Small Gains on Stanford 9 Scores Cut Across All Levels of Language Ability

Lack of English fluency played a greater role than poverty in how well California’s students performed on this year’s standardized tests, but neither was a significant factor in how much their scores improved over the prior year, a Times computer analysis found.

Overall, the modest gains made by California’s public school students on the Stanford 9 test showed a remarkable similarity, inching up at a frustratingly slow pace but with all groups rising equally, the analysis shows.

Whether large or small, wealthy or poor, predominantly English-proficient or dominated by students who were not fluent, school districts across the state fell into a pattern of slight gains in reading and more robust gains in math.

Even though limited ability to speak English presented the greatest impediment to success on the standardized tests, students who were not proficient in English gained just as much in English and math as those who had mastered English or were native speakers.

An analysis of how much school districts’ scores changed from 1998 to 1999 showed that, on average, reading scores were up 1.8 percentile points for those with limited proficiency in English and 1.9 percentile points for the fully proficient. In math, the gains were 3.9 percentile points for both groups.

Although there were no comparative scores available specifically for low-income students, the analysis showed that districts with higher percentages of the poor gained as much, on average, as those with wealthier populations.

The pattern that emerges highlights the disappointment felt by those who had hoped for some spectacular sign that the state’s intense focus on educational reform was already paying off. But it also carries the consolation that no group is yet being left behind in what can, at best, be viewed as a long journey out of educational mediocrity.

Activists had hoped this year’s testing would show benefits from major changes such as Proposition 227, the state initiative that eliminated most bilingual education. Others had hoped to see a major impact from class size reduction in the lower grades. The data failed to show a major impact from either change.

“It’s really tough to draw any major conclusions from what we’ve seen, other than it isn’t bad news, and we’re moving in a good direction,” said Ron Dietel, director of communications for UCLA’s Center for Research on Evaluation, Standards and Student Testing. “Things do seem to be a bit better. We have a long way to go.”

To conduct its analysis, The Times created a composite score intended to provide an overall picture of the success or failure of each district in the state. The composite includes all grade levels averaged across the three main tests: reading, math and language.

Among the findings of the analysis:

* Language poses a far greater barrier to performance than poverty. Although the negative effects of both are well-documented, the state’s roughly 1.5 million students whose families were poor enough to be eligible for federal lunch subsidies scored better than its 900,000 who were limited in English ability. The scores for low-income students averaged in the 35th percentile, compared with the 24th percentile for those with limited English. (Some part of that difference results from the fact that students who start off with limited English and then master the language are reclassified as English-proficient, thus skimming off many of those with the highest test scores each year.)

* But districts with higher percentages of limited English-proficient or disadvantaged students improved just as much on average

as those with lower percentages of those groups.

* District size also had no demonstrable effect on how well students scored or how much they improved from 1998 to 1999. The giant Los Angeles Unified School District, at the 30th percentile, was 15 points below the state average, but its gain of 2 percentile points over last year nearly kept pace with the average district gain of 2.7 points statewide. Overall, smaller districts were slightly more likely to gain, but the difference was too small to be statistically significant.

In a general picture of sameness, a handful of districts stood out, either for showing signs of beating the odds of poverty and language difficulty or, at the other extreme, for sinking even deeper into trouble.

One of the brighter cases was the Lynwood Unified School District, where scores rose sharply. Overall, the district’s score remained low--the 26th percentile--but that was an improvement of 4 percentile points. The district’s low-income students actually performed better than its other students. Students with limited English, meanwhile, gained more this year than the English-proficient, and are now just a point behind them.

Barbara Johnson, director of curriculum and instruction, said the Los Angeles County district has made a concerted effort to establish a reading intervention program for at-risk students. Many teachers were specially trained in a program to help kids with reading and language difficulties.

“We had to do something in general, but specifically for the underachieving students,” said James Williams, the district’s director for technology, research and evaluation.

Researchers at UCLA are preparing to spend several months combing the data for some insight into the effects of educational reforms and how much individual classroom practices can account for cases such as Lynwood’s.

The type of English immersion used in individual classrooms and the number of students in a school who continued in bilingual education are some factors that must be considered, said Barry Gribbons, project director for UCLA’s research center.

Further hindering analysis of the scores is variation among grade levels, such as the persistent dip in grades eight and nine, Gribbons said. If the phenomenon reflects a problem with the way the test is structured, rather than actual student achievement, year-to-year conclusions would be impaired.

“It’s not a very interesting conclusion to say that the findings weren’t conclusive, but I’m afraid that’s kind of where we’re headed,” Gribbons said.

In the quest for answers, clues may be found in some districts that distinguished themselves by making strong gains despite large numbers of students who are disadvantaged or who speak limited English.

Such better-than-average gains can indicate that schools are succeeding in teaching students more of what is being tested. Smaller gains of 1 or 2 points are typical of what happens as students become more familiar with a test.

In far northern Butte County, the tiny Oroville City Elementary School District has a big population of Hmong and Mien students whose refugee families fled Laos, Thailand, Vietnam and Cambodia. Hmong is the primary language other than English.

Oroville jumped 5 points on this year’s Stanford 9, reaching the 39th percentile.

Supt. Donald L. Remley said a key focus has been on building the ranks of teachers specially trained in bilingual education and teaching English as a second language. Of the district’s 193 teachers, 45% have such specialized credentials.

The district’s Wyandotte Avenue School, which runs from kindergarten through sixth grade, used federal funds to start a Mosaic program to help acculturate bilingual students. The district has also adopted “Into English,” a program for teaching English as a second language, in all seven of its schools.

“We put a lot of time into working with students who are limited English-proficient,” Remley said.

*

Richard O’Reilly, Times director of computer-assisted reporting, and data analyst Sandra Poindexter contributed to this story.

(BEGIN TEXT OF INFOBOX / INFOGRAPHIC)

How the Scores Were Analyzed

The Times created schoolwide and districtwide percentile rankings from the 1998 and 1999 Stanford 9 test results published by the California Department of Education.

The state’s average percentile ranks for schools and districts for reading, math and language tests in each grade were converted to numbers that can be combined across grades and tests. The results were weighted by the numbers of students tested and converted to an overall average percentile rank.

The work was performed by Richard O’Reilly, Times director of computer-assisted reporting, and Sandra Poindexter, data analyst. Times staff writer Doug Smith subjected the Times-created scores to a series of statistical analyses.

* L.A. COUNTY NUMBERS

A statistical snapshot of how school districts performed on statewide standardized test. B2