Alexandria school Superintendent Robert Peebles is often described as a soft-spoken man. So on the rare occasions he raises his voice, people listen.

And last week, when newly released scores from the state-administered testing program showed Alexandria students far below their Northern Virginia peers for the 10th year in a row, Peebles felt compelled to raise his voice.

"This a subject I have some emotion about," Peebles told school board members. "I'm not saying this defensively, but we are doing fine. We can do better. We are analyzing the test results."

In the past decade, many Alexandrians may have heard the same comments. But this time, Peebles says, the school system is working on a program that will match what goes on in the classroom with areas where students need the most help. And according to several national educators, Peebles is on the right track.

The heart of Peebles' plan consists of two interdependent approaches: Realigning the elementary program to focus on helping students think through complicated problems, and instituting more thorough teacher evaluations to make sure what schools say they are teaching is being taught.

"I don't think the information derived from norm tests has been adequately applied to instructional programs," he said, stopping short of blaming his predecessors. "I want to look ahead to what we can do, not so much at what we may not have done in the past."

Results from the Scientific Research Associates (SRA) achievement tests showed that last year fourth, eighth and 11th grades in Alexandria schools ranked between the 44th and 53rd percentiles in the nation in the six areas covered by the tests: reading, math, language arts, science, social sciences and reasoning ability. By comparison, students in the same grades in Falls Church and Arlington and Fairfax counties ranked between the 60th and 85th percentiles.

These results are not unusual. In the past decade, Alexandria students have dipped below or barely topped national norms on the SRA tests. When Peebles came to Alexandria two years ago, one of his priorities was to take a hard look at the school system's academic programs.

One result is an evolving program in which Peebles and his staff use the SRA scores to pinpoint areas of weakness, and then tailor the curriculum to emphasize improvement of those skills. Among the areas administrators are focusing on are reading, reading retention, communication, basic math skills and problem solving.

Donald Dearborn, assistant superintendent for elementary schools, explained the process. "We look at the tests and say we find out we are lacking in an area such as multiple-step problem solving," he said. Administrators then meet with teachers to restructure the daily schedule, he said, so more time is spent teaching those skills. In addition, specific teaching approaches that have proved successful are outlined and they are implemented by teachers.

Both Peebles and Dearborn concede that even the best program cannot succeed without effective teachers. To that end, the school system plans to intensify teacher evaluations this year to ensure that teachers are spending a good part of the school day helping students in the areas where improvement is needed.

"We are starting teacher and program audits to see, is the teacher teaching the skills we need to have stressed? Is she doing it effectively? Is she sticking with the program we've worked out?" said Dearborn. "The goal of these audits is to improve the instruction by improving the teacher."

Dearborn said teachers have been evaluated in the past, but that administrators will now sit in on classes to see which subjects the teacher is teaching effectively and which subjects he or she needs to spend more time on.

Peebles said he plans to concentrate on the first through third grades because youngsters who learn at an early age to understand what they read, to communicate and solve problems are more confident as they grow older. He also said this emphasis eventually should prevent older students from falling into what he calls the remedial cycle, in which they are dragged from higher-level courses to be tutored in subjects they should have learned earlier.

National educators said the changes in Alexandria reflect a trend in several urban schools systems in the nation, including cities such as Baltimore and Birmingham. Traditionally, educators say, students in urban areas rarely do as well on standardized tests as those in suburban areas. Recently, however, several urban cities have broken that trend, primarily by setting more specific goals for students and teachers.

"In a lot of urban centers, there is a move toward more systematic curriculums," William Spady, an official at the American Association for School Administrators in Arlington.

Spady said he has found teachers and students benefit from clearly defined curriculums that spell out the skills and goals students are expected to master.

"Teachers must then instruct directly in relation to those objectives," Spady said. "It is very precise, almost simple, but it works."

In Baltimore, which began a program similar to Alexandria's in 1977, school officials report steady improvement in two basic areas, reading and math. In 1977, the average reading level for a Baltimore student was 21.6 months below grade level. This year, the average was two months below grade level. In math, the average math skills of a Baltimore student in 1977 were 18 months below grade level; the average now is 3.1 months above grade level.

In Birmingham, which uses the California Achievement Tests to check student performance, school officials report similar results. Randy Glaze, coordinator of testing for the district, says the schools began a program similar to Alexandria's in the mid-1970s.

In 1976, Glaze said, fourth graders scored one year below grade level in reading and eight months below grade level in math. This year, fourth graders were tested at four months below grade level in reading, he said, and three months above grade level in math. Eighth graders' scores in reading, which were 1.9 years below grade level in 1976 were seven months below grade level this year, he said. And in math, eighth graders' scores are now three months above grade level, compared with 1.8 years below grade level in 1976.

"We spent five years studying what makes a good school system effective," said Gilbert Austin, director of the Center for Education Research and Development at the University of Maryland, Baltimore County. "No one program works in every city for every child, but there were some correlations."

Two factors seem to be most important, Austin said: effective teachers, and administrators who work closely with them.

"In a sense it is as simple as that," he said. "Teachers who have a good idea what they are supposed to teach, who inspire confidence and love their jobs, make for a first-rate school."

Even with the changes, Peebles warns that no one should expect overnight miracles.

"These things take time, our plans are good, but they will take time," he said. "I wouldn't trade our school population for their Arlington and Fairfax school population any time for anything. They are wonderful. We are doing our best by them."