Don't Count Out Sampling

October 3, 1997

Republicans in Congress seem bent on preventing the Census Bureau from using a statistical method that promises a more accurate count of Americans in 2000. Both House and Senate have tried to restrict spending on so-called sampling. The House bill would effectively kill the project.

Sampling, however, has always made good sense. It's simply a way of arriving at a better enumeration - to use the Constitution's word - at a more reasonable cost than past attempts to directly contact residents who don't mail in census forms.

Republican concerns spring from their distrust of the Clinton administration. Specifically, some Republicans are worried that the sampling technique, whereby some hard-to-count neighborhoods are thoroughly canvassed and the findings projected to adjust counts in other neighborhoods, could lend itself to political manipulation. With such things as apportionment of seats in Congress and distribution of federal grants at stake, the concern is understandable.

But it's not justified. The sampling plan is statistically sound, and a more solid count can only help the country. In a sprawling land of at least 265 million people, every count is an estimate. Sampling promises a better estimate.

If fears of political manipulation persist, they could be allayed by appointing an independent panel of experts to monitor the Census Bureau's sampling processes.

Funding for sampling should be protected when the House and Senate meet to reconcile their bills.