A widening problem with the scoring of the state's $34-million student-testing program has forced a second delay in the release of the much-anticipated results, officials said Monday.
The state Department of Education now hopes to release the school- and district-level scores for the Stanford 9 tests of reading, writing, math and other subjects sometime next week, three weeks after state law says they were to have been made public.
The delay involves the possible miscalculation of scores for 44 of the state's districts with schools on year-round calendars, including those in Santa Ana and San Juan Capistrano.
The latest glitch from test publisher Harcourt Educational Measurement throws into question the scores of 500 students at Santa Ana Unified's Pio Pico Elementary School. A scoring error would have a negligible effect on overall marks in the district, but it erodes confidence in the test among educators and parents, Supt. Al Mijares said.
"I believe strongly in standardized testing and the process, but I have a hard time with what we are finding with our current test publisher," he said. "The impression I have now [from Harcourt] is this lackadaisical, laissez faire, 'Oh well, let's just round off' attitude. And we're over here waiting for precision."
Even as they announced the delay, officials with the state Department of Education and the testing contractor tried to shore up confidence in the program, the key barometer of student achievement.
"This is a roadblock that needs to be cleared, but the state's testing program is still incredibly valuable, and we mean to get this right and move on," said Doug Stone, the Education Department spokesman.
The most recent problem involves scores for year-round schools in the 44 districts.
The Stanford 9 data will provide the central measurement for the state's effort to hold educators accountable for student performance, a main component of Gov. Gray Davis' education reform plan.
Secretary of Education Gary Hart also expressed his continued support for the program, which tests students in grades 2 through 11, despite the questions that have arisen about the accuracy of the scores.
"The state of California is wedded to this, and it's important not to change horses once again in midstream, and we're not going to do that," Hart said.
Hart was a key force behind the creation of the state's last academic testing program, known as the California Learning Assessment System. Technical problems with that test the first year it was given, in 1993, fueled criticism that eventually led to its demise. The state was without any statewide testing program until last year, when the Stanford 9 was given for the first time.
Ed Slawski, a senior research scientist with Harcourt, said the data have been undergoing a careful review since the June discovery of a flaw affecting scores statewide. Harcourt has acknowledged that the source of that snafu was a decision to pool the scores of students fluent in English with those who were not, thus inflating the group scores.
When the correct scores are released, the data for students not fluent in English will be among the most closely watched. That's because they will provide the first glimpse of the impact of Proposition 227, a measure approved by voters a year ago to virtually end bilingual education.
While Harcourt was recalculating those scores, an additional error was uncovered by the Long Beach Unified School District. In that case, the performance of students attending 21 schools that operate on a year-round schedule was incorrectly compared to that of students who had received many more days of instruction.
Slawski said Monday that in the past few days the company rechecked the results of all districts with year-round schools.
Slawski said he remains confident that the data, which the company plans to deliver to the state on Friday, will be correct.
"Everybody is looking very, very carefully at these data," he said. But, he acknowledged, "part of the problem is that our credibility is in question."
That's for sure, said Capistrano Supt. James A. Fleming.
Tests for roughly 1,000 students at San Juan and Las Palmas elementary schools in his South County district might have been improperly scored.
"I can't believe it," he said. "This doesn't give us much confidence in the scores we're reporting. . . . We've seen not one or two, but three or four shoes dropped on the Stanford 9. We're just wondering what shoe will drop next."
The bright spot for the affected districts is that, if anything, scores for affected students should nudge upward, once they're recalculated. In Capistrano Unified, students may have been graded on 136 days of instruction, when they had been in class for only 124.
"Fewer days, bigger scores," Capistrano testing analyst Jeff Bristow said. "It's too early to say, but this could make a big difference. Or it could make no difference, if the scores weren't processed incorrectly. We were glad to see scores in those two schools go up already, which they did. If they go up even farther, great."
Harcourt has agreed to print correct reports for the districts that received incorrect scores for year-round schools at no additional cost. Some districts may already have sent inaccurate individual reports home to parents, however.
Officials with Harcourt, which is paid about $23 million annually to operate the testing effort, were due to begin meeting with members of the state Board of Education today to discuss the mounting problems.
The state has a $2.3-million performance bond it can hold over the company's head to make sure the problems are solved. Harcourt has a five-year contract with the state, but the board must decide next year whether to renew it.
Times staff writer Kate Folmar contributed to this report.