Glitches to Delay Release of Some State Test Data

Stanford 9 test results will be released in two stages this summer because about two-thirds of the state’s school districts provided incomplete demographic data to the test publisher.

Overall results for schools, districts, counties and the state will be issued July 17, as dictated by a recent state law. But results by sub-group, which can reveal intriguing trends among children who are not yet fluent in English and other categories, will not be available until mid-August.

“It’s a little disconcerting that we have so much cleanup to do,” said Robert Rayborn, point man in Sacramento for test publisher Harcourt Educational Measurement. He added, however, that “it wasn’t a complete surprise,” given the amount of information required from districts and the high threshold of completeness--95% to 97%--demanded by the state’s contract with Harcourt.

Individual student reports that districts mail to parents will not be affected by this problem.

The state’s 3-year-old Stanford 9 program and a related accountability measure have been plagued by reporting problems.

Last summer, California officials delayed release of full Stanford 9 test scores after Harcourt provided incorrect group scores for children not yet fluent in English as well as for those who were fluent.

Then, in March, the California Department of Information Technology pulled data from its Internet site that ranked schools against others with similar ethnic and economic traits.

The information, part of the state’s new Academic Performance Index, was deemed faulty after many schools realized that they had provided incomplete or erroneous data about the portion of students qualifying under federal guidelines for free or reduced-price lunches. For the time being, the API is based strictly on Stanford 9 results.

As for the latest snafu, Rayborn said districts were hampered by the lack of a centralized database that could track students. California has been incrementally developing such a statewide student information system, but it is not slated to be fully up and running until 2005 at the earliest.

The “header sheets” for the Stanford 9, a standardized basic skills test given to students in grades 2 through 11, require schools or districts to provide a range of information.

Enrollment is broken down by gender, ethnicity, English fluency and economic status as measured by the federal lunch program, among other characteristics. Some data are readily available in computers; others must be calculated and penciled in.

Schools and districts vary widely in their handling of such data, said Doug Stone, a spokesman for the state Department of Education.

“I hate to beat up districts,” Rayborn said. “This is a very difficult task that no other state has been able to pull off without a centralized database.”

Still, in a memo to be distributed to district and county superintendents, the Education Department admonishes them to work with Harcourt to provide missing data as quickly as possible.

Of the more than 1,000 districts in California, about 165 are still having data corrected, Rayborn said. But since some of those are among the largest in the state, he added, that means about one-fourth of the 4 million test-takers are affected.

Districts will have to pay $1.25 to Harcourt for each test form that needs to be corrected.