Study Stings KIPP on Attrition Rates
KIPP charter middle schools enroll a significantly higher proportion of African-American students than the local school districts they draw from, but 40 percent of the black males they enroll leave between grades 6 and 8, says a new nationwide study by researchers at Western Michigan University.
“The dropout rate for African-American males is really shocking,” said Gary J. Miron, a professor of evaluation, measurement, and research at the university, in Kalamazoo, and the lead researcher for the study. “Kipp is doing a great job of educating students who persist, but not all who come.”
With 99 charter schools across the country, most of which serve grades 5 to 8, the Knowledge Is Power Program network has built a national reputation for success in enabling low-income minority students to do well academically. And some studies show that KIPP charter schools have succeeded in significantly narrowing race-based and income-based achievement gaps between students over
time. While not disputing that track record, the new study attempts to probe some of the more unexplored factors that might play into KIPP’s success.
It concludes, for instance, that KIPP schools are considerably better funded on a per-pupil basis than their surrounding school districts. The KIPP schools received, on average, $18,500 per pupil in 2007-08, about $6,500 more per student than the average for other schools in the same districts, according to the researchers’ analysis of federal 990 tax forms filed by schools reporting both public and private sources of funding. The study reports that nearly $5,800 of that per-pupil amount is private donations and grants.
Mr. Miron said the “$6,500 cost advantage” raises questions about the sustainability of the KIPP model.
The study also faults KIPP for not serving more students who are still learning English or who have disabilities.
“The limited range of students that KIPP serves, its inability to serve all students who enter, and its dependence on local traditional public schools to receive and serve
the droves of students who leave, all speak loudly to the limitations of this model,” the report says.
Luis A. Huerta, an associate professor of public policy and education at Teachers College, Columbia University, praised the study for exploring indicators of KIPP’s operations other than student achievement, which, while important, doesn’t tell the whole story, he said.
“If we can start speaking about these more nuanced layers, and move beyond this discussion of student achievement, we tend to get a real picture,” he said. “Here we have schools receiving upwards to $6,000 or more than traditional schools, and that’s not even accounting for the fact they have fewer services than traditional schools, yet the gains they’ve shown in student achievement are quite modest.” Mr. Huerta is a faculty associate of the National Center for the Study of Privatization in Education at Teachers College, Columbia University, which had a hand in distributing the study but did not take part in the research.
The study came in for criticism from KIPP officials, as well as from two other researchers not involved in it. They questioned its methodology and said that while Mr. Miron is asking the right questions about KIPP schools, he hasn’t provided adequate evidence to answer them.
“We see this report as having significant shortcomings in the methodologies and reject the core conclusions the report is making,” said Steve Mancini, the public-affairs director for the San Francisco-based KIPP network, which was started in 1994.
The study by the Western Michigan researchers used the federal Common Core of Data as its primary source. The researchers were able to obtain data from 2005-06 to 2008-09 for 60 KIPP schools across the country. The KIPP schools were compared with averages for other, more-traditional schools in the same districts. Besides the 990 forms, the researchers drew financial data on KIPP schools from the same federal database, which had financial data for 25 of those schools.
Robin Lake, the associate director of the Center on Reinventing Public Education, at the University of Washington in Seattle, was one of the scholars who questioned the study led by Mr. Miron.
“It seems he’s trying to explain away the KIPP effect rather than explain it,” she said. “More work needs to be done to get real answers.”
“The main point to make is the kind of data they are looking at is quite different from the kind of data we’ve been looking at,” said Brian P. Gill, a senior fellow for the Princeton, N.J.-based Mathematica Policy Research and a co-author of a study of 22 KIPP middle schools released last June. That study was commissioned by KIPP.
Mr. Gill said that Mathematica based its conclusions, including a finding that attrition of students from KIPP schools is about the same as from neighboring regular public schools, on data from individual students, not on aggregate data sets, as Mr. Miron’s study has done.
The study led by Mr. Miron found that approximately 15 percent of students disappear each year from the KIPP grade cohorts, compared with 3 percent per year in each grade in the local traditional school districts. Mr. Miron said that finding doesn’t contradict Mathematica’s finding that attrition rates are comparable between KIPP schools and local district schools on average, because his research team compared only KIPP “districts”—the cluster of kipp schools in a particular district—and the rest of the schools in districts as a whole, not individual schools with schools.
Mr. Mancini, Ms. Lake, and Mr. Gill share the view that the comparison groups used in the Western Michigan study don’t provide reliable information about student attrition. It’s not appropriate, they contend, to make conclusions about attrition by comparing the proportion of students who leave a KIPP district with the proportion of students who leave the entire surrounding school district, which might have hundreds of schools.
“You want apples-to-apples comparisons. This is like apples to watermelons,” said Ms. Lake.
Mr. Miron said that the Mathematica approach to determining student attrition is “superior” to his. But his study explores an issue that he said Mathematica hadn’t addressed: How does the fact that KIPP schools tend not to replace students that leave, particularly in the upper grades, affect attrition?
“The low-performing students are leaving KIPP schools, but they are still in the public school sector,” Mr. Miron said.
Mr. Gill said Mr. Miron’s study doesn’t account for how grade retention, a hallmark of the KIPP model, may account for some of the shrinkage in cohorts of students moving from 6th to 8th grade.
The Western Michigan study doesn’t challenge KIPP’s positive student outcomes. It says that the nonprofit network’s claims that its schools improve students’ test scores at a faster rate than regular public schools are backed by “rigorous and well-documented studies,” such as Mathematica’s.
Mike Wright, who oversees KIPP’s network growth and sustainability, characterized the report’s findings on the financing of KIPP’s schools as misleading.
He focused on the finding that KIPP schools receive nearly $5,800 more per pupil from private donations than do their surrounding school districts. One problem, Mr. Wright said, is that the finding is based on a sample of 11 KIPP districts that isn’t representative of all KIPP schools. (Mr. Miron said he used those 11 districts because they were the only ones that reported public revenues on the 990 tax forms.)
Also, Mr. Wright said of the study’s authors, “they are including everything under the kitchen sink, whether starting a school from scratch or investing in facilities” in the figure for private per
pupil funding. He contends it’s a “misrepresentation” to imply that KIPP schools are overflowing with resources, when, unlike regular public schools, they are often left on their own to pay for buildings.
Mr. Wright contends that the average funding advantage from private sources for KIPP schools in comparison with their local school districts is closer to $2,500 per pupil.
Mr. Huerta, however, said Mr. Miron’s methodology is strong, even though there are “complications in trying to dig out some of this information.”
Vol. 30, Issue 27, Pages 1,24-25