Post Enumeration Survey- 2000 Census

1
Post Enumeration Survey- 2000 Census

• Presented
• By
• Batista Chilopa
• Central Statistical Office
• Zambia

2
DATES

• Census 2000 dates 16th October 2000 to 30th
  October
• PES Date 23rd February 2001 for three weeks
• At least three months after the Census Enumeration

3
OBJECTIVES

• To Present census data to users with results of
  quantitative evaluation of the accuracy of the
  census data
• To identify the extent of over-coverage or
  under-coverage of population at all levels
  including National ,rural, urban and province.

4
OBJECTIVES Continued

• To identify levels of agreement for responses
  related to questions on sex, relationship to head
  of household, marital status age and type of
  housing units
• To identify areas of improvement for future
  censuses

5
PREPARATORY ACTIVITIES

• Identified Key Survey Personnel/Planning team
• Survey coordinator
• Sampling Statistician
• Field Coordinators
• Data Processing Specialist
• Data analysts
• All were Central Statistical Office Personnel

6
PREPARATORY ACTIVITIES Cont

• Decided on level of aggregation and sample size
• -National Level
• -Sub-national Levels
• -Provinces, Rural Urban
• -250 EAS
• Calculated the number of interviewers required
  and days needed in the field
• Established a work plan
• Estimated Budget (part of the National Census
  Budget)

7
PREPARATORY ACTIVITIES Cont

• Preparation of instruments
• Questionnaires
• Interviewers Manual
• Supervisors Manual
• Matching Instructions
• Field re-visit instructions
• Pretest and Pilot preceded main fieldwork

8
RECRUITMENT OF FIELD STAFF

• All Enumerators did not participate in Census
  enumeration (regular CSO Staff School Leavers)
• Some supervisors had taken part in census
• All trainers and field coordinators had taken
  part in census

9
TRAINING

• Enumerators training manual was prepared
• Training took 1 week
• Central training
• Supervisors Enumerators trained together
• Training manual for supervisors was also
  available Last day of training for supervisors

10
SAMPLING DESIGN

• Stratified cluster sampling
• Sample size determination based on previous
  experience 1 of the total population
• Stratification Provinces, Rural and Urban
• 250 Primary Sampling Units selected with
  probability proportional to size (pps)
• All selected Clusters were to be covered on 100
  basis (institutionalized population was not
  covered)

11
FIELDWORK

• Trainers field coordinators observed initial
  implementation of fieldwork
• Every Supervisor was assigned three enumerators
• Supervisors remained in the field through out
• Fieldwork ran for three weeks

12
MATCHING EXERCISE

• Matching was done manually using procedure C
  (matching non-movers out-movers only)
• A case-by-case comparison of Census records with
  that of the PES after data collection
• Pairing each household and each person in PES
  with the corresponding Census records

13
CASES FOR FIELD RE-VISITS

• A number of individuals were found in the census
  but not in PES and vice versa
• Entire PES enumeration Area (EA) failing to match
  with respective Census EA even when the
  geographical IDs matched

14
Field Re-visit

• Follow up visits for un-matched households and
  individuals were made
• To allow for identification of erroneous census
  enumerations and resolve some doubtful cases
• A reconciliation exercise in the office was
  conducted afterwards and data entered

15
Data Analysis

• The match rate generated after all the needed
  reconciliations was only about 70.
• There was a huge PES population that had
  non-match status code
• The office was prompted to seek technical
  assistance through UNFPA

16
TA recommendations

• All cartographic maps that did not benefit from
  the pre-enumeration up-dates needed to be
  reviewed to assess their usefulness by the PES
  Enumerators.
• All PES EAs that were updated in the field during
  the census should also be reviewed, with the aim
  of assessing their application by the PES
  enumerators.
• .

17
TA recommendations Cont

• Households should be matched before individual
  matching i.e person matching should be undertaken
  only for matched households
• Cases where households in the whole SEA do not
  match should be assumed the PES enumerator is
  covering the different geographic area and the
  solution can only be possible by field work.
• Effort should be made including relaxing rules
  so that the match rate is raised to more than 80
  percent.

18
TA Recommendations Cont

• The PES team should be involved in the data
  processing activities-formulas should be
  discussed with the data processing team before
  application.
• Data should be exported to SPSS and eventually
  excel which the PES team is familiar with which
  will be of great use when estimating content
  error and also help in checking input.
• Due to technical and financial constraints not
  all recommendations by the TA were implemented
• The office decided to compile only an
  administrative report

19
Lessons Learnt

• Pre enumeration cartographic works should be
  completed by the time of Census
• Questionnaire formats for both PES and Census
  should be similar to ease the matching
• Electronic matching should be considered
• Timely flow of resources is required
• Implementing appropriate time frames

20

• THANKYOU !