Computer 'Glitch' Forces Recalculation of NAEP Scores
The Department of Education has gotten some unwelcome news about "the nation's report card."
Department officials found out last month that a computer program made mistakes in scoring results from both the 1992 and 1994 versions of the National Assessment of Educational Progress.
The mistakes, which affect the 1992 assessments in mathematics and reading and the 1994 reading and world-geography tests, had only a negligible impact on final scores, an official at the department's National Center for Education Statistics said last week.
Nevertheless, officials are recalculating and reissuing scores that already have been made public.
Gary Phillips, the associate commissioner of the education-assessment division at the nces, said there is no need to change any interpretation of the data as a result of what he called a computer glitch.
When scores were recalculated correctly, average scores for a grade level typically went up or down by just one point. On the world-geography test, for example, the largest change by far was that some revised 12th-grade scores went up four points on a scale of 500 points.
Naep, known as the nation's report card, is the only national, ongoing measure of U.S. students' academic achievement. It has been administered since 1969 in a variety of subjects to a sampling of students.
The error, discovered by chance in the process of preparing data for the statistics center, was made by a computer program used by the Educational Testing Service in Princeton, N.J., the company under contract to conduct the tests. Ets officials declined to comment last week, referring questions to Mr. Phillips.
The error occurred in the way the computer program scored test items that required a student to construct a response or write an essay. Some omitted answers were scored simply as missing instead of as wrong. Multiple-choice test items were not affected.
The discovery of the errors meant that the NAEP results in world geography had to be recalculated before they could be released.
They had been scheduled to be released this month, but as of last week they were expected to be released next month, along with the results of the 1994 NAEP test in U.S. history, said Larry Feinberg, an assistant director of the National Assessment Governing Board, which oversees naep. The computer-scoring glitch did not affect the history results and was not to blame for the delay, officials said.
Results from the 1994 reading test were released this year in the "1994 First Look Report," which will be revised and re-released in a few weeks, officials said.
The more comprehensive reports on national- and state-level reading performance from the 1994 NAEP are being completed and will be delayed, officials said. The revised 1992 and 1994 reading results may be ready by Oct. 20, Mr. Phillips said.
The Math Factor
The full impact on the 1992 NAEP in math is not yet known, but there were relatively few open-response items on the test.
The statistics center plans to include the revised numbers for the 1992 math assessment in its 1996 NAEP reports, Mr. Phillips said.
The mistakes also have affected the National Education Goals Panel, which has been preparing its annual report on the nation's progress in meeting or approaching the eight national education goals.
The report, which uses NAEP data extensively, will include footnotes on those pages with data on reading achievement, said Bill Noxonsic?, a spokesman for the goals panel.
The glitch has not delayed the planned Nov. 9 release of the report, he said.
Late last week, officials at the Education Department's statistics center were looking into another possible mistake--not related to the ets glitch--that could affect NAEP results.
Vol. 15, Issue 04