FOR THE PAST couple of years, the Census Bureau has taken a lot of criticism for undercounting blacks and members of other minority groups. Before the 1980 Census, the bureau changed some of its procedures, put on an extensive publicity campaign and worked hard in other ways to see that fewer people were missed.

These efforts seem to have paid off. The bureau recently compared its 1980 results with population estimates obtained by a number of methods. The comparison shows that the undercount was, for the whole population, only four-tenths of 1 percent. Among blacks, the undercount was reduced from 7.6 percent in 1970 to 4.8 percent in 1980--a significant difference. The figures the bureau has tabulated at the present time do not permit conclusions on whether other ethnic or economically defined groups were also undercounted. But in all likelihood, the undercount was greatest among the poor. There is no reason to believe that there was any deliberate undercount of blacks for racial reasons.

Why do blacks continued to be undercounted? The black undercount is concentrated among males age 25 to 64 and among both males and females under age 10. Adult males of every race tend to be undercounted; they are the hardest people to keep track of. The Census Bureau can make even greater efforts to find adult black males in the future, but the undercount of this group will never be reduced to zero.

No one is sure why black children continue to be undercounted. The best the Census Bureau can do is to say "it is clear that there were certain factors which tended to keep the omission of black children high." The lesson for the 1990 Census is clear: concentrate even harder on making sure that all children are enumerated. But the fact that we still can't explain this undercount is a reminder that the workings of our society often remain tantalizingly difficult to understand--even when our statistical tools, like the 1980 Census, are the best that can be devised.