Federal auditors last week dramatically revised their estimates of how many students with disabilities were excluded from taking national reading tests in 2002.
The U.S. Government Accountability Office, in a July report, said that states on average excused 5 percent of students with disabilities from taking the 2002 National Assessment of Educational Progress tests in reading. But last week, in a letter posted on its Web site, the GAO said the average was closer to 40 percent.
is available from the .
Read also the GAO鈥檚 .
The change stems from crossed signals between the GAO, which is the auditing arm of Congress, and the U.S. Department of Education, which reports the national exam results.
The issue is potentially important, according to experts, because the congressionally mandated tests are intended to be nationally representative, which means they should include students with disabilities as much as possible.
By the GAO鈥檚 new methods, for instance, an average of 35 percent of students with disabilities sat out the 2005 reading tests.
According to Marnie S. Shaul, who supervised the GAO study, the confusion arose when investigators misunderstood the original numbers as representing the percentages of students in the total test-taking population who were excused from the test because of their disabilities, rather than the percentages of students with disabilities sitting out the tests.
The National Assessment Governing Board, which sets policy for NAEP, has been trying to encourage states to be more inclusive. Rates of exclusion for students with disabilities have crept down, as a result, from 57 percent on the 1992 4th grade reading tests to 40 percent in 2002, and then to 35 percent this year.
The Education Department does not normally highlight those numbers in its NAEP reports to the general public, but instead gives the proportions of overall test-takers who sat out the assessment because of their disabilities.
Education Department officials did not bring the GAO error to federal investigators鈥� attention until two months after the July report was published, according to Ms. Shaul, who is the auditing agency鈥檚 director of education, workforce, and income-security issues.
She said GAO mailed the department draft copies of the report three weeks before it was published.
Department officials, for their part, said they began alerting low-level agency officials to the error as soon as they picked up on it.
鈥淚n this case, we missed the error, and it should鈥檝e been caught,鈥� said Grover J. 鈥淩uss鈥� Whitehurst, the director of the department鈥檚 Institute of Education Sciences.
鈥楧isappointingly High鈥�
The congressional agency posted the revised numbers in an Oct. 28 letter to U.S. Sen. Edward M. Kennedy, the Massachusetts Democrat who had requested the original report.
School principals are allowed to exclude students with severe disabilities from the federal tests and to provide some accommodations, such as longer testing times, so that students with less severe learning problems can still take part. Exclusion rates vary widely among states. (鈥淪tates Vary on Students Excluded From NAEP Tests,鈥� Nov. 2, 2005.)
An expert on educating children with disabilities said the 40 percent exclusion number was 鈥渄isappointingly high.鈥�
鈥淚t cries out for NAEP to pay attention to this and to really figure out a way to address the issue,鈥� said Martha L. Thurlow, the director of the National Center on Educational Outcomes, based at the University of Minnesota-Twin Cities. 鈥淭o get an accurate picture of what鈥檚 going on in reading and math achievement, we really need to include all our students.鈥�
