What the report shows is a mixed picture of some good news and some bad, of gains and disappointments, of gaps narrowing and then widening again.

• In mathematics and science, there have been substantial gains since around 1980, which followed declines during the 1970s. Thus, the efforts of many schools and states to improve math and science education, which received a major boost from the "Nation at Risk" report in 1983, seem to have paid off.

• In the 1990s, student performance in mathematics has continued improving, though generally at a slower pace than in the 1980s. In science, particularly since 1992, gains in student achievement appear to have stalled.

• In reading, there has been little change in achievement over the past three decades despite considerable ferment in the curriculum. The one point at which some patterns are discernable is age 9, which usually is at grade 4 when basic instruction in how to read normally ends. During the 1970s, there was a clear improvement in early reading, followed by a clear decline in the 1980s. Since 1990, the average score for 9-year-olds has risen by three points, but that's not enough to be statistically significant.

• For 9-year-olds in the lower quartile of achievement, reading improved substantially from 1971 to 1980. These gains were lost, however, during the slide in the 1980s. Since 1990, there has been a significant recovery, though the scores seem unchanged from 1996 to 1999, and the 1999 reading level of 9-year-olds in the bottom quartile was below where it was in 1980.

• Overall, students in the lower quartile of the score distribution have made gains since the early 1970s. But with the exception of mathematics, these improvements have not continued in most age groups since 1990.

• Over the three decades of NAEP reporting, the difference in average scores between white and black students and between whites and Hispanics has narrowed—in some cases dramatically—because of gains in minority achievement. This is encouraging news.

• Virtually all of the reduction, however, in the gaps between whites and blacks took place before 1986 or 1988. Since then, the gaps have either widened somewhat or stayed the same. The average scores of black students have remained well below those of whites, and at age 17 the reading achievement of black students was lower last year than it was in 1988, a depressing reversal of the gains made over the previous two decades.

• Among Hispanic students, average scores are somewhat higher than those for blacks. Hispanics were not reported separately in NAEP's first few assessments in the early 1970s. In the past decade, the gaps between whites and Hispanics have fluctuated—some narrowing, some widening—but overall little has changed.

These racial gaps are an important subject. They are difficult and sensitive to talk about. But, as difficult as it is, the discussion is now underway in many communities and states across the country. More states and districts are finding that, as a first step toward reducing gaps in student achievement, it is important to disaggregate test scores by race.

Often in order to improve performance it is necessary to know first what the performance levels are. NAEP has been reporting its results by race since the program began three decades ago. It shows a pattern of substantial change that was immensely encouraging for two decades and now somewhat discouraging in the most recent decade.

Among 17-year-olds, for example, the black-white gap in reading was cut by more than half, from 53 points when the assessment was first administered in 1971 to 20 points in 1988. In math for 17-year-olds, the gap was reduced from 40 points in 1973 to 21 points in 1990. In more recent years, though, the gaps have widened again—back to 31 points in reading and 32 points in math.

Another way to look at the data is that the average scores for 17-year-old black students in reading and math are about the same as the averages for 13-year-old whites. While these differences are not as large as they were three decades ago, the pattern is disturbing and demands careful attention.

The average scores, however, are not the only way to understand what is happening. The NAEP long-term trends report also has performance levels, describing what students can do at 50 point intervals along the scale. At the basic skills levels in both math and reading the racial gaps have diminished sharply as the performance of black students has nearly caught up to whites. But when you move up the NAEP scales to more complicated work, the differences remain wide.

For example, at age 17, almost all students can add, subtract, multiply, divide, and solve one-step problems. About 99 percent of whites reach this level 250 on the NAEP math scale, compared to 89 percent of blacks. In 1978, when these anchor points were first used, about 96 percent of whites could reach 250, compared to 71 percent of blacks. Thus, a gap of 25 percentage points has shrunk to 10 percentage points.

At level 300, however, which requires students to solve moderately complex problems and use decimals, fractions, and percents, the black-white gap is 43 percent. Both whites and blacks have improved since 1978, but the whites have gone up from 58 percent reaching this level to 70 percent, while the proportion of blacks attaining level 300 has climbed from 17 percent to 27 percent. The gap is unchanged.

There is a similar pattern in reading.

For the basic skill level of finding facts in simple stories and drawing inferences based on short passages, 95 percent of black 17-year-olds reach the anchor point up from 82 percent in 1971, compared to 98 percent of whites in both years. At the level of understanding and analyzing relatively complicated information, whites have moved up only 3 percentage points to 46 percent over the three decades. Blacks have gone up from 8 percent to 17 percent, but a 29 percentage point gap remains.

Undoubtedly, all these disparities reflect differences in income, family, and societal factors, as well as in schooling. But, according to a special tabulation of the long-term trends data, the gaps are just as wide or even slightly wider among black and white students with college-educated parents as they are for students whose parents have much less education. The same is true even more strikingly for the gaps between whites and Hispanics. Having better-educated parents is related to higher achievement for students of all races, but it does not seem to close the racial gaps. This may raise questions about the quality of the college education that students of different races receive and the different impact college may have on incomes, careers, and families.

What should we make of all this?

Because of its nature as a survey, NAEP cannot tell us why these changes occurred. My own view, given the trends in academic performance and attainment elsewhere, is that the gains of the 1970s and 1980s probably reflect the success of the civil rights movement and the great efforts to improve the education of low-income children that started with the Head Start and Title One programs in the late 1960s. The stagnation and even reversal in some areas since then indicate we have not built adequately upon these gains.

We should not lose sight, though, that our schools have been successful in closing the gaps at the important basic skills levels of performance. This is the point at which many states have focussed the greatest attention by actions such as the passing scores they have set on high school graduation and grade promotion exams. The racial differences on many of these tests have been reduced, and NAEP provides a confirming bit of evidence that change has occurred.

That is not an inconsequential accomplishment. It means that thousands of students are graduating from high school each year with a minimum set of reading and math skills that many students three decades ago did not have.

At the same time, however, this is clearly not enough. The same disparities we see at the higher skill levels on NAEP show up in enrollments in high school honors classes and more rigorous curricula and in the upper score ranges of college entrance exams that are used for admission to the most selective and rigorous colleges. The resulting inequalities in education and employment are great.

Certainly, more resources must be directed at dealing with the problem. There must also be clear standards and expectations for all students that are much higher than the minimum competencies that have been put in place. But I think values and attitudes must be taken into account as well.

Parents generally and African American parents in particular must insist that their children have access to high quality schools, high quality curricula, and teachers who are the very best. They must also insist that the schools challenge their children and that the children work hard to meet the challenges. There must be pride in academic and scholarly accomplishment and the discipline and focus that successful academic work demands.

None of this will be easy. None of it will happen overnight. But until these academic disparities are greatly reduced, the other inequalities in our society will persist, and the nation will continue to fall short of developing the full talents and abilities of so many students.