Post-Enumeration Survey

1
Post-Enumeration Survey Requirements,
Planning, Designing and Executing
Carlos Creva Singano, National Statistic Office,
Mozambique -Maputo

• Adis Ababa Workshop 14-18 September 2009

2
1. Requisite qualities of PES

• All Census questionnaires must be returned from
  the field in all EA to National Statistic Office,
  before the PES field enumeration.
• The processing of the PES data must be separate
  from the general Census of Population and Housing
  processing Data
• The PES staff must go to the field without
  knowing the preliminary results of the respective
  EA
• To take the advantage of General Census
  enumerators and supervisors, they must be used
  as the field staff

3
2. Costs /Other resources

• The costs must be fixed based on the sample size,
  mainly at the first stage of sampling ( that is,
  nº of PSU to be selected) as well as the nº of
  Domains of Estimation
• The items to be included in the global cost are
• -Transportation (Vehicles its
  maintenance)
• - Field staff (PES enumerators PES
  supervisors PS Drivers)
• - Office Personnel (Matching
  clerks, matching Supervisors, Data Entry
  operators, Data Entry supervisors)
• - Printing and Reproduction
  (Questionnaires Manuals, Maps, List of EA, etc)

4
3.Organization structure/Technical planning

• Depend on the Political and administrative
  structure
• Mozambique
• - NSO (Central Level- Sample Design, Budget
  and its distribution by province according to the
  sample size)
• - Provincial Level ( Only for execution of
  the central directives controlling the field
  staff and financial matters) To control the
  Quality in the enumeration in the dual system
• In Mozambique, the Province is a domain of
  estimation. The sample must be representative of
  strata within each province
• In each 11 Provinces, the NSO has a Provincial
  NSO for Operations issues.

5
4.Procedure for Covarage analysis

• There are Three Procedures for evaluation Census
  coverage in PES Procedures A, B and C ( See
  Manual Evaluating Censures of Population and
  Housing
• Mozambique used the procedure C, where the goal
  was to identify all current member of the sample
  HH as the 2007 PES date in addition, any other
  resident as of the Reference Date of the General
  2007 census of Population and Housing.
• These persons are classified as either non-mover,
  out-mover and in-mover (that is, to identify
  immigrants and emigrants), with regard to their
  HH memberships status as of the 2007 Census Date.
• In the 2007 Census records, only the Residents as
  the Date of the 2007 Census (the Non-movers and
  the Out-Movers), were matched.

6
4. Procedure for Covarage analysis (Cont.)

• 2007 PES Status Mozambique
• - Non-mover
• - Out-mover
• - In-Mover
• - Bor After
• Some Definition
• Those who belonged to the HH as of the 2007
  General Census and still as of the 2007 PES are
  the Non-Movers
• Those who became members of a different HH during
  the interim period between the 2007 General
  Census and the 2007 PES are the Movers. There
  are two types of Movers
• The In-Movers - those who arrived in the HH
  after 2007 Census
• The Out-Movers - those who have left the HH
  since the 2007 Census.

7
4. Procedure for Covarage analysis (Cont.)(Some
Definitions)

• Persons who do not belong to the target
  population, that is, the population as of the
  2007 Census date are considered Out-of-scope.
  This includes persons born after the 2007 census
  or persons who died before the 2007 Census,aliens
  and nonresidents. No match is attemted or Out-of
  Scope.

8
5.Organization structure/Technical
planning (Sample Size and Domain of estimation)

• Depending on the main objectives to be measure
  (the coverage error or, the net omission rate),
  the geographic domains of analysis and financial
  resources, the sample size is fixed
• In Mozambique, the Domains of estimation are
  Provinces National and national Urban and
  national Rural
• The 2007 PES of Moz follow a Stratified Cluster
  sampling of 340 PSU Minimal of 30 PSU by
  Province and Maxim of 40 PSU in Maputo, Capital
  of Mozambique. All unit of analysis (HH,
  persons, etc) in each sample PSU were
  interviewed
• Sample designing for 95 percent of Confidence

9
5.Sample Size and Domain of estimation (Cont.)
10
5.Sample Size, Domain of estimation and
estimates(Cont.)

• Within each Province, the sample was stratified
  by urban and rural areas
• The Sample selection within each strata was
  selected independently with PPS. Measure of size
  was the nº of household within each strata
• Using the Dual system methodology (Procedure C),
  coverage and sampling errors were estimated. The
  national coverage error was 2.1 per cent. The
  Programme used were CENVAR (The IMPS Component)
  to main variables (eg, Age, sex, marital status,
  etc).
• David Megill, The U.S.Census Bureau statistician,
  did help the Mozambican staff to do it

11
5. Sample Size, Domain of Estimation and
estimates(Cont.)

• 95 percent confidence intervals for the Census
  omission rate were estimated

12
6. Sampling Frame and Units of Analysis

• Methodology of 2007 Moz PES Two-way HHs and HHs
  Members
• The target universe included only the individual
  households in Mozambique
• Quarters groups were excluded of the scope
• The units of analysis in Mozambique 2007 PES were
  Households members

13
6.Sampling Frame and Units of Analysis

• Distribution of the HH in Mozambique by Province
  and area of residence. Preliminary Results

14
6. Sampling Frame and Units of Analysis

• Distribution of the PSU or EAS by Size oh HHs in
  Mozambique by Province and area of residence.
  Preliminary Results

15
6. Sampling Frame, Units of Analysis, Sample
Selection Methodology

• The Sampling frame is stratified by Province
  Urban and rural
• The objective of this stratification is to
  improve the PES Sample Design
• The PSU or EA within each strata was selected by
  PPS. The number of HH within each was the measure
  of size.
• An implicity stratification by political
  subdivision (OER Cod, CA_Cod and EA_Cod) was done
  before sample selection within each strata

16
7. Development of Questionnaire and field
manuals

• The Moz PES questionnaire was based on the final
  2007 General census of Population and Housing
  Questionnaire
• The 1997 PES (Moz) and the 1991 PES (Ruanda) were
  used as Reference
• Demographics characteristics included in 2007
  PES
• Marital status, age, sex and relationships to the
  head of the HH and
• the section A of the General Census, that is,
  the location and area of residence of the HH to
  facility the matching process
• To facility the comparability between 2007 Census
  data and 2007 PES data, same response categories
  and precoding system were used

17
7. Development of Questionnaire and field
manuals (Cont.)

• To facility the field team (enumerators and
  supervisors), a small Manual with instructions,
  definitions was prepared
• The PES data collection aims were to identify all
  usual residents at the time of the 2007 Census in
  addition to those as of the date of the 2007 PES
  ( Procedure C in Slide 5-6).

18
8. Budgeting/ Pilot test

• The 2007 PES Budget Was planned as part of the
  2007 census program
• The 2006 PES Pilot was care out to test the Dual
  System methodology (Procedure C) of the 2007
  Census of Population and Housing. This was done
  during the Pilot General Census of Population and
  Housing in 2006

19
9. Selection and training of field staff/ data
collection

• To take the advantage of 2007 General Census
  enumerators and supervisors, those were used as
  the field staff of 2007 PES
• They were trained during two days in each
  province.
• The 2007 PES data collection did take place 15
  days after 2007 General Census data collection.
  The objective was to minimize the matching
  difficulties processes related to the changes to
  the composition of the HHs between the two Dates.
  

20
10. Omission rates for 2007 Census of Population
and Housing
21
End

• Kanimambo