Archive


home
in espanol
mission statment
news
board of directors
A.
Dr.
Joe
reports to congress
contact us
                

 

Endnotes

  1. 23 February 1999, Updated Summary: Census 2000 Operational Plan, p. 2.

  2. There is a distinction between “direct” estimates and “synthetic” estimates of population.  Direct estimates of an area’s population rely on data collected exclusively from that area.  Synthetic estimates use data from an entire region to make estimates about smaller areas within that region.  Therefore, a direct estimate of population in one of the 1990 PES block clusters uses observations made only in that block cluster.  A synthetic estimate of population in one of the 1990 PES block clusters may use observations from hundreds or thousands of clusters in a region.  For a more detailed discussion of the Bureau’s synthetic estimation, see Section V: Methodology.

  3. Although the file provided to the Board contains data for 5,180 block clusters, there were 10 block clusters where the value for the dirDSE (see page 16) is zero.  Since this variable is the denominator in coverage calculations, the calculations cannot be performed.  In mathematical terms, the result of a division by zero is "undefined."  In addition, the Bureau previously reported 5,290 block clusters included in the 1990 PES. Bureau staff informed the Board that the clusters not included in the file had no recorded population.

  4. Chicago Housing Authority, Statistical Profile:  The Chicago Housing Authority, 1991 to 1992, prepared by CHA Management of Information Systems Department and the Department of Research and Program Development, (Chicago, IL, 1992), 5.

  5. Adjustments calculated as a percentage of the official census count, as reported at  www.census.gov.

  6. That is, census coverage rate was equal to or higher than 102 percent.

  7. U.S. Census Monitoring Board, Young & Rubicam briefing, 8 July 1999.

  8. U.S. Census Monitoring Board, Public Hearing, 16 December 1998.

  9. Department of Commerce, Bureau of the Census, Committee on Adjustment of Postcensal Estimates (CAPE Committee), Assessment of Accuracy of Adjusted Versus Unadjusted 1990 Census Base for use in Intercensal Estimates (Washington, DC, 7 August 1992), p. 15.

  10. A block cluster is a single block or a group of two, three or four blocks, or a portion of a large block.  Although the file provided to the Board contains data for 5,180 block clusters, there were 10 block clusters where the value for the dirDSE (see page 16) is zero.  Since this variable is the denominator in coverage calculations, the calculations cannot be performed.  In mathematical terms, the result of a division by zero is "undefined."

  11. The data used is the revised 1990 PES file, with 357 poststrata.  This file contains information on those areas surveyed in the 1990 PES.  To our knowledge, its release is the first time the Bureau has identified the exact blocks used in the 1990 PES to an agency outside the Bureau

  12. We performed identical analysis, substituting the “Census Plus” estimate for the direct DSE.  Census Plus is an alternative population estimate calculated by the Census Bureau.  Use of this substitute did not yield appreciably different results or conclusions.  Here we report our results using the direct DSE as a standard.

  13. Data is the revised 1990 PES file, with 357 poststrata.

  14. The Bureau adjusts population counts via synthetic estimation using poststrata.  Poststrata are subpopulations that result from crossclassification of the population by age, sex, race or origin, tenure (owner or renter), and geographic type (urbanized area).  In the revised 1990 PES data used in this analysis, there were 357 such poststrata.  Poststrata may span thousands of blocks and several states in a region.  Each census block generally contains members from many poststrata. Synthetic estimation multiplies the census count of all members of each poststrata in each block by a common adjustment factor.  In a given block, the resulting adjusted counts for each poststrata are then added together, to determine an adjusted total for each block.

  15. Due to subsampling (surveying only parts of some blocks), the PES Block file slightly overstates the number  of imputations in the sample.  In addition, the PES Block file appears to overstate the total population of the sample.  For details, see Section V: Limitations.

  16. The estimates for any given block cluster are subject to random variation.  By sorting the block clusters into these groups, that random variation is reduced to a minimum.

  17. Although the file provided to the Board contains data for 5,180 block clusters, there were 10 block clusters where the value for the dirDSE (see page 16) is zero.  Since this variable is the denominator in coverage calculations, the calculations cannot be performed.  In mathematical terms, the result of a division by zero is "undefined."

  18. Department of Commerce, Bureau of the Census, Report to Congress – The Plan for Census 2000 (Washington, DC, August 1997), 6.