced than the teachers to whom past studies compared Teach for America staff. She also that the difference could be a result of efforts nationwide to recruit more qualified teachers, and cautioned that any explanation for the similarity in math scores is just speculation.

The report compared test results for the students of 156 teachers at three dozen schools in 10 states, matching Teach for America personnel with regular staff working in the same schools, at the same grade levels and under similar conditions. Principals agreed to randomly assign students to teachers of both types, to ensure that they weren't favoring one group or the other with better students.

Although the teachers in the study were all similar in terms of their students' performance on tests, there were other important distinctions between those in and out of Teach for America.

Seventy-six percent of the Teach for America staff had graduated from a selective college or university, as ranked by Barron's, compared to 40 percent of the regular teachers. Eighty-four percent had majored in a field besides education, compared to 26 percent of regular staff.

Teach for America's teachers were also much less likely to say they felt prepared for their first job and less likely to describe their training as useful in response to a survey.

And the Teach for America staff were less satisfied with their work in a number of ways. They were less likely to say that they felt camaraderie with their colleagues, and they rated the administrations of their schools and the professional caliber of the other staff less favorably. They were more likely to say that their work did not offer prestige, an intellectual challenge or opportunities for advancement. An overwhelming majority said they planned not to continue in teaching for their entire careers, compared to just a quarter of the regular staff.

Beard and Kramer acknowledged these problems, saying that the organization was looking for ways to do more to help its teachers. "During their training, we’re giving corps members more classroom time in settings that more closely match where they’ll be teaching in the fall," they wrote. "We’ve also strengthened our real-time coaching for corps members, giving them another avenue of support for the always-difficult early years of teaching."

Teachers in both groups did give similar responses when asked whether they felt they had a chance to help students succeed, both in and out of school, and whether they thought the job was personally fulfilling. Broadly speaking, though, the dissatisfaction in the Teach for America corps points to the difficulty of attracting well qualified, ambitious young people to an occupation with limited pay and professional autonomy.

The study did contain one piece of very encouraging news for Teach for America's early-childhood staff. Clark's group found that kids in second grade and younger who had Teach for America teachers performed markedly better on tests of reading. The organization's early-childhood program is relatively new, having been established in 2006, and hasn't been studied in detail before.

Clark wondered if principals might be assigning their best teachers to the higher elementary grades, as federal law requires testing in grades 3-8. If further research confirms her theory, lawmakers debating when students should be tested will want to pay careful attention to the results. Yet Clark cautioned again that the results for reading scores in early childhood were still preliminary.

Another difference between the two groups was that Teach for America staff were slightly less likely to say that a lack of support at home was a problem for their students.

Although the difference was not statistically significant, a tenet of Teach for America's philosophy is that poverty is not a barrier to academic achievement. That may be true in the abstract, but the findings in this report suggest that Teach for America staff -- on their own -- can't overcome whatever barriers their students do face.