Caught in the middle of a political dogfight over whether statistical sampling should be used in the year 2000 census, a three-judge federal panel Monday ruled in favor of the Republicans–but required some fancy legal footwork to do so.
Left unresolved by the court was the constitutionality of statistical sampling, in which scientific methods are used to account for populations that are underrepresented in census counts. The Clinton administration ought to take the matter up to the U.S. Supreme Court, which could well reverse the panel’s decision.
At issue is whether the Census Bureau should use statistical sampling to get a more accurate count of the population. According to the bureau’s own calculations, the 1990 census missed about 8.4 million people, while 4.4 million other folks were counted twice.
Most of the errors occurred in large cities and involved African-Americans and Hispanics, two groups notoriously reticent about filling out census forms and difficult to track down in all the urban nooks and crannies.
The bureau wants to do an actual count of 90 percent of the population and then do sampling-based projections on the rest of the people. There is ample evidence that sampling would achieve a more accurate final count.
That’s where partisan politics barge in: Most of the undercounted are likely to lean toward the Democratic Party, and the GOP–nervous about its majorities in Congress–adamantly opposes anything that would give the Democrats an edge, either in reapportionment battles or distribution of federal funds.
House Speaker Newt Gingrich filed suit in February to stop the Clinton administration’s plan to use sampling on grounds that it would violate the constitutional mandate for an “actual enumeration” of the population, a literal head count. That is a thin thread because the Constitution also provides for the census to be conducted “in such manner as (Congress) shall by law direct.”
The federal panel ducked the constitutional issue, basing its decision on a reading of federal laws that seem to offer contradictory directives on sampling. The panel’s reasoning isn’t outlandish, but it’s also quite possible the high court would reach a different interpretation of the federal laws.
The bottom line is that sampling would produce the fairest, most accurate count of the population. The Supreme Court should take the opportunity to examine if that, indeed, is what federal law will allow.
