1 Update on the Coverage Improvement Program for the 2010 Census Dave Sheppard Decennial Statistical Studies Division Census Information Center Program.

1

Update on the Coverage Improvement Program for

the 2010 Census

Dave Sheppard Decennial Statistical Studies Division

Census Information Center Program

October 9-11, 2007

2

Coverage improvement

Many operations or aspects of programs may lead to improved coverage in the Census

We’ll focus on two of them today…

3

Some plans to improve coverage for the 2010 Census

• Review and possibly revise census enumeration residence rules and how we communicate these rules to respondents.

• Determine ways to identify households with within-household coverage error and then develop and test methods to reduce the error.

In addition, Census Bureau staff will also provide an update on plans for measuring coverage error in Census 2010.

4

Some major coverage improvement topics not covered today:

• Improve housing unit coverage by identifying the best methods for keeping the Master Address File (MAF) up to date

• Develop methods for identifying the duplication of persons and housing units in the census

5

Coverage?

• Coverage Improvement• Methods to address census error• Impacts census counts of people and housing units

• Coverage Measurement• Evaluates the error in the Census• Does not change census counts

6

How do we end up with everyone* counted only once and in the right place?

* or at least as many people as possible

7

2010 Census coverage improvementresearch - two approaches

Get it right the first time Help respondents to initially include the correct

people on their Census form

Fix it laterIdentify households with coverage problems, get

back in touch with them, and make corrections

8

Improve residence rules instructions to respondents

• Self response paper forms • Enumerator wording for roster building

Simplify residence rules • Review of rules from Census 2000

To get it right the first time …

9

To fix it later, we identify households …

We identify households with potential coverage problems because:• They missed counting someone

(omissions), or• They counted someone who should not

have been counted there (erroneous enumerations)

10

Identify omissions • Count discrepancies (low person count) • Use of undercount coverage question• Use of administrative records

Identify erroneous enumerations • Count discrepancies (high person count)• Use of overcount coverage question• Use of unduplication within the census

For both omissions & erroneous enumerations• Large households

… for specific reasons

11

The CFU Interview

1. Verify the household and the validate the address

2. Review the list of people reported in the initial enumeration

3. Update the roster• Remove anyone who is unknown• Add anyone who is missing

4. Ask questions about possible living situations for each person on the updated roster

5. Collect any missing demographic information

12

Past Present

• Census 2000 – About 2% of all households– Large households (LHH)– Count discrepancies (CD)

• 2010 research – LHH/CD and ???– Undercount coverage question– Overcount coverage question– Use of administrative record matching– Use of person matching to identify duplication

• Coverage Followup• Field Verification`

13

Large households (LHH)

• Household contains more people than fit on the form– Mailback cases with 7 or more people

• Conducts CFU interview to:– Improve coverage– Collect demographics for persons 7+

14

Count discrepancies

High data defined person discrepancies (HDDP) • Reported more people than the reported POP count

on the census return

Low data defined person discrepancies (LDDP)• Reported fewer people than the reported POP count

on the census return

15

Count discrepancies (cont.)

• Conduct CFU interview to improve coverage– Even if no one is added or deleted, we still

determine the correct HH size

16

Undercount coverage question

• Asked once at the household-level

• Directly after question 1 about household size

• To identify housing units with potential omissions

• Probes include types of people historically missed in the Census

17

Undercount coverage question for the 2008 Dress Rehearsal

18

Overcount coverage question

• Asked once for each of the first six people listed on the form

• To identify housing units with potential erroneous enumerations

• Probes include places where people are historically overcounted in the Census

19

Overcount coverage question for the 2008 Dress Rehearsal

20

Use of administrative records

• First time we used this methodology was in the 2005 NCT, then in the 2006 Census test

• Returns identified that contain fewer people than the administrative records database has for that address

• Designed only to identify potential omissions

21

Use of person matching to identify duplication

• Research from Census 2000 continued this decade

• Identifies both potential person and housing unit duplication

• Resolution through Coverage Followup or Field Verification

22

Residence Rules Instructions

One little box … so many rules

23

Residence rules instructions

Explains how to apply the Census Residence Rules on paper forms:– Three sets of bullets detailing

• who to include • who not to include • how to count people without a permanent

place to stay

– A question about the number of people in the household

24

25

Ongoing & Future Research

26

Ongoing & future research for coverage followup

Streamline coverage followup selection criteria– Comparing efficiencies of all types of cases– Primary measure: Cost per change to a roster

• Added people and deleted people

– Additional consideration possibly given to other factors as well

• Effects on differential undercount • Effect on net census error

27

Ongoing & future research (cont.)

• Improve the coverage followup interview

• Identification of cases through the use of administrative records

• Identification of duplicates through person matching

28

Questions?

Dave Sheppard

[email protected]