Scanning the results of three years of standardized testing in D.C. public schools, a person can find almost whatever he or she is looking for.

There are dramatic successes: The percentage of children scoring at the lowest level in math dropped from 92 percent at Browne Junior High School in 1997 to 43 percent last spring; those scoring that poorly in reading at Van Ness Elementary dropped from 63 percent to 32 percent.

And there are continued, troubling failures: More than 90 percent of Dunbar High School students scored "below basic" in math all three years, and the percentage scoring "below basic" in reading at that school has been rising, from 52 percent in 1997 to 61 percent last spring.

Overall, most elementary, middle and junior high schools show welcome declines in the percentage of students scoring "below basic" on the Stanford 9 Achievement Test, which testing company Harcourt Brace Jovanovitch defines as having little or no mastery of grade-level skills. Many high schools, however, have seen little or no improvement.

A few schools saw significant numbers of students climb into the highest two categories: "proficient" and "advanced." But most movement is from the "below basic" to the "basic" level--still slightly below where curriculum experts say strong students should be.

Superintendent Arlene Ackerman has said the slowly improving scores are proof that her two-year-old campaign to improve the struggling school system is working.

But parents who are critical of her tenure point out that the most dramatic improvement in scores came between 1997--the first year the rigorous exams were given--and 1998. Scores in the first year were dismally low in part because the test was brand-new to teachers and students, and administrative chaos had prevented school officials from ordering and distributing test preparation materials in time.

The 1998 tests saw a jump in scores because of an unprecedented flurry of test preparation, the parents say. They note that at many schools, scores seemed to level off in 1999 or even dip slightly, and they question whether Ackerman's reforms have done any real good.

School officials--and the testing company--beg to differ. They insist that test scores rarely jump dramatically in consecutive years and say the important gauge is whether the District maintained its new baseline in 1999--which it largely did--or slipped back to 1997 levels.

"I'm pleased at what I see because I see steady progress," said Associate Superintendent Searetha Smith. "I don't see significant dips overall."

Smith and Ackerman say test scores will continue to rise slowly over the next three to five years if efforts to retrain and refocus teachers, sharpen curriculum and hold principals accountable are effective.

The school-by-school scores released by the school system two weeks ago contain many lessons and a few surprises. Smith said she is puzzled, for example, that the city's 10 junior high schools in general performed better than its 11 middle schools, despite recent research showing that children are more likely to thrive in a middle school than in a junior high school.

"I really want to know more about that, because it's usually the opposite," said Smith, who has been with the school system for less than a year.

The discrepancy may stem from the fact that the District's junior high schools include Alice Deal and Paul--both large, strong schools in neighborhoods that are relatively stable economically--and traditionally high-achieving Jefferson.

Scores are better at elementary schools than at secondary schools--a reflection in part of how many strong students leave the system for private schools or suburban systems as they get older.

Math scores are worse than reading scores, a weakness Ackerman says she will address with classroom and tutoring initiatives this year. And, in keeping with national research, girls outperform boys on the Stanford 9 through grade 8, at which point boys edge ahead.

The scores listed on Pages 6 and 7 of today's District Weekly include scores from public charter schools for the first time. Most of those schools, which are publicly funded but operate independently of the school system, opened just over a year ago.

Their scores are provided by the D.C. Board of Education and the D.C. Public Charter School Board, both of which authorize and oversee charter schools. The school system reported test scores for the charter schools as well as the regular public schools, but the data for the charter schools was wrong.

School officials said they have requested corrected information from Harcourt Brace, but the Washington Post decided to use the data from the chartering agencies instead of further delaying publication of the scores.