Newfangled Ways to Count Heads : Fiscal and technological changes for the census

It has been apparent for some time that the decennial American census is in trouble. The 1980 census was plagued by technical problems and damaging delays in reporting results, and the 1990 census was flawed because of undercounts in major urban areas. Now, facing severe budget cuts, Census Bureau officials have begun to wonder whether they can conduct the 2000 census at all.

Even without budget woes, the time has come for the Census Bureau to devise new ways of counting heads. The increasing difficulties of taking an old-fashioned census in urban America are all too apparent. The bureau spent $2.6 billion to conduct the 1990 census, and overlooked millions of people. Will spending as much as $4.8 billion on the next census produce markedly better results?

BROAD EFFECTS: Few functions of government have so pervasive an impact on daily life as the census, mandated by the Constitution. Every school, city council, state legislative and congressional district is redrawn every 10 years to reflect shifting population. All manner of government programs, from housing to health care, are funded under census-based formulas. Businesses use the data to make key marketing decisions.

The Census Bureau has embraced new technology, feeding huge amounts of data to eager users on the Internet. And it plans to increase efficiency in reporting and counting census returns through computer technology.

However, the tougher task is to conduct a census without sending a form or a census taker to every household. Under study are various alternatives, including ones that would exploit modern sampling techniques. For example, rather than try to count every head in hard-to-canvass urban areas, the bureau would select a few typical areas, flood them with census takers and get a fairly accurate count there. Then that data would be used to extrapolate counts for similar areas. Other ideas include using administrative data like driver’s licenses and electric meter hookups to make informed indirect estimates.

ENDLESS COUNT: One of the most promising notions is that of a “continuous” census. The American people are so mobile today that the 10-year census is obsolete even before results are published. Under a continuous census, the bureau would use the 2000 census as a base line and then update it yearly or more often with a survey of 300,000 households. This would substitute for the onerous census long form that goes to 15% of households and is so crucial in gaining economic and social data.

All of these approaches have their drawbacks. While a fairly good population estimate can be had by sampling, will Americans trust the results if they themselves are not enumerated directly?

The House has slashed $69 million from the bureau’s $340-million request for next year. That is too much. But the reality is that the head counters will have to live with less money. They should seize on this as an opportunity to show that high technology holds promise for a quality census at lower cost and inconvenience to the public.