Accuracy of the 1990 census and undercount adjustments.

Area of Science:

• Demographics
• Statistical Analysis
• Government Policy

Background:

• The U.S. Census Bureau conducted a Post-Enumeration Survey (PES) to assess the accuracy of the 1990 census.
• The Bureau recommended adjusting the 1990 census data to correct for an estimated undercount.
• The Secretary of Commerce ultimately decided against implementing the recommended adjustment.

Purpose of the Study:

• To analyze the Census Bureau's total error and loss function analyses.
• To identify and discuss the criteria used by the Department of Commerce in deciding not to adjust the census.
• To compare the Department of Commerce's decision criteria with aggregate loss functions.

Main Methods:

• Review of Census Bureau analyses regarding census accuracy and PES-based undercount adjustments.
• Examination of total error analysis and loss function analysis methodologies.
• Identification and discussion of the specific criteria influencing the Department of Commerce's final decision.

Main Results:

• The Census Bureau's analyses indicated a need for census adjustment due to undercount.
• The Department of Commerce's decision not to adjust relied on criteria distinct from standard aggregate loss functions.
• The specific criteria used by the Department of Commerce are detailed and evaluated.

Conclusions:

• The decision-making process for census adjustment involves complex analyses and specific policy criteria.
• Understanding the criteria beyond statistical loss functions is crucial for interpreting census accuracy decisions.
• Further examination of the criteria used in the 1990 census adjustment decision provides insights into statistical policy.