NAEP Reanalysis Finds Lag in Charter School Scores
A federal reanalysis of 2003 test-score data from the National Assessment for Educational Progress finds that charter schools trailed regular public schools that year in student achievement in both reading and mathematics.
The average reading score for the 150 charter schools examined was 5 percentage points lower than in a far larger pool of more than 6,700 regular public schools, according to the report released today by the National Center for Education Statistics, an arm of the U.S. Department of Education. In mathematics, the charters scored 6 percentage points lower. The report used data from the 2003 administration of the 4th grade NAEP, which included a special oversampling of charter schools to allow for comparisons between regular and charter public schools.
Many researchers and charter advocates urge great caution in using data from NAEP, the congressionally required testing program known as the nation’s report card, to examine charter schools. For one, they note that the data represent a snapshot at one point in time, with no consideration for students’ prior academic achievement.
Using a statistical technique called hierarchical linear modeling, the new report sought to go further than a controversial 2004 report from NCES by taking into account multiple student and school characteristics simultaneously. This approach, also used in a recent analysis comparing public and private schools, has faced considerable controversy, with critics questioning the reliability of some of the variables used.
After making those adjustments, the differences in average school scores remained lower for charters in math and reading, but the differences were reduced by about 1 percentage point.
A key difference between the new report and the 2004 NCES report is that in this case the focus is on the average scores of schools, while the earlier analysis focused on student-level scores, said Henry I. Braun, a researcher at the Educational Testing Service who helped conduct the new analysis.
Mr. Braun said that overall, the findings of the new report are relatively consistent with the earlier NCES study.
“They’re actually numerically very close,” he said of the two studies’ results.
“The statistical significance is more extreme here,” he said, “but in terms of the sort of policy results, there’s not much change after doing this adjustment.”
Vol. 26, Issue 01