Richmond Mayor Says Undercounting Issues Critical to Accurate Census Count

By Larry Jones

Testifying on behalf of the United States Conference of Mayors, Richmond Mayor Timothy Kaine told members of the House Census Subcommittee Feb. 11, that a proposal to reinstate "Post Census Local Review," (PCLR) will have very little effect on reducing the high level of undercounting that many cities have experienced in past census counts. To address the problem of undercounting, Mayor Kaine said the U.S. Census Bureau must be allowed to "...use all available scientific methods, including statistical sampling." He explained that "it is critically important to America's cities that the undercount issues be given primary attention as we attempt to make the 2000 census the most accurate it can be."

Mayor Kaine was invited to comment on a proposal (H.R. 472, the Local Census Quality Check Act) sponsored by Subcommittee Chairman Dan Miller (FL). The proposal would reinstate PCLR, a program used in the 1990 decennial census which provided local officials the opportunity to review housing unit counts (not address lists) for their jurisdictions and submit evidence of missing households. In discussing the city of Richmond's experience with PCLR in the 1990 census, Mayor Kaine told members of the subcommittee that the "...review had a very minor impact on correcting the housing unit discrepancies that we experienced. But, the review had virtually no effect on the significant undercounting of people in our city."

The city of Richmond experienced two problems with the 1990 decennial census: the first was a discrepancy in the number of housing units actually counted by the Census Bureau; and the second was a significant undercount of citizens, particularly those who are poor and live in minority neighborhoods. At present, the city is participating in the "Local Update of Census Addresses" (LUCA) program, which was initiated by the Bureau following the 1990 census to build a more comprehensive Master Address File before the next census. The program has allowed local officials to work with the Bureau to harmonize the Bureau's census address records with a variety of city address records. Mayor Kaine said "while there is still much work to be done..., we believe that participation in LUCA should drastically reduce housing unit discrepancies. He further explained that "the ability to correct mistakes after the census is taken is an inferior method to doing everything up front to make sure the count is accurate as possible."

During the 1990 census, about 25 percent of all local governments (about 9,800 of the 39,000 total) participated in PCLR. The program was responsible for adding 125,000 people living in 81,000 housing units to the 1990 census at a cost of $9.6 million. Subsequent evaluations by the Census Bureau revealed that an estimated 11.7 percent of those housing units were added in error. In the end, the review efforts added only one tenth of one percent to the final population count. In contrast, a post enumeration survey (PES) conducted by the Bureau after the 1990 census revealed a net national undercount of 1.6 percent or 4 million people, which was 50 percent higher than the 1980 undercount. The Bureau is planning to use this scientific statistical sampling method in the 2000 census to measure undercounts and overcounts for all geographic levels and correct the final census numbers based on the PES measurements. The U.S. Conference of Mayors strongly supports statistical sampling as a supplement to traditional counting methods. Past experience has proven that traditional counting methods alone, including intensive field follow up, will not make a significant impact on the persistent differential undercounting of racial minorities, the rural poor and children.

To demonstrate the impact of the undercount problem, Mayor Kaine submitted for the record a copy of a report published by the Conference on January 28 which shows the findings from a recent 34-city survey. The survey was conducted to determine the impact of the 1990 undercount on cities across America and the likely impact of similar inaccuracies in the 2000 census. The survey found that:

¥ The estimated number of undercounted people in the survey cities represents an average of four percent of the cities' total population.

¥ Of the 34 cities responding to the survey, 20 estimated a total loss of $536 million in federal and state funds during the 1990s as a result of the 1990 undercount. The estimated average loss to cities during the 1990s averaged $1,230 for each person not counted. For each person included in the city's 1990 population count, the loss averaged $56 per person.

¥ The estimated loss in federal and state funds during the 1990s varied from a high of $184.4 million in Chicago, to $120 million in Los Angeles, $40 million in Cincinnati, $1.3 million in St. Petersburg and $1 million in Charleston.

¥ Twenty of the cities also estimated that they would lose a total of $677 million in federal and state funds in the next decade if the 2000 census maintains the same level of inaccuracy as the 1990 census. This represents an average of $2,263 for each person not counted in the city and $129 for each person included in the city's 2000 census count.

¥ The estimated loss for the census also varied among cities from a high of $227 million in Los Angeles, to $184 million in Chicago, $80 million in St. Joseph (MO), $24.5 million in McAllen (TX) and $7.6 million in Pembroke (FL).