Methodology Knowledge Base

This knowledge base contains a collection of resources (documents, presentations, videos) to assist anyone wishing to embark on a population-based census or survey. All material is structured around the two timelines below, each listing the key stages in the collection and containing material to assist in the preparation and execution of the respective stages.

This material includes :

  • Project plans and budget templates,
  • Training material to help conduct each of the activities where training of field staff is required (household listing, pilot census, main census)
  • Tutorials to help with anything from updating map boundaries in a mapping application to conducting a census with tablets

Census Timeline

From planning to dissemination : key stages from 18 months before to 6 months after census night.

(click on each section in the timeline to expand)

Planning

Usually, census planning should start 18 months or more before the census is conducted (Census Night).  Unfortunately, this is not always the case, due to conflicting government priorities, unavailability of government/donor funds etc

This phase focuses on carefully planning all the different activities which are required to ensure a successful census.

Submission to Government

First and foremost is the project submission to Government which explains the intention to undertake the Census and includes the proposed budget, and a detailed work plan of activities. (see below)

Work Plan and Budget

Due to the remoteness of islands in the Pacific and therefor accessibility issues, preparing a workplan and budget is always a challenge.  Coupled with inconsistencies in boat/flight schedules, high costs for hiring of transport in rural areas and sometimes unfavourable weather conditions, most plans and budgets have to be reviewed progressively to accommodate these changes. Poor planning can result in delayed census activities and increased costs. Timeframes for all activities should be documented including clear deadlines. (Download a work plan template here)

A budget should clearly specifying the cost of all operations – this includes the cost of staff and project workers, equipment, stationery, transport, printing, training venues,  data processing, analysis and so forth (Download a budget template here).

Its important to remember, than even once Cabinet approves the census budget, lengthy Government administrative processes can result in a slow release of funds. These tasks should be undertaken well in advance to ensure activities are carried out as scheduled.

Procurement of equipment is also a slow process, so this also needs to be planned well in advance. For countries using CAPI, the purchase of tablets takes about 2-4 months, depending on the number of tablets.

Setup of Census Committee

The Census committee is established as an advisory group which overseas the management of the census project, providing guidance and advice, as well as high-level monitoring and evaluation. The committee usually comprises stakeholders who are part of the National Statistical System. Potential members could include the National Statistics Office, Education, Health, Agriculture, Finance, Planning and various NGOs.  

Choice of Technologies

There are several "technology" decisions required during the planning phase.

For the household listing (if there is one), will the listing forms be paper only, will GPS units be used to record locations, or will everything be done on a tablet (questionnaire + GPS location). Most countries are now using tablets for the household listing.

For the pilot/main census, will data collection be undertaken with paper + manual data entry, paper + scanning, or on a tablet (fully automated data entry) . Most countries are using tablets for the data collection. (For more information about CAPI on tablets, download this presentation) 

Find out more about CAPI using Survey Solutions here

Questionnaire Design + User Consultation

With the adoption of CAPI for most Pacific countries, data entry systems + questionnaire design need to take place well in advance, and tested.

A Census is only a snapshot of the country at a certain period of time, and made up of a limited number of questions. They should be well thought out, tested, and based on national priorities.  In the past many censuses have included questions that were not well tested and analysed resulting in wasted time and costs during collection, and limited usefulness for planning and policy development. (Table specifications like this one should be considered at this stage)

User Consultation Workshops should be conducted with stakeholders to discuss and finalise the questions that are to be included in the questionnaire. There is often a tendency for stakeholders to push for questions that are relevant to their own needs and requirements. They need to be prioritised along with all other questions to ensure ease of completion for the interviewers and so there is not excessive burden to the households (respondents).

Recruitment / Appointments

Recruitment of project staff can be a lot of work, typically for larger countries with large teams.  Normally, the first recruitment is the appointment of the Census Commissioner, who will oversee the full census project.  The census Commissioner will then appoint/recruit his or her staff which will include:

•    Deputy Commissioner
•    Financial Controller
•    Administration officer
•    Project Manager
•    Area Coordinators
•    Enumerators, Supervisors, Headquarters (for pilot and main census)
•    Coders, Data Entry operators, Editors
•    Others – drivers, packers etc…
 

See here for the SDD Census and Survey Calendar with all collections past, present and future.

Household Listing + Mapping

A listing is undertaken when there is a need to update the Enumeration Area (EA) boundaries due to movement of people. This may be as a result of housing developments, natural disasters or urban/rural migration. EA boundaries  are revised to ensure they are all roughly the same size enabling a standard workload for each Enumerator during field work.

