PES PC2010

1
PES PC2010

• Post Enumeration Survey of
• Indonesia Population Census 2010

Directorate for Development of Census and Survey
Methodology May 10, 2010
2
Introduction

• Non sampling error always exists in data
  collection.
• This error is bias source which is caused by
  field enumerator and respondent responses.

3
Introduction (2)

• Field staff error
• coverage error under coverage or over coverage,
  and
• content error
• While, respondent error response error which is
  part of content error.

4
Introduction (3)

• The Indonesia Badan Pusat Statistik (Statistics
  Indonesia, BPS) is conducting the 2010 Population
  Census in May of 2010. In order to evaluate the
  quality of the 2010 Census, the BPS will conduct
  a Post Enumeration Survey (PES) in a sample of
  census enumeration blocks (CBs) about one month
  following the census enumeration.

5
Objectives

• The objective of the PES will be to evaluate both
  the census coverage as well as the content errors
  in the census data
• Census coverage indicators (i.e., omission rate,
  net coverage error rate)
• Reliable estimates for 33 provinces
• Based on dual system estimation
• Measure census content errors
• Cross-tabulate census and PES responses for each
  variable, using records for matched persons
• Indices of inconsistency measure of response
  error

6
PES2000 VS PES2010

• PES SP2000
• Was not designed to estimate the population
• Sample size
• 4 provinces ( 11 regencies/cities)
• 40 census blocks
• Without field reconciliation

• PES SP2010
• Designed to estimate the population (Province
  level)
• Sample size
• 33 provinces
• 1200 census blocks
• With field reconciliation

7
Coverage

• PES SP2010 will be conducted in
• 33 provinces
• 1.200 census blocks (CBs)

8
Activity Time Table

• Activity Schedule
• Master Natl Instructors Workshop Week III
  April 2010
• Document Delivery May 2010
• Instructors Training Week I June 2010
• Field Enumerator Training Week II June 2010
• HH Population Enumeration Week III-IV June
  2010
• Matching Coordinator Data Processing Briefing
  Week III-IV June 2010
• Desk-match) HH population Week I July 2010
• Reconciliation visit Week II July 2010
• Final Matching Week III July 2010
• Data Processing in BPS Province July-September
  2010
• Doc data delivery to BPS HQ August-September
  2010
• Statistical Counting September 2010
• Data Analysis report writing September 2010
• Publication printing October 2010

9
Sample Allocation

• Sample allocation
• Proportional to square root of population
• Min 24 CBs Max 80 CBs per province

10
Selection of CBs Sample

• One Stage PPS, size estimated number of
  household from census mapping ( KK ), Urban/Rural
  as implicit stratification variable
• Exception for Provinces of Maluku, Maluku Utara,
  Papua Papua Barat
• Two stage sampling
• Select 6 districts
• Select 4 CBs per district

11
Training in BPS Province

• Field enumerator training (3 effective days)
• - Participants are CP2010 best enumerators
• Matcher training (2effective days)
• - Matcher from BPS Province
• Data entry training (3 effective days)
• - Persons form BPS Province

12
Diagram 1. Organization Structure of PES PC2010
13
Field Organization

• Headquarter
• General Person In Charge (PIC) is head of BPS HQ,
  helped by Deputy of MIS.
• Technical PIC is Director for Development of
  Census Survey Methodology.
• Province
• General PIC is head of BPS Province.
• Technical PIC is Head of IPDS BPS Province

14

• District
• General PIC is head of BPS Reg/City
• Technical PIC is Head of IPDS BPS Reg/City

Diagram 2. Field Organization of PES PC2010
15

• Field
• A team consists of 1team coordinator 2
  enumerators.
• Team Coord is Kepala Seksi Or Coord of SubDist
  Statistics (KSK) or staff of BPS Reg/City with
  best capabilities,
• Enumerator is statistical partner of staff BPS
  Reg/City appointed.
• Team coord coordinated with KSK for filed
  enumeration and responsible to Kasie IPDS.
• Since PES is part of Population Census, then
  Kasie IPDS should coordinate with Kasie Statistik
  Sosial.

