By Stephanie Simon
(Reuters) - Charter schools across the United States have improved in recent years, but on average, they still offer little advantage over traditional public education, according to a new study released on Tuesday.
Analyzing millions of test scores from 26 states, the Center for Research on Education Outcomes at Stanford University found that students at charters learned about as much math in a year as they would have at their local public school.
The study also found charters gave students a minimal boost on their reading scores, the equivalent of perhaps an extra eight days in the classroom.
Those results are a marked improvement over the charter sector's performance in the last big CREDO study, from 2009. That report found that charter students in 16 states lost a bit of ground in both reading and math against their peers in neighborhood schools.
CREDO director Margaret Raymond credited the gains in part to states' shutting down the worst charter schools. In addition, the performance of traditional public schools has slipped, making charters look better in comparison, she said.
CREDO's analysis, whose methodology has come under fire from charter fans and foes alike, found that 25 percent of charters outperformed nearby schools at teaching reading, while 19 percent did worse, and 56 percent were about the same. In math, 29 percent of charters did better, 31 percent did worse, and 40 percent were on par.
"I'm not giving up yet on the idea that charters will outperform district schools in the long run," said education analyst Michael Petrilli of the conservative think tank Thomas B. Fordham Institute.
"It's a mixed bag" for now, though, he added.
The nation's roughly 6,000 charter schools are funded with tax dollars but privately run by community groups, nonprofit organizations or for-profit companies. They are free and open to everyone, but they can cap enrollment and limit class sizes, unlike most neighborhood schools.
Charters often squeeze more instructional time into the year through longer days, mandatory weekend and summer sessions, or fewer electives such as music and art.
Charters enroll more than 2.3 million children, or about 4 percent of the nation's public school students.
VARIATIONS IN QUALITY
The CREDO study found vast differences in quality. Charter students in the District of Columbia, Rhode Island, Tennessee and Louisiana posted strong gains in reading and math, far outpacing peers in local schools. But charter students in Nevada, Pennsylvania, Oregon, Texas and Arizona lost significant ground.
Critics, including unionized teachers, often complain that charters skim the most motivated and disciplined students, leaving traditional public schools with those who are the toughest to reach.
The CREDO analysis found charters in the 26 states studied enrolled significantly fewer special-needs students and children not fluent in English than neighborhood schools around them, but about the same number of students in poverty.
The study found that poor black and Hispanic students gained the most from enrolling in charters.
The study attempted to match each charter student with a "virtual twin," a composite of up to seven children in nearby public schools who shared similar demographic traits and initial test scores. The researchers then tracked the charter students and their twins for several years to see who did better, and by how much, on subsequent tests.
That methodology has come under sharp criticism.
Skeptics of charter schools say the twin-matching is imprecise. For instance, a homeless child could be matched with student from a stable four-person household with annual income of $43,000, since both qualify for subsidized lunches and thus are deemed low-income.
The Center for Education Reform, a pro-charter group, also takes issue with the study. President Jeanne Allen said it was impossible to compare results across state lines.
CREDO is "improvising at best," she said, and probably "jumping to erroneous conclusions."
(Reporting by Stephanie Simon; Editing by Richard Valdmanis and Lisa Von Ahn)