A listing involves recording a basic count of all individials in each household and sometimes ages and sexes of the household members. These days a GPS waypoint (location) is often recorded to assist in the EA boundary updates and then census fieldwork. 

A Listing is only undertaken if there is no recent census or survey data available.

See here for more material:

Pilot Census

The pilot phase of a census is an important part of the planning which is designed to test the different processes and procedures which need to be executed during the main census fieldwork. This include testing the questionnaire and interview process, data capture, transfer and monitoring (if using CAPI), logistics and administration.

The training and fieldwork during a pilot should be identical, yet scaled down to ensure everything is being tested correctly.

Last checks

After a pilot census has been conducted, the census team should review everything which happened from training, to fieldwork, to data processing and tabulation to ensure any bottle necks are resolved and last-minute changes are made.

Main Census

All censuses in the Pacific refer to a given point in time which is called Census Night. This is a "snapshot" in time. All questions refer to this point in time, even if the interview is conducted a week later. 

Training

The main census training is simply a scaled-up version of the pilot census training. Training venues need to be organised, equipment, catering, stationeries, presentations, manuals, transportation, and selected EAs for fieldwork training.  Appointment of interviewers and supervisors often takes place during the training.

Data Checking

Depending on the mode of data collection various forms of verification need to be carried out in the field. A CAPI questionnaire has built-in error checks and skips which reduces the number of data entry errors and therefor doesnt require detailed checks in the field. If the questionnaires are paper-based (PAPI) then supervisors need to check key characteristics on all questionnaires and randomly perform in-depth checks across a range of questions. For both PAPI and CAPI, Supervisors need to check enuerators are not getting lost and correctly enumerating their selected EA. This process is easier to monitor with CAPI becasue GPS locations are recorded with the questionnaires and can be overlaid on digital field maps. All households and institution (occupied or vacant) in each EA need to covered and accounted for during the census period.

Data Processing & Analysis

Data Entry, Coding and Editing

There is very minimal data editing required with CAPI collections because detailed error checking is built into the data collection software which prevents most errors. Data entry is also completed in the field. CAPI also auto-codes some of the variables, though depending on how the questionnaire is structured, industry and occupation codes may need to have additional coding.

PAPI questionnaires need significant coding to take place as part of the data entry and editing process. Data entry screens have to be designed and tested, ready for the capturing of the coded questionnaires. Training of Coders and Data Entry Operators followed by actual data coding/entry normally takes 3-6 months (or more), depending on the number of questionnaires to be captured.
 

With PAPI, once data entry is completed, data files are concatenated and prepared for final editing. Whether using PAPI or CAPI, a particular software package will be selected for statistical editing and processing such as CSPro, Stata or SPSS. Its important to have the Data Editing Specifications prepared well in advance (and ideally tested during the Pilot Census) to guide the data processing staff in doing the editing.  This provides more integrity and reliability to the quality of the data.  Recoding of variables, imputations of missing values, logic and consistency checks are part of the editing phase to prepare the final version of the data. Its important to have good data handling procedures including secure storage and backup facilities.

Tabulation

Data tabulation is the final data processing phase where cross-tabulations are generated from the final dataset. This phase is also part of the final editing check because inconsistencies in the data are easily detected from cross tabulating variables.  Data is finalised after fixing the errors and inconsistencies. Similar to data editing, a Tabulation Specification document like this one should be prepared for the data processing staff to follow.  This should include all the variables that are in the questionnaire, and data users should be consulted on their table requirements to useful data products are generated.
 

Analysis and Reporting

Analysis is performed on the final tables generated in the previous step. This generally cluminates with the preparation of a Census Report and publications such as Statistical Releases, Infographics, Monographs and Administrative Reports.

Subject-matter specialists such as Demographers often undertake this work. 

Dissemination and Documentation

Wide Dissemination

Its important to share census outputs widely so that a maximum number of users can make use of this information whether it be for developing policy, research or other purposes. Some of the main dissemination products and channels are listed below:

Data Use Workshop

These workshops are designed to help explain the census results to stakeholders and potential users of the data. Procedures on how to use and interpret the data are part of this workshop so the users are aware of what information is available. In large countries its best to organise several workshops in different parts of the country to ensure maximum participation.
 

Documentation

The whole census process should be well documented as part of a review of activities and for records.  This should fully describe what transpires throughout the different phases of the census operation, the people involved, the issues faced and how they were addressed, data processing steps and metadata.  These documented learnings can be used to improve future census activities and for comparison purposes. This documentation should ideally be upload to the Pacific Data Hub Microdata Library.