16
Document Flow

• Documents for training field enumeration will
  be delivered by BPS HQ to BPS Province c.q. Kabid
  IPDS.
• BPS Province will distribute these documents
  needed by enumerators to BPS Reg/City c.q. Kasi
  IPDS.
• BPS Reg/City will distribute documents to team
  coord, next team cooed will distribute it to
  enumerators as much as needed for field
  enumeration.

17
Document Flow (contd)

• BPS Reg/City have to keep eye on documents of
  enumeration whether it have been given to team
  coord on time, whether team coord have given the
  cleaned document to BPS Reg/City.
• After document checking in BPS Reg/City is done,
  document must be delivered to BPS Province c.q.
  Kabid IPDS immediately.
• All PES documents will be kept in BPS Province
  until final report of Population Census 2010 is
  released.

18
PES Enumeration

• 1 month following census enumeration
• All census documents already in BPS Province to
  maintain independency
• Max 6 weeks after census
• Reducing recall error
• Reducing level of population movement

19
PES Enumeration (2)

• Enumeration by team
• One team consists of 1 team coordinator and two
  enumerators
• Workload 2-4 CBs per team

20
PES Enumeration (3)

• PES enumerator should be independent of the
  census operation
• PES enumerator assigned to different CBs from
  his/her census working areas.
• PES questionnaires are designed to obtain
  information on current residents of (de jure) and
  persons who moved out after census (out-movers).

21
Moving Status of Enumerated Persons

• Important to account for moving status of
  population between census and PES
• Current PES residents
• Non-movers
• Persons who are recorded after census
• In-mover
• Out-of-scope (born after census)
• Out-movers (moved or died after census)

22
Enumeration methodology

• Procedure C1 described in reference manuals
• Match non-movers and out-movers
• Easier to find in same sample CB
• Count in-movers
• More accurate coverage
• Estimate matched movers
• Apply match rate of out-movers to number of
  in-movers

23
Questionnaire SP2010 PES1

• Will be used one for all
• HH and population enumeration
• Matching of hh/population
• Field reconciliation
• Shading Area ? will be taken from census
  questionnaire (Form SP2010-C1)
• Consists of three main blocks (of total 9
  blocks)
• Block IV ? Non-mover and in-mover
• Block V ? Persons moved/died after census
• Block VI ? Persons enumerated in census not in PES

24
Form SP2010 PES1
25
(No Transcript)
26
(No Transcript)
27
(No Transcript)
28
Questionnaire SP2010 PES2

• For household enumerated in census not in PES

29
Form SP2010 PES2
30
MATCHING FLOW
SP2010 PES1
HH ENUMERATED IN CENSUS NOT IN PES
HH enumerated in PES Census
HH Status in SP200 PES1 Non-match
No
No
SP2010-C1
Yes
Yes
FILL SP2010 PES2
PERSONS INITIAL MATCHING
PROBABLY MATCH?
Yes
No
FIELD RECONCILIATION
FINISH
PERSON FINAL MATCHING
31
Desk-Match

• Two-way matching between all households and
  persons (non-movers and out-movers) enumerated in
  census and PES for sample CBs.
• HH and persons enumerated in PES coded as not
  matched are treated as census omissions in PES
  estimation procedures.

