One in five students at Fowler High School near Fresno, Calif., cannot speak English and comes from a family on welfare. Yet, by at least one measure, such students outscored their peers at more affluent schools on the California Learning Assessment System tests.

Anastasia Hendrix and Russell Clemings of The Fresno Bee compared the socioeconomic status of students in Fresno County schools with students' performance on the California Learning Assessment System (CLAS) tests.

Through their analysis, Clemings and Hendrix found that some schools did better than would be predicted given their economic backgrounds, which has been shown to have significant impact on academic performance.

Five of the schools in the Central Unified School District scored significantly lower than their statewide peers. Central High School in this district is located in a suburban area west of Fresno.

CLAS tests were given to over 1 million students in spring 1994. The test measured students' ability in reading, writing, and math. Students received a score for each subject area; the Bee averaged these scores for analysis, Clemings said.

"Generally, we published the data by ranking raw scores," Clemings said. But this time, the Bee included a comprehensive graph ranking the schools' performance against similar schools in California, measured by parent education levels and occupation.

Clemings said the data was readily available on the California Department of Education bulletin board system on the Internet. He converted the raw text files to .dbf files for manipulation.

Interpreting the data was relatively easy, Clemings said, because the state department of education calculated the socioeconomic scores for each school.

"We had no major problems other than logistical," he said. In order to create the graph, Clemings downloaded the numbers into FoxPro and graphed them using Microsoft Graph.

Clemings' and Hendrix's efforts did not go unnoticed by parents and teachers.

"They were generally pleased with where their school came out," Clemings said. "A lot of people called asking more questions about what the data meant."

Advice

Clemings had one piece of advice to offer reporters wanting to try a similar story: Use some other form of data to confirm that socioeconomic status is the most significant predictor of test scores. Clemings used Limited English Proficiency and percentage of minority enrollment to double-check his assumption.

"The idea that socioeconomic factors influence academic performance is not controversial, but it assures you don't go to the trouble to find data that it's the most important predictor and then learn it's something entirely different."

Russell Clemings can be reached at (209) 441-6371, or send e-mail to clemings@cris.com