The black and Hispanic groups are concerned that blacks and Hispanics will not be fully counted. This is not a new issue. Census statisticians have known since the 1970s that there have been undercounts of people in neighborhoods with high crime rates or large numbers of illegal immigrants. Census Bureau professionals have worked to measure these undercounts and to minimize them by using official records and enlisting local volunteers to locate residents. Their efforts have had some success, as the undercount was lower in 2000 than in 1990.
Nonetheless, there have been demands that the Census numbers be adjusted by statistical sampling. The Supreme Court ruled in 1999 that sampling could not be used to apportion House districts among the states, but left open whether it could be used for other purposes. But after an intensive three-year study, Census professionals in 2003 said they could not guarantee that sampling would produce a more accurate count than the enumeration decreed by the Constitution.
As then-Census Director Louis Kincannon said, "Adjustment based on sampling didn't produce improved figures." Sampling might produce a more accurate number for large units but not for smaller units -- just as the sampling error in public opinion polls is small for the total population but much larger for small subgroups. At the block level, sampling would result in imputing people who aren't actually there.
The potential for political mischief, political overrepresentation and greater federal funding for favored groups is obvious, just as Congress' refusal to reapportion after the 1920 Census resulted in political overrepresentation of low-growth rural areas and under-representation of then-booming big cities.
The better procedure is to trust the professionals at the Census Bureau. "I found the Census personnel to be among the most conscientious of any group I'd encountered in government service," Bruce Chapman, census director in the Reagan administration, recently wrote.
"Whatever their personal political views (I suspect that most voted for Obama), their allegiance is to the integrity of the positions of public trust they hold." This comports with my own observations of Census personnel over the years. Like other federal statistical agencies, the Census Bureau has a proud culture, developed and nurtured over many years and in many administrations, of independence from political manipulation and dedication to statistical rigor.
So it's dismaying that the Obama White House, in response to political pressure, would consider overseeing the 2010 census. A better approach, endorsed by seven former Census directors and embodied in a bill sponsored by Rep. Carolyn Maloney, a New York Democrat, would be to set the Census Bureau apart as an independent agency. That would preserve, protect and defend the census that the framers of the Constitution took pains to establish.