Barbara Bryant, director of the Census Bureau, was absolutely right when she said, "We cannot afford to plan to take the next census as we have taken recent censuses." Certainly not the way the bureau took the 1990 one.
To begin with, the 1990 census was, at $2.6 billion, far too expensive. The General Accounting Office told Congress last month that the 1990 census cost 65 percent more in constant dollars than the 1980 census. Of course, there were more people to count, but the 1990 cost per household was $25, compared to $20 (constant dollars) in 1980.
Nor was the product worth the added expense. It was less reliable than the 1980 product. This marks the first time since comparative records have been kept that a census was less reliable than the one before. Demographic analysis shows that there was an under-count of 1.8 percent; in 1980, it was 1.2 percent. There was a 4.4 percent under-count of blacks -- the worst record since 1940.
These missing persons are important. For purely political reasons, the Bush administration has refused to use adjusted figures, which would be more accurate by far, in calculating federal payments to cities and states for programs that are based on population, age and poverty. This probably does not hurt Maryland much, but it probably does hurt Baltimore City a lot. The city's population was under-counted by about 5 percent. Most big cities are in the same boat.
There are some easy procedural ways that the next census can be less costly. For example, relying on the Postal Service rather than the Census Bureau to identify vacant and nonexistent addresses would produce "substantial" savings, the GAO study concluded. And using a "radically streamlined questionnaire" in 1990 would have saved nearly a half billion dollars.
A shorter, less intrusive questionnaire would help deal with a core failing in the 1990 census -- millions of Americans are distrustful of or antagonistic to government requests for information. A simpler set of questions would surely increase participation.
Knowing that change is needed in the planning for the 2000 census is one thing. Identifying the precise changes and getting them implemented is another. The sooner Congress and the Census Bureau begin that planning, the better. It takes a long time to do it right. Planning on the 1990 census started in fiscal 1984, and, obviously, that was not early enough.