Report: Nevada and Texas among tops in census undercounts

03/28/2001

Associated Press

WASHINGTON – Greater proportions of Americans were missed by the census in states with booming minority populations or large urban centers, say Democrats who want to adjust the tally used to redistrict legislative seats.

But while California, Texas, New York and other traditional destinations for immigrants had high percentages of overlooked people, some less-populous states also had comparable net undercount rates, said a report from members appointed by former President Clinton to a bipartisan board overseeing the census.

The estimates were the latest development in a fight between Democrats and Republicans over whether data adjusted using statistical sampling provide the most nearly accurate count of America.

The report was based on an analysis of 2000 census data, but was not sanctioned by the Census Bureau or by members of the U.S. Census Monitoring Board appointed by the GOP-controlled Congress.

It was immediately criticized by Republicans as "inaccurate and irresponsible."

The estimates were expected to be addressed Wednesday at a hearing of the Senate Commerce Committee.

Overall, the Census Bureau has estimated a net national undercount in 2000 of 1.2 percent, or 3.2 million of the country's 281 million people. That is lower than the net national undercount of 1.6 percent in 1990, or 4 million people.

Bureau estimates show more minorities than whites are typically missed.

The Census Bureau has so far released race breakdowns for the populations of 44 states. State lawmakers will use the figures to redraw congressional, state and local legislative district lines. The census also is used to distribute over $185 billion annually in federal funds among the states and local governments.

All 50 states must have figures in hand by April 1.

The bureau has not produced undercount estimates for states or cities in 2000, in part because they are still studying the statistics.

The report from the Democratic members said Nevada had the highest net undercount among states at 2 percent of its population. The state experienced tremendous population growth in the 1990s, partly from a 217 percent increase in its Hispanic population.

Texas' undercount percentage was put at 1.7 percent, while California and New York were both at 1.5 percent, the report said. The undercount in two smaller states – Idaho and Wyoming – also was estimated at 1.7 percent.

Thirty-five states had lower net undercount estimates in 2000 than in 1990.

Gilbert Casellas, co-chairman of the presidential board, said the Democrats' figures prove the Census Bureau should allow adjusted data to be released in order to parcel federal money.

"We know millions of people were missed, we know where they live, Casellas said. "With billions of dollars at stake, why can't we fix the problem if we know we have the means to do so?"

Commerce Secretary Don Evans had cited the declining national undercount rates in deeming the head count the official numbers that states will receive for redistricting.

Evans, whose department oversees the Census Bureau, agreed with a bureau recommendation calling the raw head count numbers more nearly accurate than the adjusted figures. He called the 2000 census the "most accurate in the country's history."

Democrats insist an adjustment would protect against traditional undercounts of minorities and children.

John Thompson, associate director for the bureau, said officials were only beginning the long process of studying the adjusted numbers and had not seen the Democratic estimates. It was unclear when, or if, the agency itself would release comparable data.

"We have to look at the methodology (of the report) more carefully," Thompson said. "We don't know yet if it's sound."

Since the bureau itself has been unable to adjust the raw data, "It's fantasy to think that the presidential side of the board could do it with any accuracy at all," said Wick Caldwell, executive director for congressional board members.

Republicans note the Constitution calls for an "actual enumeration" and say adjustment could only lead to more errors.