1 2010 census coverage measurement survey update 2010 state data center/ census information center...
TRANSCRIPT
1
2010 Census Coverage Measurement Survey Update
2010 State Data Center/Census Information Center
Steering Committee Meeting
February 23, 2010
Thomas MuleUS Census Bureau
2
CCM – Census Coverage Measurement
• Survey provides estimates of coverage error for people in housing units and for housing units:• Excludes group quarters and remote
Alaska• Includes Puerto Rico
• No intent to adjust the census
3
Net Error
• Uses dual-system estimation, a sample capture-recapture technique
• Used in 1980, 1990, and 2000• Provides estimates of net undercount
and overcount• Key demographic groups• Specific geographic areas
• Requires independence from census
4
Components of Census Coverage Error
• New objective for 2010• Provides estimates of:
• Correct enumerations – those included correctly by the census
• Erroneous enumerations – those included incorrectly by the census
• Imputations• Omissions – those missed by the census
• Key demographic groups and by census operations
5
2000 A.C.E. Measurement Error
• 2000 coverage measurement program failed to detect erroneous enumerations• Many were census duplicates
• CCM Goal: Reduce nonsampling error from the 2000 Accuracy and Coverage Evaluation (A.C.E.)
6
Sample Design of 2010 CCM
• Area sample of block clusters – groups of census collection blocks
• Population (P) Sample: Independent sample of housing units in block clusters. Recapture sample.
• Enumeration (E) Sample: Census enumerations in the same sample block clusters
7
Independent Listing
• Independent listing of area sample• Canvass block cluster• List addresses of housing units and
potential housing units• Timing: August-December 2009
8
Person Interview
• Laptop interview at independently listed addresses• Personal visit• Gather independent roster of people
• Probing questions about places where the person:• Should have been counted• Could have been counted
• Timing: August-October 2010
9
Person Matching and Followup
• Match people collected in CCM to census person list• Computer• Clerical
• Search throughout the country• Computer nationwide search• Clerical search at places the person could have
been counted from the person interview• Person Followup (personal visit and telephone,
paper-based)• Timing: October 2010 – April 2011
10
Current Plans for Estimation Domains for 2010
CCM• Estimation domains have not been finalized so products shown here are current plans
• Break into 3 Major Groupings:• Demographic/housing unit characteristic
groups• Geographic areas• Census operation results
11
Examples of Estimation Domains for Demographic Groupings for People in
Housing Units
Demographic Groupings
Net Error Components of Census Coverage
Correct Erroneous Imputations Omissions
Race and Hispanic Origin
Yes Yes Yes Yes Yes
Tenure Yes Yes Yes Yes Yes
Age/Sex Groupings
Yes Yes Yes Yes Yes
Individual Years of Age
Yes No No No No
12
Examples of Estimation Domains for Geographic Areas for People in Housing
Units
Geographic Areas Net Error
Components of Census Coverage
Correct Erroneous Imputations Omissions
Type of Enumeration Areas
Yes Yes Yes Yes Yes
Bilingual Mailing Blocks
Yes Yes Yes Yes Yes
Region Yes Yes Yes Yes Yes
Each State Yes UnderReview
UnderReview
UnderReview
UnderReview
Substate Estimates(county, places)
UnderReview
UnderReview
UnderReview
UnderReview
UnderReview
13
Examples of Estimation Domains for Census Operations for People in Housing
Units
Census Operation/Results
Net Error
Components of Census Coverage
Correct Erroneous Imputations Omissions
Mail Return No Yes Yes Yes No
NRFU enumeration No Yes Yes Yes No
Operation Check-in Date Groupings
No Yes Yes Yes No
Coverage Followup No Yes Yes Yes No
14
What about those people who should have been enumerated
somewhere else?
• A table to show people who should been enumerated outside our block cluster search area.• Same county• Different county in the same state• Different state
15
Examples of Estimation Domains for HU Characteristics for Housing Units
Demographic Groupings
Net Error
Components of Census Coverage
Correct Erroneous Omissions
Occupancy Status Yes Yes Yes Yes
Tenure of Occupied Units
Yes Yes Yes Yes
Type of Structure(Single, Multi-units)
Yes Yes Yes Yes
16
Examples of Estimation Domains for Geographic Areas for Housing Units
Geographic Areas Net Error
Components of Census Coverage
Correct Erroneous Omissions
Type of Enumeration Areas
Yes Yes Yes Yes
Region Yes Yes Yes Yes
Each State Yes UnderReview
UnderReview
UnderReview
Substate Estimates(county, place)
UnderReview
UnderReview
UnderReview
UnderReview
17
Examples of Estimation Domains for Census Operations for Housing Units
Census Operation/Results
Net Error
Components of Census Coverage
Correct Erroneous Omissions
Field Verification No Yes Yes No
Nonresponse Followup Vacant/Delete Check
No Yes Yes No
18
Housing units that were geocoded incorrectly
• A table to show housing unit geocoding results in relation to block cluster search area• Geocoded to block cluster• Geocoded one block outside of the cluster• Geocoded beyond one block outside of the
cluster
19
Reducing Nonsampling Error in the CCM
• Part of Decennial review conducted by Director Groves
• Identified areas in the CCM where efforts can be made to reduce nonsampling error
20
CCM Sample Size
• Reduced from 300,000 to 170,000 housing units
• American Indian Reservation and Hawaii sample remain same
• 1,000 target minimum state housing unit sample size
21
2010 Evaluations and Assessments
• Evaluations• Goals
• Examine results of 2010 • Improve 2020
• Assessments• Every operation