in espanol
mission statment
board of directors
reports to congress
contact us

Monday, March 1, 1999
Kenneth Blackwell

Two sets of census books?

The United States Supreme Court, two federal courts, the majority of Congress and expert statisticians across the country have all agreed that the Clinton administration plan to add and subtract people from the 2000 Census - via statistical adjustment - is either misguided or illegal. Yet Census Director Kenneth Prewitt last week announced plans do so anyway.

The Census Bureau would take an actual count to meet a Dec. 31, 2000, deadline, then adjust the counts according to a statistical survey and release a second set of numbers, to be used for redistricting within states and the disbursing of federal funds.

In my time as mayor of Cincinnati and treasurer and secretary of state of Ohio, I have never come across a legitimate organization that was better served by two sets of books.

As the co-chairman of a bipartisan panel charged by Congress with overseeing the preparations for Census 2000, I also know how statistical adjustment works - and how it doesn't: It gets increasingly less accurate at smaller levels of geography.

A survey can give you a pretty good idea of the total national population, and is still pretty good at the state level. Big cities - it's a gamble. Small cities - it's even worse. When you get down to neighborhoods, the data are simply unreliable. Want good, hard census information about a block in Brooklyn? Fuhgeddaboudit.

New York's Rep. Carolyn Maloney insists that "modern statistical methods" can fix large undercounts in New York City. The truth is that a bad adjustment would just cover up the problem.

For example, statistics show the 1990 census missed roughly one in 12 African-American men, nationally. This could easily be "corrected" by going through the census and adding an African-American male for every dozen African-American men actually counted. But that would do little to improve the undercount, or the quality of life, in East Flatbush, where the undercount of Black men was significantly higher than one in 12.

Why? Political representation and public funds are distributed to geographic or political areas, not demographic groups. If the census determines how many African-American men are in New York, but fails to determine how many live in East Flatbush, the people of East Flatbush still won't get their fair share of representation or funding.

Rather than relying on imprecise statistical adjustments, why not use targeted solutions to find real people where they really live? Restore the Post-Census Local Review, and let local governments quality-check census numbers before they are made final.

The Census Bureau contends that Local Review is too costly and has been replaced by a pre-census update of the mailing list. In fact, the bureau updated the mailing list in 1990 and still offered Local Review. As to expense: Pay now, or pay later. Correcting a mistake afterward would require a locally funded census, a troop of lawyers and more.

Local Review will actually find people where they live, and also give local governments a say in the process that determines their political representation and public funding. Compared to a statistical adjustment that produces unreliable data and two conflicting numbers, that's a pretty good deal.

to Web ...
People, not statisti...