Test Firm Admits it Bungled

Stanford 9 scoring foul-ups may cost company millions

SACRAMENTO — The publisher of California’s problem-plagued achievement exam and its subcontractor conceded Thursday that they lacked proper quality controls and adequate staffing and had failed to train employees – admissions that could result in financial penalties of more than $ 10 million.

For the third time in as many weeks, state officials revealed a serious problem with test results for the Stanford 9: nearly 10 percent of student scores will be missing when scores are broken down by English fluency because districts failed to report children’s status.

Angry state Board of Education members listened to explanations after a series of foul-ups caused serious delays and questions about the credibility of the exam that was intended to serve as the linchpin for education reforms.

”So much depends on this test we have to try to make it work. We owe it to taxpayers and kids,” said state board member Yvonne Larsen. ”It’s not up to us to extract blood from Harcourt, but I realize there needs to be an equitable settlement. It’s the right thing to do.”

Representatives for publisher Harcourt Educational Measurement said they would hire the consulting firm Deloitte & Touche to audit their procedures, and a new management team would be created to make changes. Officials from its subcontractor, National Computer Systems, said they would be adding staff and training employees.

”We apologize for the logistical difficulties and inaccuracies,” said Jack Dilworth, president of Harcourt, the testing firm’s parent company. He added they were doing everything they could to ”do it right” in the future.

The state board will vote on penalties at its next meeting, Aug. 2. At stake is a contractual performance penalty of $ 1.1 million and a performance bond of $ 2.2 million. In addition, each of the state’s 1,055 school districts have individual contracts with Harcourt and could vote to withhold a third of their costs – a total of $ 7.3 million.

”I think the board will impose some sort of penalty, especially after hearing things like the companies didn’t have the proper staff or training,” Delaine Eastin, the state superintendent for public instruction, said in an interview.

Harcourt officials privately acknowledge that they expect some financial penalty, but sources said the board is leery of being too hard on the publisher. Establishing a comparison base-line is key to Gov. Gray Davis’ accountability reforms, and backing out would delay that.

Still, board members and the superintendent demanded that Harcourt fix its problems and stressed that no more problems were acceptable. Although the test is used by more students across the country than any other, Harcourt and NCS had serious problems in their efforts to distribute and calculate scores for the 4.2 million students in California who took the exam.

Revelations of problems for the Stanford 9 continued Thursday when Department of Education officials announced that 10 percent of student scores would be missing when results were broken down by English fluency.

This time the problems stemmed from districts that did not indicate whether students were fluent or not. Harcourt decided to throw those numbers out, against the wishes of the Department of Education staff.

Board members, as well as their legal counsel, were upset by the disclosure. Not only did Harcourt and the state staff fail to let board members decide how to handle the crucial policy issue, but more questions are now inevitable about test results.

”This was quite a surprise to me. I’m terribly concerned,” board member Monica Lozano said in an interview. ”One of the most important indicators we are going to be looking at is language proficiency and now, if 10 percent of the students aren’t classified, it could skew the entire measurement.”

The breakdown by fluency will be intensely scrutinized because of the state’s dismantling of its bilingual education program through Proposition 227.

Board members said they are anxious to learn which districts failed to report fluency. Some worried that districts might have intentionally decided not to report the information.

Department of Education officials said it would be typical for 2 percent to 3 percent of the fluency information to be blank. Eastin told the board that they were surprised by such a high number.

”We are continuing to inspect and examine this with districts,” Eastin said.

Eastin, the board and Harcourt support pending legislation that would require Harcourt to contract with the state instead of individual districts.

State officials agree that it is crucial to recapture the public’s faith in the exam results after a series of debacles:

First, deliveries of materials were late to more than 100 schools. In June, a Harcourt programming error resulted in the misclassification of 250,000 students. Students now fluent in English were mistakenly put into the nonfluent category, inflating the scores of children with limited English proficiency. As a result, scores by county, district and school were not posted on the Internet as promised.

Corrected information was supposed to be posted Thursday, but earlier this week another error was found. Scores for 190,000 students in 44 year-round districts had not been calculated correctly, and once again, posting of scores was delayed.

Department of Education officials also have discovered other technical deficiencies, including incomplete reporting for some math scores, misstatements of student enrollment and scores of zero that were unreported.

Kevin Gordon, assistant executive director of the California School Boards Association, said school districts throughout the state shared the state panel’s anger. He asked the state to give advice to districts on what their rights are if they decide to withhold payment from Harcourt.

On Wednesday, Assemblyman Scott Wildman, D-Glendale, chairman of the state’s Joint Legislative Audit Committee, asked the state’s auditor to conduct a full financial and performance probe on the exam.



Comments are closed.