32
Desk-match ( 2 )

• Matching documents PES and census will be
  conducted in BPS PROVINCE.
• a. 1 matcher will match 3 CBs
• b. Matcher is staff of BPS Province
• Quality control of matching operation will be
  better.
• Documents need to be sent back to BPS Reg/City
  for reconciliation visit

33
Matching of households

• First level of matching households
• Availability of census stickers for most
  households will make matching more effective and
  faster
• First step match households based on census
  sticker information
• Next remaining households matched based on
  names of heads of household
• Possible to use PES listing sheets to help with
  matching

34
Matching of households (continued)

• After first part of household matching, remaining
  households matched based on names of individual
  persons
• Households considered matched if at least one
  person matched
• Multiple households can be matched
• Matched 1 to 2 or more
• Matched 2 or more to 1

35
Matching of households (continued)

• 7 (seven) possibilities found in initial matching
  of household
• 1. Matched 1 to 1
• 2. Matched 1 to 2 or more
• 3. Matched 2 or more to 1
• 4. Non-match
• 5. In-mover
• 6. Out-of-scope
• 7. Possible match

36
Matching of Persons

• Only for hh that is matched
• Matching of persons only done for
• Non-movers
• Out-movers (moved or died after census)
• Matching establish numbers of rules, allow some
  flexibility in range of ages especially for older
  persons.

37
Matching of Persons ( 2 )

• Generally, person will be match in matched hh.
  Non-movers and out-movers will be matched with
  person in census.
• Non-movers are persons who are listed in Block IV
  Form SP2010 PES1, out-movers are persons who are
  listed in Block V Form SP2010 PES1.

38
Matching of Persons ( 2 )

• Possible match ? name is consistent and some
  other characteristics match, but not all
• When matcher consider there is a chance the PES
  and census persons match.
• Matching coordinator may convert some possible
  matches to final matches before reconciliation

39
Rules Guidelines of Matching

• In assinging person match status, matcher has
  match 4 variables of PES census
• 1. name of person,
• 2. sex,
• 3. relationship to head of hh, and,
• 4. age.
• Matcher has to establish rules in assigning
  person initial match status.

40
Table 4.2 Criteria of Initial Match Status for
Person
No. Criteria Conditional Initial Match Status
I. 4 variables are the same Match

II. 3 variables are the same, one variable is different
a. Sex Possible match
b. Name within tolerance Match
Out of tolerance Possible match
c. Age within tolerance Match
Out of tolerance Possible match
d. Relationship to head of household Not contradictory Match
Contradictory Possible match
41
No. Criteria Conditional Initial Match Status
III. 2 variables are the same Name OUGHT TO BE THE SAME
a. Name sex Possible match
b. Name Relationship to the head of household Possible match
c. Name Age Possible match
d. Others Non Match
IV. 1 variable is the same
Name Possible match
b. Others Non Match
V. None of 4 variables is the same Non match
42
Tabel 4.3 Tolerance for Variable Name,
Relationship to head of hh, and Age
No. Variable Tolerance
1 Name Minor spelling difference that does not change the sound. Sound is phonetically the same but the spelling differs. Shortening of names and use of initial and nicknames in common use will be considered agreements, for example, Tony for Antonio, Fata for Fatimah, Ed for Eduardo, or Ben for Benikana.
43
No. Variable Tolerance
2 Relationship to head of hh Not contradictory (see Table 4.4). For example PES may show the relationship as head, spouse head, daughter, and mother census may show the mother as head, PES head as son and all the other household members as other relatives. In this example, Census and PES entries are not contradictory.
3 Age Tolerance
Under 15 years 2 tahun
15 29 years 4 tahun
30 44 years 7 tahun
45 64 years 10 tahun
65 years or older 15 tahun
44
Tabel 4.4 Changes of Relationship to Head of HH
Between PES Census
RELATIONSHIP TO HEAD OF HOUSEHOLD RELATIONSHIP TO HEAD OF HOUSEHOLD RELATIONSHIP TO HEAD OF HOUSEHOLD PES PES PES PES PES PES PES PES PES PES
RELATIONSHIP TO HEAD OF HOUSEHOLD RELATIONSHIP TO HEAD OF HOUSEHOLD RELATIONSHIP TO HEAD OF HOUSEHOLD 1 2 3 4 5 6 7 8 9 0
RELATIONSHIP TO HEAD OF HOUSEHOLD RELATIONSHIP TO HEAD OF HOUSEHOLD RELATIONSHIP TO HEAD OF HOUSEHOLD Head of HH Spouse Child Adopted/Step child Son/daughter-in-law Grandchild Parents/Parent-in-law Others family Housemaid Others
Census 1 Head of HH P P P P P P Ð Ð Ð
Census 2 Spouse P Ð Ð P Ð Ð Ð Ð Ð
Census 3 Child P Ð Ð Ð Ð Ð Ð Ð Ð
Census 4 Adopted/Step child P Ð Ð Ð Ð Ð Ð Ð Ð
Census 5 Son/daughter-in-law P P P P Ð Ð Ð Ð Ð
Census 6 Grandchild P Ð P Ð Ð Ð Ð Ð Ð
Census 7 Parents/Parent-in-law P Ð P Ð Ð Ð Ð Ð Ð
Census 8 Others family Ð Ð Ð Ð Ð Ð Ð Ð Ð
Census 9 Housemaid Ð Ð Ð Ð Ð Ð Ð Ð Ð
Census 0 Others Ð Ð Ð Ð Ð Ð Ð Ð Ð
Notes P not contradictory Ð
Contradictory
45
Reconciliation visit

