UN Banner
HomeAboutExperiences and LessonsCase StudiesRecommendationsResourcesChecklist
Home
About▶
BackgroundPurposeRationaleStructureTarget Audience
Experiences and Lessons▶
1. Project Planning and Management2. Geospatial Mapping and EA Database3. Enumeration Instruments and Tools4. Data Capture and Management5. Census Testing and Pilots6. Recruitment and Training7. Deployment and Supervision8. Enumeration and Logistics9. Quality Assurance and Risk Management10. Census Analysis and Products11. Partnerships and Collaboration12. Advocacy and Publicity13. Procurement and Financial Management14. Post-Enumeration Survey15. Alternative Approaches16. Planning Checklist
Case Studies▶
1. Project Planning and Management2. Geospatial Mapping3. Enumeration Instruments4. Data Capture5. Census Testing6. Recruitment and Training7. Deployment and Supervision8. Enumeration and Logistics9. Quality Assurance10. Census Analysis11. Partnerships12. Advocacy13. Procurement14. Post-Enumeration Survey15. Alternative Approaches16. Planning Checklist
Recommendations
Resources
Checklist
Skip to main content

Introduction​

A census post-enumeration survey (PES) is a survey carried out in a sample of Enumeration Areas (EA) soon after the census enumeration has been completed. The main purpose of a PES is to assess the coverage of the census i.e., to estimate how many usual residents have been missed from or double-counted in the census. The PES can also be used to verify the accuracy of responses for some key census questions. The census figures may or may not be adjusted according to the result of the PES analysis, but in either case, the estimates of the true population should feed into subsequent population projections, making these more accurate. Finally, lessons learnt from the PES can inform the improved design of future censuses and surveys. However, whilst a well conducted PES will yield valuable insights into the quality of the census coverage and accuracy, if not properly planned, funded and staffed, as an integral part of the census operations, the PES risks becoming a waste of resource. If the results are not used, because for example, the necessary assumptions regarding independence and matching quality are not met, or there is too much uncertainty in estimates, then the time and money spent on the PES may have been better used elsewhere e.g., in improving the quality of the census enumeration. If a PES is going to be conducted, the objectives should be clear from the outset regarding its purpose and how the PES fits into the overarching statistical design and analysis plan of the census (see UN P&R, Rev 4, para. 4.119 to 4.126)

PES during the non-digital versus digital era

In a non-digital census, the data is collected on paper forms and the essential stage of matching the census records to PES records is carried out manually which takes a long time – usually many months. However, in a digital census the data will be collected electronically, probably using a computer-assisted personal interviewing (CAPI) device like that used during census enumeration. Various automatic matching techniques can be applied to match the data together saving a significant amount of time and enabling faster release of census outputs. The size of the clerical matching team can also be reduced thereby saving money. In a digital census, multiple options for completing the census may be offered e.g., CAPI or computer assisted web interviewing (CAWI). This may lead to more instances of over-enumeration than in a non-digital census if people respond using more than one method. The PES data can be used to detect and analyse duplicate responses in the PES EAs and thus estimate the size of the overcount in the census. The PES analysis can be carried out using Excel, Stata or other proprietary software, or using python or other open-source solutions. Boot-strapping techniques can be used to provide variance estimates around the population estimates. The biggest decision regarding the PES during a digital census, is whether a PES is necessary and good value for money. Alternative methods of quality assurance, such as using administrative records as a comparator to census counts may be viable. See UN P&R, Rev 4, para. 3.57 to 3.574 of the results for more details on alternative methods of census quality evaluation. The rest of this chapter assumes that it has been decided to go ahead with the PES.

Considerations for a digital PES

  • The PES should be conducted independently from the census, as there is an assumption made that a person is no more or less likely to respond to the PES given whether they responded to the census. This assumption will always be violated to some extent, because a person who does not wish to be counted is unlikely to respond to either the census or the PES. However, consider steps that can be taken to ensure that independence is maintained as far as possible, and that when the independence assumption is broken, how this can be accounted for in the subsequent analysis. In the context of a digital census where CAWI self-completion online enumeration methods are used, these methods should not be used in the PES enumeration. However, whilst independence should be maintained, it is also important to recognise that it is sometimes best for the PES and census teams to work together. Consider when your PES team and census team should be integrated and when separated to maintain independence in the PES. For example, consider working together on the CAPI design so that questions are harmonised, making data matching easier.

  • Consider data management and how you will version control data after processing steps such as remove false persons and resolution of duplicate persons.

  • It is essential to match the PES and census data together with a very high accuracy. Despite using automatic matching techniques, this is a challenge and will require time and resource. Data matching can either be carried out back at headquarters once all the PES data is collected, or alternatively can be matched in the field as enumeration in each PES EA is completed. In the second case it is important to ensure that independence between the census and the PES is maintained. PES enumerators must not be given census information about the areas in which they are working (UN P&R, Rev 4, para. 4.123). Consider which is the best approach in your country. See case studies from Rwanda and Togo for for examples of these two different approaches.

  • Consider steps you can take to make household matching easier. Using stickers is a good low-tech solution (see Rwanda case study). Collecting geolocations at the front door of each household during census and PES enumeration is an alternative (although may not be as successful for high-rise apartment blocks, or compounds where the ‘front door’ is not uniquely defined).

  • Decide on the size of the geographical area for which you want to estimate the coverage before creating your PES sample, since this will determine the required sample size. For example, if you want to report on coverage nationally, then the sample size required will be smaller than if you want to report on coverage by region. Typically, the smaller the area you want to report on, the larger the overall sample size.

  • Consider whether reconciliation visits will form part of your PES enumeration plan, and if so how to carry out such visits, what information you expect to gain and how this information will be used. Should reconciliation visits be carried out in all PES areas, or only in those where the matching is showing unexpected results?