REVIEW & OUTLOOK
Sense and the Census
Forget bean counting and stick to head counting.
The phrase statistical sampling doesn't get the heart racing in too many people, but it's been the Holy Grail for a lot of folks who make a living from politics. For years various factions have been pounding on the U.S. Census Bureau to statistically "adjust" the number of various minorities in America who presumably are "undercounted" in the Bureau's decennial head count. On Tuesday, Commerce Secretary Don Evans, affirming a decision by the Bureau's pros, said the U.S. will stick with the raw head count. Unless you make your living off of racial and ethnic tensions, the Bureau's decision is a most hopeful one.
The statisticians and demographers at the Census Bureau determined that statistical sampling, which involves estimates and guesswork vs. actual head counts, can be much less accurate than traditional methods, sometimes off by as much as 35%. (By contrast, the 2000 Census, using nonsampling methods, was off by an estimated 1.2%, down from 1.6% in the 1990 count.) And no doubt, making sampling "adjustments" would have been a partisan nosegay to Democrats and race-based special-interest groups, who prefer Census numbers that expand their voter rolls and warrant more federal dollars for their parochial agendas.
This alone is enough to applaud the anti-sampling decision. But consider the larger portrait of America on view here. The 2000 Census offered 63 racial options and two ethnic possibilities, Hispanic or non-Hispanic, which allowed for a total of 126 combinations. Nonetheless, some respondents still couldn't, or wouldn't, fit themselves into this racial pigeon house.
In one sample group, 11% of the respondents either couldn't find the right mix or weren't interested in being racially classified. They marked "other." It's not hard to understand why. In California, the nation's most populous state, 14% of all children born in 1997 were reportedly multiracial. According to Princeton University researchers, 8% of Hawaii's population and 10% of Oklahoma's is of mixed racial heritage. The Census Bureau reports that marriages between blacks and whites have quadrupled to nearly 300,000 over the past 20 years. In 1995, one in 12 marriages (8.4%) was interracial.
We submit that such figures represent, in part, a maturing of racial attitudes in America. It's a feat made all the more impressive when one considers that our nation is barely three generations removed from a civil war over race, and just one generation beyond legalized segregation. Steve Lewis, who operates the Web site MixedFolks.com, sees a trend. "More and more people don't want to claim either side," he told the Fort Worth Star-Telegram last week. "They don't want to be one or the other. I'm mixed and that's what I am."
Back in 1995 a Newsweek survey found that 48% of blacks and 47% of whites agreed that the Census Bureau should stop collecting information on race and ethnicity "to move toward a more color-blind society, even if it becomes more difficult to measure progress on civil rights and poverty programs." Amitai Etzioni, writing in the current issue of Policy Review, says the arrival of the 2000 Census puts us at a cultural crossroad. "At issue is how we view ourselves as a nation. Are we going to continue to be divided by race? Or will we welcome the blurring of the lines that divide the races?"
Statistical sampling, which accentuates racial and ethnic differences for political purposes, would have worked to thwart this trend. The federal government allocates more than $200 billion every year based on Census figures. That's largely why groups like the NAACP, the Hispanic Congressional Caucus and the Japanese Citizens League favor sampling and won't simply agree to a "multiracial" category on the Census form.
Last week, The Wall Street Journal reported that other racial- and ethnic-interest groups have lobbied (unsuccessfully) for separate "Creole," "German American," "Cape Verdean" and "Middle Eastern" categories, to name a few. The possibilities of course are almost endless, but each group fears that its money and political influence will decline if the numbers do. For this reason the Census has become a primary source of America's regressive identity politics. The irony is that the broader body politic is moving away from rigid racial classifications even as the civic and political leadership digs in its heels, unable to free themselves of the monetary incentives of divisive racial categories.
This needs to change. The Constitution requires a decennial count of the people for the express purpose of apportioning seats in Congress. Racial classifications are, indeed, a legitimate interest for demographers and anthropologists. And a large body of law depends on us having such information available. But tight reins should be kept on those who would exploit and manipulate Census data for political ends. The anti-sampling decision looks like a good place to start.