• Reconciliation visit is critical part of PES
  methodology
• Necessary to follow up all households and persons
  enumerated in the census but not in the PES
• Determine whether each household and person was
  correctly enumerated in the census
• Erroneously enumerated persons subtracted from
  census population in the estimation procedures
• Follow up cases of possible matches to determine
  final match status

46

• The objectives of reconciliation visit is not to
  make any correction to the result of census or
  PES enumeration (because this would break the
  assumption of independency), but only to verify
  information which have been gathered.
• To maintain independency between PES and census,
  enumerator is not allowed to add person or hh, or
  to change demographic characteristics of
  person/hh, either in census or in PES
  questionnaire.

47

• After reconciliation visit, there are no more
  cases of possible match. All outcomes will be
  mutually exclusive.

48
Dual system estimation
Asumption PES and Census are independent.
PES Census Census Census
PES Enumerated Omitted Total
Enumerated M N12 NP
Omitted N21 N22 N2.
Total NC N.2 NT
RM proportion of matched person from PES data
49
Weighting PES data

• Necessary to attach weights to the PES data for
  estimates to be representative
• Weights based on inverse of probability of
  selection
• Weights will vary by stratum and CB
• Weights may be adjusted based on final census
  results

50
Census Population Correctly Enumerated

• The total census population correctly enumerated
  ( ) is estimated by subtracting an estimate
  of the census erroneously enumerated population
  from the PES data (based on the E-sample), ,
  from the official census figure for the total
  population.

where is the official census total
population
51
Dual System Estimator of the Total Population

• The PES results will be used to evaluate the
  census coverage errors, but they will probably
  will not be used to adjust the official results
  from the census.

estimation of total population
52
Net Census Undercount Rate
estimation of total population
Net Census Undercount Rate
53
PES Estimate of the Total Population

• PES estimate of the total population
• Weighted estimate of non-movers plus in-movers
  enumerated by the PES.

• The matched population is estimated as

54
The Dual System Estimates

• Preliminary estimate of
• Once the official census population ( )
  is available, the estimate of could
  then be ratio-adjusted as
• The final estimate will the be used to
  calculate the final net census undercount rate (
  ), using the formula specified
  previously.

weighted estimate of matched
population census correctly enumerated
population (missed by PES) census erroneously
enumerated population PES survey
estimate of the total population enumerated in
census
55
Census Content Error Evaluation

• Based on the data for matched persons
• Non-movers, and
• Out-movers

56
Reference manuals

• Evaluating Censuses of Population and Housing,
  U.S. Census Bureau, 1985
• Design and Implementation of a Post-Enumeration
  Survey Developing Country Example, U.S. Census
  Bureau, August 1993
• POPSTAN (A Case Study for the 1980 Censuses of
  Population and Housing), Parts A, B and D, U.S.
  Census Bureau, 1979

57
THANK YOU

• THE END