SHOULD THE 1990 Census count be "adjusted" to account for people the Census takers are believed to miss? Some statisticians who believe they can measure the Census undercount accurately, and politicians whose states and districts stand to gain from adjustments, are arguing that adjustments should be made. But any monkeying with the actual head count is open to political abuse and threatens the integrity of government statistics.

It is generally accepted that the Census Bureau never quite succeeds in counting everyone and that it misses some kinds of people -- notably young black males -- more than others. The Census Bureau worked hard in 1980 to reduce the undercount, and statisticians believe that it now amounts to about 1 percent of the nation's total population. It means that poor urban areas tend to be slightly underrepresented in legislative districting and that central cities lose money in federal aid pegged to Census data.

The Census Bureau is currently required to report within one year of its April 1990 head count the results that will be used for political redistricting and for allocating federal monies. The problem with adjusting these results to reflect any undercount is that there will always be disagreement over just how they should be adjusted. Statisticians have their arguments, and even their best estimates have margins of error, which grow larger when you move from statewide figures to block-by-block results. The head count, in contrast, is a specific number not subject to manipulation. Once the Census Bureau starts down the slippery slope of adjusting Census returns, where will it stop? It's not hard to imagine a congressional subcommittee chairman insisting on a new adjustment method that would swell his district's population and his city's coffers. Big-city mayors and central city congressmen are already the strongest political constituency for adjusting the head count results, and they have strong political and fiscal incentives to get the populations of their constituencies revised as far upward as possible.

The 1990 Census will be the 21st conducted by the federal government. The Census Bureau has a notable record of freedom from political interference; in World War II it refused to breach the confidentiality of its records when the Army wanted to find residents of Japanese descent. The proposals for adjustment of Census results could lead to political manipulation of the Census counts and destruction of the integrity of an admirable governmental institution. Those who feel the big cities are not getting their due share of federal funds will have to find a remedy that does not hang on tampering with the Census.