Five mayors -- Edward I. Koch of New York, Richard Berkley of Kansas City, Mo., Tom Bradley of Los Angeles, Henry Cisneros of San Antonio and Harold Washington of Chicago -- say a head count alone disenfranchises minorities.

In an Aug. 22 editorial, The Post argued that the Census Bureau should not use statistical techniques to adjust for the substantial undercount of minorities that undoubtedly will occur in the 1990 Census. The editorial exhibits a distressing misunderstanding of what "adjustment" is and of the depths of the injustice that makes it necessary.

In every Census since at least 1940 there has been a severe and disproportionate undercount of minorities, especially those in large urban areas. The editorial correctly reports that about 1 percent of the population of the United States was not counted in the 1980 Census. It does not report, however, the Census Bureau's own calculation that the national undercount of black citizens was 7.2 percent and the national undercount of Hispanic citizens was 5.9 percent. The combined undercount for blacks and Hispanics in urban areas was close to 11 percent.

The result of this disparity in the undercount is that the most disadvantaged people in our society are being disenfranchised. They are losing political power and their fair share of the services bought by those revenues that are distributed according to population. This is grossly unfair. These are, after all, real people, even if the Census Bureau has failed to count them.

The disproportionate undercount is an inevitable result of the Census Bureau's exclusive reliance on the head count. There are many homeless people in our central cities who will not be found. There are also many families illegally doubled up in apartments who fear that cooperation with the census will betray their marginal, possibly illegal status. Finally, many undocumented aliens live in our cities; they too are reluctant to cooperate with the Census.

If the Census Bureau uses only the head count in 1990, the disproportionate undercount will certainly recur. Improved counting techniques will not solve the problem. Thebureau attempted such improvements in 1980 -- broad advertising, expanded numbers of counters -- but the disproportionate undercount persisted.

Fortunately, statisticians have devised scientific methods of counting these uncounted people. These "adjustment" techniques involve extrapolating from the same data that the Census Bureau already uses to determine the extent of the undercount. Committee after committee of distinguished scientists, including panels of the American Statistical Association and the National Academy of Sciences, have supported the use of adjustment as a valid way -- in fact, the best way -- to improve the accuracy of the census.

The Census Bureau has itself acknowledged at various times that adjustment is technically sound and feasible. The bureau adjusted the head-count results in the 1970 Census to correct an undercount. In post-Census studies since then, it has also used adjustment techniques. Moreover, experts in the Census Bureau reportedly endorse the use of adjustment for the 1990 count. The question is why the political officials appear to be resisting.

The adjustment methodology that has been urged is at least as rigorous as the head count itself. Conversely, the vast array of judgments that must be made in conducting the head count -- how to deploy resources, where to make the most intensive efforts, how to resolve uncertainties and ambiguities -- are more subjective and more subject to "monkeying," to use The Post's term, than are any of the techniques of statistical adjustment.

We are sympathetic to The Post's concern that the Census not be excessively politicized. But what could be more political, and cynical, than refusing to use what is generally accepted as the best method of ensuring accuracy and instead using a method that all acknowledge will disenfranchise and otherwise cheat large portions of our population?

The Constitution requires that the Census be as accurate as is practicable. We know that the head count alone is not accurate, and we know also that adjustment will improve the accuracy of the Census. The bureau should agree now to use adjustment in 1990 Census. If it will not agree, Congress should pass legislation requiring it to do so.