Botched scoring by a private testing company prompted California officials to delay release of full Stanford 9 test scores Wednesday, but partial results show that California students made only modest gains in reading and stronger gains in mathematics in most grades in 1999.
State officials said they believe the numbers they released for the state as a whole were unaffected in the blunder by Harcourt Educational Measurement, the contractor for the $34-million program. They emphasized that individual student reports sent to parents are accurate.
But they blocked the release of data for individual counties, districts and schools, saying group scores reported for students not fluent in English as well as for those fluent in English are inaccurate. The questionable data could not be easily separated from the rest, officials said.
The delay was particularly galling to many because it will hinder the state’s effort to assess the effects of Proposition 227, the voter-approved measure that last year effectively banned most instruction in languages other than English.
Many school districts around the state already had received–and publicized– the scores for students with limited English skills, which state and company officials now say could be incorrect by as much as 10 percentile points. Harcourt vowed to take responsibility for the error and to have corrected numbers by July 15.
“The problem is certainly very, very serious, and we are taking it very seriously and will do our best to rectify the situation,” said Joanne M. Lenke, president of Harcourt.
Harcourt, whose contracts with California school districts are worth $22 million, had posted a $1.2-million bond as a warranty of its work. Officials said Wednesday that it remains to be seen whether the company will be forced to give up any or all of that amount.
State Supt. of Public Instruction Delaine Eastin, who was traveling to a family reunion, said the trend in test scores was positive for the state. She conducted a news conference by phone from Williamsburg, Va.
“Overall and on balance . . . our scores are pretty much up, and . . . it demonstrates that California is moving in the right direction,” Eastin said.
Gov. Gray Davis, who has said he will not run for reelection in 2001 unless test scores rise, was pleased. Still, he said, “we have a very long way to go.”
The gains in reading were strongest in the second and third grades, which were part of the state’s $4-billion-plus program to reduce class sizes to no more than 20 students. Those grades have also been the focus in recent years of a sweeping effort to revamp teaching methods, textbooks and teacher training to put greater stress on the fundamentals of phonics.
The reading gains disappeared by high school, however. Students in ninth and 10th grades made no progress compared with those grade levels last year, and 11th-graders even dropped a point.
As was the pattern last year, ninth-grade reading scores were sharply lower than the eighth grade. That trend, called the “ninth-grade dip,” prompted a contention from the state that the scores in those grades were affected by a technical problem with the test or by statistical procedures used to establish the national median.
“This year’s ninth-graders were last year’s eighth-graders,” said Gerry Shelton, who directs the state’s testing programs. He added that it does not make sense that these students would “go from being high achievers to being dumb.”
Despite the improvement in reading results, fewer than half of California students in all grades scored at the national median–the point that half of students score above and half below.
Student performance was substantially stronger in math. Fifty percent of the state’s second- and sixth-graders scored above the national median in that subject.
The state’s return to basic skills instruction was perhaps best reflected in the scores for spelling, which educators acknowledge had not been widely taught in California until the past few years. The percentage of students at or above the national median rose 6 points in both the second and third grades.
Former Gov. Pete Wilson, who championed both the return to basic skills instruction and smaller class sizes during his term that ended in January, said the new scores show those policies are “paying real dividends” across the board.
“It’s not surprising that the kids who have had these advantages are showing the biggest improvement,” Wilson said. “When they get to the seventh and eighth grade, they are going to significantly outperform the seventh- and eighth- graders of today.”
It was Wilson who, over the objections of Eastin and many educators across the state, insisted that the state launch a statewide testing system so that parents could make comparisons between schools and school districts.
Many had urged Wilson to wait to develop a testing system until the state had adopted new academic standards. This year’s test also included questions based on those standards, and the results show that, as predicted, California students have a long way to go.
In language arts, for example, students generally were able to answer about half the standards-based questions correctly. In math, first- and second- graders got better than half the questions right, but 11th-graders were able to correctly answer only a third.
On Wednesday, Eastin charged that some of those questions were far too difficult or were otherwise flawed and would be thrown out next year. But Lenke said that test questions are routinely replaced to preserve security and that the decision to change questions reflects a desire by the state for fresh material.
That was just one of many areas in which officials from Harcourt and the state Education Department differed sharply.
Fingers were pointed on both sides as state education officials and Harcourt executives attempted in separate conference calls with reporters to explain what had gone wrong and why.
“Actually, the department notified us” Tuesday, Lenke said. “We did not know about it” until then.
Individual school districts–including Anaheim–first called the numbers into question. District officials correctly surmised that students who had been redesignated as being fluent in English were mistakenly lumped in with “limited English-proficient” students.
Lenke blamed the problem on a judgment error. She said it might have been caught if the company had more than two days to analyze data from 4.3 million test forms at company headquarters in San Antonio before Wednesday’s deadline for posting results on the Internet.
The bungled scoring is a further embarrassment for Harcourt, which has been bedeviled by delays in distributing materials and scoring tests.
However the inaccuracies occurred, on Wednesday they were undermining the credibility of the state test, which will soon be the key factor in the state’s plans to judge whether schools are poor performers.
“Our public loses trust in the school districts, and school personnel lose faith in the state Department of Education and the test itself,” said Ken Noonan, superintendent of the Oceanside Unified School District. That district was among those that released scores showing dramatic gains by students with limited English skills.
The scoring flubs also angered the superintendent of the Santa Ana Unified School District, where more than half the 54,000 students are not fluent in English.
“This is not philanthropy. It’s for-profit testing,” said Supt. Al Mijares. “There is no excuse for incompetency or malpractice.”
Mijares said he expected his school board to seek a legal opinion on whether the district could opt out of using the Stanford 9 in the future.
Times staff writers Louis Sahagun in Los Angeles and Kate Folmar in Orange County contributed to this story.