Some of the data come from both students admitted to KIPP and those not admitted in random lotteries, a scientific way of making sure a study is comparing similar groups.
Mathematica said it used a version of the nationally normed, low-stakes TerraNova test with items "assessing higher-order thinking skills" to show that the higher KIPP scores on state tests were not a fluke.
The most original part of the study was comparing higher-performing to lower-performing KIPP schools to ascertain what characteristics had the most impact on learning.
Achievement was greater in KIPP schools "where principals report a more comprehensive school-wide behavior system" and where more time was spent on core academic activities.
Compared with similar students in non-KIPP schools, middle-schoolers gained 11 months of learning in math, eight months in reading, 14 months in science and 11 months in social studies in their first three years at KIPP.
KIPP has 125 schools, including some elementary and high schools, in 20 states and the District of Columbia, but the study looked only at its fifth- through eighth-grade middle schools, the core of the network.
Mathematica found that the schools had no significant impact on persistence and educational aspirations, based on surveys of students and parents.
But KIPP students were more likely to report misbehavior such as losing their temper or giving teachers a hard time.
Was that because they were more prone to mouth off or more prone to admit they had mouthed off?
Those of us immersed in the debate are grateful for a new issue, as we figure out just how good those KIPP teachers are.
Mathews is education writer at The Washington Post.