Charter Schools Study Criticized for Flaws

Published September 1, 2009

Charter schools nationwide report an array of achievement levels, according to a new national assessment, and some of that achievement is less than stellar. While charter proponents agree lawmakers should strengthen accountability to improve quality, they are also pointing out the study’s flaws.

The study, released in June by Stanford University’s Center for Research on Education Outcomes (CREDO), claimed most charter students do not score higher than they supposedly would have in conventional public schools, with less than 20 percent doing so.

‘Virtual Twins’

CREDO looked at math scores of charter students from 15 states and Washington, DC and compared them to those of conventional public school “virtual twins” the researchers created to represent the scores the students supposedly would have had if they had stayed in conventional public schools.

“We looked at the possibility of what [the charter student’s] score would have been if he had stayed there,” coauthor Devora Davis said.

The study’s authors claimed 17 percent of charter schools educate students better than their government-run counterparts. They also said 46 percent of students in charters don’t do any better than their counterparts, and 37 percent do worse.

Flawed Comparison

Lead author Margaret Raymond declared, “The study shows that we’ve got a 2 to 1 margin of bad charters to good charters.”

But other researchers and the National Alliance for Public Charter Schools dispute that claim. The alliance noted CREDO’s researchers failed to create a legitimate control group for comparison, thus drawing unnecessarily pessimistic conclusions about the state of charter schools.

Bob Holland, a senior fellow for education policy at The Heartland Institute, which publishes School Reform News, said, “Note that [Raymond] deemed as ‘bad’ all those charters [the study admitted were] doing just as well as regular public schools. One could accurately express the findings this way: 63 percent of charter students scored as well as or better than their [imaginary] public-school peers in math.

“What’s more, the study did not actually compare charter students to real-life peers. Instead, CREDO fiddled with demographic data—ethnicity, age, socioeconomic status, and such—to create virtual public-school ‘twins’ of the charter kids for comparison. That falls far short of the gold standard for researching education results.”

Shortly after the study was released, the alliance published a model charter school law to guide future legislation in focusing on strengthening charter school accountability. Last year, more than 1.4 million children in 40 states attended more than 4,700 charter schools, while 365,000 students waited to enroll.

Whitney Stewart ([email protected]) writes from Minnesota.

For more information …

“Multiple Choice: Charter School Performance in 16 States,” Center for Research on Education Outcomes, Stanford University, June 2009:

“CREDO Reconsidered,” by Nelson Smith, National Alliance of Public Charter Schools, June 19, 2009: