A computer error caused “serious inaccuracies” in scores on the state’s Stanford 9 student achievement exam and threw in doubt evidence that non-English speaking students made dramatic gains, state education officials said Tuesday night.
Results have been trickling out from school districts around the state for several days and Los Angeles, Burbank and Newhall school districts released results Tuesday showing spectacular improvement for students with limited English who were in the first year of programs revised after Proposition 227 abolished bilingual education.
While Los Angeles and other districts showed slight improvement in overall student performance by all students, test results for students with limited English skills were shown to have more than doubled in some grades and subjects.
Those scores were definitely wrong, officials said. Whether overall results for all students tested were inaccurate was unclear and the state Department of education plans to go ahead with releasing statewide results today with the warning they might be revised.
Results were miscalculated when computer programmers for the test’s maker incorrectly categorized results for fluent and nonfluent students known as “Limited English Proficient” or LEP pupils, said Doug Stone, spokesman for the Department of Education.
“There are serious inaccuracies in the LEP and non-LEP scores,” Stone said. “We are extremely disappointed and will attempt to move heaven and earth to get the data as soon as possible. The public has been waiting for these results.”
Stone said results show that the scores for the nonfluent students are inflated because of the computer error, and the scores for fluent students are deflated.
Officials from the test’s publisher, Harcourt Education Measurement, said an unexplained computer error caused the problem. However, they are not convinced that the overall scores are incorrect.
Ed Slawski, a senior research scientist for Harcourt Educational Measurement said the spikes in the LEP student scores could have come from a year of English-immersion instruction, as ordered by the passage of Proposition 227.
“They’ve had a full year of English — the scores look right to us,” Slawski said.
However, he said the company cannot guarantee that the scores are correct.
“We’re sorry to say we can’t say that with any certainty,” Slawski said.
Stone said Harcourt told state officials a different story.
“They confirmed that there was an accuracy problem with the LEP and non-LEP scores, by school, by district, by county and by state,” Stone said.
Slawski said that the programming error occurred when test results were broken down by the degree of fluency. Some students who had been “redesignated” or deemed fluent and moved into traditional classes were instead grouped with nonfluent students, he said.
Slawski said that until Harcourt does a complete investigation, they will not know the extent of the problem, and if all scores are incorrect. He noted that results sent to each district might be correct, and problems might have occurred only when the company compiled statewide information.
Harcourt discovered the problem on its own, Slawski said, when it checked the total numbers of students taking the test and the numbers of students in each fluency category. “The numbers didn’t add up,” Slawski said.
Stone said individual districts noticed flawed information when Harcourt sent it to them. Then, when the state received complete data at 11 a.m. Tuesday, they saw the results were flawed.
Want it fixed
“I want to strongly refute that Harcourt brought this to our attention,” Stone said. “This might have been an honest mistake, but it was a real mistake and we expect Harcourt to fix it and fix it expeditiously.”
The company has until July 15 to recalculate the data, Stone said.
The controversy clouded the results that earlier in the day had given strong indication that ending bilingual education paid dividends much more quickly than expected.
“People thought that we were going to take a big hit because we were going to test the (limited English proficiency) kids. And that didn’t happen,” said Los Angeles schools Superintendent Ruben Zacarias.
Los Angeles Board of Education member David Tokofsky said an error by Harcourt alone could not account for the jump in scores for English learners.
Since the test results show relatively slow improvement among students fluent in English, he said, it is unlikely that accidentally counting some of those students as children with limited English proficiency could lead to such a sharp spike in scores.
“I don’t think the bump is bogus,” he said. “The bump will still be there, just maybe not as large.”
Results from Burbank schools for overall performance indicated students are performing below the national average in a number of subjects and grades but improved in nearly all areas overall.
“I’m very pleased with the evenness of the growth,” said Caroline Brumm, district coordinator for student and program assessment and evaluation for the Burbank school district.
Brumm said the district’s strategy over the past year has been to work with teachers on targeting weak areas in past tests, hiring early literacy specialists to train teachers, publicizing the test with parents and holding meetings with them, and spending money to teach teachers strategies for helping students with limited English skills.
In the Santa Clarita Valley, Saugus Union School District, which has 11 elementary campuses, reported improved overall results.
Class size reduction in grades 2 and 3 helped to improve results, said Joan Lucid, assistant superintendent of curriculum and instruction.
Saugus teachers were trained to simulate testing as practice, teaching pupils techniques to answer multiple-choice questions and working from a test booklet.
“Test taking is a skill,” Lucid said.
In the seven-campus Newhall School District nearly all the scores increased since 1998. Second-grade spelling and third-grade language results made the largest gains. Sixth-graders, however, fell one percentile point in reading, language and spelling, and gained one point in language.
Superintendent Marc Winger called those shifts insignificant.
The Newhall district has the highest percentage in the Santa Clarita Valley of non-English-proficient youngsters, which Winger said brought averages down because the test was administered in English.
Oak Park Unified School District Superintendent Marilyn Lippiatt called the district’s growth phenomenal, and said she was proud of the students’ scores that were for the most part higher than last year’s.
“We’re continually honing our programs,” Lippiatt said Tuesday. “The teachers worked hard this year to make sure students were taught materials that they knew would be assessed. They took last year’s scores, figured out where we needed improvement and built on our programs.”